Distortion is only important ting in an amp, better stability is important too.
OITPC improves Phase Margin and Gain Margin, and something higher ULGF is not problem in this case.
TPC and TMC are equivalent for OPS errors (I was the one who did the mathematical proof 🙂 ). For IPS errors, TPC is superior.
Using a CFA topology as Dadod did doesn't make the comparison easier, since the compensation options are less than for a VFA. In fact, other than shunt, Miller and the Ahuja compensation I can't think of any other CFA compensation topology (I could be wrong, though).
Actually, unless it is external (i.e. around RF) any VFA compensation works on a CFA.
No, first math proof was done way before your time here, by megajocke (who used more of a brute force approach) and myself (I used Delta-Wye transformation). The forum search engine is your friend.
Nope, for example you cannot use much of a Miller input inclusive compensation (MIC) in a CFA. Those three methods (shunt, Miller, Ahuja) are first order compensation methods, I should mention this.
Dadod: don’t fool yourself about increased phase margins with OITPC, everything else being equal. Audio amplifiers are minimum phase systems (at least as long as you stay away from the residual poles) and, as such, gain and phase are directly related. Otherwise said, there is no free lunch in stability vs. loop gain.
BTW, TMC is patented since 1978, and used probably well before that.
US4145666A - Multistage amplifier circuit
- Google Patents
Nihil novum sub sole.
US4145666A - Multistage amplifier circuit
- Google Patents
Nihil novum sub sole.
Other thing I noticed with OITPC that THD does not increase much with frequency or output power. That shows simulation and it was measured by Richard Marsh on real built amp.
My simulations show different. In OITPC THD rise when output level rise. In Miller and TMC, THD decrease when output level rise and at near clipping, THD will rise. I do not do many sim using TPC and I do not remember it.
I did mine because I did not like Megajoke. It was brute force, 0 intuition. You can find mine using search engine advice. It is in a yellow handwritten paper.
That is what I meant with "Actually, unless it is external (i.e. around RF) any VFA compensation works on a CFA."
In this very interesting discussion I agree that harmonic distortion is more of a technical issue with the ppm distortions of modern amps.
But a square wave is a nice way to get a good impression of the system's stability with various loads, not that unimportant IMO and not just important for video systems.
I favour the sound of an (almost) critically damped resistive loaded system, based on experience with just one amp without anything else but my own very subjective aural system, but that's exactly why I'm interested in this more general discussion.
So very important as seen in that light, you suggest a side by side comparison, side by side by in LTSpice or side by side in a listening test, I hope for the last.
And when proposing this, you must have your opinion and expectations on the outcome.
So It would be most welcome to hear what these expectations are as subjective as they may even be.
And how you did achieve these results, are they repeatable and of general value ?
Hans
Agree that the square wave test is a powerful tool for assessing stability with various loads. I always do this as part of the comp design.
In the sx-Amp (classic CFA design) write up there are some LTSpice screen shots as the comp design is evolved
In the sx-Amp (classic CFA design) write up there are some LTSpice screen shots as the comp design is evolved
Attachments
Last edited:
I have no idea what you are talking about, please explain. I never heard about either "external" and "around RF" compensation(s).
Guys, there is always a tendency in this forum to answer one thing for another and it beats me why...
Settling time is not that important, i.e. not a parameter to optimize. This des not mean square waves are not important. As Hans and Bonsai point out, they are important to measure stability.
So very important as seen in that light, you suggest a side by side comparison, side by side by in LTSpice or side by side in a listening test, I hope for the last.
I hope for the former... LTSpice or lab measurement. Since the latter would take a lot of effort, LTSpice is ok. Too hard to be objective in the latter.
Settling time is not that important, i.e. not a parameter to optimize. This des not mean square waves are not important. As Hans and Bonsai point out, they are important to measure stability.
So very important as seen in that light, you suggest a side by side comparison, side by side by in LTSpice or side by side in a listening test, I hope for the last.
I hope for the former... LTSpice or lab measurement. Since the latter would take a lot of effort, LTSpice is ok. Too hard to be objective in the latter.
Thank you for answerring.
So comparing with LTSpice is what you suggest.
But you never disclosed what your expectations are when comparing different topologies.
THD is not important as you mentioned earlier and the way of settling (not settling time) is not important, so what else remains that will make the big difference ?
So this discussion is probably rather pointless from an audio perspective.
Hans
So comparing with LTSpice is what you suggest.
But you never disclosed what your expectations are when comparing different topologies.
THD is not important as you mentioned earlier and the way of settling (not settling time) is not important, so what else remains that will make the big difference ?
So this discussion is probably rather pointless from an audio perspective.
Hans
If you comp an amp correctly and feed it with a bandwidth limited signal ie a proper audio signal there won’t be any overshoot and thus no settling time issues arising from ringing with a square wave stimulus.
(Resistive load)
(Resistive load)
I never said THD was not important. I said settling time is not important as a reaction to someone pointing out that settling time got improved due to OITPC. I pointed out that this was incidental and not because OITPC is geared towards improving settling time. It is geared towards imprving distortion.
After that I said, it would be nice to do a comparison of the different compensations side by side using THD as the barometer. This is tough to do though, since it is hard to agree what the amp to comp should be. I.e. I can show X compensation is better, but someone can then say, it is better on that amp and not this amp. Worse yet, we then get the guys who will say, but this compensation sounds better ... at that point we move from eng. to fiction and all bets are off.
Compensation schemes cannot be generalized for improving distortion in this way, or that way. Individual implementation determines how the phase rotates. At 90 degrees, feedback is not having much of an effect. The loop gain phase response indicates the distortion characteristics of the circuit. Dadod's OITPC can be used to alter the trajectory of phase by creating a valley followed by a hill. This enables a better phase margin. However, the valley often results in worse distortion at those frequencies.
And to my limited understanding, if you have a single pole the slope often starts in the audio band, and you have the 90deg in the audio band too. By implementing two (or more) poles, the slope can be moved higher, and there is less (or no) phase shift in the audio band?
Putting a low pass filter in front of a high BW amp with a low phase margin in order to get a critically damped response is like cheating yourself, it doesn't make a bad amp one bit better.
And there is no evidence whatsoever that a 1ppm THD amp, although possibly technically brilliant, sounds better than one with 1000ppm.
So as mentioned before, from an audio perspective this is all pointless.
My initial point was that one single TPC experiment with a Class A amp turned out to be in favour of single pole because the sound became a bit harsh leading to listening fatigue despite a much lower HF distortion.
All I could further measure was a difference in the way of settling, but that may or may not having anything to do with the perceived auditory differences.
So because this experiment has none general value, I was curious to hear listening experience from others with two pole compensation against single pole.
Hans