The Why And How Of Bias And Offset Adjustments For Audio Amplifiers

[@afzaalkhan.m]

Excellent video and clearly explained. Thank you

[@alreaud]

Excellent hands-on explanation!

[@imthelastone]

Very good explanation!

[@YOSEFPE]

one of the best explanation thanks !

[@chrysalide_electronics]

Hi, I have rarely seen this type of schematic with NPN to vcc and PNP to gnd, thanks for sharing. Jean-François

[@MadScientist267]

@chrysalide_electronics That's a standard "totem pole" configuration, two common emitter stages in series with each other. Also known as a "push pull".

[@VoidElectronics]

As @MadScientist267 said, this is standard. I found it a bit strange at first too, but it makes sense and it's the only way to use a complementary pair. If you put the PNP up and the NPN down, you get a dead short from Vcc+ to Vcc- through the base-emitter junctions.

[@MadScientist267]

Indeed. Maybe useful to point out the general purpose of it... it is used as a bipolar current amplifier, typically found in a driving stage. The lower current but "high" voltage signal comes in on the bases, and the emitters just follow what the bases are doing, less the Vbe drop. When one transistor is conducting, the other is in hard cutoff and vice versa. The emitters will be carrying more current than the bases, providing current (and in the case of a final, power) amplification for an AC signal... just what audio needs to drive a speaker. The Vbe drop mentioned causes a ~1.4V "gap" between the transistors, 0.7V x2, resulting in the mentioned crossover distortion in the video, which is what bias corrects, by bringing these two points together at zero in the output.

[@erikdenhouter]

I found that these settings sometimes are dependent of mains voltage, because the amp power supply is unregulated. So if you have a variac it is an idea to check that (let it again settle for 20 minutes), and set it to the proper voltage before setting the amp.

[@VoidElectronics]

@@erikdenhouter I didn't think of that, but it makes sense. Thanks for pointing this out! 😁

[@erikdenhouter]

@VoidElectronics Don't take my word for it, it is something I saw when I used an analogue moving coil meter for the bias, and traced it down to strong mains fluctuations (that time of day for washing machines). But it is worth looking into.

[@MadScientist267]

Ooof yeah that's oversight/minimalist design there. Happens when the bias is derived from the final's supply rails. A well designed amp will have the bias being formed from a regulated supply, such as the rails for the small signal components upstream. Very good point.

[@bobisyouruncle1]

Great explanation. Thank you.

[@louco2]

Great video, tank you.

[@TB-nz4kf]

Very good!

[@Leshiy_52]

I will make a small addition to the first circuit: add a resistor between the bases and the emitters of the transistors. Its resistance should be close to the load resistance. With low power, the pre-output cascade will work, which is quite capable of this task. As the power increases, a powerful output stage will be connected to the operation, without causing any visible signal distortion.

[@kennethcoutinho4439]

Awesome

[@keijojaanimets4460]

Make me a magic carpet?😁

[@keijojaanimets4460]

Good vid!

[@MadScientist267]

When you're looking at bias/crossover distortion and negative feedback is in play, reading from the output of the long tail will fly very obviously away from a sine when crossover is present in the final. When you have that waveform clean, it is set right as the feedback isn't having to compensate for the crossing. Do check for heat in the finals and verify quiescent current is reasonable (I've found "around spec" is fine as long as the heatsink stays cool at idle.) Well designed amps run a touch above lukewarm at the heatsink at room temp ambient conditions when the bias is set correctly. Then yeah your offset. That one is touchy indeed... should have been padded better heh. Oh well. Beats no pots lol Also, the scope can be made "more sensitive" to crossover distortion by feeding it the input vs output (speaker, long tail) on x/y... set up a 45 deg slope, and the crossover shows up as a "flat" spot in the trace that results (a straight, flat diagonal line indicates no crossover distortion, or compensation for it, in the signal). Other distortion is visible with this method as well, as any deviation from a straight line indicates output not tracking input accurately.

[@VoidElectronics]

@@MadScientist267 100% correct. I only figured out that I don't see distortion due to the feedback when editing the video. Maybe I'll do a followup.

[@MadScientist267]

@VoidElectronics You've got a good thing going here. Went ahead and subscribed. Was into audio design and repair in the 90s, always like to see someone's take on the core aspects of things. And someone who really understands old school analog audio at the design level is getting to be a thing of the past. Looking forward to more.

[@VoidElectronics]

@@MadScientist267 Thank you so much! Yeah, whenever I fix something I want to have a systematic approach just like shango066 so that the viewers actually learn how to troubleshoot, instead of swapping capacitors and hoping for the best. When doing design, I want to explain core concepts using the bare minimum circuitry. I will try to do some more up to date topics too, but I also think that old school stuff is better for teaching purposes.

[@waynegram8907]

What is the difference between CTAT compared to transistor thermal runaway? I'm not sure how biasing or setting the biasing point is affecting the CTAT and thermal runaway of transistors?

[@VoidElectronics]

The fact that Vbe is CTAT is inherent to bipolar transistors and has nothing to do with the bias settings. However, the fact that Vbe is CTAT could cause thermal runaway. If you drive the bases of the power transistors at constant voltage, the larger the temperature, the larger the collector current. As collector current increases, so does temperature, which causes even more collector current. Of course, this is an extreme case, because no amplifier drives the bases at constant voltage. However, if the Vd of the bias diode and the Vbe don't track very well due to different thermal coefficients, the amplifier could either lose bias at high temperature or go into thermal runaway, depending on which thermal coefficient is larger. I don't expect the coefficients to match exactly and the thermal coupling is not ideal either, so what I'm trying to say is that biasing the amplifier at a high current is risky. Also, the bias point drifts, as you can see in the video, so it's a good idea to keep an eye on it for like an hour.

[@LousyPainter]

Totally cool! Subbed with the bell.

[@VoidElectronics]

@@LousyPainter Thank you so much! 😁

[@sahhaf1234]

wont the offset come from the mismatch of output transistors?

[@VoidElectronics]

@@sahhaf1234 It could, but if the amplifier has DC coupled negative feedback, the output stage is in a feedback loop and the offset is compensated. On the other hand, the offset of the input stage is not compensated, so it propagates to the output.

[@sahhaf1234]

@VoidElectronics That's very correct. Thanks a lot for your reply. I have a question: I have never seen the mismatch of the push-pull pair analyzed anywhere. One should expect such a mismatch due to the fabrication and/or temperature differences between two output transistors, but the people doing programs about push-pull pairs treat the topic as if such differences never exist. In short, if you do a program on these imbalances you have one viewer. 🙂 Actually, such differences may not be that important in the last analysis because, as you have mentioned, they are within a feedback loop. But, still, it would be nice to see a mathematical analysis of them.

[@AIexanderHartdegen]

If you don't want a cross distorsion, please use a CLASS D AMP 🤣

[@MadScientist267]

Class D is great for "lotta power in a small space"... at the sacrifice of accuracy/fidelity. This isn't to say they aren't useful, they absolutely are, but have no place in a hifi stereo amplifier such as in the video. Class AB/+ for the win. Sometimes quality matters more than output.

[@AIexanderHartdegen]

@@MadScientist267 for sure. Class D have very agressive sound. I have both of them, class AB amp with H class (double power supply). Anway, class D sound metallic whereas AB amp (or class A before AB) sound warm and soft. Class D are nice for.. handeld bluetooth speaker. I must add something. Talking about 40 dollars class D amp has nothing to compare with a real high brand 3000 dollars class D amp. Some class A lovers heard a 3000 Dollars class D amp and... switched to class D. Highe price make sometimes the exception.

[@MadScientist267]

@@AIexanderHartdegen Personally other than the high efficiency/tiny package cases, I see them as useful for handling the bottom end with subs and such... which is where most of the power goes anyway. It doesn't need to have a high nyquist limit for that... and can save AB+ for definition, mids and highs where the power needed is much less. I don't doubt they are capable of making an amp that is pretty true to accuracy, but maybe not worth the flow.

[@AIexanderHartdegen]

@@MadScientist267 all tastes are in the nature... agree with you.