This need not be the case. If loading of the TPC network on the TIS's output is an issue, simply connect the larger of the two capacitors to the input of the TIS. Problem solved.
Note quite, Mike, at least in regard to TPC and TMC behaving the same way with the same network, as you assert.
TMC suffers with a large capacitor for C1 and a significantly smaller capacitor for C2. Even for C1 = C2 there will be a difference between TMC and TPC in global loop gain. Read Doug Self's article on TMC in vol. 0 of LA and you will see why. He used C1 = C2 and saw the effect, but he did not go into the matter of non-unity ratios of C1 to C2.
TMC likes it just the opposite: small C1 and large C2.
Cheers,
Bob
Oops! I should have said that the open-loop gain between TMC and Miller compensation with the same total series capacitance will differ, and that is why TMC does not like large C1 and small C2. TPC and TMC global loop gain obviously differ.
Cheers,
Bob
Actually, I was refering to the issue of optimising TPC in respect of the current drawn by the compensation network from the output of the TIS. Nothing at all to do with the comparison between TPC and "TMC".
So, Bob, when are you going to respond to this:
http://www.diyaudio.com/forums/soli...lls-power-amplifier-book-373.html#post3547822
Last edited:
Link doesn't work for me because I have threads set to show more posts per page than the default (for me this thread is currently 75 pages long). Please could you provide the post number?
Dont forget 3733 if you have some time left , it would
be interesting to have your conclusions about said observation.
Wahab, I am not persuaded that the loading of the TPC network on the output of the TIS is a significant issue in respect of distortion in the audio band. It might be an issue with regard to slew rate if the larger of the two TPC capacitors is connected to the output of the TIS.
With regard to distortion, however, the frequency at which the loading of the network becomes significant should be well beyond audio frequencies and should therefore not be a problem.
With regard to distortion, however, the frequency at which the loading of the network becomes significant should be well beyond audio frequencies and should therefore not be a problem.
Hi Mike,
Yes, I accidentally responded to the wrong issue. So, if the matter is confined just to TPC and VAS loading, you are correct; a large C1 and a small C2 will obviate the VAS loading concern. In fact, Harry Dymond specifically mentions TPC arrangements with C1 large compared to C2 in his AES TPC paper.
The question you did not ask that I inadvertantly responded to is that large C1/C2 ratio, if used with TMC, will result in a different loop gain around the output stage and a lower ULGF for that. Therefore, this is a case where the same network will not give the same loop gain behavior for TPC and TMC.
I'll try to respond to that other post later this evening, but over here we are about to celebrate July 4, so family may intervene and delay things.
Cheers,
Bob
Cheers,
Bob
When using the same components values , here 220pF ,
47pF and 5k , TIS/VAS loading induce 20dB lower total gain
for TPC in comparison of TMC and that is above 1kHz ,
inverting the capacitors as to present the lower value
to the output will forcibly reduce said loading but still ,
TMC will enjoy 10dB more total gain above said 1kHz.
The simuation is done with output stage connected ,
so we can see that compensation induced loading has
a significant influence.
Of course , as you point it , distorsion results are
practicaly the same , not because of an alleged
identical loop gain around the output stage , wich
is not the case since the available OLG at audio freq.
is not the same , but surely because of lower distorsion
of the input stage with TPC , wich compensate for the
reduced available gain.
I'm not going to take the bait, Mike. I don't trust your simulations and I'm sure nothing I simulate will convince you.
I do notice that you and Waly seem to disagree on this matter. so I suggest that the two of you duke it out. That will provide us some needed entertainment.
Cheers,
Bob
That's ungallant of you Mr. Cordell.
I'm sure MikeK's sims of his own circuits are 'accurate' and his 'real life' designs exhibit the wonky stability and poor THD that they show.
That's an excellent idea! Us ex-colonials, both loyal & rebel, will make it a holiday and provide beer, BBQs & fireworks
Then you seem to have jumped to a conclusion.
The approximation is often made that the distortion reduction must be proportional to the Loop Gain.
This is not true universally. Bode realised this back in 1939 and discussed it in detail.
Cherry worked a practical example in his 1982 JAES articles and showed it does not apply to most parameters for a typical audio amplifier.
I know Richard Lee is a Cherry fan and I assume his question is because he has picked up on this too.
This is reason for my frequently repeated comment that TPC and TMC are only equivalent to first order.
All discussed in next issues of Linear Audio. I do less horrific maths than Bode but I think it stands up.
Best wishes
David
Last edited:
This is a classic Bob Cordell cop-out: you reject hard evidence provided without providing grounds for doing so, or even checking its reproducibility or veracity, and then refuse to provide evidence to back up your unsubstantiated averments.
Case in point: on what grounds, precisely, would you aver that you don't trust my simulations? Have you even tried to reproduce them? I doubt it.
Therefore, I am wholly persuaded that your assertion that "...simulations with different ratios of C2 to C1 where the ULGF has been held the same and the amount of loop gain about the output stage at 20kHz has been held the same..." is merely a figment of your imagination with no foundation in fact: pure science fiction.
Last edited:
This is a classic Bob Cordell cop-out: you reject hard evidence provided without providing grounds for doing so, or even checking its reproducibility or veracity, and then refuse to provide evidence to back up your unsubstantiated averments.
Case in point: on what grounds, precisely, would you aver that you don't trust my simulations? Have you even tried to reproduce them? I doubt it.
Therefore, I am wholly persuaded that your assertion that "...simulations with different ratios of C2 to C1 where the ULGF has been held the same and the amount of loop gain about the output stage at 20kHz has been held the same..." is merely a figment of your imagination with no foundation in fact: pure science fiction.
This kind of personal insult is the reason I generally don't like to have anything to do with you.
I've done hundreds of simulations, with many combinations of C1, C2, and R1, many of which I have also done with excess phase added. Also built a test amplifier where I was able to compare TPC, TMC and straight Miller on the very same setup.
You just need to do your own due diligence before you continue to shoot off your mouth with irresponsible statements. I'm not going to do your due diligence for you. You should convince yourself by doing lots of simulations where you keep an open mind about what combinations to try, how to evaluate the results, and how to draw fair, unbiased conclusions.
Many others here have questioned your simulations. You need to do better if you want some credibility.
Bob
Can't blame you, Bob. Irrespective of who's right or wrong, his bedside manner is atrocious.
Absolute rubbish! Where exactly have I insulted you?
You've clearly not done the simulations you claim otherwise we would no doubt have seen the results here.
As for building an amplifier to test loop gain of TPC vs "TMC", I don't believe a word of it! Why then didn't you enlighten us before on your "findings"?
Instead of responding to a reasonable request to back up your assertions, you attempt to sidetrack us by claiming I am "shooting off my mouth".
This is hardly the first time. You, for instance, responded with precisely the same diversionary gambit when I presented you with irrefutable proof, courtesy of Solomon, in respect of the "VAS" vs TIS debate.
The rest of your response doesn't merit further analysis.
Last edited:
