diyAudio (
-   Solid State (
-   -   Some questions about amplifier classes, distortion and efficiency (

Plecto 13th January 2013 10:45 AM

Some questions about amplifier classes, distortion and efficiency
Hi. I've recently began looking into the different classes of amplifiers and I've made a class A amp as well as a class B push-pull amp (I believe), but I'm struggling to see the differences between them.

I believe that when an amp is operated in class A, current will flow through the transistors even when there is no output. This eliminates crossover distortion which is why this design has been popular among the audiophools, right? What about this push-pull design:

Am I right to say that this is a class B push-pull amp and that a clean "Class B" amp is one where only half the waveform is amplified?

If the above schematic is a class B, what is a class AB then? I've seen someone saying that a class AB amp is an amp that has it's input transistors biased to a certain voltage so that the transistors are on up to a certain output level making it class A for the first few watts and then class B for the rest. This sounds reasonable, but isn't there a theoretical efficiency of class AB amps of 78.5% or something like that? Surely the efficiency would be worse when it's operated in class A and then much better when it's operated in class B? And the "avarage" efficiency throughout the whole output power range would then depend on how the transistors are biased? Let's say it's biased so that it's operated in class A for the first mW and then class B up to 50W, that would make in class B in practice, but technically a class AB amplifier, no?

I managed to calculate the theoretical efficiency of a class A amp to 25% which I'm happy about, but I can't get the same result for the amp in the image I posted. If the amplifier is ran at full power, at the point of clipping, won't then the voltage and current through the transistors be the same as the voltage and current through the load? When the voltage across the load is equal the the supply voltage, the voltage across one transistor is 0V. When the voltage across the load is 0V, the voltage across the transistor is equal to the supply voltage. I can't understand how this doesn't make the efficiency 50% as the load power is equal to the transistor power (when the power from both transistor is added together that is).

I also have a question about crossover distortion. When I try to simulate the class B push-pull amp that I linked, I clearly saw the crossover distortion, but when I made the amp (with BJT's) and scoped the output, I saw absolutely no crossover distortion and I couldn't hear any difference either when listening to a sine wave either. On wikipedia it says that because of this crossover distortion, class B push-pull amps aren't that viable for audio, but when using BJT's with a threshold voltage of ~0.7V and an op-amp input stage with a high slew rate, can this really be audible?

And about the difference between a class AB design (with it's transistors biased) and class A when it comes to crossover distortion. If the transistors are biased above the threshold voltage, I can't understand how there can be any crossover distortion at all regardless of output power. Even when the output power is high enough to completely turn off one of the transistors, that transistor will still start to turn on before the load voltage reaches 0V and begins to rise (or lower) to the other half of the sine wave, so there is no need for an input stage to 'overcome' this threshold voltage. What is then the point of a pure class A design?

Another question I have is about my newly bought Rigol DS1052E scope. It has a TTL function which I was happy about, I could finally see how much distortion my amps were having :D It didn't quite turn out that way since every amp I make has no visible distortion :( The other harmonics are down at the noise level and they become visible at the same moment I can see the sine wave beginning to clip and at the same moment I can faintly hear the sine wave becoming a little distorted. So I guess the question is, how much distortion is actually there when a sine wave looks more or less perfect on the little screen of an oscilloscope?

I didn't plan on asking this many questions, I sort of got carried away :P I appreciate any help :D

Ian Finch 13th January 2013 11:21 AM

It's time to read some educational articles rather than picking away with necessarily brief forum replies.

Start with this article and cover the different classes in following articles. There is plenty of relevant reading there.

Vostro 13th January 2013 12:33 PM

2 Attachment(s)
Here are two circuits to sim and build.
The transistors should have small heatsinks, incase :)

These circuits are intended for QUICK demo,
Don't leave them on, and keep an eye on transistor temperatures !!

They should work when breadboarded.

The second circuit allows to bias the bases abit pushing it into slight class AB.
Turn potentiometer from min to max and observe output waveform on scope.

In slight Class AB the crossover distortion is still there, because there are times that only one transistor is conducting, but it now happens above cuttoff (The zero volt line (gnd)).
This distortion if biased correctly is low enough for a slight Class AB amplifier to be considered HIFI.

Biasing into heavy Class AB isnt necessary for BJTs, others might disagree.


Plecto 13th January 2013 09:52 PM

Ian Finch: I have read the guide you posted, I've also read a guide about amplifier efficiency along with several other articles. I posted here because I still don't understand it. I was hoping that someone here might help me understand it by putting it in a different language :)

Vostro: Yeah, I've seen those circuits before. By "slight class AB", do you then mean an amp where the transistors are biased below the threshold voltage? Because I can see that the more bias you have, the less crossover distortion there will be, what I don't understand is how there can be any crossover distortion at all when the base/gate is biased above the threshold voltage. I can't see any crossover distortion in my simulations atleast.

CBS240 14th January 2013 05:30 AM

Because when you have both transistor's Gm (transconductance) contributing to the output impedance, and then one turns off so Zout comes from only one transistor. This change in output Z is what gives you the crossover distortion. It is much worse with mosfets because they loose significant % of Gm at lower conducting currents. This is why we use emitter (source) resistors of specific calculated value to minimize the % change in output Z as the output changes from small signals to large signals.

Vostro 14th January 2013 12:27 PM

The crossover distortion in correctly biased Class AB is so small, i doubt youll hear the distortion, hence (HIFI) and that means you won't see it on a scope either, but its there even if you can't see it :)

Plecto 14th January 2013 12:48 PM


Because when you have both transistor's Gm (transconductance) contributing to the output impedance, and then one turns off so Zout comes from only one transistor. This change in output Z is what gives you the crossover distortion. It is much worse with mosfets because they loose significant % of Gm at lower conducting currents. This is why we use emitter (source) resistors of specific calculated value to minimize the % change in output Z as the output changes from small signals to large signals.
Yes, I can see this being a factor, but with a proper input stage and feedback, how can this still be an issue?

DF96 14th January 2013 01:00 PM

Feedback reduces distortion. It does not eliminate it. Whether distortion is audible is a matter of argument, hearing ability and personal taste.

GoatGuy 14th January 2013 01:38 PM

Recall WHY we're "even remotely concerned" about all this class / distortion business!

Class "A" amplifiers have the nominal (no signal, quiescent) conduction of the amplifying elements (transistors, mosfets, fets, tubes, whatever...) set to a point of "half-conduction" - where they can conduct UP to twice as much before signal levels approach the power supply limits, and DOWN to zero in the opposite direction. The "magic" of either output transformers or DC blocking capacitors renders the output purely alternating current, it drives the speakers (or next stages), and life is good.

However, "Class A" power dissipation is HIGHEST when there is no signal! Essentially it is so both by definition and design... just the nature of it. There is no intrinsic distortion apart from the nonlinearity of the amplifying devices, and subtractive (negative) feedback can reduce that further (at the expense of overall gain). Power dissipation though doesn't change. When there is a powerful driving signal, power dissipation in a class "A" barely changes, on the average from quiescent. When powerful, such amplifiers run hot, and stay so indefinitely.

Again, recalling "history", the almost immediate realization of electronic engineers 'way back was ... well, if we pair amplifying devices, and have each of them running only half the cycle, then they could be much closer to cut-off ... which would both save a bundle of power (when quiescent) and double-to-quadruple the amount of output power!

PERFECT for "public address" situations, where POWER is needed to get sound out to the far reaches of a theater, stadium, public or large social event. Distortion - in the chaotic sound-stage of the theatre - hardly matters to a significant degree. So, classic "class B" near-cut-off designs ruled.

Audiophiles since recognized that their quest wasn't for efficiency, or ultimate power, but sweetness of the sound, coupled with speakers far more expensive than would ever be sacrificed on the altar of a public address system. They (we) were searching for musicality, for transparency of tone, of accuracy in reproducing the faint acoustic rarefactions and compressions of air we enjoy so much.

The CLASS-B design you posted first (direct-connected bases and upper-emitter to lower-collector) would work perfectly if the transistors were themselves perfect devices, and not subject to little quantum realities such as 0.7V emitter-base bias. But they are real, physical devices, with real, physical gotchas ... that need to be engineered out for the produced signal to be musically pure of tone, and relatively free from confounding distortions.

Vostro above points out two designs using a synthetic electronic-design "lab bench" program. His left-most design shows an oscilloscope that clearly draws the cross-over gap, the distortion of direct-connected transistors. The second diagram has a more sophisticated design that using resistors and capacitors works to bias the transistors further into their nominal conduction regions, which brings "the gap" toward non-existence. For the most pure signal though, both transistors should be biased WELL into their conduction region. It is the purists approach, and it works. it also consumes more power both on the average, and quiescent.

But ... when one spends thousands and thousands of dollars over years experimenting with building things, and getting better vinyl record players, better speakers, more elaborate preamplifiers and so on ... what's the cost of "wasting" a little more power to just dumb-as-a-rock heat? I say it is cheap. Budget about 15% to 25% of the potential peak power of the amplifier to simple Class A/B heat (or better, just class A heat at a lower power level), and be happy. The sound will be sweet.


CBS240 14th January 2013 02:25 PM


Originally Posted by Plecto (
Yes, I can see this being a factor, but with a proper input stage and feedback, how can this still be an issue?

Feedback fixes everything......this is non-sense.:rolleyes: Much of the crossover distortion components are well above the bandwidth of the global loop. Feedback does not get rid of the distortion but merely shifts it to higher order frequencies. Distortion components that are well above the 20KHz audio band limit can affect the sound for sure. You want the feedback to have to 'correct' for as little as possible. IOW, the goal should be that the global loop sees as linear of a transfer through the stages of the amplifier as possible.;)

All times are GMT. The time now is 10:42 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2