• Check out the Annual AK Secret Santa exchange Program, signup before November 23rd in the AK Charities & Christmas forum!

Examples of Class A push/pull

The term "transient" is refered to a lot, "a lot" but it never seems to get much inspection. Never really defined in the conversations that use it when it's being used to characterize some factor in amp output performance and is usually blamed for bad affects to sound quality. What does the transient do? Does it create a clipping event or is it just part of the normal signal amplitude envelope? Is it a random noise event with no repetition? What is a transient's duration? Is it from a hot recording that has percussion signals above 0dB? ....How do we make sure tranients that are negatively affecting the (audible) SQ don't get into the inputs? When should we decide the source material is crap, ....so many transients?

Transient just means an abrupt change. Such signals can provoke issues you won't see by testing amplifiers exclusively with repetative wave forms like sine or square waves. One example, would be power supply ringing. Another would be some types of oscillation.
Another would be a compression effect, caused by a reduction in gain related to power supply sag.

The idea of eliminating them from the input signal is flawed, unless you plan to listen exclusively to test tones, because music is full of transients.

The ability of an amplifier to respond to transients is part of its overall fidelity.
 
I'm sure there is some technical definition of transient, but basically short duration, high intensity is it. The technicals might include specific durations and db above average or whatever but I do not know what those numbers are.

What happens in cathode bias is the more current you pull, the more negative the bias gets. Too much negative bias generates crossover distortion. If you start ramping up the power on a cathode bias amp, what you're likely to see as it gets close to clipping is a distinct "step" in the sinewave as it crosses zero. Thats the under-bias condition. Brief pulses that discharge the cathode bypass cap (or even worse if it doesn't have one) can spit out a waveform with that step in it where it shouldn't be.
 
Again, not defined. You're making my point.

I suggest consulting a textbook or glossary to get a definition of unfamiliar terms - here's one I found online, but there are a lot of others. As always, context plays a role.

Transients — An element of a sound where the spectral content changes abruptly. Most natural sounds start with a transient element before settling into something more steady-state, and it is often that transient element that provides most of the recognisable character of the sound source.

Provoking is certainly a bad thing, I guess. Especially if it becomes an issue. Why wouldn't a square wave or a sine be able to draw out the affects, why only a music signal? Can I just scope the B+ with a music signal at 100% and if it's stable, (or what would be a failure point), it passes the transient test? Is there a song that has all the requisite transients needed to do a proper test?

A sine wave at full power will just reduce the rail voltage. As soon as you stop the signal, the rail voltage will take time to recover as the filter capacitors charge. There are establish tests for transient distortion which use bursts of full power followed by no power, at different intervals. Have a look at the Audio Cyclopedia, available online for an example. It is not necessary to use music or guess work to do this testing, but it may exceed what you can do at home with basic test gear.

https://archive.org/details/AudioCyclopedia/page/n1451

See 23.49 for an explanation.
 
Nothing more abrupt you can apply than the front of a square wave.

Yes, but a square wave is periodic... over time it puts a continuous load on an amplifier, and the effect on power supply and bias voltages is averaged out by the time constant of the power supply and cathode bypass capacitors.

When you remove the load entirely, by switching it off, all these voltages now start to recover and change, where under a constant square wave, they remain stable. I really suggest to look at the reference I provided above, it explains it quite well.

I suppose an easy analogy if you're a car guy, is carburettors - have you ever had a car, where works fine when you gradually increase the throttle, but any abrupt changes make it misbehave? For example, if you floor it, and it almost dies.. or if you're going along full throttle and then completely let off the gas. That's a mechanical analogy to how an amplifier sees a transient.
 
Last edited:
We still have no lab quality definition of a transient and if it's only available in a music stream, what's the test to determine whether we should jirk, jerk, jurk, .

I sent you a link above to an excellent section from a very well known book about all things audio on how to test for transient distortion using off the shelf test equipment. I'm not sure I understand the source of confusion.

Also - the fact that transient distortion exists, isn't a guarantee that any cathode bias AB amp is definitely going to "sound like crap". It's possible to use negative feedback, appropriately sized bypass capacitors, and engineering judgment to minimize this problem to the point it really doesn't matter too much. Going Class A is the "brute force" approach, there's all sorts of reasons why you would rather use AB.
 
Last edited:
Just quote it with a citation. It's OK. Be the one with specifics. My original point up above is still holding.

I gave a link to the page, and provided the section number... re-typing a textbook for someone too lazy to click a link exceeds the level of effort I'm interested in putting into something I'm not getting paid to do - sorry.
 
Mind you this really is only an issue with AB amps since the current swings. The bias points don't shift much since the current draw remains pretty constant with class A amps.

So let's agree clipping is at the (+) peak and we stay just under it. That's nowhere near the sine 0v point.

The problem with AB amps and cathode bias is that under high output conditions, you can have both tubes in cutoff at the same time. This isn't peak output clipping, its actually the other direction. Say the idle bias voltage is 25v, under high power you might see that at 40v or so, which is not a voltage where at least one tube is conducting at all parts of the output wave. Peak clipping is at the top of the wave, crossover distortion is at the zero-crossing.
 
In push-pull amplifiers, a cathode bypass cap is mandatory in a class AB design -- but will invariably hurt performance in a true Class A design.

Dave
 
In push-pull amplifiers, a cathode bypass cap is mandatory in a class AB design -- but will invariably hurt performance in a true Class A design.
In the "class A" amp I built last year, that I made out of that pair of Fisher TA-800 OPTs, I tried very hard to get it right, I STILL needed a cathode bypass cap to get lowest THD out of it at full power, as measured with the THD meter. I'm still not sure why I needed that bypass cap. But now that I can look back and see the bigger picture, it might be that with fixed bias like I used, it was easier for the two tubes in the PP pair to amplify at different levels--i.e., the tubes were not perfectly balanced, nor could I make them so at low power and at high power at the same time. I could force balance at one end or the other, but not both. I couldn't have it both ways it seemed, so I settled for a cathode bypass cap.

So I'm not sure if I should have called it Class A, or Class AB1. I believe I had full 360 degree conduction for both tubes, it's just each side couldn't be balanced at both low power and high power simultaneously. I wonder if that is an artifact that is more easily seen with fixed bias than with cathode bias?
 
Its not just bad supply regulation, though that doesn't help any. The more stout the power supply, the more current can flow through the cathode resistor, and the more the bias ends up changing. The resistor does a lot to limit how much current can be drawn though, which is why fixed bias amps make more power.

A bad power supply makes for a bad amp though, no two ways about it. Everything that goes to the speaker comes out of the wall and through the power supply.

I think that condition or operating point would only exist by accident where an amp normally biased near B class would be dropped into deeper B for some odd reason. If both tubes are not conducting that is the lowest point of current draw the amp could produce. How can you have high output with no conduction from both tubes?

The only way to completely prevent this is to make idle current high enough that it can't dip fully into class B operation. While it can be done, it also beats up the tubes. Bias things at a more happy spot for long life and you get distortion just by the nature of the design. As for making power output with gaps in the conduction, thats not too hard. The peaks are high enough that it averages out the small gap near zero. Unless the bias point is so screwed up that it has big gaps, it really doesn't affect the RMS voltage output. Typical crossover distortion looks something like this, its not a big missing hunk of the waveform.

xover_550_61.gif


Precisely how audible this is, I don't honestly know but I know the distortion meter can hear it.
 
Quote: "There is no idle current ever in class B biasing for either tube." This is technically not true. RCA explains Class B operation this way in most of their Receiving Tube Manuals:

"A Class B amplifier is an amplifier in which the grid bias is approximately equal to the cutoff value, so that the plate current is approximately zero when no exciting grid voltage is applied, and so that plate current in a specific tube flows for approximately one-half of each cycle when an alternating grid voltage is applied." (Bold is my emphasis).

Therefore, there is no absolute defining value for the quiescent current of a Class B amplifier, except that by definition, it is (obviously) very low -- no doubt just high enough to ensure that neither tube does in fact cut off. If either or both tubes do reach cutoff or beyond, the distortion produced would be extreme. A practical Class B amplifier cannot operate with cathode bias.

Based on this definition, Bogen made good use of Class B1 operation in (at least) the MO-200A (and its siblings) and MX-60A amplifiers. In these units, the 8417 output tubes typically operate with a quiescent cathode current of just 12-15 mA, with no provisions to adjust the bias in these units -- likely just enough current to ensure that any typical example of the tube will conduct if even minimally so when installed. This is certainly well below the quiescent current required for proper Class AB1 operation -- and while those who are sticklers for detail can still claim this represents Class AB1 operation, it must also be considered such poor execution of it as to for all intents and purposes be more properly considered as Class B1 operation based on the intentions of RCA's definition.

Dave
 
In order to make sure you get no crossover distortion near full power you end up over-biasing at idle so the current is higher than it needs to be and that reduces tube life. If the bias points didn't shift this would not be needed. Cathode bias was used because it was cheap and easy, not because it had performance advantages.

so yes, it can be done, and from a technical standpoint its not even that hard, but its basically putting a band-aid on what is fundamentally a less than ideal design.
 
Cathode bias was used because it was cheap and easy, not because it had performance advantages.

Ttwo additional factors argue in favor of cathode bias:
Safety factor. A fixed-bias amplifier which loses its bias supply will rapidly destroy the power tubes and output transformers. Cathode bias is essentially foolproof, even though fools are ingenious.
Simplicity of Ownership. As the tubes age the bias adjusts without the need for manual intervention. This is important for most owners as biasing a tube exceeds the knowledge or skillset. Most couldn't change a plug on a line cord.[1]​

In practice, a power supply should not fail and changes in tubes with age are minor. But power supplies built using electrolytics, inherently unreliable, do fail and the the smallest unit of time perceivable by a human is the amount of time between "WTF, the sound just stopped" and "WTF, my output pair is magma, the output transformer is in flames, the rise in voltage chernobyled the other channel and all the tubes, my rectifier given to me by Schade himself is slag, my power transformer winding just went to Valhalla, and I'm only halfway through Inna-Gadda-Da-Vida!".[1]


[1] Caution: Hyperbole is not to be seriously or internally taken. Use Hyperbole only as directed. As your doctor if Hyperbole is right for you. If you are suffering from humor impairment or sleep deprivation do not use Hyperbole as it may aggravate these conditions. Do not combine Hyperbole with Sarcasm, Irony, or Snark as these may intensify its effects. Wash hands after using Hyperbole. Do not share your Hyperbole with others as they may not appreciate its subtleties.
 
In practice, a power supply should not fail and changes in tubes with age are minor. But power supplies built using electrolytics, inherently unreliable, do fail and the the smallest unit of time perceivable by a human is the amount of time between "WTF, the sound just stopped" and "WTF, my output pair is magma, the output transformer is in flames, the rise in voltage chernobyled the other channel and all the tubes, my rectifier given to me by Schade himself is slag, my power transformer winding just went to Valhalla, and I'm only halfway through Inna-Gadda-Da-Vida!".[1]
:rflmao::rflmao::rflmao:
 
Cathode bias operation using a common cathode resistance is more effective than fixed bias in true class A push-pull designs, as the common cathode connection allows for any imbalance present at the cathode connection to act as an error signal that is injected into the stage in common mode fashion to minimize distortion. Because the dynamic balance is never absolutely perfect at all frequencies and power levels, this is the reason why adding a cathode bypass cap to such (Class A) configurations invariably elevates distortion -- most particularly of the IM type -- and why Williamson specifically recommended against using a cathode bypass cap in his amp.

Dave
 
Ttwo additional factors argue in favor of cathode bias:

sure, that goes back to "easy". It doesn't produce more power or lower distortion. I'm not saying it has absolutely nothing going for it, but strictly from a performance standpoint in an AB amp it has problems.
 
sure, that goes back to "easy". It doesn't produce more power or lower distortion. I'm not saying it has absolutely nothing going for it, but strictly from a performance standpoint in an AB amp it has problems.

I'm not intending to quibble, but in this context the word "performance" is an vague umbrella term addressing how well the amplifier fulfills the intended purpose of increasing signals, not solely being limited to describing the magnitude of signal increase, which would better be described as "power output". So the meaning of "performance" best includes such concepts of "power output", "distortion", "stability", "tube longevity", etc.

Cathode bias spans many categories and thus impacts upon various aspects of the amplifier's design goals and consequently affects its performance.
 
Cathode bias spans many categories and thus impacts upon various aspects of the amplifier's design goals and consequently affects its performance.

increased performance, as in more power, less distortion. I'm a car guy, "performance" is understood in the context of "more", which only occasionally crosses paths with the question of how easy it is to live with. Cathode bias has its good points, but its got problems too. So does fixed bias for that matter. You don't buy a Ferrari to tow a boat, and you don't enter a Mack truck in a high speed road coarse.
 
Back
Top Bottom