"Like" AX84 on facebook

'EL84 class of operation. A or AB?'
Author:rufinus (guest: search)
Date: Fri, Sep 07th, 2007 @ 09:56 ( . )

I'm interested, how can you determine a class of operation for the EL84 in my HO?
For example, here are my measured voltages:

Pin9 275v
Pin7 288v
Pin3 9.3v

Screen current is considered to be about 5mA. Resistor R10 is 220 ohm. At 37mA of plate current this gives 10.3 watt of plate dissipation. Screen dissipation is about 1.37 watt then.

In this case, what is the class of amplification? A or AB? How do i determine it?
Another question, can i look safely with the oscilloscope at the output of the OT?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Sat, Sep 08th, 2007 @ 04:53 ( . )

The HO is a single ended amp, which means it MUST be class A. Only push pull amps can be AB.
(You COULD make a class B single eneded amp, but it would be no use at all for audio!)
You can hook your scope across the secondary of the OT provided you have a load connected aswell, like the speaker or a big power resistor of a similar resistance to the speaker Z.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sat, Sep 08th, 2007 @ 19:15 ( . )

Actually there are several SE guitar amps that do operate in somewhat moderate Class AB. There is a fairly long thread about this from some time back, with some very helpful clarifications from CarlB...

On 09/08/2007 @ 03:53, Merlin wrote :
:
The HO is a single ended amp, which means it MUST be class A. Only push pull amps can be AB.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Sun, Sep 09th, 2007 @ 11:08 ( . )

On 09/08/2007 @ 18:15, Zaphod wrote :
Actually there are several SE guitar amps that do operate in somewhat moderate Class AB. There is a fairly long thread about this from some time back, with some very helpful clarifications from CarlB...
Unfortunately it is wrong. You can of course bias an SE stage cool, so that it hits cut off before grid-current, but it's still Class A; the load on the valve doean't change at any time.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Sun, Sep 09th, 2007 @ 13:22 ( . )

On 09/09/2007 @ 10:08, Merlin wrote :
On 09/08/2007 @ 18:15, Zaphod wrote :
: Actually there are several SE guitar amps that do operate in somewhat moderate Class AB. There is a fairly long thread about this from some time back, with some very helpful clarifications from CarlB...
:
Unfortunately it is wrong. You can of course bias an SE stage cool, so that it hits cut off before grid-current, but it's still Class A; the load on the valve doean't change at any time. --
I would have to agree with Merlin on this one.
The label "Class A" is merely a general description for the mode of operation to differentiate between other modes of operation when a valve is used in the situation it was designed for, ie. linear .
But IMO, when a valve has the snot driven out of it (as is most often the case in guitar amplifiers) so that it may not conduct for a complete 360º of a cycle the rules for class labels can not be applied.
Or if it is biased so cold that it will conduct for considerably less than 360º but more than 180º when a large modulated voltage is applied to the grid. Though at some threshold point if you were to decrease the voltage swing to the grid the valve will be conducting for 360º and therefore class "A". However this may not be linear as the bias point may already be in the start of the cut-off region.

So to sum up, if the valve is operated with in its design centre maximums and conducts for 360º of the cycle it can be labelled class "A". Once it goes outside those limits the class labels are fairly pointless.

Perhaps we should invent some new labels to better describe the mode of operation for guitar amplifiers?
I propose : Class "S"

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Sun, Sep 09th, 2007 @ 15:48 ( . )

That's true, the classic definitions do get a little wearisome where guitar is concerned, though the terms Clas A, Class B, and Class AB are quite straightforward in any situation. Only the suffixes of Class AB1, AB2 etc., are pointless for guitar.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Pete Allen (registered user: 895 posts )
Date: Sun, Sep 09th, 2007 @ 15:48 ( . )

On 09/09/2007 @ 10:08, Merlin wrote :
On 09/08/2007 @ 18:15, Zaphod wrote :
: Actually there are several SE guitar amps that do operate in somewhat moderate Class AB. There is a fairly long thread about this from some time back, with some very helpful clarifications from CarlB...
:
Unfortunately it is wrong. You can of course bias an SE stage cool, so that it hits cut off before grid-current, but it's still Class A; the load on the valve doean't change at any time.
--



Argh!!! not this discussion again :-)

Can we just use "Hot biased", "Warm biased" and "Cold biased" and not go over this ground again?

Pretty please? :-)

Pete

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 09th, 2007 @ 18:03 ( . )

Yeah, I agree with Pete.

Merlin's absolutely correct, though. I think the confusion comes in because you can bias such that cutoff is reached before saturation, so it sort of acts like a PP stage does in class AB. But as the load stays the same at all times, it's in Class A. But tell someone it's Class A and let them look at the waveform on a scope and they won't see what you'd normally expect for a Class A signal at max chat. The definitions just don't work properly when you want the stage to distort.

And what would you call it if it hits saturation before cutoff? It's still IN class A but again wouldn't LOOK like it is on a scope.

Never mind! It's a pointless discussion. ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Francis Vaughan (registered user: 712 posts )
Date: Sun, Sep 09th, 2007 @ 18:22 ( . )

On 09/09/2007 @ 17:03, Zoe N Iain wrote :
: Never mind! It's a pointless discussion. ;-}


Actually it occurs to me that some sort of nomenclature that covered the bases for the different sonically interesting choices might be useful. We do sort of understand when cold biasing is talked about, but that only covers a small part of the possible range, and just how cold etc isn't really there.

Imagine being able to explain a Wreck in terms of: Class-Q stage into Class-R into Class-S driving a Foo mode PI and a PP output stage biased to operate at Mode-Z ..... for instance. :-)

OK, this is perhaps not the whole story, but you get the idea. Also perhaps a few more differentiators on basic topologies. We have Anode and Cathode followers, Cathodyne and Schmidt PIs, but not much more. Somewhere Zoe and Iain get to christen some of their creations :-)

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 09th, 2007 @ 19:50 ( . )

Heh, see your comment from the old thread I just posted below... ;o)

On 09/09/2007 @ 17:03, Zoe N Iain wrote :
:
Yeah, I agree with Pete....
:
: Never mind! It's a pointless discussion. ;-}
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 09th, 2007 @ 20:17 ( . )

Like I said below, any definition can be argued to be correct in the context we're using here. We're taking a nice neat hi-fi PA and hammering the living sh*t out of it with an enormous signal that is often clipped in the preamp as well. We value OTs that aren't especially linear. We nail HORRIBLY non-linear speakers on the end. The actual load on the PA valves is changing every time we move a finger on the fretboard, because the reflected impedance is constantly changing with frequency.

So does it really f*ckin' matter? Seriously, who gives a sh*t what class the PA is operating in? We design for tone, not mathematical perfection. So screw it! I'll happily operate a PA in class D if I can get a good sound out of the damn thing.

Like I said, it's a pointless discussion.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 09th, 2007 @ 20:22 ( . )

But does the discussion have to have a point? Sometimes it's just fin to talk around things....

On 09/09/2007 @ 19:17, Zoe N Iain wrote :
:
: Like I said, it's a pointless discussion.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 09th, 2007 @ 20:42 ( . )

Nope. I'm happy to discuss the subject for the fun of it. I have been.

I suppose I pretty much agree with what Pete and Francis said... let's kick the discussions of which class things work in into touch and come up with some useful terminology for the job we're doing.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 09th, 2007 @ 19:38 ( . )

It's not wrong. The basic definition of the various classes is related to what part of the duty cycle the amplifying element conducts for. Class AB, Class B and even Class C SE power amplifiers have been accepted in the RF field for many years. The change in load is an additional resulting effect of Class AB operation in PP amps, which of course could never occur in an SE amp. It's false logic to say that since that particular effect of Class AB PP operation, doesn't occur in a SE amp, therefore it can't operate in Class AB!

On 09/09/2007 @ 10:08, Merlin wrote :
:
: Unfortunately it is wrong. You can of course bias an SE stage cool, so that it hits cut off before grid-current, but it's still Class A; the load on the valve doean't change at any time.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sun, Sep 09th, 2007 @ 18:57 ( . )

Folks, there are SE Class AB amps out there, quite accepted by the EE community as ... well, class AB SE amplifiers. Typically the folks who think of this mode are microwave power engineers, but the definition applies and the class is useful for guitar amps as well.

Here's a good explanation from Randall Aiken:

"Finally, can a single-ended amplifier be class AB or class B? The simple answer is, yes. Many RF (radio-frequency) amplifiers are single-ended class B or class C (current flows for less than 180 degrees of an input cycle). However, for audio amplification, these are of little use. Technically speaking, you can have a a class AB single-ended amplifier, which was biased away from the linear portion of the curves, but that amplifier would (hopefully!) be rated at the unclipped output power, so plate current would be flowing at all times at the max undistorted output power. This would stretch the definition a bit, particularly since the amp would undoubtedly be able to be driven into asymmetrical cutoff (where us guitar players like to hang out!) which would effectively be class AB single-ended operation, because the amp is not biased in the middle of the transfer curves and is capable of being driven into cutoff for a portion of the input cycle. As long as the manufacturer isn't rating the amplifier for its output power in this clipped state, the amplifier would normally be called class A, single-ended."

Let's look at the P1eX output stage. Biased well below the midpoint, to be able to accept the higher plate and screen voltages. Designed for about 20 watts useful guitar output power. It can't achieve that without reaching cutoff before clipping, and so we aren't *intending* it to be a class A amp - *just the same as the microwave guys aren't intending their circuits to operated class A.* They want that extra power from the SE class AB operation, and so do we. Neither group cares much that the power available in that mode isn't as clean as SE class A. It's clean enough for the application.

For a guitar amp, you can design - and intend for - the useful 'relatively' clean output power available from an SE amp to be higher than an SE amp biased at midpoint. It's not cleaner by measurements, but our ears are AOK with much of the 2nd order stuff that you start to get. Once you intend for that and design to make use of that, you have an SE class AB amp on your hands.

As an aside, for guitar, pentode SE class AB sounds better to my ears than pentode SE class A operation. Definitely more lively and interesting tone.

Good discussions. Although the class definitions may be complete, they cause much confusion because of what's inferred about their intended use. SE class AB is useful for guitar, and so we should start thinking about that as the actual class we are often using.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Old thread - Here are the definitions, and how to determine A vs'
Author:Zaphod (guest: search)
Date: Sun, Sep 09th, 2007 @ 19:49 ( . )

I thought I would dig these definitions out, from a previous thread....

On 08/10/2006 @ 05:30, CarlB wrote :
Class A: cathode current flows for the complete 360 degrees of the waveform.
:
: Class AB: cathode current flows for less than 360 degrees, but more than half (180 degrees) of a the waveform.
:
: How to tell the difference? Crank the amp up until just before grid conduction (e.g., just before grid clipping for class A1 and AB1). If there's still cathode current flowing at the bottom of the waveform (i.e., the side opposite the one which is just before grid conduction clipping), then it's class A. If there's more than half of the signal for which cathode current is flowing, then it's class AB.
:
: For class A2 and AB2 (i.e., muscling the grids into non-clipping at grid condtion), crank the amp until just before grid conduction (verify with a scope, meter, etc.), and repeat the above.
:
: Anything beyond grid conduction is fair game (and fun) for amplifier operation, but it is not fair game for class determiniation. That's because any output stage, no matter how hot-biased, can be driven into bottom-side cutoff at some point.
:

On 08/10/2006 @ 05:44, ZoeNIain wrote :
Ah! I'd forgotten about grid current. That makes sense. Cheers, Carl.
--


-- REPLY: [With No Quote] --- [With Quoted Text]

'Old thread - Here are the definitions, and how to determine A vs'
Author:CarlB (guest: search)
Date: Thu, Sep 13th, 2007 @ 20:43 ( . )

On 09/09/2007 @ 19:49, Zaphod wrote :
I thought I would dig these definitions out, from a previous thread....
:
:
On 08/10/2006 @ 05:30, CarlB wrote :
: Class A: cathode current flows for the complete 360 degrees of the waveform.
: :
: : Class AB: cathode current flows for less than 360 degrees, but more than half (180 degrees) of a the waveform.
: :
: : How to tell the difference? Crank the amp up until just before grid conduction (e.g., just before grid clipping for class A1 and AB1). If there's still cathode current flowing at the bottom of the waveform (i.e., the side opposite the one which is just before grid conduction clipping), then it's class A. If there's more than half of the signal for which cathode current is flowing, then it's class AB.
: :
: : For class A2 and AB2 (i.e., muscling the grids into non-clipping at grid condtion), crank the amp until just before grid conduction (verify with a scope, meter, etc.), and repeat the above.
: :
: : Anything beyond grid conduction is fair game (and fun) for amplifier operation, but it is not fair game for class determiniation. That's because any output stage, no matter how hot-biased, can be driven into bottom-side cutoff at some point.
: :
:

:
On 08/10/2006 @ 05:44, ZoeNIain wrote :
: Ah! I'd forgotten about grid current. That makes sense. Cheers, Carl.
: --
:

:
--



I'm seeing I wasn't completely correct in the above. For A2, you go out to the saturation point, not the grid conduction point. It's all about what happens first: saturation-clipping, cutoff, or both?

-- REPLY: [With No Quote] --- [With Quoted Text]

'Old thread - Here are the definitions, and how to determine A vs'
Author:Merlin (registered user: 5226 posts )
Date: Fri, Sep 14th, 2007 @ 06:23 ( . )

On 09/13/2007 @ 20:43, CarlB wrote :

: I'm seeing I wasn't completely correct in the above. For A2, you go out to the saturation point, not the grid conduction point.

Eh? The "2" suffix always indicates you're running into grid current, whether only a little or a lot. (But the valve should never hit saturation.) I think you were right in the first instance.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Old thread - Here are the definitions, and how to determine A vs'
Author:CarlB (guest: search)
Date: Sat, Sep 15th, 2007 @ 11:18 ( . )

For A2 or AB2, the idea is that the grids are driven into conduction by a source with low impedance, so as to avoid grid clipping.

For such an amp the top side of the output is only limited by saturation, that place on the left side of the plate graph where all all the current traces are below the "knee."

So, for class A operation of this type, the tube must be biased such that either saturation happens first, or saturation and cut-off occur at the same time.

It's the same situation as for a mosfet. Since the device doesn't have a gate conduction point (well, without having inadvertently destroyed the part it doesn't), class A is setup such that the device hits saturation first, or saturation and cut-off at the same time.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Old thread - Here are the definitions, and how to determine A vs'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Sat, Sep 15th, 2007 @ 14:57 ( . )

Thank you. That's what I've been saying for years, albeit not as lucidly...

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 09th, 2007 @ 20:08 ( . )

Sorry, Carl, I STILL disagree. If the load is constant through 360 degrees of the cycle, the stage is operating in Class A. It might not be behaving as though it is, because it's clipping, but it's still class A.

And to further iterate my earlier comment... Who cares? If the damn thing works, who really gives a sh*t what class of operation it's working in? If this was a radio, microwave, radar, etc I might give a toss. But it isn't, so I don't. Bias it so it sounds good and doesn't red-plate/red-screen and leave the discussion to the audiophools.

The problem is... all of the points of view are equally valid. Phil/Your/Randall's assertion that you can operate a SE stage in AB are perfectly valid, because the stage ACTS as though it is. So even me turning round a saying it CAN'T is bollocks, though true, because it ACTS as though it is. The noise that comes out the end SOUNDS as though it is, so for all practical purposes (tone wise) it IS in AB. But it still ISN'T operating that way. The load stays constant (or very nearly so. If the air gap is insufficient in the OT core, it probably doesn't) and the current draw is nearly constant. There HAS to be a notable variation in current throughout the cycle, observed from the point of view of one valve in the stage, for it to be operating in AB.

Like I said, who cares! ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 09th, 2007 @ 20:15 ( . )

The load changing isn't part of the textbook definition of amplifier Class, although it is something that results from operating a PP amplifier in Class AB. And once the power element starts cutting off (clipping) for part of the cycle, you will get a change in current draw.

And if you don't really care, that's fine as well. According to your previous post below, you seem to be happy with CarlB's definitions. :o)


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 09th, 2007 @ 20:34 ( . )

I AM happy with Carl's definition. And yours. And Mine, Randall's, etc. It depends on your point of view as to which class of operation a stage is operating under at any given instant, which a fluid thing anyway once you start clipping the bast*rd.

I guess you could even go as far as to say that NO stage is operating in class A unless it's being run off a 100% stable constant current source. That would be the definition for a cross-field amplifier, if I'm not mistaken (Mike Sulzer will know), so the definition even changes with application.

I guess my point is... this discussion pops up every so often and we all pile in trying to nail it down to an exact science. And it ISN'T. It's an art. We want what should be a beautifully smooth, clean gain chain to turn our input signal into a horrible distorted mess. One that just happens to sound good. You can't do that with textbook definitions and science. I'm the first person to admit that I'm a sim-junkie. Nothing wrong with that, saves having to do a bunch of tedious sums. But there's still a point where you decide you've got it as good as you're going to and get out a soldering iron.

You build a good amp with definitions, sums, sims and sticking to the textbook. You build a great one with your heart, soul and those weird sticky out things attached to the side of yer head. I think you once told ME that, Phil.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 09th, 2007 @ 21:36 ( . )

Yes, I think we're pretty much in agreement over that. However, I'm sick to death of some peoples' point-blank refusal to think outside the box, just repeating the old "SE amps can only work in Class A" mantra. Says who??? I also think that CarlB deserves credit for a set of definitions that get down to the fundamental issue and therefore are widely applicable in all kinds of applications, and not just traditional audio or hifi. I like your open-mindedness about all this....

On 09/09/2007 @ 19:34, Zoe N Iain wrote :
I AM happy with Carl's definition. And yours. And Mine, Randall's, etc. It depends on your point of view as to which class of operation a stage is operating under at any given instant, which a fluid thing anyway once you start clipping the bast*rd.
:
: I guess you could even go as far as to say that NO stage is operating in class A unless it's being run off a 100% stable constant current source. That would be the definition for a cross-field amplifier, if I'm not mistaken (Mike Sulzer will know), so the definition even changes with application.
:
: I guess my point is... this discussion pops up every so often and we all pile in trying to nail it down to an exact science. And it ISN'T. It's an art. We want what should be a beautifully smooth, clean gain chain to turn our input signal into a horrible distorted mess. One that just happens to sound good. You can't do that with textbook definitions and science. I'm the first person to admit that I'm a sim-junkie. Nothing wrong with that, saves having to do a bunch of tedious sums. But there's still a point where you decide you've got it as good as you're going to and get out a soldering iron.
:
: You build a good amp with definitions, sums, sims and sticking to the textbook. You build a great one with your heart, soul and those weird sticky out things attached to the side of yer head. I think you once told ME that, Phil.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 09th, 2007 @ 21:59 ( . )

Yup, quite happy to be open minded about it. But I still agree with Merlin.

If there's no change of load, and by extension current flow, in a PA, it's operating in Class A. Single ended stages have to do that to work clean before the onset of clipping, so they are inherently class A. Different matter once you bias them such that they'll hit cutoff before saturation or vice versa but that situation only happens when driven by an appreciably large signal. The stage is still Class A but it's being overdriven. Maybe we should adopt A1-C (cold), A1 (dead centre) and A1-H (hot) as the accurate definitions for those cases?

That's the trouble here, isn't it? We're being limited by the definitions that 'proper' amps would use. But we aren't building proper amps. We're abusing the valves, as far as the original designers would be concerned.

Actually, that begs an interesting question.. has anyone ever designed a valve that's specifically intended to be overdriven into audio frequency distortion? One or two computer valves probably were, as they were designed for relatively low frequency operation as square wave switches. 5965 and such. Which would explain why my GE 5965 absolutely kicks ass in the 12AU7 position of the Blues preamp, cos it's MEANT to be overdriven and designed to do so.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sun, Sep 09th, 2007 @ 22:25 ( . )

We're kind of mixing in other points of discussion here ...

There's a few of us (well, probably just Phil and I) who do want to get to the bottom of this definitively - and think it just might be possible in an undeniable way.

If you don't think that's possible Iain, wish us luck, roll your eyes, and have a good chuckle. It's probably well deserved!

I don't see the definitions in the RC-XX manuals. They do describe them, but don't give the definitions.

Hmmm. Can we all accept the following?

"Web definitions for Class AB
an amplifier operating with the grid bias adjusted so plate current flows for greater than 180 degrees, but less than 360 degrees of the input waveform, by biasing the tube above cutoff, but below the point required for class A operation. The distortion is higher at low signal levels than true class A, but the efficiency is higher, although not as high as class B, allowing more output power than class A for a given plate dissipation.
www.aikenamps.com/AmpTerms.html"

If we can accept that definition, then we can definitely bias an SE amp into AB.

And we do see the RF guys accepting that as amp's class of operation.

As an guitar amp example, the P1eX gets higher output power - higher efficiency - from a given plate dissipation by accepting that its max power dissipation is generated when the plate current is conducting less than 360 degrees.

That's an SE amp ...




-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sun, Sep 09th, 2007 @ 22:30 ( . )

For the last paragraph and sentence of that post:

'As a guitar amp example, the P1eX gets higher output power - higher efficiency - from a given plate dissipation by accepting that its max power dissipation is generated when the plate current is conducting less than 360 degrees.

That's a class AB SE amp ...'

(Don't post when distracted.)

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 09th, 2007 @ 22:49 ( . )

Exactly my point. The 'traditional' definitions of which class it's in don't apply when you drive it to clipping. So it's a pointless discussion. If you look at it that way, EVERY guitar amp is Class A some of the time, AB some of the time and even Class B a bit of the time. But a SE amp is biased, at idle, in class A or you wouldn't be able to get a clean signal out the end at all, no matter how low the drive. And it's the idle condition that matters.

Enough! It DOESN'T matter. Describe it in terms of the maximum possible dissipation or something, because trying to describe the class of operation is useless and only holds true for one given signal level anyway, due to the reflected impedance of the speaker.

Same for hi-fi, for that matter, but they're too stupid to understand that.


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Mon, Sep 10th, 2007 @ 00:22 ( . )

Yes, and no. An amplifier that's built by design to cut off during part of its duty cycle, isn't the same as a true Class A amp design that's simply having the snot driven out of it, and therefore being forced to cut-off as well as saturate.

And all Class AB amps appear to be working in Class A when operating at lower amplitudes, whether they're PP or SE. This is originally where the classic Vox Class A confusion came from - the fact that the EL84s are usually biased at around 100% dissipation at idle and conduct through the entire duty cycle when not being cranked hard.

On 09/09/2007 @ 21:49, Zoe N Iain wrote :
:
Exactly my point. The 'traditional' definitions of which class it's in don't apply when you drive it to clipping. So it's a pointless discussion. If you look at it that way, EVERY guitar amp is Class A some of the time, AB some of the time and even Class B a bit of the time. But a SE amp is biased, at idle, in class A or you wouldn't be able to get a clean signal out the end at all, no matter how low the drive. And it's the idle condition that matters.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Mon, Sep 10th, 2007 @ 01:34 ( . )

Well it all comes back to my point, we need a label for a Single-Ended power-amp that is specifically going to have the snot driven out of it most of the time, whether it be cold biased or hot biased, whether it conducts for 360º or not.
After all it is only name we are talking about, so as to describe to someone in one word a class/mode of operation.
I think Francis put it fairly well.

Of course this label wouldn't apply to the high-fidelity community, they wouldn't even dream of over-driving their precious power valves in the fear that the mojo would escape.
To the contrary my valves only contain snot 8^O

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Tue, Sep 11th, 2007 @ 02:44 ( . )

And that is why there's this common, wrong, but perfectly understandable horde of people who speak of an amp running in Class A until it's driven into Class AB.

Is there anything wrong with saying that if we can drive it into cutoff before we reach saturation, it's biased Class AB?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Mon, Sep 10th, 2007 @ 01:21 ( . )

Iain, you wrote, "Exactly my point. The 'traditional' definitions of which class it's in don't apply when you drive it to clipping. "

But it's not going into clipping. The traditional definitions apply. The plate conducts for less than 360 degrees before clipping happens. (And that's what does happen in the P1eX output stage.)

If you have an SE amp that clips before cut-off (or I suppose clips and cuts-off at the same time), that would be class A: there is no current cessation under normal (non-clipping) operations.

I don't disagree with you about the total usefulness for the term 'class' (which has limited usefulness for us in my opinion), it's just that when the 'class' question gets brought up we should strive to give the correct answer, regardless of what that is.

Or, I suppose we could answer that it's nearly irrelevant for guitar amps, and so we tend to characterize output stages by SE or PP, and then by hot, medium or cold biased. If that's where you want to leave it, I suppose we might all be able to agree on that.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Mon, Sep 10th, 2007 @ 15:00 ( . )

To suggest that a single ended stage can be class AB demonstrates a gross misunderstanding of either; how a PP amp works, or what those old textbook definitions are trying to explain.
It's an easy trap to fall into, - I did myself when I was a newb, but Class AB DOES NOT simply refer to a bias point somewhere between cutoff (Class B) and mid-point (perfect Class A). It refers very specifically to the way the load on a PP amp makes a transistion from a class A load, to a class B load. It really does mean a combination of class A AND B. Single ended stages don't exinit this change in loading.

I agree that the old definitions that use "degrees of conduction" are useless for guitar, and should be banished forevermore, but definitions of class using the position of the operating point DO work, provided you understand what is happening in ush pull first.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Mon, Sep 10th, 2007 @ 16:27 ( . )

I will again say that this looks like circular logic to me. You are using a definition for Class AB which only applies for PP amps, and then on that basis, saying that therefore SE amps can't ever be Class AB. That's just not a valid argument, as all it really proves is that an SE amp isn't a PP amp. IMO the phenomenon you're using as your definition IMO is an effect of Class AB operation, not the underlying cause.

And as already stated Class AB SE amplifiers are widely accepted in the RF world, as are Class B and Class C. It only seems to be the audio crowd that struggle with the idea, because traditionally they've only see Class AB used in PP amps. But I can't really criticize anyone, since when I was a newb I used to think that way as well. :o)

On 09/10/2007 @ 14:00, Merlin wrote :
:... but definitions of class using the position of the operating point DO work, provided you understand what is happening in push pull first.
:
On 09/10/2007 @ 14:00, Merlin wrote :
:
To suggest that a single ended stage can be class AB demonstrates a gross misunderstanding of either; how a PP amp works, or what those old textbook definitions are trying to explain.
: It's an easy trap to fall into, - I did myself when I was a newb, but Class AB DOES NOT simply refer to a bias point somewhere between cutoff (Class B) and mid-point (perfect Class A). It refers very specifically to the way the load on a PP amp makes a transition from a class A load, to a class B load.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mark L (registered user: 1458 posts )
Date: Mon, Sep 10th, 2007 @ 16:49 ( . )

Of course a SE amp can be class AB or even B or C, it's just not desirable in (hi-fi) audio application. The main reason they never biased SE amps class AB is because the clean, undistorted output is lower than class A and distortion was deemed undesirable in audio amps. The same reason you see circuits of class AB audio amps in PP configuration, max undistorted output.

But for RF (and to some extent guitar amps) it's a different story. In RF they bias for efficiency and the unwanted harmonics are filtered out with high-Q filters, in some cases used as frequency doublers or triplers. With guitar, in most cases, we want harmonic distortion.

Classes of operation are based around what period of the cycle the device (semiconductor or valve) is conducting (or not conducting) at maximum power, thats's all!
When it reaches saturation the device is still conducting, so saturation doesn't come into it, i.e. when a class AB PP amp clips due to saturation it is still a class AB PP amp.

I did a quick seach for references with SE class AB and B amps to check my recollection on this topic (it's been a while since I studied Electronics Engineering).
[link]
[link]
[link]
[link]




-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Mon, Sep 10th, 2007 @ 17:41 ( . )

Certainly SE stages can be Class C, D, etc. (not that they'd be any use to us), but they cannot be two classes at once. The name only applies to a PP phenomenon. It's not a "cause and effect" thing like Zaphod is trying to suggest, its just a name. A name for a stage having a class A, and a class B loadline all at once. Simple. No circles there. I fail to see why some find that so hard to grasp.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mark L (registered user: 1458 posts )
Date: Tue, Sep 11th, 2007 @ 01:14 ( . )

All class AB amps behave as class A at low power (i.e. when the valves are conducting 360 deg.) and up to the point where it starts to go into cut-off a class A load line (Zplate-plate/2) is apparent. After that point a class B load line is apparent (Zplate-plate/4).

To determine the class of the operation of an SE amp to describe the way it is biased, you need to look at it at full output voltage swing up to the point of saturation. If it's not conducting for some of the cycle but less than 180 deg, it's class AB, simple.

If you just simply increase the input and look at the output up to the point of distortion then stop, of course it's class A. Audiophiles do this because they are not interested after it starts to clip, so to them an SE amp is only of any use to them operated in the class A region, so they bias accordingly. From this, somehow they came to the conclusion that SE can only be class A, which is not true like many other myths in the audiophile world.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Tue, Sep 11th, 2007 @ 01:46 ( . )

So we have another hole in the logic. When there's conduction for 360 degrees at rated output an SE amp is in Class A. When there's conduction for 180 degrees, it's in Class B. When there's conduction for less than 180 degrees its in Class C. So what class is it in when it conducts for less than 360 degrees but more than 180 degrees? It doesn't have an operating class? This clearly shows that the definition being used falls short, and a more universally applicable one needs to be used instead.

To me a PP amp operating in Class AB is a special case within the general category of Class AB amps. As we know, it exhibits the very interesting phenomenon of changing load impedance, or as you put it, having two different load lines. That is a description of the *behavior* of this particular one type of Class AB amplifier, rather than the fundamental definition of what all Class AB amps are.

On 09/10/2007 @ 16:41, Merlin wrote :
:
Certainly SE stages can be Class C, D, etc. (not that they'd be any use to us), but they cannot be two classes at once. The name only applies to a PP phenomenon.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Tue, Sep 11th, 2007 @ 01:49 ( . )

Of course, I ultimately agree with Zoe that this becomes somewhat of an ivory tower thing, when all we really care about is how a guitar sounds through any given amplifier. :o)

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:yecrom (registered user: 1199 posts )
Date: Tue, Sep 11th, 2007 @ 10:17 ( . )

I wish this discussion was happening on 18-watt or one of it's related sites.

I can't do the little popcorn smiley here. :-D

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Tue, Sep 11th, 2007 @ 10:22 ( . )

On 09/11/2007 @ 10:17, yecrom wrote :
I wish this discussion was happening on 18-watt or one of it's related sites.


Oh God, it'd probably get banned or something!

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Tue, Sep 11th, 2007 @ 10:57 ( . )

It would actually sit quite comfortably at SEwatt.com or wattkins.com, which are part of the 18watt.com family (or vice versa). But I don't see any reason not to have the discussion on AX84.com.... /-:

On 09/11/2007 @ 10:22, Merlin wrote :
:
: Oh God, it'd probably get banned or something!
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Humor Break'
Author:Doug H (registered user: 5455 posts )
Date: Tue, Sep 11th, 2007 @ 12:18 ( . )

Tastes Great! Less Filling!

Bah... Who cares?

This thread needs a humor break:

[link]

Enjoy...

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Tue, Sep 11th, 2007 @ 09:55 ( . )

On 09/11/2007 @ 01:46, Zaphod wrote :
So we have another hole in the logic. So what class is it in when it conducts for less than 360 degrees but more than 180 degrees? It doesn't have an operating class?

Ha, no the hole is quite clearly in your logic:
Suppose you took an SE stage, and biased it hot. Really hot. Clearly it's class A. Now overdrive it. More, and more. Eventually it will hit cutoff, so it's no longer conducing 180 degrees. By your logic, it must be class AB. Therefore, by your logic, an SE stage cannot be class A at all! Which is of course, absurd.

There is nothing wrong with the class system, it makes perfect sense provided you define it using the load line, and where the bias point falls on it, which some textbooks do.
It only falls apart when you try and define it using "degrees of conduction" which I agree only works for hifi, or repeating, symmetrical waveforms. So forget "degrees of conduction". Simple. Just use the other definitions.

That is a description of the *behavior* of this particular one type of Class AB amplifier, rather than the fundamental definition of what all Class AB amps are.

Er...no... that IS the definition of Class AB. That's where the name comes from. I think that's pretty obvious.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Tue, Sep 11th, 2007 @ 11:57 ( . )

Merlin and Phil,

Things are getting a bit heated, so let's strive to keep things civil.

The disagreement we have seems to be about the definition itself.

Does Class AB, by definition, refer to only push pull amps and/or amps where load-lines change? I've never seen that. I've only seen the distinction between the conduction angles as defining the classes.

Aside from the not having seen anything in the definitions of class AB about load-lines or push-pull exclusivity, there are many examples of SE class AB as used by the professionals. A few examples:

1) [link]

2) Patents for tweaks to Single-Ended Class AB circuits (RF, etc., domain).

3) Randall Aiken's citing of class definitions, and how they apply:

[link]

So, again, I don't think it's question of not seeing each others side in the discussion. I think we need to look at what definitions we're using, and making certain they are they the same. I'd say we're not using the same definitions. Let's strive to find out which is the accepted *definition* for amplifier class.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Tue, Sep 11th, 2007 @ 13:02 ( . )

Hmmmmm, I thought we were having a reasonably sensible discussion here. I don't see any reason that can't be done in a civil way... :o)

Anyway, to add to the links already posted, I just went over to my bookcase and grabbed a few electronics textbooks, to see what they would say about the definition of Class AB. Interestingly, a couple of the more modern ones didn't mention it at all and instead went into great detail about how to build Class B amps that don't suffer from crossover distortion - all done in SS with lots of NFB, of course. So I ended up with just two books from my quick random grab, which had anything much to say on the subject.

The New Penguin Dictionary of Electronics by E.C. Young 1988
A linear amplifier in which the output current flows for more than half but less than the whole of the input cycle, ie the angle of the flow is between pi and 2pi. At low input-signal levels class AB amplifiers tend to operate as class A and at high input-signal levels as class B amplifiers.

Modern Electronic Circuit Design by D.J. Comer 1980
It is sometimes advantageous to cause each stage to conduct more than one-half cycle, but less than the full cycle. This mode of operation is called Class-AB.


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Tue, Sep 11th, 2007 @ 17:46 ( . )

There is the key to our problem of definition " Linear amplifier". We are not (for the most part) building linear amplifiers.

Anyway if any of you have access to any of the ARRL handbooks, I suggest you read those. There are plenty of references to "Classes of amplifier service" explained very well by some very learned authors. Some almost as old as the thermionic valve itself.

Here is one paragraph you may find of interest

ARRL handbook 1992 p.3-17
A single-ended class-B stage is suitable
for linear RF amplification if the load is
a tuned circuit having a Q of five or better.
The flywheel (ringing) effect of the tank cir-
cuit supplies the missing half of the output
cycle. The efficiency of this type of
amplifier is typically 60 percent.


This of course doesn't apply to audio amplifiers, though one could bias a SE audio amp into class-B, it just wouldn't sound all that good.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Wed, Sep 12th, 2007 @ 03:00 ( . )

I don't think we should put to much weight in the "linear amplifier" term, as there is no exact criterion for an amplifier to be linear. For example, does a linear amplifier have less than 1% THD? Or less than 20%? No, it's just a general term, which distinguishes from such things as Class D PWM amplifiers, and a guitar amp can also be considered to fall within this general category, even one that distorts.

On 09/11/2007 @ 17:46, Dan N wrote :
:
There is the key to our problem of definition " Linear amplifier". We are not (for the most part) building linear amplifiers.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Wed, Sep 12th, 2007 @ 08:54 ( . )

On 09/12/2007 @ 03:00, Zaphod wrote :I don't think we should put to much weight in the "linear amplifier" term, as there is no exact criterion for an amplifier to be linear.
But one criteria when we set to make a linear amplifier is, we strive for perfection.
The point I'm trying to make is, that the engineers/scientists of the time whom developed the class of operation standards had set their goals to make the perfect amplifier. So that the output wave form is a facsimile of the input wave form. Were as we set our goals to distort and even add harmonics to the wave form.
They wouldn't even have giving a thought of someone trying undo all their efforts in the pursuit of perfection.

So can we always apply the existing terms of class to a guitar amplifier ?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Wed, Sep 12th, 2007 @ 10:33 ( . )

We are on the whole still striving for perfection in our amplifiers. It's just what we mean by perfection has shifted a bit. However, I think you're quite right to suggest that some of the class definitions may also have to be adjusted to apply more generally. Also the suggestion by some folks that many SE guitar amplifiers aren't operating in any class at all is clearly nonsensical.

Now the conduction angle definition works well for all types of amplifiers, so we can hold on to that. However definitions based on load impedance change or absence of distortion are limited in their usefulness, and are really only helpful in the traditional audio and hifi fields.

So far, I still believe that CarlB's revised definitions are the most useful and generally applicable, since they're based on how the valves themselves behave and don't depend on such things as lack of distortion or load change.

:
On 08/10/2006 @ 05:30, CarlB wrote :
: Class A: cathode current flows for the complete 360 degrees of the waveform.
: :
: : Class AB: cathode current flows for less than 360 degrees, but more than half (180 degrees) of a the waveform.
: :
: : How to tell the difference? Crank the amp up until just before grid conduction (e.g., just before grid clipping for class A1 and AB1). If there's still cathode current flowing at the bottom of the waveform (i.e., the side opposite the one which is just before grid conduction clipping), then it's class A. If there's more than half of the signal for which cathode current is flowing, then it's class AB.
: :
: : For class A2 and AB2 (i.e., muscling the grids into non-clipping at grid condition), crank the amp until just before grid conduction (verify with a scope, meter, etc.), and repeat the above.
: :
: : Anything beyond grid conduction is fair game (and fun) for amplifier operation, but it is not fair game for class determiniation. That's because any output stage, no matter how hot-biased, can be driven into bottom-side cutoff at some point.
: :
:


On 09/12/2007 @ 08:54, Dan N wrote :
:
: The point I'm trying to make is, that the engineers/scientists of the time whom developed the class of operation standards had set their goals to make the perfect amplifier. .... So can we always apply the existing terms of class to a guitar amplifier ?
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Wed, Sep 12th, 2007 @ 10:48 ( . )

On 09/12/2007 @ 10:33, Zaphod wrote :

:
: Now the conduction angle definition works well for all types of amplifiers, so we can hold on to that. However definitions based on load impedance change or absence of distortion are limited in their usefulness, and are really only helpful in the traditional audio and hifi fields.

It's the other way round. The definitions using change in load are nice and intuitive- you can see WHERE the class names come from, and what they mean. The conduction angle thing is totally useless for guitar because we consistently overdrive our valves to the point where they don't conduct for 360 degrees, even when they're in irrefutable Class A. There's no need for a "new system", it will only serve to complicate matters. "cool class A", or "hot class A", or "mid point biased class A" work perfectly well, and the same goes for "hot class AB" etc.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Wed, Sep 12th, 2007 @ 12:09 ( . )

I tend to agree with you on that point. There is something quite neat and elegant with this particular definition. Unfortunately, it's not the generally accepted one. However I think CarlB's conduction angle-based definition, which looks at grid current, works well for guitar amplifiers as well as hifi and RF applications.

On 09/12/2007 @ 10:48, Merlin wrote :
:.... The definitions using change in load are nice and intuitive- you can see WHERE the class names come from, and what they mean.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Wed, Sep 12th, 2007 @ 16:31 ( . )

Yes, I too agree there is no need re-define the existing classes, and the terms "hot" and "cold" work for me as well.
However I'm not convinced that you can give an SE guitar amp the label AB.
A normally bias SE Class-A1 amp is quite capable of also being a A2 (provided the grid driving valve can flow current) and also class-B simply by the amplitude of the grid signal.
But for a guitar amp, at what power level do we use to define the class?
Normally it would be the maximum signal that is most likely to be seen by the grid.
However... are we always going to be playing geetar with the amp at full power ?
Some will, but I think that for the most of us that we will be playing at less than half power. Hence the valve will always be conducting. Unless of course you are using a very low power amp.

On 08/10/2006 @ 05:30, CarlB wrote :: Class A: cathode current flows for the complete 360 degrees of the waveform.
: :
: : Class AB: cathode current flows for less than 360 degrees, but more than half (180 degrees) of a the waveform.
I'm just wondering whether if we can use "cathode current" ?
The ARRL handbook uses the term "Power taken from the power supply by the device"
So measuring cathode current flow may be fine for a common cathode, anode loaded amplifier, it may not work for other types.
I just can't think of which ones though, hmmmm :?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 12th, 2007 @ 17:16 ( . )

I swore I wouldn't re-enter this pointless discussion, being as I actually agree with Merlin and screw what the textbooks say. But..

The only useful way you can assess the class of an amplifier is to define it at idle. So a SE amp is CLASS A, because you CAN'T bias it any other way at idle, unless you set it up so that it only conducts for 180 degrees of the signal, NO MATTER HOW SMALL THE SIGNAL HAPPENS TO BE, then it's class B. BUT IF you bias it in class B, then it is no longer capable of clean amplification AT ALL and it IS OPERATING ON A DIFFERENT LOADLINE, AS WELL!

A SE amp can't swing between one load condition and another. There are no additive or cancelling currents in the OT core to swing the loadline from one to the other, like you get with an AB push-pull amp. It can't change class of operation! IT IS PHYSICALLY and ELECTRONICALLY IMPOSSIBLE!

The reason you can ONLY define it at idle is that there is NO loudspeaker in existence that presents a consistent load to the output device through it's entire frequency range. If the reflected impedance changes, the point where the device is currently hanging out in it's class of operation chnages. If it changes ENOUGH, and let's be honest - cranking the shit out of it is going to change stuff - then it is no longer fixed at an operating point. So even saying that something is operating in Class A, B, R or Z is a stupid thing to say with regard to an amplifier that is DESIGNED to spend most of it's time on a totally non-linear portion of it's output device characteristic curves.

The closest we can manage is to say that it's Class Whatever at idle and clips at a given voltage. So you can use the Hot, Mid, Cold terms as described above.

I don't give a shit what the textbooks say, they were written by people who would never have IMAGINED that someone would want to drive such a big signal into a PA as to make it clip, at audio anyway. Radio is totally different. There were no JCM800s when those definitions were written and THAT is why they specified conduction angle as the determinant.

A single ended amplifier can be operated, for useful audio purposes, in only one way. And it's class A at idle.


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Jon Anderson (registered user: 1491 posts )
Date: Wed, Sep 12th, 2007 @ 17:45 ( . )

On 09/12/2007 @ 17:16, Zoe N Iain wrote :
: The only useful way you can assess the class of an amplifier is to define it at idle.


This confuses me a bit. If you only define at idle, then wouldn't all amps be class A? How would you get a class B amp with no signal?

Jon

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 12th, 2007 @ 18:01 ( . )

You can't get Class Anything with no signal, Jon. But the instant that there is a sniff of a signal a Class A amp will amplify it through 360 degrees. A Class B amp is INCAPABLE, even at 1uV of doing so. It can only amplify 180 degrees.

The reason we are having this argument is that certain people have neglected ONE simple thing about PP AB amps. BOTH sides of the phase inverter are feeding a signal to BOTH sides of the PA. The currents in the OT primary cancel, so you can bias the bloody thing any way you like.

In a SE amp, you can't DO that. There is only one driver signal. The PA HAS to amplify all of it. You can bias such that one end clips before the other, sure. But the PA can NEVER shift it's load from class A to class B. They require different loads. A pure Class A (will clip at the SAME time on both positive and negative half cycles) or pure class B (will only EVER be able to present one half cycle or less). The ability of a PP PA to do so is based on the change of loadline and current that the driving devices see. There is no significant change for a SE stage to see, hence no ability to change class of operation. It can only clip.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 12th, 2007 @ 18:32 ( . )

(Iain) Sorry, I have to agree with Merlin, John, RDH4 and Zoe on this one. I think the textbooks are just missing out 'at idle', because if they are the textbook definitions and ours agree 100%.

Class A = can conduct 360 degrees of the cycle at ANY input level

Class B = can NEVER conduct more than 180 degrees at ANY level

But a shift in operating point requires a shift in current. And a SE stage can't do that.

Sorry guys, you're wrong. And so are your textbooks if they state it as truth.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 12th, 2007 @ 17:46 ( . )

Well, I'm certainly not the experienced technician or engineer that some of the others in this thread are, BUT...

The only reference I have to hand is RDH4. I'm not trying to be pedantic, draconian, or a strict True Believer in Big Red, but for what it's worth, RDH4 agrees with Zoe, Iain, Merlin, et. al., in that class of operation includes the load and not just the angle of conduction in its definition. Start on about page 545 if you want to look for yourself.

I could well be reading this wrong, but heck, there's my $0.02 in the (agreed) pointless debate.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Wed, Sep 12th, 2007 @ 19:36 ( . )

From page 545:

"Class A operation* is the normal condition of operation for a single valve, and ..."

It's not saying it's the only class.

And what F. Lagnford-Smith, B.Sc., B.E. (the author of that section) thought of as normal is in line with "Audio Frequency Power Amplifiers," the section title.

That is to say, where faithful audio reproduction is the goal. So, yes, class A would be the normal mode for someone wishing to maximize the non-distorted power available from a Single-Ended amplifier.

What was normal for good ol' 'F. Langford' is not normal for us, however. We want an amp that generates more power from a given plate than you'd get from class A operation. So, we bias it for class AB.

The P1eX starts to cut-out before it clips, but that's OK with us. We get more power out of it this way, and we like the tone of it. And nope, it isn't a faithful audio reproduction amplifier right before clipping, because it's already in cut-off.

But us guitarists don't care. It *sounds* good.

And so there you have it.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 12th, 2007 @ 20:24 ( . )

I get what you're saying, Carl. We are certainly using the things in ways Mr. Langford would never have envisioned.

However, I was only trying to point out something about what the definition of amplifier class entails. Reading the rest of that section, from pages 572-around 585, describes class AB as a condition in which half of the OT primary has a current of 0 for part of the cycle, and further, how the grids are exitied in an alternating fashion from the phase inverter.

It seems to me, from a purely "definition of terms" standpoint, that AB operation is therefore only possible in a push-pull configuration.

Not that I think that matters in any way, for the reasons you and everyone else have stated before.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Wed, Sep 12th, 2007 @ 20:53 ( . )

OK, but you're missing some of what I was trying to point out.

He's not giving the definitions here, just how the classes are used in hi-fi type audio reproduction devices.

But again we're not doing hi-fi style reproduction here. When he writes, "Class A operation* is the normal condition of operation for a single valve ..." he is acknowledging that it is not the only operating condition.

When he writes, "Class AB operation* indicates overbiased conditions, and is used only in push-pull to balance out the even harmonics," he is again writing in the context of the section. If the aim is faithful audio reproduction, then you'd only use SE class A, and reserve class AB soley for push-pull. Otherwise you'd have an SE amp with max clean power that's lower than for class A. And he correctly indicates that you just wouldn't do that for this application.

But again we would do that for ours. Although its maximum clean power is lower than class A, our SE class AB achieves more power before clipping. And to us that sounds like more clean power. We don't care that the meters and scopes say, it's our ears that we need to please.

And yes, 'you say potato, I say potahto.' But I do think there is a correct answer. But it's not worth getting getting annoyed at other good folk here over, which is what I'm afraid is happening to some degree.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 12th, 2007 @ 21:01 ( . )

(Iain) Not at all, Carl. We've repeatedly said that your definition could well be considered correct, at the same time we said any definition was suspect, depending on the application.

I agree with my good lady wife that your the facts of the matter are fairly inconsequential.

But I also agree with Merlin about the defintions.

To echo Zoe, who cares?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Wed, Sep 12th, 2007 @ 21:40 ( . )

Oh ... I do care. I really want to see this community use the correct definitions, whatever those are.

I do realize that others think they're right and I'm wrong, and regardless of who is right, the paramount concern is having respect on all sides for all who are willing to slog through the effort of working this through.

Back to the discussion ... In this instance, I "know" there's only one correct definition. Heh!

The definitions are solely about the degree of conduction discriminators. There is nothing in them about changing load-lines or ... what was that other part?

Anyway the other stuff, in some mighty lofty tomes no less, is always there to help understand the definitions ... as they apply to the applications at hand. Few do a great job of providing a comprehensive survey of the classes as applied to the many fields of amplification that use them.

And so in lies the misunderstandings. Since Class AB isn't talked about in terms of SE, or something to the fact that "AB only applies to PP," (note, "AB only applies to PP" is true only for the goal of faithful audio reproduction), then people start to assume that an SE amp can't be Class AB.

Those darned RDH4, ARRL, RC-24, etc., guys!!! Quite correct, the lot of them, but they sure weren't very complete on this issue.

Maybe we need to look in the RF sections of the tomes for more about SE class AB. It certainly wouldn't have occurred to the hi-fi guys to include it.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Wed, Sep 12th, 2007 @ 23:15 ( . )

On 09/12/2007 @ 21:40, CarlB wrote : Maybe we need to look in the RF sections of the tomes for more about SE class AB. It certainly wouldn't have occurred to the hi-fi guys to include it.

Does anyone know roughly what year(s) that the classes came to being ?
I will try to look through some old journals at the library to see what sort of discussions took place during that period.

I wouldn't discount that any such discussion took place, but the thought of the day may have been "Who in their right mind would do that to a valve?", "We won't bother to give that operating condition a class descriptor."

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Stephen Keller (registered user: 6220 posts )
Date: Thu, Sep 13th, 2007 @ 01:34 ( . )

The earliest reference I found to it was this one:

J.H. Morecroft and H.T. Friis, "The vacuum tube as a generator of alternating current power," Trans. A.I.E.E., 38, No. 2, October 1919.

I'm not able to tell you if Morecroft and Friis use the "class" terminology, but they are cited with respect to the analysis Class B and Class C amplifiers in various I.E.E.E. journal articles. It's where I'd start looking for the origin of the terms.

Stph

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Stephen Keller (registered user: 6220 posts )
Date: Thu, Sep 13th, 2007 @ 11:50 ( . )

I found this article on Google. Morecroft and Friis do not use the "class" terminology, rather they talk about efficiency at various operating points. In other words they are exploring the phenomenon which forms the organizing principle behind the classes.

Stph

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 12th, 2007 @ 17:48 ( . )

Let me put it more simply.. Phil, upto the onset of clipping, in what Class of operation is a P1eX PA operating?

It's Class A, even you can't disagree.

Does it, at any point, do some magical transformation into an amp that that is NO LONGER CAPABLE of conduction through 360 degrees of it's input cycle if you lower the input amplitude sufficiently again but could hold up the current flow through the OT primary, which a PP AB amp does?

The load is inductive. Current and voltage are NOT in phase. But the currents cancel, hence no need for an air gap to prevent saturation of the core. I'm assuming you still agree?

A SE amp set to clip on one side of the waveform before the other will, if the bias offset is large enough, saturate the OT core even with an air gap.

A SE amp set in class B WILL NOT! There is 180 degrees of conduction, causing saturation, and 180 degrees of nothing, allowing the relaxation of said saturation. It's how you get a relatively good sine wave out of a SE Cl. B amp, since the charge is dumped into the other side of the waveform, which the amp isn't actually supplying.

In practice, the relaxation of the OT charge takes longer than the driven cycle, hence you can get more efficiency from class C.

So the closest you can manage with a SE audio amp is Class A but 'badly' biased, such that the airgap is adequate to dump the core saturation that, in a true PP AB amp, would be cancelled by the other half of the output cycle.

If it doesn't have such a wondrous power, then it is operating in CLASS A BUT CLIPPING! And the airgap is doing it's job.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Wed, Sep 12th, 2007 @ 19:06 ( . )

"... upto the onset of clipping, in what Class of operation is a P1eX PA operating?

It's Class A, even you can't disagree."

The P1eX plate stops conducting on the bottom side well before it starts clipping. That's why we're saying it's Class AB.

"Does it, at any point, do some magical transformation into an amp that that is NO LONGER CAPABLE of conduction through 360 degrees of it's input cycle if you lower the input amplitude sufficiently again but could hold up the current flow through the OT primary, which a PP AB amp does?"

No. But that's not the definition of a class AB amp as I've seen it in print. I've only seen it defined in terms of conduction angle. If we can get a definition

You're effectively saying there are a whole realm of professional engineers, including RF engineers and patent engineers, who have missed the definition. They refer to a whole family of output stages as Single-Ended Class AB. And, this source wrong would be wrong too:

[link]

(About a third of the way down.)

So ... these folks are all wrong?




-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 12th, 2007 @ 19:17 ( . )

(Iain) Yup.

Class A = can amplify 360 degrees of an input signal

Class B = Can NEVER amplify MORE than 180 degrees.

SE = NO change of current, hence no change of load.

PP = a change of current, thus a change in effective load impedance.

Therefore, there is NO change from class A to class B in a SE amp. It is electronically impossible. Argue all you like. It's Ohm's Law.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Wed, Sep 12th, 2007 @ 19:45 ( . )

On 09/12/2007 @ 19:17, Zoe N Iain wrote :
(Iain) Yup.
:
: Class A = can amplify 360 degrees of an input signal
:
: Class B = Can NEVER amplify MORE than 180 degrees.
:
: SE = NO change of current, hence no change of load.
:
: PP = a change of current, thus a change in effective load impedance.
:
: Therefore, there is NO change from class A to class B in a SE amp. It is electronically impossible. Argue all you like. It's Ohm's Law.
--



Careful ... I most respectfully believe you have an erroneous understanding of the definitions are here.

I'll let that go for a bit however, because why do you say an SE amp does have changing plate current? When I look at those, the cathode starts climbing pretty quickly for increased power. At the precipitous moment, it then goes into cut-off for the bottom of the output signal.

Big change in current there. Which I think you agree with. So, you're indicating something slightly different? If so, describe that so I can see the core of your logic here ...

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 12th, 2007 @ 20:12 ( . )

(Iain) OK, in a single ended amp, the current goes up in the cathode circuit as a response to grid voltage. Standard valve theory, yeah?

In a push-pull amp it does too, but both sides are out of phase. But applied across opposite sides of an OT that has it's middle connected to the supply voltage.

So you have the real/apparent power thing. In SE the PA draws the current but then gives it back. In PP it takes it, then doesn't. I'm admittedly crap at at explaining this shit. I understand it but explaining it is a bit harder.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 12th, 2007 @ 20:20 ( . )

(Iain) Ok, that's bollocks, even to me. I understand what happens but I have trouble explaining it. So explain what the situation is in Class A and B and why you assert that operation changes from one to the other in a SE amp.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Wed, Sep 12th, 2007 @ 21:10 ( . )

Let's not do that just yet ... I'd like for us to come to an concurrence about what's happening in the plates first.

Does this sound correct:

SE Amp. Output stage biased on the cold side. Signal starts out low, and the plate is conducting for the full cycle. (Don't worry about class yet, just what's happening to the current through the plate.) Start upping the signal. Since we're biased on the cold side, at some point the plate goes into cut-off, but the top side hasn't reached grid conduction.

Picture a sine wave input, and next the plate current as reproducing that sine wave. If the whole signal can be reproduced with the plate current, then there is 360 degree conduction going on. As soon the bottom part of that current reproduction hits hit cut-off there's no current for that portion of the waveform. There's less than 360 degree conduction now.

Do we agree on the above? Either way, tell me more about the picture you have of the SE current not changing. That would be something different than what's going on above, and I think it's at the core of our non-concurrence.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Thu, Sep 13th, 2007 @ 05:27 ( . )

OK, I'll give it a go.

Firstly, just to quote someone brighter than either of us...

[link]

Having absorbed that (and noted that Mr. Aiken ALSO says that a SE amp can only operate in Class A for audio), my reasoning that the change of current is a determinant AND that the definition is only designed to be applied at the instant a tiny signal takes the PA out of it's quiescent condition is this;

In a Class A amp, the change of current is only instantaneous. Obviously, the current is changing because a tube is turning change of voltage at the grid into change of current at the anode, in a ratio defined by the mu of the valve. But the average current, observed over a few complete cycles, doesn't change. You can prove it by building one and trying to get the PSU to sag. Can't be done, unless you cheat and do it synthetically, I'm sure you agree. A pure class A amp will amplify 360 degrees of the signal until it runs out of headroom, then it clips the top, bottom or both.

In a Class B amp, the idle current is near or at zero when no input signal is present. Only when you give it something to amplify does it demand current from the power supply. In the absence of a tuned circuit to permit flywheel effect to fill in the other half of the waveform, a pure Class B amp can never amplify more than 180 degrees of the input signal. And it doesn't matter how small the signal is.

A Class AB amp is obviously somewhere between the two. It's biased above zero, so there's always some idle current. But, in a PP amp the current demand changes when you ask it to do some work. Because it IS biased partway to class B. It's not idling at zero so the current flow isn't 'lots,none,lots,none..' but it does require that increased current is supplied as input signal level is increased.

In a SE amp, the average current demand doesn't change, unless it's biased into class B at idle. In which case it can NEVER give you a clean output, no matter how small the signal is. Not true for an eX. Perfectly capable of a clean output.

But, you are about to argue, I'm not saying an eX is biased in Class B, I'm saying it's in AB. Yes, I reply...

But the definition of AB is that it operates each valve with a conduction angle of less than 360 degrees but more than 180. AT IDLE! It does that ALL the time. So in single ended operation, you wouldn't EVER be able to get a complete, symmetrical sine wave out of it. In push pull you can, because the other half is taking up the load when the first half starts to slack. The two signals are resolved in phase across the secondary to produce a composite that does have a complete and symmetrical waveform.

A SE amp CAN'T do that. So it would never be able to give a complete symmetrical output, no matter how small the signal is.

So there you go. You can operate SE as A or B, but not somewhere in between. You CAN in an RF circuit because you've got a resonant load (and quite often a servo system that constantly adjusts the idle bias).

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Thu, Sep 13th, 2007 @ 08:53 ( . )

This thread is getting quite disjointed, so I'm not sure where is best to place it, but here goes.

I've been reading more of the ARRL handbook. There are descriptions of amplifier class in several chapters throughout the book.
I believe they are far better and more comprehensive descriptions than that of RDH4 or anything else I have read on the internet.
I'm not going to type out a few thousand words quoted from the book because that would take me several weeks to accomplish.

But here are some key points to consider:
There is a difference between the criteria for class of service of the audio amplifier and the RF power amplifier.
"Class-AB audio amplifiers are Push-Pull"
Class-AB RF power amplifiers can be single ended, in fact most of them are.
"A voltage amplifier is always Class-A1"

I know this is probably is of little use to you all with out reading it in context, but what else can do ?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Stephen Keller (registered user: 6220 posts )
Date: Thu, Sep 13th, 2007 @ 11:09 ( . )

I think the most telling distinction between the classes in the ARRL handbooks (I have six editions from early 40's to the mid 60's) is the expressed relationship of output power and plate efficiency to grid-excitation voltage. None of the class definitions that I've found in the ARRL books make mention of the load or its type, only the idle operating point and whether at idle current flows on the resultant plate current curves. I would agree that in ARRL's view, the classes AB and B are explicity restricted to push-pull topologies at audio frequencies. I don't think that means you can't set the idle bias point at some AB-like position in an SE output stage, just that the term "class AB" doesn't apply to it by definition.

Part of the problem we have with the "class" definitions is that they were old and hoary 50 years ago. By the 50's everyone just knew what they meant. Things got further complicated with the advent of solid state, the class definitions took on new twists. For example, I've seen references transition journal articles that explain that a class B transistor circuits is a little like a class AB vacuum tube circuit, because they leave a little current flowing in the output during the negative excursion of the input signal.

Question for you Dan?

Which ARRL book are you looking at? I'm willing to scan and post relevant pages (assuming I have them) if you think they will help clarify the discussion.

Stph

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Thu, Sep 13th, 2007 @ 11:36 ( . )

Here's where I see that falling down:

If the class definition hinges upon whether current flows on the resultant plate current curves at idle, then there's a problem trying to include most of the PP colder-biased amps in that category. At idle, current is there, and the both tubes amplify 360 degrees of the signal via their plate current.

So, is that a class A amplifier? If not, why would you put that class category upon an SE amp biased, and amplifying in the same way?

Class AB amps *do* conduct for the full 360 degrees at lowish output levels. It's just that before grid conduction happens (remember when these definitions came out, there were no MOSSFETs or BJTs to think about), the plate current stops for a portion of the waveform.

So, the only reason these old guys didn't talk about class AB for SE is that they didn't think you'd want to use if for that.

The definition is based upon conduction angle before grid conduction, not the plate current curves at idle. I don't see how the idle interpretation could work for all classes. Am I just misinterpreting your explanation?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Stephen Keller (registered user: 6220 posts )
Date: Thu, Sep 13th, 2007 @ 11:57 ( . )

On 09/13/2007 @ 11:36, CarlB wrote :
Am I just misinterpreting your explanation?
--

More likely I wasn't starting out clearly. This evening, I'll scan and post the 1942 ARRL pages that define the classes, so we have some common talking points.

Stph

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Thu, Sep 13th, 2007 @ 17:12 ( . )

On 09/13/2007 @ 11:09, Stephen Keller wrote :: Question for you Dan?
:
: Which ARRL book are you looking at? I'm willing to scan and post relevant pages (assuming I have them) if you think they will help clarify the discussion.

Stephen,
I have the 1992 edition and have access to others at various libraries.
Because it is under copyright and published by a non-profit organisation I am hesitant to scan and put them up on the public internet.
But it would help to clarify a few points I think, as the RDH is rather vague in places and open to interpretation.

But the text that I'm referring to are:
1992 Ed.
p.3-17 Radio Design Technique and Language, Classes of Amplifier Service.
pp.4-23,24,25 Solid State Basics, Transistor RF Power Amplifiers.
pp.5-4,6,7 Vacuum Tube Principles, Class-A Amplifiers (through to and including) Class-C Amplifiers.
pp.15-1,2 RF Power Amplifiers, Classes of Operation.

And then further in chapter 15 there are some design examples giving.

NB. Am I quoting the page numbers correctly ? eg. p.3-17 means Chapter 3 page 17 because that is how they are numbered in the book :)

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Thu, Sep 13th, 2007 @ 19:05 ( . )

"I would agree that in ARRL's view, the classes AB and B are explicity restricted to push-pull topologies at audio frequencies. I don't think that means you can't set the idle bias point at some AB-like position in an SE output stage, just that the term "class AB" doesn't apply to it by definition."

Before I'm even looking at it, I'm pretty sure what you're seeing is an explanation of why they wouldn't use SE class AB in their idea of an audio amp.

The definition is about the conduction angle, and is orthoganal to if the amp is SE or not. The rest is their explanation of the classes applied to the application at hand. They thought some classes were useful to an application, talk about them, some weren't, so don't - or even say they aren't useful or something to that effect.

What difference is there between RF frequency concepts and Audio frequency concepts that would allow SE class AB for RF, but not for Audio? It's not the definition (i.e., conduction angle). The only thing that makes any sense is that for their idea of audio, you want to maximize the clean power for audio, whereas for RF they knew that maximizing the clean SE power wasn't the goal.

Neither is it the goal for us guitarists. And so, we should be talking about these SE amps for what they're intended and designed for -

1) Maximum clean power: bias to the plate current Vg midpoint, and talk about SE class A.

2) Maximum useful power: bias somewhat colder to accept higher B+ and screen, accept the reduced conduction angle before grid clipping, and talk about SE class AB.

Pretty straightforward ...

But you won't see them defining Class AB in terms of push-pull only. You'll just see them omit it from the audio section or say something to the effect of, "

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mcdadez (guest: search)
Date: Thu, Sep 13th, 2007 @ 19:08 ( . )

But...but....SE cant be class AB....

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 16th, 2007 @ 10:13 ( . )

Yeah, and the earth is flat. :o)

On 09/13/2007 @ 19:08, Mcdadez wrote :

But...but....SE cant be class AB....

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Randall Aiken (guest: search)
Date: Thu, Sep 13th, 2007 @ 20:12 ( . )

On 09/13/2007 @ 05:27, Zoe N Iain wrote :
OK, I'll give it a go.
:
: Firstly, just to quote someone brighter than either of us...
:
: [link]
:
: Having absorbed that (and noted that Mr. Aiken ALSO says that a SE amp can only operate in Class A for audio)



I'm sorry, but I have to disagree with myself here. :)

Contrary to what that paper says, what I intended to say was that SE audio amplifiers are *generally* designed to run at class A bias points to attain maximum undistorted output. I didn't intend to say that you *can't* design an SE output stage for another class of operation, just that you wouldn't normally do this because you would limit the amount of output power you could attain before distortion.

SE class AB is very real, indeed, and as others have noted, is used in RF output stages all the time (well, not that often, because class C is more efficient). You could very well design your audio output stage as SE class AB if you wanted, but as mentioned previously, you'd hit clipping a lot sooner.

There is nothing in the class definitions that says that class AB has to be push-pull, and the "changing load line in AB operation" has nothing to do with the definition, nor is it defined as "transitioning from class A to class B", it is simply a matter of the conduction angle. A class A stage is operated around the midpoint between cutoff and saturation. A class B stage is operated at cutoff, a class AB stage is operated somewhere in between cutoff and midpoint, and a class C stage is biased beyond cutoff.

If you bias an SE amp stage asymmetrically enough, it will be considered operating in class AB. This is not limited to output stages, either - go look closely at the Soldano-type "cold clipper stage" using a 39K cathode resistor. That's SE class AB operation in a preamp stage.


Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Thu, Sep 13th, 2007 @ 20:29 ( . )

Proving once and for all, that you really aren't dead yet (see my post). I hope we can all agree upon that.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Randall Aiken (guest: search)
Date: Thu, Sep 13th, 2007 @ 22:18 ( . )

On 09/13/2007 @ 20:29, CarlB wrote :
Proving once and for all, that you really aren't dead yet


I hope not! Although I feel that way sometimes...

Ra

-- REPLY: [With No Quote] --- [With Quoted Text]

'Been doing some reading, I have to agree with CarlB et al.'
Author:Stephen Keller (registered user: 6220 posts )
Date: Thu, Sep 13th, 2007 @ 20:38 ( . )

Okay,

So I've been doing some reading, quite a lot actually. Perhaps the clearest statement of definitions for this is found in William Orr's The Radio Handbook. It is located on Pete Millett's site here:

[link]

(Look at pages 107 and 108.)

Orr defines the classes on the basis of conduction angle at the given operating point with respect to the maximum grid excitation voltage. That means that an SE class AB is quite possible. His definitions are clear, and they don't muddy the water with matters of where the different operating biases are best applied.

Given Orr's definitions, then Carl's outline of how to determine the class of an amplifier in an earlier thread are reasonable.

Stph


-- REPLY: [With No Quote] --- [With Quoted Text]

'Been doing some reading, I have to agree with CarlB et al.'
Author:CarlB (guest: search)
Date: Thu, Sep 13th, 2007 @ 20:49 ( . )

Note, my observations need a bit of tweaking. For A2, grid conduction no longer produces clipping. So instead, you go out toward saturation.

If the tube hits saturation first or saturation and cutoff at about the same time, it's Class A2.

If the tube hits cutoff before saturation, it's Class AB2.

Note, Class A2 needs a big honking plate, because now you need to bias half way between saturation and cutoff, not just Vg=0 and cutoff ...

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Thu, Sep 13th, 2007 @ 23:02 ( . )

"SE class AB is very real, indeed, and as others have noted, is used in RF output stages all the time (well, not that often, because class C is more efficient). You could very well design your audio output stage as SE class AB if you wanted, but as mentioned previously, you'd hit clipping a lot sooner. "

Well, actually you reach cutoff sooner. But you do get to increase the output power achieved before clipping (given the same plate dissipation at idle). It turns out these SE class AB amps sound pretty clean at the same apparent output volume to our ears. The neat thing is, they then go on to produce more nearly clean output volume, and finally, more distorted output volume.

Heck of a trade, I'd say. Flush the THD measures those are for all the misguided Hi-Fi guys. Let your ears be the guide, and an SE class AB stage beats an SE class A amp. Well, at least to my ears.

By the way, the old Fender Champs were SE class AB. They had a 6V6 with well beyond 250V plate/screen (about the max you can on a 6V6 for class A and not over-dissipate), and so they had to be lowering the bias to keep the plate from over dissipating.

Phil rediscovered this for his P1eX output stage. Pretty darned good output stage, I think.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Thu, Sep 13th, 2007 @ 20:19 ( . )

"Firstly, just to quote someone brighter than either of us...
:
: [link]
:
: Having absorbed that (and noted that Mr. Aiken ALSO says that a SE amp can only operate in Class A for audio), "

Hey, we could just ask him why hes says that. Unlike that ARRL guys, he's not dead ...

"my reasoning that the change of current is a determinant AND that the definition is only designed to be applied at the instant a tiny signal takes the PA out of it's quiescent condition is this;
:
: In a Class A amp, the change of current is only instantaneous. Obviously, the current is changing because a tube is turning change of voltage at the grid into change of current at the anode, in a ratio defined by the mu of the valve. But the average current, observed over a few complete cycles, doesn't change. You can prove it by building one and trying to get the PSU to sag. Can't be done, unless you cheat and do it synthetically, I'm sure you agree. A pure class A amp will amplify 360 degrees of the signal until it runs out of headroom, then it clips the top, bottom or both. "

Agreed ...

"In a Class B amp, the idle current is near or at zero when no input signal is present. Only when you give it something to amplify does it demand current from the power supply. In the absence of a tuned circuit to permit flywheel effect to fill in the other half of the waveform, a pure Class B amp can never amplify more than 180 degrees of the input signal. And it doesn't matter how small the signal is."

Practically, the current must be above zero toward achieving the 180 degree conduction.

"A Class AB amp is obviously somewhere between the two. It's biased above zero, so there's always some idle current. But, in a PP amp the current demand changes when you ask it to do some work. Because it IS biased partway to class B. It's not idling at zero so the current flow isn't 'lots,none,lots,none..' but it does require that increased current is supplied as input signal level is increased."

Agreed ...

"In a SE amp, the average current demand doesn't change, unless it's biased into class B at idle. In which case it can NEVER give you a clean output, no matter how small the signal is. Not true for an eX. Perfectly capable of a clean output."

Disagree. First, the average current demand in a P1eX output stage does change from no signal to max conditions. It has to, because it's biased on the cold side. For large input signals, the top half of the output signal starts having more area (volt-amps) in it than the bottom half, and so the average current starts to climb. You see that on the bypassed cathode resistor. Second, you note that for this colder-than-midpoint biasing, small signals are amplified cleanly, whereas large signals go into cutoff first (before grid conduction and/or saturation). That's what class AB is about, and the P1eX is class AB.

"But, you are about to argue, I'm not saying an eX is biased in Class B, I'm saying it's in AB. Yes, I reply...

OK ...

"But the definition of AB is that it operates each valve with a conduction angle of less than 360 degrees but more than 180. AT IDLE! It does that ALL the time. "

Disagree. The definition of class AB is that the conduction angle is less than 360 degrees. This is measured/observed at just before you reach clipping (either grid clipping or saturation).

"So in single ended operation, you wouldn't EVER be able to get a complete, symmetrical sine wave out of it. "

Disagree. If you're biased above class B, then there will be smallish signals that will amplify cleanly just fine.

"In push pull you can, because the other half is taking up the load when the first half starts to slack. The two signals are resolved in phase across the secondary to produce a composite that does have a complete and symmetrical waveform."

Agree, kind of. Note that in push-pull class AB, both plates are conduction 100% for quite a range of smallish input signals ... Same as we see for SE class AB. But they're haven't reached either saturation/grid-clipping or cutoff, which is where the class of the amp is observed at.

"A SE amp CAN'T do that. So it would never be able to give a complete symmetrical output, no matter how small the signal is."

Disagree ...

"So there you go. You can operate SE as A or B, but not somewhere in between. "

Disagree. And so did the U.S. Naval electrical engineering training series (NEETS), which is where the following training is taken from:

[link]

"You CAN in an RF circuit because you've got a resonant load (and quite often a servo system that constantly adjusts the idle bias)."

Agree. Note you can also have SE class AB without the resonant load, see the above.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Fri, Sep 14th, 2007 @ 06:30 ( . )

I think there has been a change in the general ionterpretation of the class names. The Name 'AB' originally grew from the change in load witnessed in a PP valve amp. (Otherwise they'd have made class B, "Class C", and the one between A and C would be "Class B".)
However, in modern SS complementary designs there is no change in load, even when biased to a class AB operating point. In that sense, the modern interpretation of "AB" can be argued for SE. I just don't like it. Because for valve design, which is what I'M interested in, it detracts from the original, intuitive definitions.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Fri, Sep 14th, 2007 @ 11:32 ( . )

Where do you get the history about the origin of the term from? Reading as far back as far as 1932 I can't find anything that suggests that. I would be genuinely interested to know your sources. AFAIK Class AB was simply called that because it lay between Class A and Class B, which presumably were the first two modes that got defined.

While the idea of the amp that changes load impedance and has two load lines is a neat concept (which I quite like too), I don't see it as being any more intuitive than the idea of being between Class A and Class B. Intuitive seems to be a very subjective thing, and what's intuitive to me isn't intuitive to you, and vice versa.

On 09/14/2007 @ 06:30, Merlin wrote :
:
I think there has been a change in the general interpretation of the class names. The Name 'AB' originally grew from the change in load witnessed in a PP valve amp. (Otherwise they'd have made class B, "Class C", and the one between A and C would be "Class B".)
:
: However, in modern SS complementary designs there is no change in load, even when biased to a class AB operating point. In that sense, the modern interpretation of "AB" can be argued for SE. I just don't like it. Because for valve design, which is what I'M interested in, it detracts from the original, intuitive definitions.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Thu, Sep 13th, 2007 @ 15:11 ( . )

Heh, but like a moth drawn to the flame...! ;o) I honestly don't think the discussion is entirely pointless. I agree that at the end of the day all we want is great guitar tone, and who the hell then cares what Class the amplifier is running in. Some people may be horrified to at the idea that I also like great guitar solos, and to be honest I don't really care much about what scales they're using either. :o)

Personally I love it when we cover new ground on AX84. We've had such things as DC drive between gain stages, doing all sorts of funky things with pentodes and parallel triodes, and now we have the opportunity to re-examine the whole matter of operating Classes in guitar amps. So I really welcome your involvement in the discussion. I hope that at the outcome of this discussion, all of us will have gained a better understanding of how guitar power amps really work, and hopefully more usefully definitions of operating classes.

Looking at what you've said, about SE amplifiers, I think we need to remember that Class AB PP amplifiers also operate like Class A amplifiers at low signal levels, just like SE ones do. So you can't really build a valid class definition from how an amp behaves at idle or low signal levels.

On 09/12/2007 @ 17:16, Zoe N Iain wrote :
:
I swore I wouldn't re-enter this pointless discussion, being as I actually agree with Merlin and screw what the textbooks say. But..
:
: The only useful way you can assess the class of an amplifier is to define it at idle. So a SE amp is CLASS A, because you CAN'T bias it any other way at idle, unless you set it up so that it only conducts for 180 degrees of the signal, NO MATTER HOW SMALL THE SIGNAL HAPPENS TO BE, then it's class B. BUT IF you bias it in class B, then it is no longer capable of clean amplification AT ALL and it IS OPERATING ON A DIFFERENT LOADLINE, AS WELL!
:
: A SE amp can't swing between one load condition and another. There are no additive or cancelling currents in the OT core to swing the loadline from one to the other, like you get with an AB push-pull amp. It can't change class of operation! IT IS PHYSICALLY and ELECTRONICALLY IMPOSSIBLE!
:
: The reason you can ONLY define it at idle is that there is NO loudspeaker in existence that presents a consistent load to the output device through it's entire frequency range. If the reflected impedance changes, the point where the device is currently hanging out in it's class of operation chnages. If it changes ENOUGH, and let's be honest - cranking the shit out of it is going to change stuff - then it is no longer fixed at an operating point. So even saying that something is operating in Class A, B, R or Z is a stupid thing to say with regard to an amplifier that is DESIGNED to spend most of it's time on a totally non-linear portion of it's output device characteristic curves.
:
: The closest we can manage is to say that it's Class Whatever at idle and clips at a given voltage. So you can use the Hot, Mid, Cold terms as described above.
:
: I don't give a shit what the textbooks say, they were written by people who would never have IMAGINED that someone would want to drive such a big signal into a PA as to make it clip, at audio anyway. Radio is totally different. There were no JCM800s when those definitions were written and THAT is why they specified conduction angle as the determinant.
:
: A single ended amplifier can be operated, for useful audio purposes, in only one way. And it's class A at idle.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mcdadez (guest: search)
Date: Thu, Sep 13th, 2007 @ 17:08 ( . )

Well my thoughts are..I agree with Zoe@Ian,Merlin etc.But like Zaphod i do agree that Guitar amps are a special breed in the aduio world where what type of class,mojo capacitors etc mean nothing to tone unlike the Hifi world.Its great reading for a noob like me..but IMO it doesnt matter whos right or wrong in this type of discussion its just *good* old fashioned debating.Where one person puts their point across and the other theirs.I still think that *some* people should admit when their wrong in a debate and stop all this clutching at straws to look intelligent and correct...then again im a noob and only have RDH4 and google to help me understand this debate.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Thu, Sep 13th, 2007 @ 17:22 ( . )

It's one of those debates that can really ONLY be settled by building a PA setup such that if it is possible for it to operate in AB it will if you drive it a bit. And hook up a sh*tload of test gear to measure everything that happens when you do.

I guess you could do it relatively simply by sticking a current meter in series with the plate supply and cranking on it. If there's an appreciable change in current, it's in class AB. If not, it's class A. I can't decide whether a regular multimeter could be used for the job, though. In theory there would be two current mechanisms in play... the instantaneous currents caused by the signal itself and the average current drawn from the PSU. I'm not sure which one a meter would try to measure, or whether you'd get screwy results as it tried to measure both.

If you set it to measure DC current, which is presumably the average current delivered from the PSU, it will still be seeing the AC from the signal. And vice versa.

Anyone know whether you can do that with a normal meter, or knows of a way to do it that would let you use a standard piece of test gear? My brain mustn't be working properly today, cos I can't suss it out.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Thu, Sep 13th, 2007 @ 17:27 ( . )

On 09/13/2007 @ 17:22, Zoe N Iain wrote :
: Anyone know whether you can do that with a normal meter?


Makes sense to me. Just measure the anode current and turn up the signal level, and watch for a point where it increases from an average. Sound kosher?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Thu, Sep 13th, 2007 @ 17:49 ( . )

On 09/13/2007 @ 17:22, Zoe N Iain wrote :: If you set it to measure DC current, which is presumably the average current delivered from the PSU, it will still be seeing the AC from the signal. And vice versa.
:
: Anyone know whether you can do that with a normal meter, or knows of a way to do it that would let you use a standard piece of test gear? My brain mustn't be working properly today, cos I can't suss it out.

The average DC flowing from the supply rail will increase as the AC signal is increased. This is because of the inefficiencies of the valve, the power is lost in heat.

Back in the ol' days... when one went up to a shop counter and asked for a pure class-A push-pull amplifier . The sales assistant's suggestive selling line would be "Would you like a power station with that sir ?" .

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Thu, Sep 13th, 2007 @ 18:48 ( . )

Here's how I'd do it:

SE: sneak a 1 ohm resistor between the cathode and the cathode resistor/bypass-cap pair. Use a scope, and put the two probes across the 1 ohm resistor in differential mode. Scope the plate current directly, well, with the screens there too. You could put the 1 ohm and probes up at the plate, but watch the DC and AC maximums for the probes and the scope.

PP: same thing as above. If on a cathode, just make sure the 1 ohm resistor is not shared by both cathodes, i.e., put it at just one or the other of the cathodes before cathode resistor and bypass cap. Or, if it's negative grid biased, just sneak the 1 ohm between one of the cathodes and ground.

What you'll see:

An Idle current of something - say 75mA

Introduce a small signal, watch the current follow the signal.

Introduce a large enough signal and see one of three things -
a) Grid clipping, no cutoff.
b) Cutoff, no grid clipping
c) Cutoff and grid clipping at the same time.


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Thu, Sep 13th, 2007 @ 17:33 ( . )

On 09/13/2007 @ 15:11, Zaphod wrote :: Looking at what you've said, about SE amplifiers, I think we need to remember that Class AB PP amplifiers also operate like Class A amplifiers at low signal levels, just like SE ones do. So you can't really build a valid class definition from how an amp behaves at idle or low signal levels.
That is part of the Class-AB criteria as stated in ARRL H. .

Class-B conduction angle is 180º, not more, not less, but a few º here and there is tolerable. A class-B device also behaves like a rectifier because half the wave is missing.
Hence if we place two class-B devices together driven 180º out of phase, we have a class-B push-pull amplifier.
Then it should follow that definition for class-AB is somewhere between class-A & class-B.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Tue, Sep 11th, 2007 @ 22:28 ( . )

On 09/11/2007 @ 13:02, Zaphod wrote :

:
: The New Penguin Dictionary of Electronics by E.C. Young 1988
: A linear amplifier in which the output current flows for more than half but less than the whole of the input cycle, ie the angle of the flow is between pi and 2pi. At low input-signal levels class AB amplifiers tend to operate as class A and at high input-signal levels as class B amplifiers.

That would agree with my take on the definition. An SE amp, cool biased, never operates like a class B amp at high signal levels. It acts like a clipping Class A amp (which is what it is).


: Modern Electronic Circuit Design by D.J. Comer 1980
: It is sometimes advantageous to cause each stage to conduct more than one-half cycle, but less than the full cycle. This mode of operation is called Class-AB.
Whereas that definition doesn't seem to refer to any particular topolgy- an over simplification perhaps? Bare in mind that many modern textbooks are quite poorly written, and not without spurious or misleading information (although not always INCORRECT informaton).

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Tue, Sep 11th, 2007 @ 23:21 ( . )

What's needed is the strict class AB definition ...

I think what you're seeing are textbooks that describing the push-pull instance for class AB.

The question is: is there an accepted definition by all engineers? Is that definition based exclusivily pon conduction phase, or, does it also include push-pull and/or load-lines discriminators?

If the definition does not include the latter discriminators, then the practice, and the texts, and the patents that refer to Single-Ended Class AB amplifiers are correct ... If not, then they are wrong.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Tue, Sep 11th, 2007 @ 23:48 ( . )

Every definition I've ever seen simply discussed conduction angle.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Wed, Sep 12th, 2007 @ 00:26 ( . )

On 09/11/2007 @ 23:21, CarlB wrote : I think what you're seeing are textbooks that describing the push-pull instance for class AB.

Reading through the ARRL handbook I can only find references to Class-AB, AB1 and AB2 "Push-Pull" amplifiers.
No instances of SE Class-AB, however there is a SE Class-C amplifier described.
And it seems that conduction angle is the main criteria for class designation. The word "linear" being another.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Wed, Sep 12th, 2007 @ 00:49 ( . )

But, what definitions are given in the ARRL?

The examples may only be of push-pull, but from those you can't conclude that the definitions must exclude SE - unless this is stated somewhere in the definition, or the exclusion can be concluded from the definition. But such an exclusion can't be done simply from the conduction angle discriminators ...


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Wed, Sep 12th, 2007 @ 01:23 ( . )

On 09/12/2007 @ 00:49, CarlB wrote :But, what definitions are given in the ARRL?
:
: The examples may only be of push-pull, but from those you can't conclude that the definitions must exclude SE - unless this is stated somewhere in the definition, or the exclusion can be concluded from the definition. But such an exclusion can't be done simply from the conduction angle discriminators ...
It first describes Class-A and then Class-B, in the way we all understand it here, and then on to, "if two Class-B devices are combined and driven 180º out of phase in a Push-Pull circuit".
But, no it does not exclude Single-ended devices (I should have been more clear about that).
The text is more orientated towards "Power output","Efficiency" and "linearity", which can be obtained once the Class of operation is known.
eg. A class-A amplifier (either SE or P-P) can only have a maximum theoretical efficiency of only 50 percent, with 25 percent being a more realistic figure.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Balijukka (guest: search)
Date: Fri, Sep 14th, 2007 @ 09:16 ( . )

Oh yes it can. We are talking about AMPLIFIERS.
In theory an amplifier just amplifies, it is not a clipper or DSP. The output should be just like the input.
In a p-p amp it is possible to have identical input and output with one of the sides not conducting.
In an SE amp it is impossible. So an SE amplifer can only be class A. It has nothing to do how hot or cold it is biased. At maximum undistorted, unclipped signal the current flows 360 degrees. If it does not, the amp clips. Simple and easy as that.

jukka

On 09/12/2007 @ 00:49, CarlB wrote :
But, what definitions are given in the ARRL?
:
: The examples may only be of push-pull, but from those you can't conclude that the definitions must exclude SE - unless this is stated somewhere in the definition, or the exclusion can be concluded from the definition. But such an exclusion can't be done simply from the conduction angle discriminators ...
:
:
--


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Fri, Sep 14th, 2007 @ 10:13 ( . )

On 09/14/2007 @ 09:16, Balijukka wrote :
In a p-p amp it is possible to have identical input and output with one of the sides not conducting.
: In an SE amp it is impossible. So an SE amplifer can only be class A.

Precisely. *applauds*

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Evan (guest: search)
Date: Fri, Sep 14th, 2007 @ 11:43 ( . )

On 09/14/2007 @ 10:13, Merlin wrote :
On 09/14/2007 @ 09:16, Balijukka wrote :
: In a p-p amp it is possible to have identical input and output with one of the sides not conducting.
: : In an SE amp it is impossible. So an SE amplifer can only be class A.

: Precisely. *applauds*
--



Man, I was *ridiculed* for making this point some eons ago in another thread. Where were you guys :-)?

At this point, I really don't give a fig for the fight over definitions. I'd happily use Carl's definitions -- they are clear and well reasoned. Doesn't change physical reality a bit.


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Fri, Sep 14th, 2007 @ 12:18 ( . )

No it doesn't. It's an argument over pedantry, really.

What arrives at the speaker looks as if it came out of something that's biased AB. We just don't agree that the stage is technically operating IN AB. Doesn't matter for anything. What we all see on a scope is still what we'd expect to see, no matter which side of this debate we're respectively on.

Hopefully, there have been people reading the thread that have taken the trouble to look up the reasons for our disagreement over terms for themselves and formed their own opinion about which side they come down on, learning something about the techniques involved in the process. If so, then it's been worth doing for the educational value.

Which 'world view' you hold on the matter actually matters very little. Either viewpoint will let you come up with a SE PA that rounds off one peak before the other, which is presumably what you're trying to achieve if you're walking down this road in the first place. Whether the truth is that the PA is in Class A but clipping, or in AB and behaving normally, the picture on the scope will still be the same.

I did, I believe, say 'Who cares?' right at the start. ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Evan (guest: search)
Date: Fri, Sep 14th, 2007 @ 14:43 ( . )

Trying to remember the name of Douglass Adams' something or other "Institute for Slowly and Painfully Working Out the Very Obvious"....

I think it's more "meta-pedantry", myself :-).

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Fri, Sep 14th, 2007 @ 20:21 ( . )

Well at this stage, having read all the arguments and the various texts, I don't know what to believe.
Too much of the text went over my head, through trying to absorb so much in such a short period of time.

However, I will concede that a single-ended amplifier can operate in Class-AB for RF service, though it is possible for audio service we could not label it as such if we were to get any sort of respectable (musical pleasing) sound out, distorted as it may be.
I still believe we need another term to better describe what we (guitar amp builders) are trying to achieve by over-driving a Single-Ended amplifier.

Personally I prefer the sound of a power valve that is biased very hot and driven into saturation.

This thread has been very educational and enjoyable for me and hopefully for everyone else as well.
If any of my posts seemed like a personal attack on anyone, I apologise for that as I didn't mean any of it to come across that way :)

We should have lots more of these in depth discussions, hey! that's the idea of this forum...

Cheers!
Danny

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mark L (registered user: 1458 posts )
Date: Fri, Sep 14th, 2007 @ 23:10 ( . )

On 09/14/2007 @ 20:21, Dan N wrote :
: However, I will concede that a single-ended amplifier can operate in Class-AB for RF service, though it is possible for audio service we could not label it as such if we were to get any sort of respectable (musical pleasing) sound out, distorted as it may be.


Ah, but that's where the problem is!
Many seem distracted by the application that amplifier is used for, rather than looking at the bias of the amp.
As far as the amp is concerned it doesn't care whether it has audio or radio frequency on the input, it's still a SE class AB amp (or whatever it is biased as).

Just because a SE class AB is not suitable for faithful audio reproduction doesn't mean it ceases to be class AB.

It's like saying if a 4-cylinder Mazda made for city driving is taken off-road down a rough dirt track, it must now be an off-road vehicle because that's what it's being use for at the in this situation. Same with a class AB SE amp, if your using it for audio doesn't mean it must now be class A.


: We should have lots more of these in depth discussions, hey! that's the idea of this forum...
:
: Cheers!
: Danny
--



Exactly!
Some see it as conflict for some reason and take it personal. I don't know why.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Right on the nail'
Author:Zaphod (guest: search)
Date: Sat, Sep 15th, 2007 @ 12:06 ( . )

Couldn't agree more. On all counts. There's no need to see any of this as personal, as it's a purely technical discussion, and a pretty fascinating one at that.

From how I see it in purely EE terms, a stage is in Class AB if it's designed, with it's voltages, load impedance and bias to operate in cutoff for some part of it's cycle which is less than 180 degrees.

Whether it's RF or audio, or what degree of distortion exists at the output, is not relevant to the definition, as that is a matter of the particular application it's being used for.

On 09/14/2007 @ 23:10, Mark L wrote :
:
: Ah, but that's where the problem is!
: Many seem distracted by the application that amplifier is used for, rather than looking at the bias of the amp.
: As far as the amp is concerned it doesn't care whether it has audio or radio frequency on the input, it's still a SE class AB amp (or whatever it is biased as).
:
: Just because a SE class AB is not suitable for faithful audio reproduction doesn't mean it ceases to be class AB.
:
: It's like saying if a 4-cylinder Mazda made for city driving is taken off-road down a rough dirt track, it must now be an off-road vehicle because that's what it's being use for at the in this situation. Same with a class AB SE amp, if your using it for audio doesn't mean it must now be class A.
:
: : We should have lots more of these in depth discussions, hey! that's the idea of this forum...
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Fri, Sep 14th, 2007 @ 23:15 ( . )

"Whether the truth is that the PA is in Class A but clipping, or in AB and behaving normally ..."

I doesn't sound like you quite have what some of us are saying. The quote would be:

"Whether the truth is that the SE stage is in class A, but bottoming out, or is in class AB and is bottoming out ..."

This is at the core of the disagreement. Some say an SE amp is class A regardless if it reaches cutoff before it reaches grid-clipping (i.e., A1) or saturation (i.e., A2); while others are saying that if it reaches cutoff before it does reach grid clipping (i.e., AB1) or saturation (i.e., AB2), then by definition it has to be class AB - regardless of if it is SE or PP.

The 'SE can be class A, B, C, ... but not AB' folks have cited a definition that is based upon a small-signal examination of conduction capability. (I have never seen this in texts, and so I can't adhere to that.) I have not seen the approach account for why an RF SE class AB amplifiers can exist while one allegedly can not exist for audio.

The 'SE can be class A, AB, B, C, ... ' folks cite a definition based upon the largest clean signal the stage can amplify. This is is consistent with the texts that I have read.

While many authors have clouded the issue by implying or stating that an audio SE amp can't be AB, they are referring to what makes sense for the goal they have assumed: maximum undistorted output from a given SE plate dissipation.

Our goal is different: maximum cleanish output and more distorted output too, from a given SE plate dissipation. With that goal in mind, we need to (and do) design SE amps that are purposefully biased class AB: they can amplify small signals faithfully, large signals clean-sounding, and large signals more loudly than class A biased SE amps (with distortion on both, of course).

Whereas SE class A amp has the higher maximum power before cutoff, our ears hear the SE class AB amp as being the same loudness - if a bit more full sounding - due to the increased bottom rounding / 2nd-order harmonic stuff.

And, the class AB is definitely louder than the class A when really cranked.

Let's give this SE class AB some credit as being pretty darned good for guitar, and let the hi-fi guys keep their SE class A. SE class A sounds duller and less interesting to my ears, "clean" or distorted.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sat, Sep 15th, 2007 @ 12:09 ( . )

Right on, Carl!

On 09/14/2007 @ 23:15, CarlB wrote :
:
: Let's give this SE class AB some credit as being pretty darned good for guitar, and let the hi-fi guys keep their SE class A. SE class A sounds duller and less interesting to my ears, "clean" or distorted.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Fri, Sep 14th, 2007 @ 12:11 ( . )

Amplifiers like that only exist in theories and dreams. The perfect amplifier doesn't exist. Even the best hi-fi amps color and distort a little. And nearly all guitar amps distort quite a lot, whether they run in Class AB or not. So by your logic are you saying that no guitar amp can operate in Class AB, or simply that they're not amplifiers? I don't wish to appear disrespectful, but it all sounds pretty silly to me.

And BTW, now that Randall Aiken has clearly stated his view that there's no reason an SE amp can't be Class AB, do you still want to buy his book? :o)

On 09/14/2007 @ 09:16, Balijukka wrote :

Oh yes it can. We are talking about AMPLIFIERS.
: In theory an amplifier just amplifies, it is not a clipper or DSP. The output should be just like the input.
: In a p-p amp it is possible to have identical input and output with one of the sides not conducting.
: In an SE amp it is impossible. So an SE amplifer can only be class A. It has nothing to do how hot or cold it is biased. At maximum undistorted, unclipped signal the current flows 360 degrees. If it does not, the amp clips. Simple and easy as that.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Evan (guest: search)
Date: Fri, Sep 14th, 2007 @ 12:17 ( . )

On 09/14/2007 @ 12:11, Zaphod wrote :
Amplifiers like that only exist in theories and dreams. The perfect amplifier doesn't exist. Even the best hi-fi amps color and distort a little. And nearly all guitar amps distort quite a lot, whether they run in Class AB or not. So by your logic are you saying that no guitar amp can operate in Class AB, or simply that they're not amplifiers? I don't wish to appear disrespectful, but it all sounds pretty silly to me.

--



That's a strawman.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Fri, Sep 14th, 2007 @ 12:31 ( . )

The point is, the class definitions were formulated to apply to a perfect amplifier (which I totally agree doesn't exist and would probably bore the sh*t out of you to listen to, if it did). The RDH4 explanation certainly says that they are intended to apply to a linear amplifier, somewhere in the first paragraph or two of that section. (I don't have a copy to hand to quote from. Mine's on CD and Audiophool John has got it at present. Anyone want to look it up for me?)

So the crux of our point is that if the amplifier it operating clean, cos the signal is small, then it is currently running in class A because it is amplifying 360 degrees of the signal.

A PP AB amp can still carry on giving you a clean (linear) output even after both halves are limiting the peak on one side of the waveform. A SE 'AB' (audio) amp cannot. And that's our point. There's a difference between the two things.

I contend that it is 'Class A But Clipping', you reckon it's Class AB. The scope picture will be the same, whatever, so it doesn't really matter.


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mcdadez (guest: search)
Date: Fri, Sep 14th, 2007 @ 13:15 ( . )

Hmm im searching for what you mean,havent found it yet.Although on Chapter 13 Audio Frequency Power Amplifiers.The first thing it discusses is the classes..ill see if i can upload it.But in a nutshell its says

Class A Operation is the normal condition of operation for a single valve and indicates the plate current is not cut off for any portion of the cycle

Class AB Operation indicates overbiased conditions and is used only in PP to balance out even harmonics

Does this help the argument probably not..ill see if i can find the comment you were talking about.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Fri, Sep 14th, 2007 @ 13:22 ( . )

It really makes no difference either way...

On 09/14/2007 @ 13:15, Mcdadez wrote :
:
: Does this help the argument probably not..ill see if i can find the comment you were talking about.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mcdadez (guest: search)
Date: Fri, Sep 14th, 2007 @ 13:34 ( . )

Gave up already anyway lol trying to look and click through a pdf is just a PITA...especially one as big as RDH4 :)

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Fri, Sep 14th, 2007 @ 22:42 ( . )

Yes Jukka we are talking about amplifiers.


The definition for a class A amp is that the output conduction current conducts for all 360 degrees at the maximum signal the amp can produce cleanly. Class A means you'll hit grid clipping (class A1) or saturation (class A2) at the same time or before you hit cutoff.

The definition of a class AB amp is that the output is conducting for more than half of the waveform, but less than all of the waveform, at the maximum signal the amp can produce cleanly. All that class AB means is that the stage reaches cutoff on the low side of a large enough signal before it reaches grid-clipping (class AB1) or saturation (class AB2).

As such, any single triode or pentode can be biased in class AB1 or AB2. As Randall notes, it can be a preamp stage (such as the famous cold-biased Soldano preamp stage) or a power amp stage (such as the 6V6 Champ, or the EL34 P1eX or Raffler).

There is no reason that a single triode or pentode can not be biased into class AB1 or AB2.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Balijukka (guest: search)
Date: Sat, Sep 15th, 2007 @ 00:48 ( . )

But these two claims of yours are contradictory.

If a SE stage hits grid clipping or saturation before cut-off clipping, it still clips, which is not any more a clean output. As long as the output is clean, it's a class A, and after that is clipped class A.

You can only have a clean output with a tube in a cut-off (or saturated)if there is a fellow tube that takes care of the job.

So class AB and B can only be in a push-pull, (or maybe cleverly actively biased parallel SE, but I doubt that.:) )

jukka


On 09/14/2007 @ 22:42, CarlB wrote :

: The definition for a class A amp is that the output conduction current conducts for all 360 degrees at the maximum signal the amp can produce cleanly. Class A means you'll hit grid clipping (class A1) or saturation (class A2) at the same time or before you hit cutoff.
:
: The definition of a class AB amp is that the output is conducting for more than half of the waveform, but less than all of the waveform, at the maximum signal the amp can produce cleanly. All that class AB means is that the stage reaches cutoff on the low side of a large enough signal before it reaches grid-clipping (class AB1) or saturation (class AB2).
:
: As such, any single triode or pentode can be biased in class AB1 or AB2. As Randall notes, it can be a preamp stage (such as the famous cold-biased Soldano preamp stage) or a power amp stage (such as the 6V6 Champ, or the EL34 P1eX or Raffler).
:
: There is no reason that a single triode or pentode can not be biased into class AB1 or AB2.
:
--


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Sat, Sep 15th, 2007 @ 04:30 ( . )

On 09/15/2007 @ 00:48, Balijukka wrote :
But these two claims of yours are contradictory.
:
On 09/14/2007 @ 22:42, CarlB wrote :
:
: : The definition for a class A amp is that the output conduction current conducts for all 360 degrees at the maximum signal the amp can produce cleanly. Class A means you'll hit grid clipping (class A1) or saturation (class A2) at the same time or before you hit cutoff.



You're right Balijukka, it is contradictory, the definitions aren't based on how the amp clips.
Carl, where does it say that for Class A it must hit grid clipping at the same time as, or before, cut-off? Why can a Class A amp not hit cut-off first?

Of course, in SE, it CAN hit cut off first- even though it's Class A (cool class A). In a PP amp, although one valve may hit cut-off first, the ultimate output signal does NOT appear clipped because there is a partner valve to fill in the spaces. THAT is why CLass AB is so special, and not applicable to SE stages.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sat, Sep 15th, 2007 @ 09:00 ( . )

I think I need to re-word that a bit.

I was taking take class A2 and class AB2 into account, but my wording could be read two ways.

The point is that for class A1 or class AB1, you need to look only out to grid-clipping.

For class A2 or AB2, the amp doesn't clip at grid-conduction. You have a low impedance driver that muscles the grid past its conduction point without clipping. So, the stage must now be looked at all the way out to saturation, where the all the plate curves run together at the left side of the graph.

When I said, "Class A means you'll hit grid clipping (class A1) or saturation (class A2) at the same time or before you hit cutoff," I really should modify that to avoid ambiguity:

Class A means you'll hit either grid clipping (class A1) or saturation (class A2) before you hit cutoff or at the same you hit cutoff.

I hope that says things better.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Zaphod (guest: search)
Date: Sat, Sep 15th, 2007 @ 11:48 ( . )

This is pretty much how I see it, but I get the feeling that there may be confusion on the part of some about what the difference is between grid clipping and cut-off, and maybe also saturation. It might be a good idea to briefly spell them out.

On 09/15/2007 @ 09:00, CarlB wrote :
:
: The point is that for class A1 or class AB1, you need to look only out to grid-clipping.
:
: For class A2 or AB2, the amp doesn't clip at grid-conduction. You have a low impedance driver that muscles the grid past its conduction point without clipping. So, the stage must now be looked at all the way out to saturation, where the all the plate curves run together at the left side of the graph.
:
: When I said, "Class A means you'll hit grid clipping (class A1) or saturation (class A2) at the same time or before you hit cutoff," I really should modify that to avoid ambiguity:
:
: Class A means you'll hit either grid clipping (class A1) or saturation (class A2) before you hit cutoff or at the same you hit cutoff.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:CarlB (guest: search)
Date: Sat, Sep 15th, 2007 @ 14:16 ( . )

I'll give it a go:

Grid-clipping happens when the grid starts to conduct. For some portion of the input waveform, the grid becomes positive compared to the cathode. Since the grid structure to the cathode structure forms a diode, when forward biased current flows. The grid is low impedance during grid conduction. For grid-clipping, the driving source can not supply the current necessary to keep the conducting grid from flattening a portion of the signal.

If the amplifying stage's driving source can provide sufficient current to keep the grid signal un-clipped during grid conduction, then the tops of the grid signal continue positive of Vgk = 0. The amplifying stage's current continues upwards for the positive portions of the waveform, and grid-clipping is no longer an issue.

If the grid signal is driven high enough beyond grid conduction, there comes a point at which the amplifying plate can no longer supply the current necessary to faithfully follow the signal at the grid. The plate's current sourcing capacity has saturated to its maximum and the stage is saturating, it flattens the waveform for those portions of the waveform that it can no longer supply more current.

Cutoff occurs on the opposite side of the output waveform. The plate current decreases on the bottom side of the output signal. There is a limit to how much the plate current can decrease from the idle conditions - the plate can not supply anything less than zero amperes. So, if the input signal becomes large enough, there becomes a portion of the output signal where the plate current drops to zero, and the stage is said to be in cutoff.

So, back to the class definitions.

If the stage in question hits grid-clipping at or before cutoff, then the stage is biased class A1.

If the stage in question hits saturation at or before cutoff, then the stage is biased class A2.

If stage in question hits cutoff before grid-clipping, and it is conducting plate current for more than 180 degrees of the output signal, it is class AB1.

If the stage in question hits cutoff before saturation, and it is conducting plate current for more than 180 degrees of the output signal, it is class AB2.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Mark L (registered user: 1458 posts )
Date: Sat, Sep 15th, 2007 @ 18:28 ( . )

Thanks Carl, good explaination.

I know I've been guilty of calling grid-clipping, saturation. I'll have to be more careful when refereing to Vg=0v.


-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Merlin (registered user: 5226 posts )
Date: Sun, Sep 16th, 2007 @ 13:21 ( . )

On 09/15/2007 @ 14:16, CarlB wrote :
: If the stage in question hits saturation at or before cutoff, then the stage is biased class A2.

That I'm afraid, is utter rubbish. The "2" suffix simply indicates that grid current flows. That's it.
Now, it doesn't matter whether that's in a REALLY well designed hifi amp where the source impedance is so low that it can indeed reach saturation, or in a slightly less capable amp where only a few loud transients are allowed for, or indeed a guitar amp where the grid signal clips quickly. If the amp is intended to routinely draw grid current, it's Class A2 (or AB2). Full stop.
Therefore all guitar amps are A2, when put into the hands of anyone wanting an overdriven tone.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Randall Aiken (guest: search)
Date: Sun, Sep 16th, 2007 @ 13:54 ( . )

On 09/16/2007 @ 13:21, Merlin wrote :
On 09/15/2007 @ 14:16, CarlB wrote :
: : If the stage in question hits saturation at or before cutoff, then the stage is biased class A2.
:

: That I'm afraid, is utter rubbish. The "2" suffix simply indicates that grid current flows. That's it.
: Now, it doesn't matter whether that's in a REALLY well designed hifi amp where the source impedance is so low that it can indeed reach saturation, or in a slightly less capable amp where only a few loud transients are allowed for, or indeed a guitar amp where the grid signal clips quickly. If the amp is intended to routinely draw grid current, it's Class A2 (or AB2). Full stop.
: Therefore all guitar amps are A2, when put into the hands of anyone wanting an overdriven tone.
--



Wrong. Any amplifier can be driven to clipping, that doesn't change it's class of operation (else all amps would be class D switching amps!).

All guitar amps are not class A2, they are usually either class A1 or AB1. If the bias and drive is such that it allows normal operation into the positive region of the grid curves, then it becomes class A2 or AB2. An AC-coupled "standard" output stage cannot be driven into the positive region of the grid curves, it takes a low-impedance or DC-driven stage to accomplish that.

Just because you can draw enough current to clamp the peaks when you hit the zero volts grid curve does not make it an A2/AB2 amp. It just makes it a clipped A1/AB1 amp.


Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:John Hynes (registered user: 2628 posts )
Date: Sun, Sep 16th, 2007 @ 14:05 ( . )

OK Randall, You've confused me here. Not that that's very hard to do! :)

Please clarify:

Are you saying that it is possible to draw grid current and not be in A2 or AB2? Because if you are, you're disagreeing with everything I've ever read or heard.

It has always seemed patently obvious to me that the presence or lack of grid current draw at any portion of the cycle, and that alone, determines the 1 or 2 subscript.

Is that incorrect, in your view?

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Merlin (registered user: 5226 posts )
Date: Sun, Sep 16th, 2007 @ 14:13 ( . )

On 09/16/2007 @ 14:05, John Hynes wrote :
: Are you saying that it is possible to draw grid current and not be in A2 or AB2? Because if you are, you're disagreeing with everything I've ever read or heard.

Quite. Randall is mistaken, because you cannot define what "normal operation" is where grid current is concerned- where does the "2" suffix begin to apply? When THD is 1%? 10%? 100%? The "hardness" of the clipping is irrelevant. Grid current = Class A2/AB2.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Randall Aiken (guest: search)
Date: Sun, Sep 16th, 2007 @ 15:04 ( . )

On 09/16/2007 @ 14:13, Merlin wrote :

: Quite. Randall is mistaken, because you cannot define what "normal operation" is where grid current is concerned- where does the "2" suffix begin to apply? When THD is 1%? 10%? 100%? The "hardness" of the clipping is irrelevant. Grid current = Class A2/AB2.
--



Normal operation is grid current without clipping. As I said before, any amp can be driven to clipping, that doesn't change it's operating class. You can choose to call clipping "drawing grid current, therefore class A2/AB2" if you like, but you are wrong.

A class A2 amp or class AB2 amp can put out *clean*, unclipped power at higher levels than a class A1/AB1 amp because it is designed to be driven to positive voltages on the grids. Just because you can hit zero and clamp there when AC-coupled and draw a bit of current while clipping doesn't define it as a class A2/AB2 amp, that is a ridiculous assertion.

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Zaphod (guest: search)
Date: Sun, Sep 16th, 2007 @ 16:04 ( . )

Quite

On 09/16/2007 @ 15:04, Randall Aiken wrote :
:... Just because you can hit zero and clamp there when AC-coupled and draw a bit of current while clipping doesn't define it as a class A2/AB2 amp, that is a ridiculous assertion.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Merlin (registered user: 5226 posts )
Date: Mon, Sep 17th, 2007 @ 15:36 ( . )

On 09/16/2007 @ 15:04, Randall Aiken wrote :
: A class A2 amp or class AB2 amp can put out *clean*, unclipped power at higher levels than a class A1/AB1 amp because it is designed to be driven to positive voltages on the grids.

Yes, that's the general idea. But thre is inevitably SOME compression (i.e., soft clipping) because the source impedance is NEVER a perfect voltage source. SO you cann't define it by saying simply "it doesn't clip", you either have to define a particular level of THD, or concede that ANY amp designed to regularly draw grid current- even a guitar amp- is Class A2/AB2.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:CarlB (guest: search)
Date: Tue, Sep 18th, 2007 @ 01:48 ( . )

From the following:

[link]

"B. Occasionally the subscripts 1 and 2 are used as class A1, class B2, etc., the subscript 1 indicating that the grid is negative at all times throughout its cycle of operation, and 2 indicating a positive grid swing during a portion of each cycle."

If the intent was for '2' to indicate just grid current, the above would have been written, 'indicating a positive grid current during a portion of each cycle.'

The grid-clipping amp you mention is incapable of positive grid swing during a portion of each cycle. So you can't call it class A2.

If you re-examine the definitions of these classes in an overall context of any output signal within the range of smallest to maximum power, given clean reproduction, it becomes clear.

Class A1: the max the element can reproduce cleanly is before grid clipping.

Class A2: the max the element can reproduce cleanly is before saturation, as the driver by design can go past grid conduction.

Class AB1: the max the element can reproduce cleanly is before plate current cut off. In push-pull, the stage can produce cleanly up until grid-conduction.

Class AB2: the max the element can reproduce cleanly is before plate current cut off. In push-pull, the stage can produce cleanly up until saturation.


-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Zaphod (guest: search)
Date: Tue, Sep 18th, 2007 @ 05:19 ( . )

So you're now discussing a practical implementation rather than a theoretical definition. In theoretical definitions we can have the luxury of such things as perfect voltage and current sources. And even in the real world, if we put our minds to it, I'm sure we could build a really good voltage source with a range sufficient for the amplitude of the driving signal, that would give us a very clean sine wave output from the amplifier.

The bottom line is that we shouldn't confuse the small grid current transients that naturally occur in a Class A1 or AB1 amp, with the sustained grid current drive of a Class A2 or AB2 one.

On 09/17/2007 @ 15:36, Merlin wrote :
:
: Yes, that's the general idea. But there is inevitably SOME compression (i.e., soft clipping) because the source impedance is NEVER a perfect voltage source. SO you can't define it by saying simply "it doesn't clip", you either have to define a particular level of THD, or concede that ANY amp designed to regularly draw grid current- even a guitar amp- is Class A2/AB2.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:John Hynes (registered user: 2628 posts )
Date: Sun, Sep 16th, 2007 @ 16:43 ( . )

On 09/16/2007 @ 14:13, Merlin wrote :
On 09/16/2007 @ 14:05, John Hynes wrote :
: : Are you saying that it is possible to draw grid current and not be in A2 or AB2? Because if you are, you're disagreeing with everything I've ever read or heard.
:

: Quite. Randall is mistaken, because you cannot define what "normal operation" is where grid current is concerned- where does the "2" suffix begin to apply? When THD is 1%? 10%? 100%? The "hardness" of the clipping is irrelevant. Grid current = Class A2/AB2.
--



From what I'm getting from my conversation with Randall below, I think we're all talking about the same thing, but possibly identifying it in somewhat ambiguous terms that make it seem like we disagree. I don't think we do.

I have on various occasions made the simple assertion that:

grid current = "2" suffix
no grid current = "1" suffix

Other than what Randall pointed out about leakage current, which though true in a strict sense is a minor technicality, I still hold this is true.

Randall and I both outlined one possible situation in which this can be implemented: A high-impedance AC coupled stage driving an A/AB stage is going to go into clamping before the grids are driven positive and draw current while operating in a fairly linear way in the positive grid curve area. It will have some leakage current, some utterly trivial amount of current as the bias hovers around Vg=0, but that's it. That's A/AB "1"

You could also drive that stage from a low impedance DC coupled source, and when the grids are driven towards the positive region, current is available, the clamping doesn't happen, and the stage operates in the positive grid curve territory. That's A/AB "2"

However, it should be pointed out that the definition of "1" and "2" is described in any number of texts as the presence of grid current being drawn as the grids go positive with respect to the cathodes. Can this occur in ways other than what is outlined above? If so, it's A/AB2.

A specific topology of driving stage isn't part of the definition, but it's capabilities are. Could a hypothetical AC coupled preamp stage be driven to the point where it's grid goes positive with respect to the cathode, thus mandating that it was able to source enough current to do so, without clamping? If yes, that stage is A2.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:CarlB (guest: search)
Date: Sun, Sep 16th, 2007 @ 17:37 ( . )

" grid current = "2" suffix
: no grid current = "1" suffix"

Modify that a bit:

No grid clipping = "2" suffix
Grid clipping = "1" suffix

In a high-impedance dc-coupled driver (such as with dc-drive), there is a bit of current supplied but it is insufficient enough for it supply enough current. But there still is current drawn during conduction. The stage being driven is still A1 or AB1.

Even an AC coupled grid will draw current as it attempts to go positive biased. That charges the coupling cap more, and when the signal dips below grid conduction the stage is temporarily biased more coldly. If this gets too bad, then the stage exhibits blocking distortion, where the cutoff becomes long enough for our ears to pick up on. It's still an AB1 stage, albeit one that is exhibiting blocking distortion.

"Other than what Randall pointed out about leakage current, which though true in a strict sense is a minor technicality, I still hold this is true."

Hope I persuaded you otherwise ...

"Randall and I both outlined one possible situation in which this can be implemented: A high-impedance AC coupled stage driving an A/AB stage is going to go into clamping before the grids are driven positive and draw current while operating in a fairly linear way in the positive grid curve area. It will have some leakage current, some utterly trivial amount of current as the bias hovers around Vg=0, but that's it. That's A/AB "1"

It will be drawing current, supplied by that coupling cap. If it's cap-coupled, you need to limit the cap charging current to less than it's discharge-back-to-quiescent current for heavy grid-clipping. This is done by increasing the grid stopper, decreasing the grid to ground resistor, decreasing the part of the signal trying to go over Vgk=0, or some mixture of these.

The grids are conducting current, but it's still an A1 or AB1 stage because the stage output can't faithfully follow the input past Vgk=0.

: You could also drive that stage from a low impedance DC coupled source, and when the grids are driven towards the positive region, current is available, the clamping doesn't happen, and the stage operates in the positive grid curve territory. That's A/AB "2"

Yup!!!

"However, it should be pointed out that the definition of "1" and "2" is described in any number of texts as the presence of grid current being drawn as the grids go positive with respect to the cathodes. Can this occur in ways other than what is outlined above? If so, it's A/AB2."

Not sure I'm following you here ...

"A specific topology of driving stage isn't part of the definition, but it's capabilities are. Could a hypothetical AC coupled preamp stage be driven to the point where it's grid goes positive with respect to the cathode, thus mandating that it was able to source enough current to do so, without clamping? If yes, that stage is A2."

Yes, but it's going to have a difficult time keeping away from blocking distortion, unless you give some place to get the grid current from that doesn't involve charging a cap. For instance, and interstage transformer (although these have their own problems for driving A2 or AB2).

Hope I'm making these pretty clear.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Randall Aiken (guest: search)
Date: Sun, Sep 16th, 2007 @ 15:08 ( . )

On 09/16/2007 @ 14:05, John Hynes wrote :
OK Randall, You've confused me here. Not that that's very hard to do! :)
:
: Please clarify:
:
: Are you saying that it is possible to draw grid current and not be in A2 or AB2? Because if you are, you're disagreeing with everything I've ever read or heard.
:
: It has always seemed patently obvious to me that the presence or lack of grid current draw at any portion of the cycle, and that alone, determines the 1 or 2 subscript.
:
: Is that incorrect, in your view?
:
: -John
:
--



I'm saying that it is possible to drive just about any amplifier into clipping, which doesn't change it's class of operation.

A true class A2 or class AB2 amplfier is designed to be driven into the positive region of the grid curves, which takes power to do. If you take an AC-coupled class A1 or class AB1 amp and drive it to clipping, the grid will clamp the peaks and draw just enough current to keep it clamped, but you aren't driving the grids positive, and you aren't getting any more clean power out of them.

There is a very big difference between a clipped A1/AB1 and and a true A2/AB2 amp.

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'No, no, no'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Mon, Sep 17th, 2007 @ 00:14 ( . )

Maybe this was mentioned earlier in this whole rats nest, but the original definitions were almost certainly meant to apply to clean amps with the tube operating in as linear a portion of the range as possible.

The engineers at the time never even *dreamed* we would *want* distortion. Distortion was the enemy, to be annihilated if at all possible. They accepted that some distortion would be present, but all the evidence is that they minimized it where possible, and ignored it after that.

So I just don't buy definitions such as 2 relating to grid clipping vs merely relating to grid current flowing.

It may be useful to you, but it's something else. So you really ought to come up with another terminology. Which has, as I recall, been suggested before now!

[I'm not responding to Randall; this was simply the point where I had to jump in with this.]

-- REPLY: [With No Quote] --- [With Quoted Text]

'No, no, no'
Author:Randall Aiken (guest: search)
Date: Mon, Sep 17th, 2007 @ 00:51 ( . )

On 09/17/2007 @ 00:14, Harrison Ford Prefect wrote :
Maybe this was mentioned earlier in this whole rats nest, but the original definitions were almost certainly meant to apply to clean amps with the tube operating in as linear a portion of the range as possible.
:
: The engineers at the time never even *dreamed* we would *want* distortion. Distortion was the enemy, to be annihilated if at all possible. They accepted that some distortion would be present, but all the evidence is that they minimized it where possible, and ignored it after that.
:
: It may be useful to you, but it's something else. So you really ought to come up with another terminology. :



I don't see the need for any new terminology - the old terminology still holds true, even for amps driven into distortion. A class AB1 amp designed in 1950 still distorts the same way a class AB1 amp designed in 2007 does. Just because you purposefully clip an amp doesn't change it's class of operation. I can't see why this is even in question, to be honest, it makes no sense at all from an engineering perspective.

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 08:31 ( . )

Randall and Carl are indeed sticking closely to the established definitions for the classes, based on conduction angle and grid conduction criteria.

Where the confusion arises is the traditional requirement in many applications (eg audio and RF) for the output of the stage to be relatively undistorted. Where an amplifying stage is operating in anything but Class A, this requirement means that you have to find a way of recreating the input waveform. One way is to use a pair of these stages operating anti-phase in what we refer to as push-pull. Another way, popular in RF, is to use a resonant tank circuit. These days there are digital wave-shaping techniques that could also be used.

But essentially these are all additions to the basic Class AB stage, which come from *requirements* imposed by the specific application- they don't form a part of the fundamental class definitions. And when the application is a guitar amplifier, then the always clean output requirement pretty much goes away (unless maybe it's Roland JC-120). However, we've all been so heavily programmed into thinking that part of the *definition* of Class AB is PP operation, that we find it very hard to get out of the box and unlearn it. In my own experience, I found I had to unlearn a lot of the electronics I used to know when I got into guitar amps. And the SE Class AB thing was probably the hardest for me to come to terms with.

On 09/17/2007 @ 00:51, Randall Aiken wrote :
:
: I don't see the need for any new terminology - the old terminology still holds true, even for amps driven into distortion. A class AB1 amp designed in 1950 still distorts the same way a class AB1 amp designed in 2007 does. Just because you purposefully clip an amp doesn't change it's class of operation. I can't see why this is even in question, to be honest, it makes no sense at all from an engineering perspective.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:John Hynes (registered user: 2628 posts )
Date: Mon, Sep 17th, 2007 @ 08:54 ( . )

On 09/17/2007 @ 08:31, Zaphod wrote :
Randall and Carl are indeed sticking closely to the established definitions for the classes, based on conduction angle and grid conduction criteria.
:
: Where the confusion arises is the traditional requirement in many applications (eg audio and RF) for the output of the stage to be relatively undistorted. Where an amplifying stage is operating in anything but Class A, this requirement means that you have to find a way of recreating the input waveform. One way is to use a pair of these stages operating anti-phase in what we refer to as push-pull. Another way, popular in RF, is to use a resonant tank circuit. These days there are digital wave-shaping techniques that could also be used.
:
: But essentially these are all additions to the basic Class AB stage, which come from *requirements* imposed by the specific application- they don't form a part of the fundamental class definitions. And when the application is a guitar amplifier, then the always clean output requirement pretty much goes away (unless maybe it's Roland JC-120). However, we've all been so heavily programmed into thinking that part of the *definition* of Class AB is PP operation, that we find it very hard to get out of the box and unlearn it. In my own experience, I found I had to unlearn a lot of the electronics I used to know when I got into guitar amps. And the SE Class AB thing was probably the hardest for me to come to terms with.
:
: On 09/17/2007 @ 00:51, Randall Aiken wrote :
: :
: : I don't see the need for any new terminology - the old terminology still holds true, even for amps driven into distortion. A class AB1 amp designed in 1950 still distorts the same way a class AB1 amp designed in 2007 does. Just because you purposefully clip an amp doesn't change it's class of operation. I can't see why this is even in question, to be honest, it makes no sense at all from an engineering perspective.
: :
:
--



Actually Phil, you've just touched on something I've been thinking, and provided me with an opportune place to bring it up. :)

A circuit is a circuit with respect to the definitions laid out for these classes however many years ago they were penned. The fact that one is a distorting guitar amp, and another is a linear HiFi amp, and another is an RF amp, DOES NOT MATTER. Certain definitions were created, they were given clear, uncomplicated meaning, and they are what they are.

If a certain circuit does not fall within those boundaries, you need to create a new definition of what you've got, not re-appropriate another definition and change its meaning.

Here is where we seem to disagree: I think (as do several others, Randall has said explicitly at least) that the old definitions do indeed apply to these circuits. To me, if that original definition of "AB" doesn't take into account your so-called modern usage of a single ended circuit that has one aspect of AB behavior, but not *all* of them, as envisioned by those who coined the definitions, THEN IT IS NOT AN AB AMPLIFIER. I think its an A amplifier driven into clipping. You disagree, so maybe you ought to call it something else, maybe "Class Az" for "A, Zaphod's Way" or something - but it isn't AB. Can't be. The definition of AB includes the characteristics of the load, which require P-P operation.

The details of "1" vs. "2" suffixes are clear as well. "1" = no grid current, "2" = grid current draw. These only make sense in the context of ignoring grid leak and some current present as drawn from a coupling cap in an AC coupled stage. That's all. I think maybe we all agree on that by now, though.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Dan N (guest: search)
Date: Mon, Sep 17th, 2007 @ 09:27 ( . )

On 09/17/2007 @ 08:54, John Hynes wrote :If a certain circuit does not fall within those boundaries, you need to create a new definition of what you've got, not re-appropriate another definition and change its meaning.
I couldn't agree with you more there John. It's like trying to fit a round peg in a square hole, you can hammer away at it 'till the cows come home but it just not going to fit. We need to drill another hole. This the type of thing that legal practitioners do, and I think we are all above that.

So for now I'm simply going to call it SE.

Even though the texts don't specifically define what grid current is, I do believe we are all in agreement over "1" & "2" ;)


-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:CarlB (guest: search)
Date: Mon, Sep 17th, 2007 @ 09:42 ( . )

John,

I don't think the texts support your assertion here.
SE class AB is very real, and we use it all the time. There is nothing out of the intent of those who envisioned defined the classes many years ago.

You write,

"To me, if that original definition of "AB" doesn't take into account your so-called modern usage of a single ended circuit that has one aspect of AB behavior, but not *all* of them, as envisioned by those who coined the definitions, THEN IT IS NOT AN AB AMPLIFIER. I think its an A amplifier driven into clipping. (...)

"The details of "1" vs. "2" suffixes are clear as well. "1" = no grid current, "2" = grid current draw. These only make sense in the context of ignoring grid leak and some current present as drawn from a coupling cap in an AC coupled stage. That's all. I think maybe we all agree on that by now, though. "

You have a body of experience that you're drawing on. However, it's time to review that experience against the texts. Please take a look back through the definitions in those texts and the examples given in those texts. See if your understanding really is in line with what is there.

If you still think it is, please let me know what texts you're looking at. I'd like to understand where the notion of 'class AB is push-pull only' come from.

Thanks much,

Carl

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:John Hynes (registered user: 2628 posts )
Date: Mon, Sep 17th, 2007 @ 11:20 ( . )

On 09/17/2007 @ 09:42, CarlB wrote :
John,
:
: I don't think the texts support your assertion here.
: SE class AB is very real, and we use it all the time. There is nothing out of the intent of those who envisioned defined the classes many years ago.
:
: You write,
:
: "To me, if that original definition of "AB" doesn't take into account your so-called modern usage of a single ended circuit that has one aspect of AB behavior, but not *all* of them, as envisioned by those who coined the definitions, THEN IT IS NOT AN AB AMPLIFIER. I think its an A amplifier driven into clipping. (...)
:
: "The details of "1" vs. "2" suffixes are clear as well. "1" = no grid current, "2" = grid current draw. These only make sense in the context of ignoring grid leak and some current present as drawn from a coupling cap in an AC coupled stage. That's all. I think maybe we all agree on that by now, though. "
:
: You have a body of experience that you're drawing on. However, it's time to review that experience against the texts. Please take a look back through the definitions in those texts and the examples given in those texts. See if your understanding really is in line with what is there.
:
: If you still think it is, please let me know what texts you're looking at. I'd like to understand where the notion of 'class AB is push-pull only' come from.
:
: Thanks much,
:
: Carl
--



Very simply put, the classes were defined within the framework of putting out a clean signal. A single-ended amplifier only puts out a clean signal if it is conducting 360 degrees of the cycle. It is class A. Overdrive it, and it's a class A amp driven to clipping.

An AB amplifier only puts out a clean signal if one side of a pair is conducting when the other isn't. It's push-pull by nature.

These definitions themselves lose their meaning if you contend that the fact that they are clipping matters. Clean, unclipped signal is the criteria for class. Example: your "I call this an AB SE amplifier", when fed a small enough signal to avoid clipping, is behaving exactly as an A amplifier should.

The fact that we drive them to clipping doesn't matter at all. The class definitions are framed in the context of faithful signal reproduction. That's all. Simple.

If we wish to have a class definition for an overdriven single-ended amplifier, that somehow recognizes that it's clipping, then we need a new class name for that. Otherwise, it's a clipping class A amp.

I'm pretty sure that all texts agree with that. That none of the classic texts describe that a single-ended amplifier, when driven to clipping, is in a class other than "A" is not an error of omission. I realize there are many, many engineering texts we refer to, but the only one I have handy is RDH4. It is very clear what the classes mean, and upon reviewing it, I still believe my contention to be true.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 12:51 ( . )

I think you're missing the fact that an overdriven Class A amp is a different beast to an SE amp that's operating in cutoff part of the time by design, and when it's not necessarily even working very hard.

Or take the situation where you have a Class AB PP amp. You pull out the power tube(s) from one side, and it doesn't suddenly become an overdriven Class A amp. All the power tubes in that amp are biased to operate and go into cutoff in the same way whether they're operating in push-pull or not. And once again, I will remind you of the SE Class AB RF power amplifiers. They're both SE and in Class AB, and the fact that they normally have a resonant tank on the output doesn't change the fact.

It is indeed clear what the classes mean, which has always been defined by conduction angle - pure and simple. Anything more than that is application-dependent. If you want to speculate about RDH4, I would suggest omitting audio SE Class AB stages may have been a graver omission. But who would have imagined any requirement for them back then? :o)


On 09/17/2007 @ 11:20, John Hynes wrote :
:
: If we wish to have a class definition for an overdriven single-ended amplifier, that somehow recognizes that it's clipping, then we need a new class name for that. Otherwise, it's a clipping class A amp.
:
: I'm pretty sure that all texts agree with that. That none of the classic texts describe that a single-ended amplifier, when driven to clipping, is in a class other than "A" is not an error of omission. I realize there are many, many engineering texts we refer to, but the only one I have handy is RDH4. It is very clear what the classes mean, and upon reviewing it, I still believe my contention to be true.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Mon, Sep 17th, 2007 @ 13:33 ( . )

I'm not going over to the dark side, or anything, but Phil's likely right about that one.

If there had been overdriven guitar amps back when the definitions were being drawn up, then we probably wouldn't be having this discussion. They'd have either implicitly said that Class AB is an acceptable description for a SE amp used for that purpose. Or there'd be another class definition to cover it, which we'd all know and accept.

Despite the fact that I only consider the classic class (sorry) definitions to apply to linear amplifiers, having never been taught anything to the contrary - thus meaning that I wouldn't describe it as AB for a single-ended amp, I do understand the reason why my esteemed colleagues on the other side of this debate think it is. And had Leo Fender, et al, been born 50 years earlier, the situation would probably be much different.

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Merlin (registered user: 5226 posts )
Date: Mon, Sep 17th, 2007 @ 15:40 ( . )

On 09/17/2007 @ 13:33, Zoe N Iain wrote :
If there had been overdriven guitar amps back when the definitions were being drawn up, then They'd have either implicitly said that Class AB is an acceptable description for a SE amp used for that purpose. Or there'd be another class definition to cover it, which we'd all know and accept.
But the nomenclature system works perfectly well for guitar amps- as I have already argued. The old masters managed to fluke it! It doesn't matter that guitar amps aren't particularly linear, they are still linear enough to the point where conduction angles can still be ascertained, and class determined. And an SE amp will only deliver a clean(ish) (i.e., linear) signal while conducting 360 degrees, which equals class A, as everyone agrees. But only a PP amp can so that beyond cutoff.

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 17:31 ( . )

The real question boils down to whether a clean output is essential to the class definition or not. Some maintain that conduction angle is the only true determinant. Others clearly still think the boundary conditions require a clean output, rather than grid current behavior.

On 09/17/2007 @ 15:40, Merlin wrote :
: But the nomenclature system works perfectly well for guitar amps- as I have already argued. The old masters managed to fluke it! It doesn't matter that guitar amps aren't particularly linear, they are still linear enough to the point where conduction angles can still be ascertained, and class determined. And an SE amp will only deliver a clean(ish) (i.e., linear) signal while conducting 360 degrees, which equals class A, as everyone agrees. But only a PP amp can do that beyond cutoff.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:John Hynes (registered user: 2628 posts )
Date: Mon, Sep 17th, 2007 @ 13:44 ( . )

I'll keep this short, as we are going to start re-hashing the same old arguments here, so here's my summary:

I think you are ignoring the whole definition, in that it is not just conduction angle. Part of the definition of AB requires two separate load lines, due to differing currents in each half of the OT primary.

Given that, can either of you tell me which "half" of the OT primary in an SE amp has more/less current flow at any given time? :)

Obviously, the problem is that we all seem to disagree on what factors to include in the definition. I hold that is is either explicit (in some) or implicit (by example) in many texts that the load is part of the definition, and not merely conduction angle.

You and others disagree. OK. That's fine. I can live with that. :)

Can we all talk about something more productive, now?

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Mon, Sep 17th, 2007 @ 14:01 ( . )

Hear, hear. The discussion has hit the point of diminishing return. Though some of the points raised have made me think about a few things slightly differently (or more deeply), to my benefit, so it's been a worthwhile discussion to that point. But let's move on.

So, what else can we argue about? ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 14:57 ( . )

M: Oh look, this isn't an argument.
A: Yes it is.
M: No it isn't. It's just contradiction.
A: No it isn't.
M: It is!
A: It is not.
M: Look, you just contradicted me.
A: I did not.
M: Oh you did!!
A: No, no, no.
M: You did just then.
A: Nonsense!
M: Oh, this is futile!
A: No it isn't.
M: I came here for a good argument.
A: No you didn't; no, you came here for an argument.
M: An argument isn't just contradiction.
A: It can be.
M: No it can't. An argument is a connected series of statements intended to establish a proposition.
A: No it isn't.
M: Yes it is! It's not just contradiction.
A: Look, if I argue with you, I must take up a contrary position.
M: Yes, but that's not just saying 'No it isn't.'
A: Yes it is!
M: No it isn't!
A: Yes it is!
M: Argument is an intellectual process. Contradiction is just the automatic gainsaying of any statement the other person makes.
(short pause)
A: No it isn't.
M: It is.
A: Not at all.
M: Now look.
A: (Rings bell) Good Morning.
M: What?
A: That's it. Good morning.
M: I was just getting interested.
A: Sorry, the five minutes is up.
M: That was never five minutes!
A: I'm afraid it was.
M: It wasn't.
Pause
A: I'm sorry, but I'm not allowed to argue anymore.
M: What?!
A: If you want me to go on arguing, you'll have to pay for another five minutes.
M: Yes, but that was never five minutes, just now. Oh come on!

[link]

[link]
On 09/17/2007 @ 14:01, Zoe N Iain wrote :
:
: So, what else can we argue about? ;-}
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Randall Aiken (guest: search)
Date: Mon, Sep 17th, 2007 @ 15:08 ( . )

On 09/17/2007 @ 13:44, John Hynes wrote :
Part of the definition of AB requires two separate load lines


And that's the heart of the matter. In all my years of engineering, and in all the books I've read, I've never heard of the definition of class AB requiring two separate load lines. You can indeed have SE class AB, as seen in many textbooks, and the ARRL handbooks (and yes, I was a ham radio operator in my younger days, so I'm intimately familiar with these handbooks, and have several old ones going back to the 1949 edition on my bookshelf).

By the "classic" definitions, SE class AB is not applicable to audio only because you cannot get linear reproduction of the input, but if you *want* nonlinear reproduction of the input, it suits your needs just fine. SE class AB does indeed exist, and there is no need for a "changing load line".

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 15:21 ( . )

I started out in short wave and amateur radio as well, well before I even learned to play guitar, and I will say the same thing. I never once came across dual load lines being said to be part of the definition of Class AB.

On 09/17/2007 @ 15:08, Randall Aiken wrote :
:
: And that's the heart of the matter. In all my years of engineering, and in all the books I've read, I've never heard of the definition of class AB requiring two separate load lines. You can indeed have SE class AB, as seen in many textbooks, and the ARRL handbooks (and yes, I was a ham radio operator in my younger days, so I'm intimately familiar with these handbooks, and have several old ones going back to the 1949 edition on my bookshelf).
:
: By the "classic" definitions, SE class AB is not applicable to audio only because you cannot get linear reproduction of the input, but if you *want* nonlinear reproduction of the input, it suits your needs just fine. SE class AB does indeed exist, and there is no need for a "changing load line".
:
: Randall Aiken
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Merlin (registered user: 5226 posts )
Date: Mon, Sep 17th, 2007 @ 15:48 ( . )

On 09/17/2007 @ 15:08, Randall Aiken wrote :
You can indeed have SE class AB, as seen in many textbooks, and the ARRL handbooks

Yes, but only at RF, where the resonant load allows a failry clean (360 degree) output signal despite the valve operating near cutoff. But this is not the vase for audio, so you cannot apply RF nomenclature to our particulr branch of electronics.

By the "classic" definitions, SE class AB is not applicable to audio only because you cannot get linear reproduction of the input, but if you *want* nonlinear reproduction of the input, it suits your needs just fine.
But even guitar amps are linear enough that the definitions still apply perfectly well- they still conduct 360 degrees, or less, while giving a quite faithful reproduction at the output. The fact that we drive them to clipping is irrelevant, and doesn't magically change their operating class.

AB does indeed exist, and there is no need for a "changing load line".


Grnated, with modern complementary designs there is no change in load like we see in a valve amp, but it at least demonstrates where the name originated. But it cannot exist in SE, and any rational EE can see that.

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 15:10 ( . )

I thought we had already scotched that idea of two different load lines being part of the basic definition of Class AB, earlier on in this thread.

Pardon me for repeating myself again, but Class AB SE RF amplifiers, which are widely accepted and understood, clearly don't exhibit this behavior. It is a phenomenon specific to PP Class AB amplifiers, which IMO are a special case within the general category of Class AB amplifiers, and an important one at that.

So whatever your view may otherwise be about Class AB SE amplifiers, this clearly shows that the two load-line concept isn't part of the fundamental definition.

On 09/17/2007 @ 13:44, John Hynes wrote :
:
: I think you are ignoring the whole definition, in that it is not just conduction angle. Part of the definition of AB requires two separate load lines, due to differing currents in each half of the OT primary.
:
: Given that, can either of you tell me which "half" of the OT primary in an SE amp has more/less current flow at any given time? :)
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:John Hynes (registered user: 2628 posts )
Date: Mon, Sep 17th, 2007 @ 15:49 ( . )

On 09/17/2007 @ 15:10, Zaphod wrote :
I thought we had already scotched that idea of two different load lines being part of the basic definition of Class AB, earlier on in this thread.
:
: Pardon me for repeating myself again, but Class AB SE RF amplifiers, which are widely accepted and understood, clearly don't exhibit this behavior. It is a phenomenon specific to PP Class AB amplifiers, which IMO are a special case within the general category of Class AB amplifiers, and an important one at that.
:
: So whatever your view may otherwise be about Class AB SE amplifiers, this clearly shows that the two load-line concept isn't part of the fundamental definition.
:
: On 09/17/2007 @ 13:44, John Hynes wrote :
: :
: : I think you are ignoring the whole definition, in that it is not just conduction angle. Part of the definition of AB requires two separate load lines, due to differing currents in each half of the OT primary.
: :
: : Given that, can either of you tell me which "half" of the OT primary in an SE amp has more/less current flow at any given time? :)
: :
:
--



I'm not going to continue this, as we've already been over it all. If you say that various radio texts describe AB in those terms, fine. I believe you! :)

As I've said many times, the only text I have is RDH4. Page 545, Section 5(ii), "Classes of Operation":

Class A operation is the normal condition of operation for a single valve, and indicates that plate current is not cut off for any portion of the cycle.

Limiting Class A push-pull operation is operation such that one valve just reaches plate current cut-off when the other reaches zero bias.

Class AB operation indicates overbiased conditions, and is used only in push-pull to balance out the even harmonics.
...
<end of quote>

All of these stated in the context of clean signal reproduction. All examples given discuss the condition of the load line in each stage.

So, let's end this. I chose to use this undeniably clear RDH4 definition. :)
You chose to use another source's definition, which seems also to be clear to you. :)

If you wish to continue to try to convince me I'm wrong, then hold a seance and ask F. Langford-Smith if he feels he was "wrong" in some way, given that another engineer wrote a different book that disagrees with him.

And in case that comes off as smarmy, I apologize. This whole debate is merely entertainment to me, as I could care less what anyone chooses to call one class or another! Reasonable people can, have, and will continue to disagree about this, and since our "source texts" apparently disagree as well, we'd then have to argue who among their authors were right or wrong, and, truthfully, I haven't the skill or knowledge to do that.

Fair enough?

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 17:20 ( . )

I understand what you've just said, but you haven't actually responded to the point I was making about the two load lines not being part of the generally Class AB definition.

The text you've quoted below doesn't mention it as being part of the Class AB criteria either. And I also have to say in passing, that the quote from RDH4 doesn't read much like a definition, so much as a brief general description of typical Class AB usage.

And whatever our respective textbooks say, Class AB SE amplifiers do exist in the RF world, and naturally don't have the dual load line behavior.

On 09/17/2007 @ 15:49, John Hynes wrote :
:
: I'm not going to continue this, as we've already been over it all. If you say that various radio texts describe AB in those terms, fine. I believe you! :)
:
: As I've said many times, the only text I have is RDH4. Page 545, Section 5(ii), "Classes of Operation":
:
: Class A operation is the normal condition of operation for a single valve, and indicates that plate current is not cut off for any portion of the cycle.
:
: Limiting Class A push-pull operation is operation such that one valve just reaches plate current cut-off when the other reaches zero bias.
:
: Class AB operation indicates overbiased conditions, and is used only in push-pull to balance out the even harmonics.
: ...
: <end of quote>
:
: All of these stated in the context of clean signal reproduction. All examples given discuss the condition of the load line in each stage.
:
: So, let's end this. I chose to use this undeniably clear RDH4 definition. :)
: You chose to use another source's definition, which seems also to be clear to you. :)
:
: If you wish to continue to try to convince me I'm wrong, then hold a seance and ask F. Langford-Smith if he feels he was "wrong" in some way, given that another engineer wrote a different book that disagrees with him.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Mon, Sep 17th, 2007 @ 17:31 ( . )

Shot yourself in the foot there Philip, my son! A SE RF transmitter operating in AB demonstrably DOES have a dual load line. Try and run an FM signal through one and watch what happens to the amplitude as you stray away from the centre frequency.

(Granted, that has no bearing on audio amps, but it's nice to pick on you occasionally ;-} )

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Tue, Sep 18th, 2007 @ 03:38 ( . )

Heh, however a somewhat special case there again. :o) If you modulate away from the centre frequency, I would actually expect to see more than a dual load line though.

On 09/17/2007 @ 17:31, Zoe N Iain wrote :
:
Shot yourself in the foot there Philip, my son! A SE RF transmitter operating in AB demonstrably DOES have a dual load line. Try and run an FM signal through one and watch what happens to the amplitude as you stray away from the centre frequency.
:
: (Granted, that has no bearing on audio amps, but it's nice to pick on you occasionally ;-} )
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Tue, Sep 18th, 2007 @ 03:51 ( . )

In practice, I suspect the loadline probably is continuously variable, so you're right. But there's one distinct step where you can see the transition from one to the 'other', because there's a sudden change in the amount of signal in the bottom half of the wave. 3dB down point, I think.

Never mind, just saw an opportunity to pick on you, Phil ;-} Has no bearing on an audio amp, unless it was using a tuned circuit (though I can't think why you'd want to, except maybe in some kind of modulator).

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Tue, Sep 18th, 2007 @ 05:01 ( . )

Yeah, all in good fun :o) And of course, the example you gave is flawed as far as having any bearing on this particular discussion, because the change in load line has nothing to do with the amplifier's own transition between its Class A and Class B like modes with on alternate half cycles of the carrier frequency. As you've noted in this case the impedance change is actually the result of a frequency shift being applied to a tuned tank circuit. If you run the same amplifier in continuous wave (ie no modulation) then it will continue to operate as an SE Class AB stage with a single load line.

On 09/18/2007 @ 03:51, Zoe N Iain wrote :
In practice, I suspect the loadline probably is continuously variable, so you're right. But there's one distinct step where you can see the transition from one to the 'other', because there's a sudden change in the amount of signal in the bottom half of the wave. 3dB down point, I think.
:
: Never mind, just saw an opportunity to pick on you, Phil ;-} Has no bearing on an audio amp, unless it was using a tuned circuit (though I can't think why you'd want to, except maybe in some kind of modulator).
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Tue, Sep 18th, 2007 @ 05:43 ( . )

Indeed it will.

It's twigged something in the back of my mind about the loading for audio amps, though, but I can't quite put my finger on it. Is there anything like a tuned circuit in play there? The inductance and resistance of the primary in parallel with the reflected impedance, or something? I'm not that well up on transformer theory. Will have to look it up, I guess.

Again, nothing to do with the matter at hand, of course. Just has me wondering about the way the OT frequency response works.

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:John Hynes (registered user: 2628 posts )
Date: Mon, Sep 17th, 2007 @ 18:00 ( . )

Since RDH4 tends to define things with a general description like that, and then gives functional examples, what I was pointing out was that in addition to that "general class description", the accompanying examples discuss the shift in currents and loads in the halves of the primary of the OT. Obviously, this example was a PP stage.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:CarlB (guest: search)
Date: Mon, Sep 17th, 2007 @ 22:01 ( . )

John,

Here's from the RDH4, chapter 13:

'(ii) Classes of operation
Class A operation* is the normal condition of operation for a single valve, and indicates that the plate current is not cut off for any portion of the cycle.
Limiting Class A push-pull operation* is operation such that one valve just reaches plate current cut-off when the other reaches zero bias.
Class AB operation* indicates overbiased conditions, and is used only in push-pull to balance out the even harmonics.
Class B operation* indicates that the valves (which are necessarily in push-pull) are biased almost to the point of plate current cut-off.
The numeral "1" following A or AB indicates that no grid current flows during any part of the cycle, while "2" indicates that grid current flows for at least part of the cycle. With Class B operation the "2" is usually omitted since operation with grid current is the normal condition.'

Let's take a look at this passage.

1) He doesn’t state the definitions, which is unfortunate. This chapter does not give the definitions, it only gives descriptions for the classes that the author felt useful for clean audio reproduction. From this chapter, many readers derive their working definitions, these working definitions erroneously exclude other valid operating classes.
2) There is nothing there or in the chapter stating or implying that class A is the only condition for a single valve, just that it is the "normal" condition. Normal as considered by Langford-Smith for an AF power amp was to maximize the available clean power for a given plate dissipation. Given that goal, class A would be the normal condition you would conclude. It is not the only normal condition that we need to consider for guitar.
3) Langford-Smith wrote, “Class AB operation* indicates overbiased conditions, and is used only in push-pull to balance out the even harmonics.” He had this right for maximizing clean power: there wasn’t anything else he would have thought to use that for in an audio amp.. So what he wrote, while correct for the section, leads many readers to incorrect assumptions about what can and can’t be class AB. Guitar amps do use SE class AB operation to maximize the overall useable power, they keep “cleanish” sound levels comparable to that of class A operation. Second order distortion levels of maximum “clean” SE class AB are much higher than for maximum SE class A, and do not meet the chapter’s limits for harmonic distortion, so that class is “used only in push-pull …” However, for guitar it sounds at about the same sound level, but just “warmer,” so we do use it. It’s a “normal condition for our application.
4) “The numeral ‘1’ following A or AB indicates that no grid current flows for at least a part of the cycle, while ‘2’ indicates that grid current flows for at least part of the cycle.” Think maximum clean output power. A1 or AB1 means that faithful reproduction is only possible below grid-conduction. The source driving the grid is incapable of sourcing current to continue upwards of Vgk=0. It is possible to grid-clip a class A1 or AB1 stage. This does not mean it changes class, however. Again, while considering maximum clean output power, A2 or AB2 means faithful reproduction is possible above Vgk=0. Later in the chapter he writes, “Some of the distortion occurring with Class AB2 or Class B(2) operation is due to the effect of grid current on the input circuit.” So, A2 and AB2 do accept distortion effects from a non-ideal source (i.e., a non-zero ohm source able to source a limited amount of current).

So, although I can see where you interpreted the author’s intentions a given way, unfortunately it is easy with only his text as a guide to get the wrong working model for class definitions. It’s a shame. So many rely on the RDH4, however it is not the only source that should be used. Other examples that explain class AB and define class AB are out there, and need to be consulted. As an example, do look through the Naval Electronics Engineering Training U.S. Naval electrical engineering training series (NEETS), which is where the following section on class AB is taken from. Notice that its class AB example is specifically given as single ended.:

[link]

Consider: if this information was published by NEETS those many years ago, then if there was a problem with including an SE class AB stage as the example would have been caught many times over by many eyes, and the text would have been corrected.


-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Tue, Sep 18th, 2007 @ 04:52 ( . )

You've stated that very well regarding the section in RDH4, Carl.

And if anyone is having trouble finding a clearly stated-definition, then they need look no further than the one you quoted from NEETS - [link] One of the fascinating things about this one is that it shows an SE transistor stage as the example of a Class AB amplifier!

The main points from the NEETS passage state:

"If the amplifying device is biased in such a way that current flows in the device for 51% - 99% of the input signal, the amplifier is operating class AB."

"Notice that the output signal is distorted. The output signal no longer has the same shape as the input signal. The portion of the output signal that appears to be cut off is caused by the lack of current through the transistor."

"Class AB amplifiers are usually defined as amplifiers operating between class A and class B because class A amplifiers operate on 100% of input signal and class B amplifiers (discussed next) operate on 50% of the input signal. Any amplifier operating between these two limits is operating class AB."

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:John Hynes (registered user: 2628 posts )
Date: Tue, Sep 18th, 2007 @ 08:26 ( . )

OK Zaph,

I've read that NEETS material, as well as the tube-equivalent one cited therein.

I have a several comments:

1) Yes, it does state in clear terms, with examples, that an SE amp can be AB based solely on conduction angle.

2) I have, in my few hours of googling last night turned up many, many articles that share that opinion, and cite that same material. All of them were taken from an RF amplifier standpoint, although I grant you that the NEETS material does not frame itself that way.

3) I have also found many articles, written by prominent audio engineers (most saliently, Eric Barbour comes to mind) who all state in clear terms that load conditions apply to the definition, and hence, an SE amp cannot be AB. These of course are framed solely in the context of AF.

Conclusion: I find definitive material on both sides of the issue. The problem is, who's definition counts? I know from earlier in the thread that at least one individual is doing some research to find the "origin" of the class definitions, and maybe that will be helpful. In the meantime, it seems that it is common in AF-only groups to hold that AB is a PP-only topology, and equally common in RF circles to hold that it can be either.

At any rate, <clack> - that's the sound of me tipping my king. I concede the argument, for the time being. You win. Can we move on now? <LOL>

;)

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Tue, Sep 18th, 2007 @ 09:25 ( . )

Gee, I didn't want you to lose any sleep over it! :o)

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:John Hynes (registered user: 2628 posts )
Date: Tue, Sep 18th, 2007 @ 09:29 ( . )

On 09/18/2007 @ 09:25, Zaphod wrote :
Gee, I didn't want you to lose any sleep over it! :o)
--



Oh I didn't really. My 18-month old son was sick, so sleeping was out of the question anyway. :(

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Dan N (guest: search)
Date: Tue, Sep 18th, 2007 @ 09:58 ( . )

Here is a list of books I have gone through in the past few days.
NOT one of them has any instance of a SE Class-AB audio amplifier.
NEETS is the only instance I have seen.
So Phil your the E.E., surely you must have some text books that have an example of a SE-AB audio amp that you could share with us?
:
ARRL handbook 1936, 1941, 1992
RDH4, RDH3
High fidelity design --Crowhurst-Cooper 1957
RCA-25
Power amplifier --Terman 1955
1938 Standards report --I.R.E.
The Radio handbook --Orr 1959
Valve amplifiers --Jones 2001
Grondslagen van de radiobuizentechniek --Philips Holland 1943


Anyway I'm tired now, so tired, so so tired... I'm going away now to play some guitar...

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Randall Aiken (guest: search)
Date: Tue, Sep 18th, 2007 @ 13:08 ( . )

On 09/18/2007 @ 09:58, Dan N wrote :
Here is a list of books I have gone through in the past few days.
: NOT one of them has any instance of a SE Class-AB audio amplifier.



You won't see any references to SE class AB for audio, especially in older references, as it is not suitable for "normal" audio amplification, because the signal will distort. You will only find references to class AB push-pull, where the other tube "fills in" the missing parts to eliminate the distortion and allow their use for "hi-fidelity" audio reproduction.

Back in the old days, they would think you were crazy for designing an audio amplifier that distorted on purpose! However, in our new "enlightened" times of guitar amplification, we may specifically want to design *for* a specific type of distortion, so an SE class AB stage may be exactly what we want, particularly because it will emphasize the second-order harmonics that you won't get out of a class AB push-pull stage. This is one of the reasons the Soldano "cold-clipper" stage and the original Marshall master volume 10K cathode resistor stages are so well-liked. They aren't biased to class A operation.

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:John Hynes (registered user: 2628 posts )
Date: Tue, Sep 18th, 2007 @ 13:31 ( . )

On 09/18/2007 @ 13:08, Randall Aiken wrote :
On 09/18/2007 @ 09:58, Dan N wrote :
: Here is a list of books I have gone through in the past few days.
: : NOT one of them has any instance of a SE Class-AB audio amplifier.
:
:

:
: You won't see any references to SE class AB for audio, especially in older references, as it is not suitable for "normal" audio amplification, because the signal will distort. You will only find references to class AB push-pull, where the other tube "fills in" the missing parts to eliminate the distortion and allow their use for "hi-fidelity" audio reproduction.
:
: Back in the old days, they would think you were crazy for designing an audio amplifier that distorted on purpose! However, in our new "enlightened" times of guitar amplification, we may specifically want to design *for* a specific type of distortion, so an SE class AB stage may be exactly what we want, particularly because it will emphasize the second-order harmonics that you won't get out of a class AB push-pull stage. This is one of the reasons the Soldano "cold-clipper" stage and the original Marshall master volume 10K cathode resistor stages are so well-liked. They aren't biased to class A operation.
:
: Randall Aiken
--



And this is the reason it seems for conflicting definitions. As you've just said, to get a clean output signal from an AB stage, it must be push-pull, and hence, push-pull becomes part of the definition, from the standpoint of an audio engineer who wishes to create a faithful reproduction of signal. That is the frame of reference that the early "definers" were coming from in the AF engineering category. So it makes sense that Class AB is defined in those terms by a certain group.

So far, it seems that whoever coined these terms is lost somewhere. I suppose an argument can be made that "common usage" should prevail. In that case, both positions are commonly understood.

...and no, I'm not trying to re-argue any specific point. I just think that the argument, and confusion, itself is justified, and that both positions have merit depending on viewpoint. Even if someone were to produce the original documents defining the operational classes, and if that definition states that AB is push-pull only, would you agree? Probably not.

It's not really that important, anyway. We all know what we're talking about when describing these circuits.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Tue, Sep 18th, 2007 @ 13:50 ( . )

My observation is that the audio engineering texts tend to focus purely on Class AB PP, almost to the point of making it appear to be part of the definition, while the radio texts focus more on the conduction angle aspect. The background reasons for that appear to be pretty obvious.

It also looks like both Randall Aiken and I have a background in radio... :o)

On 09/18/2007 @ 13:31, John Hynes wrote :
:
: And this is the reason it seems for conflicting definitions. As you've just said, to get a clean output signal from an AB stage, it must be push-pull, and hence, push-pull becomes part of the definition, from the standpoint of an audio engineer who wishes to create a faithful reproduction of signal. That is the frame of reference that the early "definers" were coming from in the AF engineering category. So it makes sense that Class AB is defined in those terms by a certain group.
:
: So far, it seems that whoever coined these terms is lost somewhere. I suppose an argument can be made that "common usage" should prevail. In that case, both positions are commonly understood.
:
: ...and no, I'm not trying to re-argue any specific point. I just think that the argument, and confusion, itself is justified, and that both positions have merit depending on viewpoint. Even if someone were to produce the original documents defining the operational classes, and if that definition states that AB is push-pull only, would you agree? Probably not.
:
: It's not really that important, anyway. We all know what we're talking about when describing these circuits.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Boiling it all done, Are the rea question to be answered'
Author:Bryan James (registered user: 450 posts )
Date: Tue, Sep 18th, 2007 @ 10:41 ( . )

let's face it, all these definitions, what really matters to the average internet surfer looking for an amp is. "If I plug in my detuned 7 string guitar, can i sound like jimmy page on the first led zeppelin CD? after all they sound killer on mp3 that i've downloaded on my computer speakers." other wise I think we should just leave the SE AB to the rf area.

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Francis Vaughan (registered user: 712 posts )
Date: Mon, Sep 17th, 2007 @ 10:07 ( . )

I really wanted to stay quiet. But whatever.

Indeed - a very useful discussion, and useful from the point of view of eliciting thoughts on how the various design techniques and sonic attributes can be described.

Just to muddy the waters a tiny bit. Class AB PP comes about for a very specific reason, which touches on a problem with the very dry definitions of class. The only reason to use AB other than B for a PP amp is because of the non-linearity at turn on. So one biases the amp so that the commutation between sides is pushed to much higher signal levels - where as a proportion of the total signal the distortion due to commutation error is much smaller - and thus represents a lower percentage of distortion. A pure class A PP amp biases itself to the point where the commutation never occurs during normal operation (in principle pushing it to the point where one side reaches cut off at the same time the other reaches saturation.) However it is clear that this use of the terminology is very much dependant upon the load. This is because saturation is dependant upon load, because you will either run out of power supply capacity or current carrying capacity. This yields a parallel set of internet forum arguments in the HiFi community about the definition of true class A, and the notion that a class AB amp transitions from class A to class AB as the signal level rises. This argument hinges on whether the signal level is large enough to push one side or the other into cut off - and hence give rise to the commutation related distortion. Consensus n those forums is that class A PP is only valid for a given range of loads. But this is also usually a semiconductor argument, so they don't care about loads in the same way.

OK, that said, I do feel that there is enough weasel room in the definitions to make it worthwhile trying to nail down what we mean, not in some way that evokes another round of 'tis 'tisn't argument about A, AB, AB1 AB2, but about the sonic benefits.

Adding "Hot" and "Cold" to AB helps. This is important because there is one place where the conventional definitions don't help enough. There has been much talk about conduction angle. But all of the definitions implicitly assume that the loss of conduction occurs at cut-off. So where a definition talks about an absolute angle, there is no notion of phase. Which isn't good. We know that the phase matters, the transfer function at saturation is significantly different from the transfer function at cut-off.

It has been implicitly suggested that the term Hot-AB be used to describe a stage biased so that it distorts mainly due to saturation, and Cold-AB a stage that distorts mainly due to cut-off. Then perhaps simple class A is one that either is intended to stay linear, or have roughly equal proportions of distortion components come from both saturation and cut-off. This is an interesting mode, since it is linear for two disjoint segments. It doesn't really meet the spirit of class A definition, at least not when distorting - and that is when we are interested. One very much doubts the inventors of the class names ever thought of that one. Sure it is less than 360, but two disjoint segments, both less than 180 degrees with a saturated lump in the middle?

Then there is the issue of grid clipping. There seems to be no good name for that. Perhaps because it generally yields an unpleasant distortion, so there is no desire to name it, just eliminate it. Blocking distortion covers at least some of the ground, but not all. And that is a condition, not a design mode.

So back to where I came in much earlier. How about a challenge? Take a few well known amps, including a SLO and a Wreck perhaps, and describe meaningfully the design, in terms of these classes, in a form that would allow someone to pretty much understand what is going on.

And a further challenge, describe the Melissa, and Zoe and Iain's Blues pre-amp in the same manner.


-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 10:12 ( . )

I agree with you that it does not matter whether one is a distorting guitar amp, and another is a linear HiFi amp, and another is an RF amp. In fact that's exactly what I'm saying.

And furthermore, along with yourself, CarlB and Randall Aiken, I'm saying that the original and fundamental definitions still very much *do* apply. Having to operate in push-pull never was one of them. That was and is an additional practical requirement dictated by certain specific applications of a Class AB (or Class B) amplifier.

I can see no solid engineering reason anywhere that says all Class AB amplifiers have to be PP.

On 09/17/2007 @ 08:54, John Hynes wrote :
:
: A circuit is a circuit with respect to the definitions laid out for these classes however many years ago they were penned. The fact that one is a distorting guitar amp, and another is a linear HiFi amp, and another is an RF amp, DOES NOT MATTER. Certain definitions were created, they were given clear, uncomplicated meaning, and they are what they are.
:
: If a certain circuit does not fall within those boundaries, you need to create a new definition of what you've got, not re-appropriate another definition and change its meaning.
:
: Here is where we seem to disagree: I think (as do several others, Randall has said explicitly at least) that the old definitions do indeed apply to these circuits. To me, if that original definition of "AB" doesn't take into account your so-called modern usage of a single ended circuit that has one aspect of AB behavior, but not *all* of them, as envisioned by those who coined the definitions, THEN IT IS NOT AN AB AMPLIFIER. I think its an A amplifier driven into clipping. You disagree, so maybe you ought to call it something else, maybe "Class Az" for "A, Zaphod's Way" or something - but it isn't AB. Can't be. The definition of AB includes the characteristics of the load, which require P-P operation.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Do they care?'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Mon, Sep 17th, 2007 @ 22:42 ( . )

I'm with you on this.

I do think, though, that there's value in pushing terms that are more important to players who are interested in amp tone, such as "warm biased" and "cold biased".

The average guitar player only *thinks* they care of their amp is Class A or not. But there are things they do care about, and I think there's a lot of value in getting useful terminology out there.

Class A is a useful term for us (assuming everyone finally understands what it means 8^) as amp designers and builders, but that's about it.

-- REPLY: [With No Quote] --- [With Quoted Text]

'The crux of the matter'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Mon, Sep 17th, 2007 @ 09:58 ( . )

Guess I wasn't clear.

I don't want to replace the terms, I think we may want to add new ones.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:John Hynes (registered user: 2628 posts )
Date: Sun, Sep 16th, 2007 @ 14:16 ( . )

Actually, I *must* be confused, as I just reviewed what you wrote in your technical article on blocking distortion,where it says:

"You will also get more power out of the output stage because it is now running in class AB2 or class A2 (the "2" suffix indicates grid current flows for a portion of the cycle)."

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Randall Aiken (guest: search)
Date: Sun, Sep 16th, 2007 @ 15:15 ( . )

On 09/16/2007 @ 14:16, John Hynes wrote :
Actually, I *must* be confused, as I just reviewed what you wrote in your technical article on blocking distortion,where it says:
:
: "You will also get more power out of the output stage because it is now running in class AB2 or class A2 (the "2" suffix indicates grid current flows for a portion of the cycle)."
:
: -John
:
--

If you read that carefully, you will note that prior to that statement I said the following:

"Add a DC-coupled cathode follower between the phase inverter and the grid of the output tubes, with the cathode follower cathode resistor returned to a high negative voltage, and the grid bias applied to the grid of the cathode follower. This effectively isolates the output tube grid circuit from the phase inverter and its associated AC coupling, and provides a very low impedance source for the output stage. This will prevent the output stage from going into grid clamp, and will eliminate the long time constant of the AC coupling."

The elimination of the AC-coupling and addition of the DC-coupled cathode follower allows you to supply drive current to the grids of the output stage, which allows you to drive them positive, which then converts the output stage to a class AB2 output stage. Grid current will flow (supplied by the DC-coupled cathode follower) the entire time the grids are above zero volts, so you are now operating in the positive grid region where grid current flows as defined by the "2" suffix.

There is a whole world above that "Vg=0" curve on the plate characteristic curves, but it takes power to drive the tubes there. AC-coupled grid clamping is not it.


-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:John Hynes (registered user: 2628 posts )
Date: Sun, Sep 16th, 2007 @ 15:26 ( . )

On 09/16/2007 @ 15:15, Randall Aiken wrote :
On 09/16/2007 @ 14:16, John Hynes wrote :
: Actually, I *must* be confused, as I just reviewed what you wrote in your technical article on blocking distortion,where it says:
: :
: : "You will also get more power out of the output stage because it is now running in class AB2 or class A2 (the "2" suffix indicates grid current flows for a portion of the cycle)."
: :
: : -John
: :
: --
:
: If you read that carefully, you will note that prior to that statement I said the following:
:
: "Add a DC-coupled cathode follower between the phase inverter and the grid of the output tubes, with the cathode follower cathode resistor returned to a high negative voltage, and the grid bias applied to the grid of the cathode follower. This effectively isolates the output tube grid circuit from the phase inverter and its associated AC coupling, and provides a very low impedance source for the output stage. This will prevent the output stage from going into grid clamp, and will eliminate the long time constant of the AC coupling."
:
: The elimination of the AC-coupling and addition of the DC-coupled cathode follower allows you to supply drive current to the grids of the output stage, which allows you to drive them positive, which then converts the output stage to a class AB2 output stage. Grid current will flow (supplied by the DC-coupled cathode follower) the entire time the grids are above zero volts, so you are now operating in the positive grid region where grid current flows as defined by the "2" suffix.
:
: There is a whole world above that "Vg=0" curve on the plate characteristic curves, but it takes power to drive the tubes there. AC-coupled grid clamping is not it.
:
:

--



Then my understanding of the "2" suffix is indeed correct.

You are saying that in an A1/AB1 stage, you can drive it to clipping. Since it is AC coupled, it cannot draw grid current. It gets clamped at Vg=0. This is, by definition, A1/AB1 operation, no grid current available or drawn.

In A2/AB2, you have coupled to the output with a low impedance source with a DC path. When driven past the Vg=0 curve, it attempts to draw current at the grid, and succeeds. Thus, it is A2/AB2, as grid current is drawn.

The way you are explaining this seems self-contradictory, though I don't think on purpose. If you actually disagree with what I've said above, then please explain how a stage can be classified by the "2" suffix without grid current. I understand that clipping isn't enough, I agree fully with that statement.

To summarize, if a low impedance DC source is provided, and the stage is biased so it can be driven past Vg=0, it will attempt to draw grid current and succeed. It is thus A2/AB2. If AC coupled, and there is no current path provided and the driving impedance is high, when driven to the point where Vg=0 is reached, the stage approaches a point where grid conduction would occur, but fails, and clamping occurs. There is no grid current drawn. It is thus A1/AB1. Clipping has nothing to do with it. The presence of grid current does. Correct?

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Randall Aiken (guest: search)
Date: Sun, Sep 16th, 2007 @ 15:51 ( . )

On 09/16/2007 @ 15:26, John Hynes wrote :

: Then my understanding of the "2" suffix is indeed correct.
:
: You are saying that in an A1/AB1 stage, you can drive it to clipping. Since it is AC coupled, it cannot draw grid current. It gets clamped at Vg=0. This is, by definition, A1/AB1 operation, no grid current available or drawn.
:
: In A2/AB2, you have coupled to the output with a low impedance source with a DC path. When driven past the Vg=0 curve, it attempts to draw current at the grid, and succeeds. Thus, it is A2/AB2, as grid current is drawn.
:
: The way you are explaining this seems self-contradictory, though I don't think on purpose. If you actually disagree with what I've said above, then please explain how a stage can be classified by the "2" suffix without grid current. I understand that clipping isn't enough, I agree fully with that statement.
:
: To summarize, if a low impedance DC source is provided, and the stage is biased so it can be driven past Vg=0, it will attempt to draw grid current and succeed. It is thus A2/AB2. If AC coupled, and there is no current path provided and the driving impedance is high, when driven to the point where Vg=0 is reached, the stage approaches a point where grid conduction would occur, but fails, and clamping occurs. There is no grid current drawn. It is thus A1/AB1. Clipping has nothing to do with it. The presence of grid current does. Correct?
:
: -John
:
--



I'm not sure what we are agreeing/disagreeing with. I didn't say anything about being able to have a "2 suffix" output stage that doesn't draw grid current.

You can drive any amp to clipping, whether it is class AB1 or AB2. The point is that you cannot drive the grids positive in the usual hi-Z AC-coupled class AB1 output stage. If the output stage were true class AB2, it would allow you to drive the grids several volts positive *without* clipping or clamping (which is how you get more output power than class AB1), and the driver stage would have to be able to supply this current while doing so.

When driving an AB1 amp to clipping, the grids clamp at the peaks, and, yes, a tiny amount of grid current flows during the peaks to keep them there (this is necessary to charge the coupling cap to the clamp point), but it is relatively small in comparison the amount of drive current it takes to make a true class AB2 output stage, and the output grids are not driven into the positive grid region, they typically clamp right at, or slightly below, the zero volt point.

A class AB2 stage requires the output tube grids to be driven positive for an appreciable portion of the cycle, and the grid clamping of a class AB1 stage does not allow this. Class AB1 amps are not operating in the positive grid region at all, they are just clipping the peaks due to the grid clamping, and are thus a clipped AB1 output stage, not an AB2 output stage.

Put simply, a class AB2 amp drives the grids positive, where they draw current, and the amplifier continues to function in a linear manner for an extended portion of the curves, and a class AB1 amp is not designed to drive the grids positive, so it cannot produce any more clean power after it hits the zero volt grid curve.

If your only criterion for class AB2 is any amount of grid current flow at all, then all amps would be class AB2 due to grid leakage. ;)

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:John Hynes (registered user: 2628 posts )
Date: Sun, Sep 16th, 2007 @ 16:00 ( . )

OK What you just said there was perfectly clear, and I completely agree with you.

Just for the record, I will state that my "take" on the definition of what the "2" suffix means is based on grid current draw in the positive grid bias region. If the grids are positive, and grid current flows, it's "2". This is in line with what you've said, and no, I don't count leakage current. :)

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Randall Aiken (guest: search)
Date: Sun, Sep 16th, 2007 @ 16:33 ( . )

On 09/16/2007 @ 16:00, John Hynes wrote :
OK What you just said there was perfectly clear, and I completely agree with you.
:
: Just for the record, I will state that my "take" on the definition of what the "2" suffix means is based on grid current draw in the positive grid bias region. If the grids are positive, and grid current flows, it's "2". This is in line with what you've said, and no, I don't count leakage current. :)
:
: -John

Yes, I think we are on the same page,then. :)

Randall Aiken
:
--


-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Dan N (guest: search)
Date: Sun, Sep 16th, 2007 @ 17:29 ( . )

The way Randall explains it is pretty much the way the ARRL handbook explains it and that is the way I understand it.

But, What I don't quite understand is what currents (as in flow of electrons) do you include or exclude to determine suffix "1" or "2" ?
Even in an AC-coupled grid, for a change in potential to occur some electrons must flow.
Also the inter-electrode capacitances of the valve need to be charged/discharged for a change in potential on the electrodes to occur.
So do we not include those currents?

-- REPLY: [With No Quote] --- [With Quoted Text]

'Definitions of terms, Carl?'
Author:Randall Aiken (guest: search)
Date: Sun, Sep 16th, 2007 @ 17:56 ( . )

On 09/16/2007 @ 17:29, Dan N wrote :
: But, What I don't quite understand is what currents (as in flow of electrons) do you include or exclude to determine suffix "1" or "2" ?
: Even in an AC-coupled grid, for a change in potential to occur some electrons must flow.
: Also the inter-electrode capacitances of the valve need to be charged/discharged for a change in potential on the electrodes to occur.
: So do we not include those currents?
:
--



No, you don't include coupling cap charging currents or electrode currents. The whole idea of class A2/AB2 was to describe an amplifier topology designed to be driven into the positive grid region for higher output power. The grid then conducts a large current, which must be supplied by a specially-designed driver stage.

Driving an AB1 amp to the edge of the zero volt curve where it hits grid clamp is not what they had in mind...

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Sat, Sep 15th, 2007 @ 12:31 ( . )

On 09/15/2007 @ 09:00, CarlB wrote :
: Class A means you'll hit either grid clipping (class A1) or saturation (class A2) before you hit cutoff or at the same you hit cutoff.


That's not quite what I meant. I meant; where does the rule book say that in class A, it can't hit cut-off FIRST? in exactly the same way as it hits grid-current or saturation first, when biased warm.
Fact is, it doesn't say it anywhere, which is why a cool biased SE stage is still class A.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sat, Sep 15th, 2007 @ 14:28 ( . )

"... where does the rule book say that in class A, it can't hit cut-off FIRST? in exactly the same way as it hits grid-current or saturation first, when biased warm.
: Fact is, it doesn't say it anywhere, which is why a cool biased SE stage is still class A."


If it is in class A, then by definition, the plate current must conduct for the full output signal at the power rating of the amp. The power rating of a class A stage is given before clipping or saturation, same as for a class AB stage. The difference between the two is that the A stage must conduct full waveform at the point before grid-clipping or saturation, whereas the class AB does not.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Sun, Sep 16th, 2007 @ 09:24 ( . )

On 09/15/2007 @ 14:28, CarlB wrote :

: If it is in class A, then by definition, the plate current must conduct for the full output signal at the power rating of the amp.

but the power rating of the amp is stated by the designer, not by the circuit. If I build an amp CAPABLE of 100W, but it is only INTENDED to be driven to 10W, then it's a 10W amp. And if it so happens that it's biased cool class A but never hits cutoff before 10W, then it's a class A 10W amp.

The power rating of a class A stage is given before clipping or saturation. The A stage must conduct full waveform at the point before grid-clipping or saturation, whereas the class AB does not.


I don't believe this is true. The power rating of ANY class is given BEFORE CLIPPING. It doesn't matter whether that is clipping due to grid-curret, cutoff, or anything else for that matter. A PP class AB amp of course doesn't clip due to cutoff (before grid current), thanks to it's partner valve, which is a special case worthy of a special designation (AB). Whereas an SE stage WILL clip at cutoff- so it's still class A, nothing more.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 16th, 2007 @ 09:34 ( . )

Sorry, everybody. We STILL agree with Merlin on this one. Not that it really matters, of course.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Derek H (registered user: 105 posts )
Date: Sun, Sep 16th, 2007 @ 09:50 ( . )

So you are agreeing that an EL84 SE amp cannot ever be biased into class AB, or did I lose the thread of the discussion?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 16th, 2007 @ 10:00 ( . )

Yup. The crux of our argument is that if the signal is clean, it's still operating in Class A. Once it starts to clip, it's operating in exactly the way 1/2 of a PP AB amp does. But it still isn't IN AB, because the class definitions apply only to linear amps...those that are delivering a clean signal to the speaker. When one half of the signal through each output device in an AB PP amp is being compressed/clipped, it is still capable of outputting a clean signal. A SE amp is not. So it's not strictly operating in AB.

As I've been saying all along, it doesn't matter, though. The design and result is the same whether you believe you're in AB or not at that point. This is just a discussion of definition of terms, nothing more.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 16th, 2007 @ 14:02 ( . )

But a Class A amp that's being overdriven so hard that it clips is NOT the same as an amp (SE or PP) that is *designed* to go into cutoff for part of the cycle. That is my point. Clipping and cutoff are two completely different things, even if to the layperson they may appear to be similar.

Now whether or not the amp is clean at the output or not is somewhat subjective issue, with criteria that will vary between different applications - eg even "clean" in RF amp, guitar amp terms or clean in hifi terms aren't quite the same.

And if you take an RF SE Class AB amp, the amplifying element in it works in exactly the same was as in a Class AB SE guitar amp (and without any change in load impedance). Although the RF amp uses a resonant tank to recreate the sine wave afterwards, that in no way detracts from the fact that the amplifying element in both are working in exactly the same way - Class AB.

But of course you're right that at the end of the day all the guitarist cares about is how the amp sounds. :o)

On 09/16/2007 @ 10:00, Zoe N Iain wrote :
:
Yup. The crux of our argument is that if the signal is clean, it's still operating in Class A. Once it starts to clip, it's operating in exactly the way 1/2 of a PP AB amp does. But it still isn't IN AB, because the class definitions apply only to linear amps...those that are delivering a clean signal to the speaker. When one half of the signal through each output device in an AB PP amp is being compressed/clipped, it is still capable of outputting a clean signal. A SE amp is not. So it's not strictly operating in AB.
:
: As I've been saying all along, it doesn't matter, though. The design and result is the same whether you believe you're in AB or not at that point. This is just a discussion of definition of terms, nothing more.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 16th, 2007 @ 10:05 ( . )

Both of your statements appear to be on shaky ground. Marshall builds a number of 50W valve amps, which when measured give out somewhat over 60W RMS clean. In this case Marshall states it to be a 50W amp, but the facts clearly show otherwise. The amplifier is capable of 60W and will put out 60W, despite what the designer says it's intended to produce.

On your second point, this appears to be an assumption on your own part, as to what the definition of "before clipping" is. First of all you are deliberately confusing clipping with cutoff. They are different things and should not be confused with each other, even though the traditional audio or hi-fi designer might consider to consider both to be equally undesirable in an SE amp. But here we're not dealing with traditional audio or hifi. And again looking at the SE RF amplifier case, the power valve does certainly go into cutoff for part of the cycle. The fact that the resonant tank flywheels enough to keep the final output to the antenna looking like a sine wave, doesn't change the fact that the stage is indeed working in SE Class AB (and experiences no load impedance change).

Or consider the example of a typical Class AB PP amp, running at maximum clean amplitude and not clipping. The fact that the pair(s) of power valves are operating in Class AB is set by the combination of anode and screen voltages, load impedance and bias. Now if you remove the power valve(s) from one side of the PP amp, it doesn't suddenly turn into a class A amp. It will still be in cutoff for part of its normal operating cycle. The main difference is that since only half the OT primary is in use the entire time, there won't be the change in load impedance at the transition between the Class A and Class B operating regions.

On 09/16/2007 @ 09:24, Merlin wrote :
:
: but the power rating of the amp is stated by the designer, not by the circuit. If I build an amp CAPABLE of 100W, but it is only INTENDED to be driven to 10W, then it's a 10W amp. And if it so happens that it's biased cool class A but never hits cutoff before 10W, then it's a class A 10W amp.
:
: The power rating of ANY class is given BEFORE CLIPPING. It doesn't matter whether that is clipping due to grid-current, cutoff, or anything else for that matter. A PP class AB amp of course doesn't clip due to cutoff (before grid current), thanks to it's partner valve, which is a special case worthy of a special designation (AB). Whereas an SE stage WILL clip at cutoff- so it's still class A, nothing more.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sun, Sep 16th, 2007 @ 10:48 ( . )

"The power rating of ANY class is given BEFORE CLIPPING."

Agree. As such, a Class A amplifier must conduct plate current for the full 360 degrees, all the way up to grid-conduction for A1, or saturation for A2.

"It doesn't matter whether that is clipping due to grid-curret, cutoff, or anything else for that matter."

Don't agree. Cutoff is a different condition, due to less than 360 degree conduction. If it is not clipping due to grid conduction or saturation, but it is in cutoff, then it is incapable of supplying 360 degrees of conduction for the output signal. By definition, it is a class AB1 or AB2 stage.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sun, Sep 16th, 2007 @ 12:00 ( . )

Well, to be as accurate and complete as I can be:

The maximum clean power is determined right at the onset of grid-clipping, saturation, or cutoff.

At that point the output signal is still following the input signal. Beyond that point, it does not, and so the amp is no longer clean.

So, right at the last point of the stage being able to follow its input signal is where class determination is made.

That is why the onset of grid-clipping, saturation, or cutoff is used to determine the class of operation.

Below onset of grid-clipping or saturation, a class AB stage can conduct for 360 degrees. It is not a class A stage however. To be a class A stage, it must be able to conduct for 360 degrees of the output signal - right up to grid-clipping (A1) or saturation (A2).

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Sun, Sep 16th, 2007 @ 12:56 ( . )

On 09/16/2007 @ 12:00, CarlB wrote :
I agree with you right up to this point:
Below onset of grid-clipping or saturation, a class AB stage can conduct for 360 degrees. It is not a class A stage however. To be a class A stage, it must be able to conduct for 360 degrees of the output signal - right up to grid-clipping (A1) or saturation (A2).

If an SE stage produces a clean output, in MUST be conducting 360 degrees of the input waveform. It doesn't matter whether it hits cutoff of grid current first, it's still 360 degrees until one side or other clips.
An AB amp continues to produce a clean output IN SPITE of each valve NOT conducting for 360 degrees. As you well know, that can ONLY happen in PP. Hence it has it's own class distinction; AB, which CANNOT apply to SE.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sun, Sep 16th, 2007 @ 18:08 ( . )


"An AB amp continues to produce a clean output IN SPITE of each valve NOT conducting for 360 degrees. As you well know, that can ONLY happen in PP. Hence it has it's own class distinction; AB, which CANNOT apply to SE."

That first sentence is true of PP class AB stages. But push-pull being the exclusive recipient of class AB distinction does not follow from that thought.

Picture an SE stage, colder biased. It amplifies smallish to mediumish signals quite faithfully. Now, push the input signal just large enough so that the plate current touches zero at the bottom of the output signal before top clipping or saturation occurs. The stage is no longer providing the full 360 degree plate current and is at its limits of faithful reproduction. By definition, this is class AB.


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 02:40 ( . )

I think there are several of us that disagree that the engineering definition of a Class AB amp is that it must produce a distortion-free clean output. From a strictly engineering perspective class is a function of how the valve (transistor, FET, etc) is set up to operate in a given circuit.

Whether or not you require a clean output (or even whether you use a resonant tank afterwards to clean the output) is related to the *application* you want to use the amplifier for. It's not universal, and therefore can't form part of the fundamental definition of amplifier class.

And once again insist that the SE stage do something that only a PP amp can do, and then use that as "proof" that an SE amp can't operate in anything but Class A is in fact a circular argument. In reality you're just proovong that an SE amp can't be a PP amp!

On 09/16/2007 @ 12:56, Merlin wrote :
:
: An AB amp continues to produce a clean output IN SPITE of each valve NOT conducting for 360 degrees. As you well know, that can ONLY happen in PP. Hence it has it's own class distinction; AB, which CANNOT apply to SE.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Balijukka (guest: search)
Date: Sun, Sep 16th, 2007 @ 10:20 ( . )

I think that's the question we have decide.

Whether the output signal is allowed to be clipped (from any end) when we do the measuring or not.

What I don't agree is Carl's clipping meaning only grid clipping or saturation. The famous Soldano clipper is usually referred as a cold clipper, instead of a saturation clipper.

Carl's classes do make sense are very clear and acceptable, but then we have to dump the clean output requirement.

This also implements that the amplifier classes are just a matter of biasing .

Anyway I think we all know what is meant if someone says :"my input triode is in class AB". :)

jukka

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:CarlB (guest: search)
Date: Sun, Sep 16th, 2007 @ 11:48 ( . )

"I think that's the question we have decide.
:
: Whether the output signal is allowed to be clipped (from any end) when we do the measuring or not."

'Clipping" at the bottom end is due to cutoff. Cutoff is used for determination of less than 360 degree conduction. As such, it is a discriminator between class A or AB operation.

Grid conduction clipping is an indication of class A1 or AB1 operation. The stage is either capable of supplying 360 degree conduction before grid-clipping (class A1) or it is not (class AB1).

Saturation is is an indication of class A2 or AB2 operation. The stage is either capable of supplying 360 degree conduction before saturation (class A2) or it is not (class AB2).

"What I don't agree is Carl's clipping meaning only grid clipping or saturation. The famous Soldano clipper is usually referred as a cold clipper, instead of a saturation clipper."

Calling that stage a clipper is OK for getting across the idea of what it does to the signal, but it is not formally correct. It is actually a "cold-cutter-offer,' most wouldn't have a clue to what you are talking about. So "cold clipper" is accepted guitarista/techie vernacular. But it is not formally correct, and even misleading, toward understanding class determination.

"Carl's classes do make sense are very clear and acceptable, but then we have to dump the clean output requirement."

These are not my classes. These are the accepted practice for determining class of operation.

You do not lose the clean output requirement. The amp class determination happens at the boundary of what happens just at maximum clean power: grid-clipping or saturation first (class A), grid-clipping or saturation and cutoff at the same time (class A), or cutoff first (class AB).

"This also implements that the amplifier classes are just a matter of biasing."

That is correct. You can bias any stage to be class A or class AB. If you hit maximum plate dissipation before you achieve class A (and you want class A), then you must reduce the plate voltage to allow you to bias the stage hotter, toward achieving that class A operation.

"Anyway I think we all know what is meant if someone says :"my input triode is in class AB". :)"

Probably. But I really hope to put any notion that statement is technically incorrect to rest. It is a correct description, according to accepted definitions, if you intend to convey that the stage hits cutoff before grid-clipping (AB1) or saturation (AB2).

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Balijukka (guest: search)
Date: Mon, Sep 17th, 2007 @ 05:40 ( . )

On 09/16/2007 @ 11:48, CarlB wrote :
: You do not lose the clean output requirement. The amp class determination happens at the boundary of what happens just at maximum clean power: grid-clipping or saturation first (class A), grid-clipping or saturation and cutoff at the same time (class A), or cutoff first (class AB).


This is where I think you are wrong.

At maximum clean output, a SE stage conducts 360 degrees. so it is class A. It does not matter what it would do or where it would clip after you drive it a bit harder.

In a true class AB, as it is in p-p amp, at maximum clean output, both tubes go quite deep in the cut-off well before we hit the clipped output.

jukka

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Mon, Sep 17th, 2007 @ 06:58 ( . )

That's my definition of it, too.

I think the point is that it's the 'shape' of the output from the amp that matters, which includes the OT in it's circuit as an anode load. Not the shape of the output signal from the active device itself. That's just a valve. It's not an amp stage, because you need the supporting components as well before you can call it that.

However, as has been said above, you can use a tank circuit to 'fill in' the rest of the cycle. And that would be a SE AB amp. Without it, though, it's class A.

I don't think it really matters for any real application which you hold to be true, though. The circuit will be the same, it's only how you describe it that is different. And we all know what is meant by 'Class AB SE', whether we think it's an accurate definition when applied to an audio amp, or not. So it's not a big deal.

FWIW, Iain and I both think SE 'AB' PAs sound better than 'smack dab in the middle' Class A ones, too. Just in case anyone was thinking we didn't think it was the 'right' way to build one, or something. It's only the terminology we disagree on. And we might well be wrong, wouldn't be the first time. ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Dan N (guest: search)
Date: Mon, Sep 17th, 2007 @ 07:54 ( . )

On 09/17/2007 @ 06:58, Zoe N Iain wrote : I think the point is that it's the 'shape' of the output from the amp that matters, which includes the OT in it's circuit as an anode load. Not the shape of the output signal from the active device itself. That's just a valve. It's not an amp stage, because you need the supporting components as well before you can call it that.
From what I can gather, it is not always clear in the definitions whether they are talking about the active device alone or the whole unit itself. Though the definitions do include voltage amplifiers, current amplifiers and power amplifiers (both current and voltage).
However an active device (valve or semiconductor) placed between two electrical points of potential difference is merely a variable resistor that impeads the flow of electrons.
Without a load to work upon it cannot amplify any signal. so I think we should include the load into the equation.
But the points of measurement needs to be taken at the nodes of the active device, as not all amplifiers have an OT.

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Stephen Keller (registered user: 6220 posts )
Date: Mon, Sep 17th, 2007 @ 11:26 ( . )

On 09/17/2007 @ 07:54, Dan N wrote :
On 09/17/2007 @ 06:58, Zoe N Iain wrote : I think the point is that it's the 'shape' of the output from the amp that matters, which includes the OT in it's circuit as an anode load. Not the shape of the output signal from the active device itself. That's just a valve. It's not an amp stage, because you need the supporting components as well before you can call it that.
From what I can gather, it is not always clear in the definitions whether they are talking about the active device alone or the whole unit itself. Though the definitions do include voltage amplifiers, current amplifiers and power amplifiers (both current and voltage).
: However an active device (valve or semiconductor) placed between two electrical points of potential difference is merely a variable resistor that impeads the flow of electrons.
: Without a load to work upon it cannot amplify any signal. so I think we should include the load into the equation.
: But the points of measurement needs to be taken at the nodes of the active device, as not all amplifiers have an OT.
:
--

Sigh....

I have found a reference by Eastman (Fundamentals of Vacuum Tubes) that lists the class definitions as recorded in the 1938 Standards Report of the Institute of Radio Engineers. The IRE is the predecessor to the IEEE and was the standards body that defined these terms originally.

Would it be helpful to this discussion if I were to scan those so we can all see what the original authors of the class definitions were thinking?

Stph

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Dan N (guest: search)
Date: Mon, Sep 17th, 2007 @ 17:10 ( . )

On 09/17/2007 @ 11:26, Stephen Keller wrote :: Would it be helpful to this discussion if I were to scan those so we can all see what the original authors of the class definitions were thinking?
Stephen, I am interested in reading those papers, it would be helpful.


BTW, the ARRL handbooks 1939 [link] and 1941 [link] are at Pete's site.
*Caution huge file sizes*

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Mon, Sep 17th, 2007 @ 22:44 ( . )

Please, please, please!!!

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Stephen Keller (registered user: 6220 posts )
Date: Mon, Sep 17th, 2007 @ 23:15 ( . )

[link]

-- REPLY: [With No Quote] --- [With Quoted Text]

'CLEAN ot NOT'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Mon, Sep 17th, 2007 @ 22:48 ( . )

IIRC, the definitions go something like, "An amplifier is said to be operating in Class A if..."

If so, the definitions would almost certainly be referring to circuits, not merely components (tubes).

A tube is not typically an amplifier, a tube circuit may be. And two tubes don't make a push-pull amp by any stretch of the imagination. They're a good start, I'll admit.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Sat, Sep 15th, 2007 @ 02:40 ( . )

On 09/14/2007 @ 22:42, CarlB wrote :: The definition for a class A amp is that the output conduction current conducts for all 360 degrees at the maximum signal the amp can produce cleanly. Class A means you'll hit grid clipping (class A1) or saturation (class A2) at the same time or before you hit cutoff.

Carl, I'm not sure if I understand your description of A1 & A2 properly?
So here is my take on it, and please correct me if I'm not correct?

Class-A1 amplifier has its (control) grid always biased negative, even with the largest signal to be seen by the grid. There is no grid conduction and there is no power lost in the grid circuit and poses a very high impedance.
Therefore, as an example, a 12AX7 anode follower voltage amplifier is quit capable of driving the grid of a EL34, KT88 etc.

Class-A2 amplifier is labelled so, if at some point during the AC cycle the grid is made positive and current flows in the grid (grid conduction). As a consequence power is then taken up in the grid circuit, the impedance drops rapidly as the voltage at the grid is further increased. Therefore to maintain the power in the grid circuit, because we now have voltage and current, we need to use a power amplifier to drive the grid. Otherwise distortion of the wave form will occur. The 12AX7 would not be able to cope with such a load.
Class-A2 is a way of increasing the power output of the valve.

Anyway that's about it. I'm sure I've left something out, need to do a bit more reading...


P.S. Jukka did you feel the earthquake the other day ? hope your OK ;)

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sat, Sep 15th, 2007 @ 09:54 ( . )

"Class-A1 amplifier has its (control) grid always biased negative, even with the largest signal to be seen by the grid. There is no grid conduction and there is no power lost in the grid circuit and poses a very high impedance."

You have the concept correct, with a slight point of clarification:

There can be grid conduction in class A1 or class AB1 stage, however it clips the output when grid conduction is reached. So you look at the operational class (A1 or AB1) with regards to what happens first, cut-off (AB1), grid-clipping (AB1), or both at the same time (AB1).

"Therefore, as an example, a 12AX7 anode follower voltage amplifier is quit capable of driving the grid of a EL34, KT88 etc."

I'm not sure, I'd probably use at least a 12AT7 anode or cathode follower to muscle a power tube grid. But maybe a 12AX7 an pull off the job.

"Class-A2 amplifier is labelled so, if at some point during the AC cycle the grid is made positive and current flows in the grid (grid conduction). "

There can be grid conduction in a class A1 or AB1 amp, it's just that the stage clips when this occurs. In a class A2 or AB2, it does not clip.

"As a consequence power is then taken up in the grid circuit, the impedance drops rapidly as the voltage at the grid is further increased. Therefore to maintain the power in the grid circuit, because we now have voltage and current, we need to use a power amplifier to drive the grid. Otherwise distortion of the wave form will occur. The 12AX7 would not be able to cope with such a load."

I agree with you, I don't think a 12AX7 could have enough power to muscle the grid past grid-conduction.

"Class-A2 is a way of increasing the power output of the valve."

You'd think correct, but ... If you're already biased at the maximum plate dissipation and maximum clean power output, you can't take advantage of the extra A2 drive. The stage will go into cutoff first. and then the amp should be classified AB2.

To take advantage of class A2 toward maximizing clean power, you have to lower the plate voltages so that you can bias the plate with more current (when compared to a class A1 stage).

Class AB2 is another matter. Let the bias stay cold, muscle the grids, you still hit cutoff before clipping or saturation.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sat, Sep 15th, 2007 @ 09:55 ( . )

Ack. Here's a correction:

"So you look at the operational class (A1 or AB1) with regards to what happens first, cut-off (AB1), grid-clipping (A1), or both at the same time (A1)."


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Sat, Sep 15th, 2007 @ 17:18 ( . )

Carl, I see you've had to type out the definitions several times today LOL!

I understand what you are saying now. What was throwing me off was the bit about "A1 grid clipping". Because the texts that I've read only state subscript 1= no grid conduction and 2= grid conduction.
Of course the texts are talking about linear amplifiers, but as I interpret it "A1" can not go into grid clipping, if it does then it can only be labelled A.
From what you are saying it can go past the onset of conduction, but providing that the source driving the grid is of high enough impedance, so as no current flows and as distorted as the top of the output wave form becomes, that it can be labelled A1 ?
Though some current must flow into the grid for a change in voltage to occur, be it only very little ?
I guess this point needs some clarification ? What amount of current is acceptable ? Even if the grid is close to zero bias there is still current flowing, be it only maybe µAmpere.
The texts are not clear on this.

I can see this is where Direct-Coupling / DC-drive has its advantages

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sat, Sep 15th, 2007 @ 23:30 ( . )

"What was throwing me off was the bit about "A1 grid clipping". Because the texts that I've read only state subscript 1= no grid conduction and 2= grid conduction.
: Of course the texts are talking about linear amplifiers, but as I interpret it "A1" can not go into grid clipping, if it does then it can only be labelled A.
: From what you are saying it can go past the onset of conduction, but providing that the source driving the grid is of high enough impedance, so as no current flows and as distorted as the top of the output wave form becomes, that it can be labelled A1 ?"

I think you may have it, but let me try to restate that to see if we're in agreement.

For Class A1, the driver can not muscle the grid into following the signal anymore past Vgk=0. Since we just said it was an A1 amp, we look for evidence of that being that the amp hits top-side grid-clipping at or before the onset of bottom-side cutoff.

For Class A2, the driver has low enough impedance and enough current source capability to muscle the grid into following the signal past Vgk=0. Since we just said this was an A2 amp, we look for evidence of that being that the amp hits top-side saturation at or before the onset of bottom-side cutoff.

"Though some current must flow into the grid for a change in voltage to occur, be it only very little ?

It's a forward biased tube diode. I'm not exactly sure the equivalent resistance, but I'll take a guess in the 2000 ohm neighborhood. If that's so, to generate 10 extra volts over Vgk=0, you'd need 5mA. That's a mini-power amp, something like a 12AT7 as a hot cathode follower or such. And since you'd want to dc-drive the thing, you'd want a negative supply so that the grid-driver output could swing above and below ground.

"I guess this point needs some clarification ? What amount of current is acceptable ? Even if the grid is close to zero bias there is still current flowing, be it only maybe µAmpere."

Depends on how hard you try to take it positive of Vgk=0. Right at Vgk=0, no current. Go above that and it starts going up via that parabolic diode curve.

"The texts are not clear on this."

And neither are the data sheets, which is a shame. You'd think they publish some equivalent resistance or something of the sort for positive grid drive. But they don't.

"I can see this is where Direct-Coupling / DC-drive has its advantages"

Note, there's a difference here. The direct coupling from a decent powered low impedance driver can muscle the grid past Vgk=0 into class A2 or AB2 territory.

The dc-drive setup does not muscle the next grid into A2 or AB2, it simply makes the bias more resistant to a temporary colder shift, the temporary colder shift tending to happen in a normal cap-coupled setup. The dc-drive thing sets up the next stage for a quick-bias recovery class A1 or AB1.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Dan N (guest: search)
Date: Sun, Sep 16th, 2007 @ 01:09 ( . )

I think you may have it, but let me try to restate that to see if we're in agreement.

Yes, we are in agreement then ;)
I think I was only looking at from the point of a linear voltage amplifier, where the output wave form is still able to increase proportionally in amplitude with the increasing amplitude of the input signal to the preceding driver stage.


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Wed, Sep 12th, 2007 @ 03:12 ( . )

The term "acts like" a Class B amp is also somewhat misleading. The amplifying elements in a Class AB PP amp will only cut-off for less than 180 degrees. So they are only "like" Class B amps in the sense that they deliberately go into cutoff for some part of the cycle. The same for a Class AB SE amps.

However, my basic point is that the texts give the class definitions based on conduction duty cycle. Any other phenomena are secondary, and don't form part of the basic definition.

On 09/11/2007 @ 22:28, Merlin wrote :
:
That would agree with my take on the definition. An SE amp, cool biased, never operates like a class B amp at high signal levels. It acts like a clipping Class A amp (which is what it is).
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mark L (registered user: 1458 posts )
Date: Tue, Sep 11th, 2007 @ 16:29 ( . )

On 09/11/2007 @ 09:55, Merlin wrote :
: There is nothing wrong with the class system, it makes perfect sense provided you define it using the load line, and where the bias point falls on it, which some textbooks do.
: It only falls apart when you try and define it using "degrees of conduction" which I agree only works for hifi, or repeating, symmetrical waveforms. So forget "degrees of conduction". Simple.


Hmm, I can't find any reference to defining class of operation by load line, they all refer to degrees of conduction.

Can you give me a reference to back it up, or even state that SE can only be class A.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Tue, Sep 11th, 2007 @ 17:02 ( . )

And that is the whole crux of the issue. Power amplifier operational classes are always and have always been defined in terms of duty cycle.

However, you will find plenty of sources in the audio electronics world that will say that SE amps have to be Class A, since within their own particular context that is indeed true. It's an assumption they're making based on the sensible idea that amplifiers are supposed to amplify the input signal without unduly distorting it.

On 09/11/2007 @ 16:29, Mark L wrote :
:
: Hmm, I can't find any reference to defining class of operation by load line, they all refer to degrees of conduction.
:
: Can you give me a reference to back it up, or even state that SE can only be class A.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Mark L (registered user: 1458 posts )
Date: Wed, Sep 12th, 2007 @ 05:42 ( . )

On 09/11/2007 @ 17:02, Zaphod wrote :

: However, you will find plenty of sources in the audio electronics world that will say that SE amps have to be Class A, since within their own particular context that is indeed true. It's an assumption they're making based on the sensible idea that amplifiers are supposed to amplify the input signal without unduly distorting it.



I agree 100% with that. If I were to build a SE for to amplify music with min. distortion and max. headroom and output power I would bias it 'true' class A.

But, saying a SE amp "should" be biased class A is different from saying a SE amp "can only" be biased class A.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Mon, Sep 10th, 2007 @ 18:48 ( . )

Very true. Other side effects are just that, and not part of the basic definition of class.

On 09/10/2007 @ 15:49, Mark L wrote :
:
: Classes of operation are based around what period of the cycle the device (semiconductor or valve) is conducting (or not conducting) at maximum power, thats's all!
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Tue, Sep 11th, 2007 @ 13:12 ( . )

Or did you mean [link] and [link] ?

On 09/10/2007 @ 16:49, Mark L wrote :
:
: I did a quick seach for references with SE class AB and B amps to check my recollection on this topic (it's been a while since I studied Electronics Engineering).
: [link]
: [link]
: [link]
: [link]
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Chris Hurley (moderator: 7842 posts )
Date: Tue, Sep 11th, 2007 @ 17:18 ( . )


: Or, I suppose we could answer that it's nearly irrelevant for guitar amps, and so we tend to characterize output stages by SE or PP, and then by hot, medium or cold biased. If that's where you want to leave it, I suppose we might all be able to agree on that.


Surely we'd have agreement about the SE or PP parts. :)

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Evan (guest: search)
Date: Tue, Sep 11th, 2007 @ 19:10 ( . )

On 09/11/2007 @ 17:18, Chris Hurley wrote :

: : Or, I suppose we could answer that it's nearly irrelevant for guitar amps, and so we tend to characterize output stages by SE or PP, and then by hot, medium or cold biased. If that's where you want to leave it, I suppose we might all be able to agree on that.
:

:
: Surely we'd have agreement about the SE or PP parts. :)
:
--



At this point I think it's completely inappropriate to bring up the fact there is no such thing as "push - pull". Valves are unidirectional. They pull or pull depending on your choice of current convention, but they darned sure don't do both.

:-)

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Merlin (registered user: 5226 posts )
Date: Tue, Sep 11th, 2007 @ 22:29 ( . )

On 09/11/2007 @ 19:10, Evan wrote :

:
: At this point I think it's completely inappropriate to bring up the fact there is no such thing as "push - pull". Valves are unidirectional. They pull or pull depending on your choice of current convention, but they darned sure don't do both.
--


I think the "push and pull" description technically refers to the instantaneous (i.e., AC) current flow in the OT, which does indeed flow in two directions.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Tue, Sep 11th, 2007 @ 23:12 ( . )

Plus, all the amps I use have a front end and a back end, so how can they be single ended?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Tue, Sep 11th, 2007 @ 23:28 ( . )

As there is no such thing as a monopole, so you may be onto something ...

Is there a Complementary-End (CE) somewhere in the universe for each of our SE output stages? And do they play the reverse of what we're hearing, wherever they may be?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Tue, Sep 11th, 2007 @ 23:49 ( . )

And does reverse in this case mean inverted, or backwards, or what?

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sat, Sep 15th, 2007 @ 17:32 ( . )

In my case, the reverse would be 'a guitar being played by someone who has some idea of how to do so' ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Sat, Sep 15th, 2007 @ 18:40 ( . )

I gotta figure that part out some day.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Sun, Sep 16th, 2007 @ 09:37 ( . )

Shame we don't know where the 'quantum amp' is. We could offer to swap. Presumably the 'quantum Zoe' sucks as badly as I do, by the standards of her own phase-space, so she'd probably be perfectly happy with getting an amp that makes her sound as though she don't. ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Sun, Sep 16th, 2007 @ 13:40 ( . )

Or we could talk about whether they blow or suck. :o)

Carrying on from this, we know that SS push-pull (or blow-suck) power amps commonly use complementary transistor pairs - ie NPN transistors paired with PNPs. What if we could do something similar with "complimentary" pairs of valves? So we would need to have positronic valve/tubes to complement the more usual electronic kind.

On 09/11/2007 @ 19:10, Evan wrote :
:
: At this point I think it's completely inappropriate to bring up the fact there is no such thing as "push - pull". Valves are unidirectional. They pull or pull depending on your choice of current convention, but they darned sure don't do both.
:
: :-)
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Sun, Sep 16th, 2007 @ 16:20 ( . )

You don't have any positronic valves?

Check with Asimov's estate; they still have a few.

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:CarlB (guest: search)
Date: Sun, Sep 16th, 2007 @ 18:19 ( . )

Shhh! That's my next circuit excursion: "Positronic Drive!" Of course, it's 'three laws safe.'

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Miles O'Neal (registered user: 13998 posts )
Date: Mon, Sep 17th, 2007 @ 00:17 ( . )

Great. Now we have to argue about how the three laws apply to distortion!

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:Zaphod (guest: search)
Date: Wed, Sep 12th, 2007 @ 06:55 ( . )

Unfortunately your question got lost in the ensuing theological debate. :o) The answer is that you can only know by plotting the load line for your power stage at those voltages and whatever load impedance you're using. Here are a couple of articles on the subject:
[link]
[link]

If the load line is fairly symmetric around the idle point, then your amp is in Class A. If it's asymmetrical towards cut-off then it's in SE Class AB.

And yes, you can look at the OT output with your scope. Be careful of looking at the OT primary without high voltage probes. You can get voltages much higher than B+ induced on the OT primary, sometimes as high as two or three times B+.

On 09/07/2007 @ 09:56, rufinus wrote :
I'm interested, how can you determine a class of operation for the EL84 in my HO?
: For example, here are my measured voltages:
:
: Pin9 275v
: Pin7 288v
: Pin3 9.3v
:
: Screen current is considered to be about 5mA. Resistor R10 is 220 ohm. At 37mA of plate current this gives 10.3 watt of plate dissipation. Screen dissipation is about 1.37 watt then.
:
: In this case, what is the class of amplification? A or AB? How do i determine it?
: Another question, can i look safely with the oscilloscope at the output of the OT?
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:rufinus (guest: search)
Date: Mon, Sep 17th, 2007 @ 09:10 ( . )

Thanks Zaphod!

And everyone else too. I think i got an answer now :)
Thank you for participation :)
WOW




On 09/12/2007 @ 06:55, Zaphod wrote :
Unfortunately your question got lost in the ensuing theological debate. :o) The answer is that you can only know by plotting the load line for your power stage at those voltages and whatever load impedance you're using. Here are a couple of articles on the subject:
: [link]
: [link]
:
: If the load line is fairly symmetric around the idle point, then your amp is in Class A. If it's asymmetrical towards cut-off then it's in SE Class AB.
:
: And yes, you can look at the OT output with your scope. Be careful of looking at the OT primary without high voltage probes. You can get voltages much higher than B+ induced on the OT primary, sometimes as high as two or three times B+.
:
: On 09/07/2007 @ 09:56, rufinus wrote :
: I'm interested, how can you determine a class of operation for the EL84 in my HO?
: : For example, here are my measured voltages:
: :
: : Pin9 275v
: : Pin7 288v
: : Pin3 9.3v
: :
: : Screen current is considered to be about 5mA. Resistor R10 is 220 ohm. At 37mA of plate current this gives 10.3 watt of plate dissipation. Screen dissipation is about 1.37 watt then.
: :
: : In this case, what is the class of amplification? A or AB? How do i determine it?
: : Another question, can i look safely with the oscilloscope at the output of the OT?
: :
--


-- REPLY: [With No Quote] --- [With Quoted Text]

'EL84 class of operation. A or AB?'
Author:James Marchant (registered user: 450 posts )
Date: Thu, Sep 13th, 2007 @ 06:51 ( . )

I swore I'd never debate on a class A thread, but after reading through this one a bit I have come up with my definition of class A.

An amplifier is said to operate in class A if the marketing department think they can charge more for it by doing so.

James
=0^0=

-- REPLY: [With No Quote] --- [With Quoted Text]

'Quote of the Week!'
Author:Zaphod (guest: search)
Date: Thu, Sep 13th, 2007 @ 09:09 ( . )

ROFL!!!

On 09/13/2007 @ 06:51, James Marchant wrote :
:
I swore I'd never debate on a class A thread, but after reading through this one a bit I have come up with my definition of class A.
:
: An amplifier is said to operate in class A if the marketing department think they can charge more for it by doing so.
:
: James
: =0^0=
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Quote of the Week!'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Thu, Sep 13th, 2007 @ 09:31 ( . )

(Iain) I'm waiting for one of the 'Class A - All the Way' companies to decide they've been doing it for long enough and it's time to shut up shop. Be great if they changed their website to tell the truth, just to piss off all the people that bought into the hype.

"The MumboJumbo 2000, not actually a class A amp at all, contrary to what we might have told you. And there was you thinking you had Golden Ears and would know the difference! Fooled ya!" ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'Quote of the Week!'
Author:bancika (registered user: 2430 posts )
Date: Mon, Sep 17th, 2007 @ 09:09 ( . )

:)
A sounds much better than B and less confusing then AB. It's like first class amp

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey! rufinus, are you still here?'
Author:Dan N (guest: search)
Date: Mon, Sep 17th, 2007 @ 05:11 ( . )

errrr...what was your question again... something about an EL84 ?

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey! rufinus, are you still here?'
Author:Evan (guest: search)
Date: Mon, Sep 17th, 2007 @ 08:02 ( . )

I think he concluded we were all a bunch of classholes.

<rimshot>

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey! rufinus, are you still here?'
Author:rufinus (guest: search)
Date: Mon, Sep 17th, 2007 @ 09:16 ( . )

Sweeet Jesus!!

I really, REALLY appreciate all your answers. Thanks a lot for everyone, who participatet. I'm not going to ask again :)

P.S. Wheee.. THAT was ONE of a DEBATE.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey! rufinus, are you still here?'
Author:Zaphod (guest: search)
Date: Mon, Sep 17th, 2007 @ 10:15 ( . )

ROFL!!!

On 09/17/2007 @ 08:02, Evan wrote :
:
I think he concluded we were all a bunch of classholes.
:
: <rimshot>

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey! rufinus, are you still here?'
Author:Balijukka (guest: search)
Date: Tue, Sep 18th, 2007 @ 06:24 ( . )

of course that depends on how you define WE.
in a debate there is WE and there is YOU.
now which side were you on....

jukka

On 09/17/2007 @ 10:15, Zaphod wrote :
ROFL!!!
:
: On 09/17/2007 @ 08:02, Evan wrote :
: :
: I think he concluded we were all a bunch of classholes.
: :
: : <rimshot>
:
--


-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey! rufinus, are you still here?'
Author:Zaphod (guest: search)
Date: Tue, Sep 18th, 2007 @ 09:23 ( . )

I think he meant "we" in the sense of anyone who contributed to this thread. :o) In my case, as I have two heads who often disagree with each other, it gets even more confusing.

On 09/18/2007 @ 06:24, Balijukka wrote :
:
: of course that depends on how you define WE.
: in a debate there is WE and there is YOU.
: now which side were you on....
:
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Randall Aiken (guest: search)
Date: Tue, Sep 18th, 2007 @ 17:22 ( . )

Okay, here's a rather poorly-scanned copy of some pages from the 1941 classic text by Herbert J. Reich "Principles of Electron Tubes".

In addition to containing a good explanation of the amplifier classes, it also explicitly states on the bottom of page 129 the point I have been arguing all along, that SE class AB is indeed possible in an audio amplifier, but you would "normally" only use push-pull class AB to avoid excessive distortion.

Notice there is no mention of changing load lines, nor any requirement of a tank circuit in the class definitions, either.

[link]

I guess this makes Mr. Reich an "irrational" engineer, too? :)

I have other old texts with class definitions that are nearly identical, I can scan those too, if anyone is interested.

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Tue, Sep 18th, 2007 @ 17:36 ( . )

Hmm, he's got us on the ropes with that one, boys! Hard to argue that Reich isn't a credible source.

Surely someone must have an earlier text that contradicts that, or we'll have to concede defeat, here! ;-}




-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Randall Aiken (guest: search)
Date: Tue, Sep 18th, 2007 @ 17:46 ( . )

On 09/18/2007 @ 17:36, Zoe N Iain wrote :
Hmm, he's got us on the ropes with that one, boys! Hard to argue that Reich isn't a credible source.
:
: Surely someone must have an earlier text that contradicts that, or we'll have to concede defeat, here! ;-}
:
:
:
:
--



Heh heh...pardon me while I gloat excessively, throw my shoulder out of joint while patting myself on the back, and just make a general ass of myself in public here. ;)

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Jon Anderson (registered user: 1491 posts )
Date: Tue, Sep 18th, 2007 @ 18:17 ( . )

On 09/18/2007 @ 17:46, Randall Aiken wrote :
: Heh heh...pardon me while I gloat excessively, throw my shoulder out of joint while patting myself on the back, and just make a general ass of myself in public here. ;)


Feel free to dance a perverted combination of the Electric Slide, the chicken dance, and the Hokey Pokey while your at it. We all know it's the natural dance for geeky engineers. =)

Jon

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Randall Aiken (guest: search)
Date: Tue, Sep 18th, 2007 @ 18:19 ( . )

On 09/18/2007 @ 18:17, Jon Anderson wrote :
: Feel free to dance a perverted combination of the Electric Slide, the chicken dance, and the Hokey Pokey while your at it. We all know it's the natural dance for geeky engineers. =)



We always call it doing the "Dilbert victory dance" that he does in the cartoons...

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Stephen Keller (registered user: 6220 posts )
Date: Tue, Sep 18th, 2007 @ 17:47 ( . )

This is essentially the same thing that Eastman said in his book "Fundamentals of Vacuum Tubes" and that Orr said in "The Radio Handbook." Two other credible sources. The definitions allow it. Reich's book, by the way is available at Pete Millett's site here: [link]

Stph

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:John Hynes (registered user: 2628 posts )
Date: Tue, Sep 18th, 2007 @ 17:44 ( . )

On 09/18/2007 @ 17:22, Randall Aiken wrote :
Okay, here's a rather poorly-scanned copy of some pages from the 1941 classic text by Herbert J. Reich "Principles of Electron Tubes".
:
: In addition to containing a good explanation of the amplifier classes, it also explicitly states on the bottom of page 129 the point I have been arguing all along, that SE class AB is indeed possible in an audio amplifier, but you would "normally" only use push-pull class AB to avoid excessive distortion.
:
: Notice there is no mention of changing load lines, nor any requirement of a tank circuit in the class definitions, either.
:
: [link]
:
: I guess this makes Mr. Reich an "irrational" engineer, too? :)
:
: I have other old texts with class definitions that are nearly identical, I can scan those too, if anyone is interested.
:
: Randall Aiken
:
--



Oddly enough, I have just acquired a copy of that very book this morning!

Geez Randall, I hope that barb wasn't directed at me. I don't recall accusing anyone of being "irrational". If I did, I hereby formally apologize, and stand chastened and rebuked. ;)

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Randall Aiken (guest: search)
Date: Tue, Sep 18th, 2007 @ 17:48 ( . )

On 09/18/2007 @ 17:44, John Hynes wrote :
: Geez Randall, I hope that barb wasn't directed at me. I don't recall accusing anyone of being "irrational". If I did, I hereby formally apologize, and stand chastened and rebuked. ;)



Nah, John, that was just a jab at Merlin for one of his earlier posts. It's all in good fun for the sake of stirring the pot... :)

Randall Aiken

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Dan N (guest: search)
Date: Tue, Sep 18th, 2007 @ 22:40 ( . )

Lol! ironically I downloaded that book yesterday but never had time to read it. So the one book that I had access to that would have clarified it; eluded for me, oh well !

So now I concede that we do indeed have a Class-AB descriptor for a Single-Ended audio amplifier.

Not only that but it also mainly applies to realm of guitar amps, because the hi-fi community would loathe to dabble in such a strange use of the thermionic valve.

In searching for the earliest citation of class of operation, I found mention of that in C.1912 Western Electric may of applied for a patent for a Class-C amplifier. But the source is unsubstantiated.
And Bell Systems Labs mentioned it before C.1919

It did get me thinking though that term conduction angle to define class may of been around before the invention of the thermionic valve. It well may of been used to describe electric motors or generators and from there simply applied to amplifiers ?
But that's another story, the library which I had access to journals from C.1886 has put all of it's pre C.1940's stuff into storage :(

Cheers
Dan

BTW... Miles, can you FAQ this (the whole thread) ? But you will need to correct all the spelling and grammar mistakes :^O

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Balijukka (guest: search)
Date: Tue, Sep 18th, 2007 @ 23:07 ( . )

ok, agreed.

the clean output requirement is not part of the definition, just the conduction cycle.

now just one more thing

.... WHERE'S THE BOOK, RANDALL ?

jukka

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:CarlB (guest: search)
Date: Tue, Sep 18th, 2007 @ 23:22 ( . )

You and I may have to write it at this rate, Jukka!

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:CarlB (guest: search)
Date: Tue, Sep 18th, 2007 @ 23:14 ( . )

There were magnetic amplifiers. The likes of Tesla and other amazing minds came up with them. They had limited frequency response, yet the first cabled trans-atlantic audio communications successfully employed them.

Interestingly, early power supplies used them and they're making a comeback.

I don't know if there were push-pull magnetic amplifiers, but I'd bet that class A could describe their operation.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:CarlB (guest: search)
Date: Tue, Sep 18th, 2007 @ 23:04 ( . )

Randall,

Although I fully followed Mr. Reich's text, I can see readers being confused about class definitions for the first (or hundredth) time when reading them. The the correct technical verbiage used is not always the easiest to follow.

The following is from edits to a few of my prior posts. Do you (or anyone else) see anything wrong with the following working descriptions? If so, perhaps simple corrections could be suggested? Is there any type of consensus that these could make understanding tube classes easier?

--------------------------------------

Class A1: the maximum the SE valve can reproduce cleanly is at the onset of grid clipping. If push-pull, the maximum the stage can reproduce cleanly is at the onset of grid clipping.

Class A2: the maximum the SE valve can reproduce cleanly is before saturation occurs. The valve's driver, by design, can drive the grid well into conduction and so the output continues to follow the input signal. If push-pull, the maximum the stage can reproduce cleanly is at the onset of saturation.

Class AB1: the maximum the SE valve can reproduce cleanly is before plate current cut off. In push-pull, the maximum the stage can produce cleanly is at the onset of grid-conduction.

Class AB2: the maximum the SE valve can reproduce cleanly is before plate current cut off. The valve's driver, by design, can drive the grid well into conduction and so the output continues to follow the input signal. In push-pull, the maximum the stage can produce cleanly is at the onset of saturation.

For the above working definitions, the following apply:

Grid conduction. If for some portion of the input waveform the grid becomes even slightly positive compared to the cathode, current will flow. The grid structure to the cathode structure forms a diode, and so when forward biased current flows.

Grid-clipping. This occurs when the driving source can not keep the conducting grid from flattening a portion of the input signal. This can occur because the driving source is not sufficiently lower in impedance than the forward biased grid, or because the driving source has insufficient power to supply the current necessary for forcing the grid to follow the signal.

Saturation. If grid clipping can be overcome by the stage's input driver, then the tops of the grid signal continue positive of Vgk = 0. The amplifying stage's current continues upwards for the positive portions of the input waveform - until the point at which the plate can no longer supply the current necessary to faithfully follow the signal at the grid. The plate is said to be saturating, and the waveform flattens for those portions that the plate can supply no more current.

Cutoff. This occurs on the opposite side of the output waveform. The plate current decreases on this side of the output signal. There is a limit to how much the plate current can decrease from idle condition - the plate can not supply anything less than zero amperes. So, if the input signal becomes large enough, there becomes a portion of the output signal for which the plate current drops to zero, and at this point the stage is said to be in cutoff.

-- REPLY: [With No Quote] --- [With Quoted Text]

'agreed'
Author:Balijukka (guest: search)
Date: Wed, Sep 19th, 2007 @ 06:06 ( . )

looks ok for me

jukka

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 07:51 ( . )

How about this Carl:

Simplest common denominator. Then, we won't have to have debates like the last one again. :)

How about:

Amplifier classes are defined by angle of conduction. The classes are:

Class A: Conduction for 100% of the input cycle
Class AB: Conduction for less than 100%, but more than 50% of the input cycle
Class B: Conduction for exactly 50% of the input cycle
Class C: Conduction for less than 50% of the input cycle
Class D: Conduction in either 100% or 0%

Additionally, these classes can be subdivided into two categories, denoted by the subscript "1" or "2". These indicate:

Subscript 1: No grid current is drawn at any point during the input cycle.
Subscript 2: Grid current is drawn at some point in the cycle

These designations ignore leakage current which is always present at the grid during normal operation.

(Maybe follow with some examples like AB2 - grid driven positive, current drawn, Then A1, grid when driven to Vg=0, cannot source current, clamps... etc.)

We could even include pretty waveform screenshots.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Okay, here's your proof, read it and weep!'
Author:Dan N (guest: search)
Date: Wed, Sep 19th, 2007 @ 08:55 ( . )

That looks good to me John ;)
However... just to be pedantic, If it is going to be based on conduction *angle* then the unit of measure should be degrees.
If we are going to use percent then we should call it duty cycle.

And you are spot on with the term for "1" & "2" being subscript and not suffix. It's only 'coz computers are stoopid and don't know how to print subscript that it appears as a suffix.
It can display superscript though ¹²³

-- REPLY: [With No Quote] --- [With Quoted Text]

'Amendements to FAQ for HFP'
Author:Zaphod (guest: search)
Date: Wed, Sep 19th, 2007 @ 08:55 ( . )


AFAIK subscript 1 actually means that *sustained* grid current isn't drawn. You can still have transient grid currents as you approach grid clamping, and this contributes toward blocking distortion.

Subscript 2 indicates sustained grid current is being drawn by (or supplied to) the grid while the input waveform is above Vg=0

I hope I didn't screw up in my explanation there. Randall and Merlin debated this somewhere further up in the thread.

Maybe could also add some additional comments along the lines of:

1 - In traditional audio applications, where minimization of distortion is essential, Class AB and Class B amplifiers normally only exist in push-pull form. Therefore many texts in the audio field make the assumption that all Class AB and Class A amps will be push-pull. This assumption doesn't necessarily apply in other fields, such as radio and guitar amplification.

2 - Typical Class AB push-pull audio amplifiers, which have a center-tapped OT primary, exhibit the interesting phenomenon of changing load impedance. While both sides of the PP pair are conducting each sees the equivalent of half of the OT's end-to-end impedance. When either side goes into cutoff, the opposite one now sees a only quarter of the OT primary impedance. This impedance change is accompanied by increased current draw.

This is off the top of my head. Please feel free to tweak and improve.

On 09/19/2007 @ 07:51, John Hynes wrote :
:
: Subscript 1: No grid current is drawn at any point during the input cycle.
: Subscript 2: Grid current is drawn at some point in the cycle
:
: These designations ignore leakage current which is always present at the grid during normal operation.
:
: (Maybe follow with some examples like AB2 - grid driven positive, current drawn, Then A1, grid when driven to Vg=0, cannot source current, clamps... etc.)
:
: We could even include pretty waveform screenshots.
:
: -John
:
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Amendements to FAQ for HFP'
Author:Zaphod (guest: search)
Date: Wed, Sep 19th, 2007 @ 08:58 ( . )

Dang I can't even type what I meant to say myself!

It should of course read, "Therefore many texts in the audio field make the assumption that all Class AB and Class B amps will be push-pull."

On 09/19/2007 @ 08:55, Zaphod wrote :
:... Therefore many texts in the audio field make the assumption that all Class AB and Class A amps will be push-pull. This assumption doesn't necessarily apply in other fields, such as radio and guitar amplification.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Amendements to FAQ for HFP'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 09:18 ( . )

On 09/19/2007 @ 08:58, Zaphod wrote :
Dang I can't even type what I meant to say myself!
:
: It should of course read, "Therefore many texts in the audio field make the assumption that all Class AB and Class B amps will be push-pull."
:
: On 09/19/2007 @ 08:55, Zaphod wrote :
: :... Therefore many texts in the audio field make the assumption that all Class AB and Class A amps will be push-pull. This assumption doesn't necessarily apply in other fields, such as radio and guitar amplification.
: :
--



Sounds good to me Zaph.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Amendements to FAQ for HFP'
Author:Randall Aiken (guest: search)
Date: Wed, Sep 19th, 2007 @ 09:21 ( . )

On 09/19/2007 @ 09:18, John Hynes wrote :
On 09/19/2007 @ 08:58, Zaphod wrote :
: Dang I can't even type what I meant to say myself!
: :
: : It should of course read, "Therefore many texts in the audio field make the assumption that all Class AB and Class B amps will be push-pull."
: :
: : On 09/19/2007 @ 08:55, Zaphod wrote :
: : :... Therefore many texts in the audio field make the assumption that all Class AB and Class A amps will be push-pull. This assumption doesn't necessarily apply in other fields, such as radio and guitar amplification.
: : :
: --
:
:

:
: Sounds good to me Zaph.
:
: -John
:
--



I think that clarifies it nicely.

RA

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Merlin (registered user: 5226 posts )
Date: Wed, Sep 19th, 2007 @ 14:31 ( . )

Chrish, is this the longest thread on AX84 yet? Funny how vehemently Americans will defend such an obvioulsy flawed idea! Oh well, it takes all sorts!

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 14:49 ( . )

On 09/19/2007 @ 14:31, Merlin wrote :
Chrish, is this the longest thread on AX84 yet? Funny how vehemently Americans will defend such an obvioulsy flawed idea! Oh well, it takes all sorts!
--



You know, I think we have yet to find the "right" definition, whatever that may be. Randall and others produced evidence that several reputable authors consider conduction angle to be the only determinant. I have found the writings of a few prominent audio engineers that disagree with that, for all the reasons we've already covered. The "Audio Perspective" has been suggested to be acknowledged in the definition by Zaphod, as radio guys obviously accept this. That's OK with me.

If anything, it seems both simple and hard at the same time: If we say "conduction angle only" and agree, then SE AB operation is possible, and simple, unambiguous definitions are possible.

The hard part is, what about Carl's example in the thread on "clipping"? He gives an example of a stage that is driven to reach saturation & cutoff equally, being center-biased. It will fail to conduct 100% of the cycle at that point, but he maintains that it is a Class A amplifier. This is contradictory to our "simple" definition. So, is Carl mistaken, or is the simple definition wrong? Does *how* the stage clips matter? Does that it clips *at all* matter? The material presented to support their case uses conduction angle solely as a determinant. By that logic, Carl's example is a Class AB amplifier if we use that text's definition.

So, are there other factors? If so, we can only derive them BY EXAMPLES in the texts cited. That would demonstrate that an AB amplifier is only shown as not conducting in cutoff. The examples all show this, but the definitions don't. Just as in an audio-only text, the examples of AB are all push-pull, but the definitions don't specify that it must be. Hmmmm.

And that changes the picture.

Let's keep going so this is definitely the longest thread ever. I'm having a blast. Not. :)

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Stephen Keller (registered user: 6220 posts )
Date: Wed, Sep 19th, 2007 @ 15:10 ( . )

It seems to me that the definition based solely on conduction-angle is effective only at an input signal that is at or just less than the point where the grid begins to draw current (for subscript 1 classes) or at or just less than the point where the driver stage can no longer supply adequate grid current (subscript 2 classes). I'm mean isn't that why the definitions refer to both the bias voltage and the grid excitation voltage? Presumably, the former is a DC offset and latter is an amplitude?

Stph

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 19th, 2007 @ 15:25 ( . )

And if you use THAT as a definition, you can't have a SE AB amp.

I'm afraid to say, coming from a decidedly radio/radar biased background, that I consider any definition that says you can have SE AB, except at a point frequency defined by a tuned load, to be utter bollocks.

The definition, as used by engineers of the world's militaries (who, and let's be honest here, invented (or at least put out the request for) most if not all of this technology), was for a system that could give out a clean signal. That was true up until the late thirties. And even after it wasn't anymore, with the advent of radar (which demands non-linear signals to work) it was another 10 years before anyone built an amplifier that didn't especially care about linearity. And another 12-15 before anyone realised that the non-linearities might actually be a good thing.

Colour me non-convinced.

But as Mr. Reich says that SE AB is OK, how can I argue? ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Stephen Keller (registered user: 6220 posts )
Date: Wed, Sep 19th, 2007 @ 15:35 ( . )

Maybe I'm not expressing myself clearly, because it was Mr. Reich's drawing of the single-sided AB1 and AB2 transfer curves that was prompting my thinking (or apparent lack thereof).

There have been enough different slants on this problem expressed here that what I once thought I understood intuitively, has become all muddled and complexified.

Stph

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 19th, 2007 @ 15:42 ( . )

I move, at this point, that we all accept the Aiken/Berger/Rowley definition and just call the damn thing AB. We all know what it means, whether we agree with it or not, and all future noobie numpties will get the point, also.

On reflection, I STILL don't agree with the Reich definition Randall posted. But hey, he knows more about this shit than I do, so who am I to argue. Let's just come to a compromise and say that for GUITAR AMP use, you CAN bias a SE stage in AB, because we don't especially care about linearity and in some cases want to avoid it. OK?

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Merlin (registered user: 5226 posts )
Date: Wed, Sep 19th, 2007 @ 15:43 ( . )

On 09/19/2007 @ 15:42, Zoe N Iain wrote :
On reflection, I STILL don't agree with the Reich definition Randall posted. But hey, he knows more about this shit than I do, so who am I to argue.

You big fat quitter!

And I think you know just as much as Randall.


-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 19th, 2007 @ 15:58 ( . )

I DO know as much as Randall, easily, but he knows it about audio amps and I know it about VHF/UHF/SHF amps. I'm sure he wouldn't try and second guess me about a radar, unless he WANTED to look a total prick. So I am compelled to return the courtesy for audio. Because pretty much any discussion I had with him would result in that. We all have different skills and knowledge.

I quietly suspect that Mike Sulzer and possibly Stephen Keller know more than me about either, but that's a different matter. ;-}

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Dan N (guest: search)
Date: Wed, Sep 19th, 2007 @ 17:04 ( . )

One thing to note is that in radio work a valve can be driven into saturation or cut-off for the purpose of harmonic generation.
So to say that early engineers never intended for a valve to be overdriven is not entirely correct.
Phil and Randall would know a hell of a lot more than me about this: but the allocation of the amateur radio frequencies is based on harmonics.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Wed, Sep 19th, 2007 @ 17:35 ( . )

Yes, you're right. But I wasn't going to muddy the waters or even clutter the frequency bands with that one! :o)

On 09/19/2007 @ 17:04, Dan N wrote :
:
One thing to note is that in radio work a valve can be driven into saturation or cut-off for the purpose of harmonic generation.
: So to say that early engineers never intended for a valve to be overdriven is not entirely correct.
: Phil and Randall would know a hell of a lot more than me about this: but the allocation of the amateur radio frequencies is based on harmonics.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Wed, Sep 19th, 2007 @ 17:39 ( . )

It does appear that Randall knows quite a bit about HF though. And I'm feeling very nostalgic with all this talk of radio, although I lose interest at wavelengths shorter than 2m. I even have a little battery valve portable waiting for me to get round to restoring it.

On 09/19/2007 @ 15:58, Zoe N Iain wrote :
:
I DO know as much as Randall, easily, but he knows it about audio amps and I know it about VHF/UHF/SHF amps. I'm sure he wouldn't try and second guess me about a radar, unless he WANTED to look a total prick. So I am compelled to return the courtesy for audio. Because pretty much any discussion I had with him would result in that. We all have different skills and knowledge.
:
: I quietly suspect that Mike Sulzer and possibly Stephen Keller know more than me about either, but that's a different matter. ;-}
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 15:46 ( . )

On 09/19/2007 @ 15:42, Zoe N Iain wrote :
I move, at this point, that we all accept the Aiken/Berger/Rowley definition and just call the damn thing AB. We all know what it means, whether we agree with it or not, and all future noobie numpties will get the point, also.
:
: On reflection, I STILL don't agree with the Reich definition Randall posted. But hey, he knows more about this shit than I do, so who am I to argue. Let's just come to a compromise and say that for GUITAR AMP use, you CAN bias a SE stage in AB, because we don't especially care about linearity and in some cases want to avoid it. OK?
--



I'm cool with that, too. I'm just having a bit of carrying on at this point.

I think that there ought to be a big 'ol asterisk next to the FAQ definition pointing to this thread, though, in case anybody actually cares about the complexities involved with clearly defining this.

...or all of the concept's supporters can just acquiesce and accept that it's not a practical concept in AF and does not exist, de facto. <light fuse and get away> :)

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Stephen Keller (registered user: 6220 posts )
Date: Wed, Sep 19th, 2007 @ 15:53 ( . )

I forget... which side am I on?

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Wed, Sep 19th, 2007 @ 17:43 ( . )

I suppose a compromise position might be to name "Class AB SE" as a special case within Class AB, much like I've been saying that Class AB PP is in its own right.

BTW on the matter of the changing load lines, I would guess that an OT-less PP Class AB amp wouldn't exhibit that behavior. What would you think?

On 09/19/2007 @ 15:42, Zoe N Iain wrote :
:
I move, at this point, that we all accept the Aiken/Berger/Rowley definition and just call the damn thing AB. We all know what it means, whether we agree with it or not, and all future noobie numpties will get the point, also.
:
: On reflection, I STILL don't agree with the Reich definition Randall posted. But hey, he knows more about this shit than I do, so who am I to argue. Let's just come to a compromise and say that for GUITAR AMP use, you CAN bias a SE stage in AB, because we don't especially care about linearity and in some cases want to avoid it. OK?
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Wed, Sep 19th, 2007 @ 15:11 ( . )

Merlin, I originally learnt the class definitions during my teens from a number of British electronics books I borrowed from the local library, among them the RSGB Handbook c1970. I was into shortwave and amateur radio at the time, and I read every book on the subject I could get my hands on.

As I remarked elsewhere, it appears that in the RF field the purely conduction angle approach is favored, whereas in audio they generally prefer to limit Class AB to PP amps. In audio Class AB SE is a useless concept, while in radio it isn't.

On 09/19/2007 @ 14:31, Merlin wrote :
Chrish, is this the longest thread on AX84 yet? Funny how vehemently Americans will defend such an obvioulsy flawed idea! Oh well, it takes all sorts!

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Merlin (registered user: 5226 posts )
Date: Wed, Sep 19th, 2007 @ 15:22 ( . )

On 09/19/2007 @ 15:11, Zaphod wrote :
In audio Class AB SE is a useless concept, while in radio it isn't.

Quite agree. Tank circuits are a differnt beast altogether.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Bryan James (registered user: 450 posts )
Date: Wed, Sep 19th, 2007 @ 15:29 ( . )

just curious, has someone built one of these SE AB amps? what does it sound like? is all this a mute point? To get a SE amp into ab is to change the bias resistor? if so can someone do it and post some clips? if someone has done it, are there any clips.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Wed, Sep 19th, 2007 @ 16:24 ( . )

Sure. Take a typical SF Champ running a 6V6 at somewhere above 400V (IIRC). Or take a P1 eXtreme, as mentioned more than once by CarlB. And there are quite a few others around. Typically these amps bias pretty cold to keep within dissipation ratings, and as a result go into cutoff quite a lot. I think most people agree that amps like this sound pretty good, with plenty of even order harmonics being generated. And in general, most people seem to like power amp distortion anyway. Maybe if these amps went further towards the Class B side of operation, and spent more time in cutoff, the effect might get a bit too much.

On 09/19/2007 @ 15:29, Bryan James wrote :
:
just curious, has someone built one of these SE AB amps? what does it sound like? is all this a mute point? To get a SE amp into ab is to change the bias resistor? if so can someone do it and post some clips? if someone has done it, are there any clips.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Bryan James (registered user: 450 posts )
Date: Thu, Sep 20th, 2007 @ 18:35 ( . )

here's my problem with that, and maybe it was the way i was taught years ago, is I would expect a SE AB amp to always produces a clipped signal regardless of input. It's what I would expect from the amp.

The fact that the champ can produce a clean output makes it class A, sure with enough signal it will be a clipped class A and becuase of the Bias be unbalanced, yes that's the same as AB but it's still a class A amp. my be a badly designed class A but still a class A

here's an example, if I plug into my SE amp a guitar with a single pickup, have it set to a nice clean signal, shows up on my scope. then switched guitars to a dual humbucker guitar. look at the scope and it looks like an AB signal now since i'm clipping the amp. That doesn't mean the amp is now Class AB.

that doesn't mean I'm saying class AB SE amps aren't possible, that's been proven, but I would say that the signal from and AB will always be AB.

or are we going to start saying this amp is class A with class AB tendiancies?


-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:CarlB (guest: search)
Date: Thu, Sep 20th, 2007 @ 21:22 ( . )

Here's how to consider it, as best as I can tell:

Some accepted definitions, in this instance from Reich, section 5-14:

"A class A amplifier is one in which the grid bias and alternating grid voltage are such that plate current in the tube, or in each tube of a push-pull stage, flows at all times."

"A class AB amplifier is one in which the grid bias and alternating grid voltages are such that plate current in the tube, or in each tube of a push-pull stage, flows for appreciably more than half, but less than the entire electrical cycle."

"The suffix 1 may be aadded to the letter or letters of the class identification to denote that grid current does not flow during any part of the input cycle, and the suffix 2 to denote that grid current flows during some part of the cycle."

Alright then, let's consider a class AB1 push-pull stage. Most of us accept that this type of stage does exist. And that it does not change class simply by changing the AC input signal applied to it. If you don't accept that, then we need to have a more basic discussion.

You said something to the effect of, "class AB1 must always have a portion of it's output wave in cutoff." Well, that's not true for small signals, where the full output waveform is conducting. But it is a class AB1 amp, we just have to figure out what is meant by the terminology.

Let's figure out what must be true. Looking at the suffix 1, we know that grid current does not flow for any part of the input signal, and so we know that that the upper limit of input signal that we must observe class AB1's output conditions to be valid for: a bit less than grid-clipping. The bottom limit that we must observe class AB1's output conditions to be valid for is somewhere above minimal input signals (which conduct all the time as we determined above), but somewhere at or below our maximum input signal. In other words, we don't care too much where the lower valid limit precisely happens, we do know it will happen somewhere between no input signal and our maximum valid input level.

Remember, these are the input limits over which the class AB1 output conditions must be true. Outside of these limits, the inputs and outputs are still there - they just don't count in the determination of operating class, by definition.

So, we know that class AB1 push-pull must conduct less than the full output cycle, but more than half of the output cycle, starting at that precipitous input signal somewhere between no input signal, and up-to just before grid clipping.

We also know that class AB1 push-pull must conduct less than the full output cycle, but more than half of the output cycle, at just before grid clipping.

So on the bench, where would we most logically determine if a push-pull stage is class AB1? Right before grid clipping. Put the input signal at just before clipping (first go into grid clipping just barely, then back off a bit), and see if the output signal is conducting for less than the full cycle, but more than half the cycle.

Now, let's consider the SE class AB1 SE stage. As it turns out, you soon realize there isn't any difference for the valid input range of this class of operation determination.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Fri, Sep 21st, 2007 @ 06:26 ( . )

What you are describing is typical Class AB behavior. On lower signal levels they appear to work like a Class A amp. With moderately increased signal levels they go into cutoff for part of each cycle. But an examination of the load line makes it quite clear that the power tube's operating conditions are set for Class AB operation.

In the PP world, this is also how a Vox AC15 or AC30 operates. And we know that a lot of (misinformed) people think those are Class A as well.

On 09/20/2007 @ 18:35, Bryan James wrote :
:
here's my problem with that, and maybe it was the way i was taught years ago, is I would expect a SE AB amp to always produces a clipped signal regardless of input. It's what I would expect from the amp.
:
: The fact that the champ can produce a clean output makes it class A, sure with enough signal it will be a clipped class A and becuase of the Bias be unbalanced, yes that's the same as AB but it's still a class A amp. my be a badly designed class A but still a class A
:
: here's an example, if I plug into my SE amp a guitar with a single pickup, have it set to a nice clean signal, shows up on my scope. then switched guitars to a dual humbucker guitar. look at the scope and it looks like an AB signal now since i'm clipping the amp. That doesn't mean the amp is now Class AB.
:
: that doesn't mean I'm saying class AB SE amps aren't possible, that's been proven, but I would say that the signal from and AB will always be AB.
:
: or are we going to start saying this amp is class A with class AB tendiancies?
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Fri, Sep 21st, 2007 @ 06:35 ( . )

(Iain) Quite true. I've come to the conclusion that we may as well call the damn thing AB. There isn't really a definition that 100% fits, because the class definitions were invented with the assumption that the biggest signal to be applied would be the largest one that would allow the amp to run clean. But, not counting a tiny (and probably inaudible. Certainly inconsequential for a guitar amp) bit of non-linearity distortion that might creep in due to the overbiasing, a SE stage biased like that is behaving like half a push-pull AB stage, so AB is the only close definition we have available. It's either that or we invent a new one, which we'd then have to explain to every new member that comes along. It's easier to just call it AB and assume that the original engineers would have, too, had there been guitar amps at the time.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Francis Vaughan (registered user: 712 posts )
Date: Fri, Sep 21st, 2007 @ 06:41 ( . )

I've come to the conclusion that we may as well call the damn thing AB.


Exactly so. :-)

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:John Hynes (registered user: 2628 posts )
Date: Fri, Sep 21st, 2007 @ 06:37 ( . )

Hey Zaph - did you just say that the load matters in determining class? <LOL>

J/K - forget that I asked.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'What about Class-BC ?'
Author:Dan N (guest: search)
Date: Fri, Sep 21st, 2007 @ 07:33 ( . )

This space is intentionaly blank :?

-- REPLY: [With No Quote] --- [With Quoted Text]

'What about Class-BC ?'
Author:John Hynes (registered user: 2628 posts )
Date: Fri, Sep 21st, 2007 @ 07:37 ( . )

On 09/21/2007 @ 07:33, Dan N wrote :
This space is intentionaly blank :?
--



I think Class BC describes an amplifier constructed of tubes made out of mammoth tusks, with filaments made out of trilobite antennae and evacuated by leaving the assembly sealed against a glacial fissure as the ice recedes.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'What about Class-BC ?'
Author:Dan N (guest: search)
Date: Fri, Sep 21st, 2007 @ 07:45 ( . )

On 09/21/2007 @ 07:37, John Hynes wrote : I think Class BC describes an amplifier constructed of tubes made out of mammoth tusks, with filaments made out of trilobite antennae and evacuated by leaving the assembly sealed against a glacial fissure as the ice recedes.:
: -John
LOL!
*Caution* Once the Heavy Metal dudes hear of this they'll be bangin' down your door to get hold of one these amps.

-- REPLY: [With No Quote] --- [With Quoted Text]

'What about Class-BC ?'
Author:Francis Vaughan (registered user: 712 posts )
Date: Fri, Sep 21st, 2007 @ 07:47 ( . )

:-)

At least for a PP amp it makes sense. Simply underbias it. Sounds dreadful. (I shouldn't say that, I'm sure it is just the sound someone out there wants.) Has the curious nature that until the input signal gets above a given level there is no output, and it gets cleaner the louder is gets (until it overdrives that is.)


-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Fri, Sep 21st, 2007 @ 09:04 ( . )

Heh, I have always said that in fact. The answer lies in the sacred oracle of the load lines. :o) Voltage (anode & screen), bias and load all determine the class a tube operates in.

The problem was whether Class AB amps, by definition, must have two load lines. Apart from the RF amplifier example, I also have a hunch that OT-less Class AB PP amps also have a single load line. Any experts on OTL amps around, who might be able to confirm or deny that?

On 09/21/2007 @ 06:37, John Hynes wrote :
:
Hey Zaph - did you just say that the load matters in determining class? <LOL>
:
: J/K - forget that I asked.
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:CarlB (guest: search)
Date: Fri, Sep 21st, 2007 @ 17:01 ( . )

Hey Phil,

I'm not an expert on the OTL designs, but from what I have seen ...

The ones I've seen are push-pull. If they're biased AB, both tubes help drive the load on both sides of the small signals. As the signal increases, each helps the other less out as they start to approach cutoff. Finally at cutoff, only one is driving the load.

I don't think those are biased class A, but I could be wrong.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Bryan James (registered user: 450 posts )
Date: Fri, Sep 21st, 2007 @ 10:36 ( . )

On 09/21/2007 @ 06:26, Zaphod wrote :
What you are describing is typical Class AB behavior. On lower signal levels they appear to work like a Class A amp. With moderately increased signal levels they go into cutoff for part of each cycle. But an examination of the load line makes it quite clear that the power tube's operating conditions are set for Class AB operation.
:


See here's part of my problem with this. is that when I was in school, it was taught that a non-balanced q setup on a class A amp was still a class A amp. even if you where in cutoff or saturation, it was still a class A amp, yes this was with solid state, but that's the way the classifications where taught. and where probably a lot of people will be come.

I understand that a small amount of signal in an AB amp will be clean. and if you on got a clean signal at the volume set to 1 or 2, then either a distorted or clipped signal out pass that, I would it's class ab cuase through the majority of it's opperation it's in class AB, but if I got a a clean signal till 6 or 7 before clipping, I would argue that it's a class A amp, even if pass 7 it would be an overdriven class A


-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Fri, Sep 21st, 2007 @ 11:06 ( . )

However, the most fundamental definitions of what the classes are state that if the amp is in cutoff between 1% to 49% of the cycle then it's Class AB. Now I got taught a bunch of other slightly misleading stuff as well, which is why it took me a while to readjust back to the fundamental definitions, when applied to SE amps. Like I said, we should look at the loadline(s) to see how the amp has been designed.

Now whether the amp is clean up to 2 or up to 9 is a completely relative thing. If the amp is a warm-biased Class AB, then it will start to hit cutoff later, but it's still Class AB, as the load line will show. As another example consider a Class AB PP amp. The tubes operate in Class AB. Pull one side of the pair out. The remaining tube(s) still operate(s) in Class AB, but now it's an SE amp. It will also tend to distort a lot, but in our case we don't mind that. :o)

On 09/21/2007 @ 10:36, Bryan James wrote :
:
: See here's part of my problem with this. is that when I was in school, it was taught that a non-balanced q setup on a class A amp was still a class A amp. even if you where in cutoff or saturation, it was still a class A amp, yes this was with solid state, but that's the way the classifications where taught. and where probably a lot of people will be come.
:
: I understand that a small amount of signal in an AB amp will be clean. and if you on got a clean signal at the volume set to 1 or 2, then either a distorted or clipped signal out pass that, I would it's class ab cause through the majority of it's opperation it's in class AB, but if I got a a clean signal till 6 or 7 before clipping, I would argue that it's a class A amp, even if pass 7 it would be an overdriven class A
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 15:34 ( . )

On 09/19/2007 @ 15:11, Zaphod wrote :
Merlin, I originally learnt the class definitions during my teens from a number of British electronics books I borrowed from the local library, among them the RSGB Handbook c1970. I was into shortwave and amateur radio at the time, and I read every book on the subject I could get my hands on.
:
: As I remarked elsewhere, it appears that in the RF field the purely conduction angle approach is favored, whereas in audio they generally prefer to limit Class AB to PP amps. In audio Class AB SE is a useless concept, while in radio it isn't.
:
: On 09/19/2007 @ 14:31, Merlin wrote :
: Chrish, is this the longest thread on AX84 yet? Funny how vehemently Americans will defend such an obvioulsy flawed idea! Oh well, it takes all sorts!
:
--



Sounds good. I say that, in light of all that you've said (and I agree) that, having been forced to recant my original statement that "AB SE can not exist", I offer a new statement:

AB SE Only exists as a useful concept in radio engineering, is useless to the discussion of audio circuits, and therefore, does not exist in AF, de facto.

Also, Randall must either disprove this line of reasoning or take back his Dilbert dance. Or buy me a beer.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 19th, 2007 @ 15:49 ( . )

That's the trouble, John.

I just ran a VERY simple simulation of an EL84 stage, biased in AB as these guys suggest. It clips the bottom before the top. Then I swapped the load for a tuned 1MHz tank and it still works the same, but the tank fills in the missing 1/3 of a cycle.

If you accept that an audio load can be swapped for a tank, at any given point of operation, the circuit is operating in AB.

I STILL say it isn't at audio, but the arguments stack up right smartly when you apply it to low frequency radio. And really, really badly at higher frequencies, but that's a different matter.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:CarlB (guest: search)
Date: Wed, Sep 19th, 2007 @ 16:59 ( . )

Is the P1eX hi-fi audio? It's not optimized for that. Guitar? Sure. There's a difference. Audio (hi-fi): maximize measurably clean power. Guitar: maximize good sounding power, the heck with the measures.

In the way that we use it, does the P1eX cut-off at higher acceptable output levels? Even before it reaches grid-clipping or saturation? Sure. So for our application, we have a use for an SE class AB1 output stage.

If we wanted to maximize measurably clean output power, we'd bias the P1eX mid-scale, so as to top clip and bottom limit at the same time. That would be SE class A1, but that doesn't sound as good, and it doesn't allow us to get the extra power that we want and can reasonably get.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zoe N Iain (registered user: 8138 posts )
Date: Wed, Sep 19th, 2007 @ 17:11 ( . )

I've probably agreed with you at some point in this thread, Carl, but in case I haven't.. I agree, biasing a SE stage off centre sounds better than hitting the thing right in the middle.

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 17:11 ( . )

On 09/19/2007 @ 16:59, CarlB wrote :
Is the P1eX hi-fi audio? It's not optimized for that. Guitar? Sure. There's a difference. Audio (hi-fi): maximize measurably clean power. Guitar: maximize good sounding power, the heck with the measures.
:
: In the way that we use it, does the P1eX cut-off at higher acceptable output levels? Even before it reaches grid-clipping or saturation? Sure. So for our application, we have a use for an SE class AB1 output stage.
:
: If we wanted to maximize measurably clean output power, we'd bias the P1eX mid-scale, so as to top clip and bottom limit at the same time. That would be SE class A1, but that doesn't sound as good, and it doesn't allow us to get the extra power that we want and can reasonably get.
--



So, just to be clear: You don't think the only defining criterion for assigning class of operation is conduction angle?

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:Zaphod (guest: search)
Date: Wed, Sep 19th, 2007 @ 17:32 ( . )

Going by that last post, I think he still very much does.

On 09/19/2007 @ 17:11, John Hynes wrote :
:
: So, just to be clear: You don't think the only defining criterion for assigning class of operation is conduction angle?
:

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:CarlB (guest: search)
Date: Wed, Sep 19th, 2007 @ 20:44 ( . )

You bet, conduction angle.


-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 21:04 ( . )

On 09/19/2007 @ 20:44, CarlB wrote :
You bet, conduction angle.
:
:
--



Then why does symmetrical clipping vs. off-center bias matter? No matter where the bias is, a signal of a certain size (that is to say, sufficiently small) would not clip at the given bias. It would conduct for the full 360 degrees. It'd be class A, then? So, if I drive the snot out of it, so 50% of the wave gets clipped with the bias sufficiently off-center, but 50% still "intact", that is to say, in cutoff for 50% of the cycle, now it's class B? Before it reaches that extreme, it's class AB?

Hmmm. Class would seem not to have any real meaning then. Unless we say that the classification is only measured by conduction angle at a certain signal level. How about at clean output? This seems to fall in line with your suggestion that an amp that clips perfectly symmetrically when driven hard enough is still class A, even though at that point it isn't conducting 360 degrees of the cycle anymore.

Of course, that would make it impossible to have a single-ended class AB amplifier, and we're back to square one.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:CarlB (guest: search)
Date: Wed, Sep 19th, 2007 @ 23:58 ( . )

"Then why does symmetrical clipping vs. off-center bias matter?"

Heh! You were just waiting for that, I could just *feel* it.

That's a very good question, let me see if I can get through the answer in a way that makes sense.

Let's consider a class A1 stage. By conditions of the A1 definition, it will grid clip. OK, so this is also the maximum limit that we can expect our 360 degree conduction condition to apply for.

So, what has to be happening to ensure we have 360 degree conduction, all the way out to our max limit that applies? Not cutoff first, because that would not be a full 360 degree conduction for the output. For class A1, the stage has to biased so as to have grid-clipping happening first, then cutoff.

Unless I'm missing something really basic, which has been known to happen ...

In practice, we do put the stage at about midpoint of grid-clipping to cutoff, to maximize clean output power, and call that "Class A1." To be pedantically correct however, we really need to be sure that the stage is biased such that grid-clipping happens first.

Let's take a look at class AB1. Well, if grid clipping were to occur first, it would be class A1, from the above. So, cutoff must happen first. Or pedantically cutoff and grid-clipping could be occuring at the same time - but that would just rub too many people the wrong way and "street practice" is "that's class A1," so I wouldn't go pushing the latter.

So ... what are your thoughts on the above?

-- REPLY: [With No Quote] --- [With Quoted Text]

'Hey ChrisH?'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 21:20 ( . )

Actually, let me think on this for a bit.

I'm currently re-reading the Reich passage that Randall mentioned yesterday. Reich considers bias in the formal definition as well. He does not, however, specifically mention signal level. At least, that I've seen yet. I'll read some more and investigate some other texts and see what I get.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'OK, Let's broaden the definition'
Author:John Hynes (registered user: 2628 posts )
Date: Wed, Sep 19th, 2007 @ 21:46 ( . )

I've just reviewed two Reich texts, and the sections on amplifier classes in RDH4, as they pertain to both SE and PP in audio.

Here's what I think the definition must include consideration of:

1) Conduction angle
2) Bias
3) This is implied by (2), but the actual measurement of conduction angle is taken at the onset of maximum power output. The condition of the waveform doesn't matter in that case, and fits both SE and PP into any class.

Reich names (1) and (2) specifically in both texts, and says so in a way that indicates that he's quoting an earlier source text (he says "the definition...") The maximum power output criterion is implied by the inclusion of bias conditions in the definition. Sooo....

...how's this sound:

"Amplifier classes are determined by conduction angle at the point of maximum power output..."

We all know the rest. This also accounts for slightly higher power outputs in "2" subscript stages.

Clean & simple. Seemingly beyond reproach. Backed up by various sources.

How's that?

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'OK, Let's broaden the definition'
Author:Francis Vaughan (registered user: 712 posts )
Date: Wed, Sep 19th, 2007 @ 22:15 ( . )

"Amplifier classes are determined by conduction angle at the point of maximum power output..."


Still no good. Power continues to rise as the the stage is overdriven. Drive any stage hard enough and you will get a square wave out of it - and it is at this point you get maximum power.

But you can't use maximum undistorted power - since that will, by definition, require 2pi radians of conduction angle. And for a SE stage is therefore class A - by definition.

I'm still trying to keep out of the argument, but this underlines the problem with these definitions for guitar amps. For anything but a squeaky clean amp, all the designs transit through the various operating conditions dynamically. The operating conditions are useful when a description of the various tone generating mechanisms is needed, but you can't take a guitar amp and proclaim through inspection of the schematic what class any stage operates in. About all you could do is mumble that class AB1 is sort of close, but that that doesn't help understand. Like I asked before. Try to describe the operation of a Wreck. By themselves these classes are not very useful.

-- REPLY: [With No Quote] --- [With Quoted Text]

'OK, Let's broaden the definition'
Author:CarlB (guest: search)
Date: Wed, Sep 19th, 2007 @ 23:33 ( . )

I think I agree, Francis. But heck I don't know at this point. Let me try to answer John ...

-- REPLY: [With No Quote] --- [With Quoted Text]

'OK, Let's broaden the definition'
Author:John Hynes (registered user: 2628 posts )
Date: Thu, Sep 20th, 2007 @ 06:38 ( . )

Thank you for that, Francis. Since the postulate I put forth prior is wrong, I agree wholeheartedly. My initial "scenario" of transiting through classes definitely applies.

-John

-- REPLY: [With No Quote] --- [With Quoted Text]

'OK, Let's broaden the definition'
Author:Dan N (guest: search)
Date: Thu, Sep 20th, 2007 @ 00:10 ( . )

In response to some of that later statements made:

Several texts state that Class definition is determined from the maximum *expected* signal amplitude to be seen at the grid.
And That an A1 amplifier ceases to be once the output amplitude can no longer increase in a constant proportion to the input.
So this would include clipping of the output wave form whether it be from saturation or cut-off.

Adhering to a strict definition of the terms: A conduction angle of 360°-(1°/infinity) can not be class-A.
Though for such a small amount the human ear can not detect it, nor for that matter could any instrument on this planet be sensitive enough to measure it, but mathematicaly it does exist.
So as Carl puts it, a SE class-AB only just go into cut-off to qualify. But because we can not measure it we would never know if it was class A or AB.

Matters on Wave form:
Most of the texts and from what we have discussed, *assume* a symmetrical Sinusoidal waveform at the input.
There is nothing in the definitions to suggest that it must be that way. The texts simple state that the output should be a faithful reproduction of the input.
Suppose we where to apply a asymmetrical non-uniform wave to the grid, eg. one side of the DC-bias is Square and has say an amplitude of 90% of the peak-peak value for 300° of the cycle and the remaining 60° is sort of sinusoidal.
The centre point of the resultant R.M.S. value will shift the DC-bias point.
So for a sustained signal of that form can we still measure and define the class?
This sort of wave form is not that uncommon for a snot driven guitar pre-amp output.


Later on I want to revisit the grid conduction thingy ;)

-- REPLY: [With No Quote] --- [With Quoted Text]

'STOP THIS THREAD'
Author:Balijukka (guest: search)
Date: Thu, Sep 20th, 2007 @ 00:35 ( . )

and start a new one
it's getting too long.
took me 14 minutes to download

jukka

-- REPLY: [With No Quote] --- [With Quoted Text]

'STOP THIS THREAD'
Author:Mark L (registered user: 1458 posts )
Date: Thu, Sep 20th, 2007 @ 06:47 ( . )

I reakon this thread would break the record for the longest 8^P

-- REPLY: [With No Quote] --- [With Quoted Text]


--- 38 Users Online --- 26 Recent Unique Posters

Q8673=1485233902 - Threads: / 1485233902