That may be because your amp is also critically damped, and that was exactly what triggered me to start the discussion.
Just like profdc9’s amp could be made critically damped with OITPC even without input filter, something that TPC can’t do.
But again it’s nothing more than a suspicion.
Hans
Just like profdc9’s amp could be made critically damped with OITPC even without input filter, something that TPC can’t do.
But again it’s nothing more than a suspicion.
Hans
I’m surprised you think a LP filter on an amplifier input is cheating Hans!
I’d argue it’s a critical part of amplifier compensation and you get the added benefit of RFI reduction.
In a VFA, it limits the input LTP max differential and when used in conjunction with emitter degen ensures that the input stage remains operating in its linear region. For CFA power amps, you can also get the input stage to transition to class B with a very fast rise time signal. The input filter prevents that. You can’t use phase lead comp in a CFA (the zero at HF causes instability problems - but you know that anyway).
http://hifisonix.com/wordpress/wp-content/uploads/2017/10/CFA-Front-End.pdf
For VFA, some pics
I’d argue it’s a critical part of amplifier compensation and you get the added benefit of RFI reduction.
In a VFA, it limits the input LTP max differential and when used in conjunction with emitter degen ensures that the input stage remains operating in its linear region. For CFA power amps, you can also get the input stage to transition to class B with a very fast rise time signal. The input filter prevents that. You can’t use phase lead comp in a CFA (the zero at HF causes instability problems - but you know that anyway).
http://hifisonix.com/wordpress/wp-content/uploads/2017/10/CFA-Front-End.pdf
For VFA, some pics
Attachments
Last edited:
It is cheating. The LP filter does nothing to fix an improperly compensated amplifier. The input LP filter is there for a different purpose, to filter out RF ingress that may saturate the input stage.
Amplifier stability is a feedback loop property, the input LP filter is not in the feedback loop. It is true that some stability problems can be triggered by a fast input signal (a classic example would be phase inversion in very old op amps, which would, as a result, latch up toward a rail) but otherwise the input LP filter has no bearing with the amplifier intrinsic stability. If for example the amplifier becomes unstable when the output approaches the rails, then no input filter will help.
That is where OITPC shine. Attached LG plot with input filter disconnected and with DC offset of 1.7V(close to clipping) at the input.
Attachments
The situation is like a balloon. Poke it one way and it will bulge in another. Compensation schemes are for achieving/improving stability. Better stability can be achieved but with often worse phase rotation within the audio band. Open loop bandwidth is really what determines phase performance. Then the trade off becomes gain vs phase shift. People chase fast input stages because subsequent stages must have lower speed in order to meet stability criteria.
Oh, that is surprise, correctly compensated
I see 1.7MHz ULGF, which is still high, but not absurd, and a 2 pole shaped phase (which is not even conditional stable). I can not evaluate all the feedback loop in OITPC, and the results as shown do not characterize the overall stability. Otherwise said, my estimate is a "black box" estimate that is agnostic of the compensation method.
I never said OITPC is wrong, only that a) doesn't seem to have any advantage over any other 2 pole compensation ref. loop gain and phase margin, in an apple to apple comparison and b) the stability is not proven, considering only the stability of the major loop(s) is not even close to proof.
In general, I reject any results based on simulation, without a satisfactory detailed explanation on how this compensation works, the design backgrounds and the design rules to extrapolate it to other amplifier topologies. Otherwise, look at your own THD results above, 0.13ppm @1W/8ohm, you must be joking aren't you? You are showing 70 something dB loop gain @ 2KHz and 3KHz, meaning your simulated open loop amp has not more than 0.05% open loop distortions (2nd and 3rd harmonic) who are you trying to fool with this result?
I am sure you measured yourself this amp and confirmed the simulation results. Please don't quote Richard's measurements, I trust those less than a simulation.
If you are hoping for my endorsement for OITPC, please don't waste your time. You have a lot of work to make OITPC a serious solution (if at all) for compensanting an amplifier.
If you are hoping for my endorsement for OITPC, please don't waste your time. You have a lot of work to make OITPC a serious solution (if at all) for compensanting an amplifier.
Who said anything about the amplifier becoming unstable at the rails? Or latch-up? Or that the filter fixes an improperly comp’d amp?
An amp gets comp’d to handle a certain BW and the filter is part of ensuring that.
You are right.
I suspected the way of settling as a possible cause of a worsened sound with TPC, but it could just as well be the phase relation between harmonics that became different.
That sounds even a bit more likely as a possibility.
But anyhow, I’m more than satisfied without TPC and was just curious to hear other peoples listening experiences.
Hans
The input stage filter cannot be totally divorced from the compensation mechanism. This is especially true of BJT input. To test the stability of the amp with a filter, a signal now needs to be injected into the feedback loop to see how it settles.
One would think settling time might have an impact on sonics. Difficult to prove. From an engineering stand point, good settling from input step and feedback step disturbance are desirable to show the amp never loses control.
One would think settling time might have an impact on sonics. Difficult to prove. From an engineering stand point, good settling from input step and feedback step disturbance are desirable to show the amp never loses control.
+1
The goal is the specification (technically) and sound good/accurate. Input filter is part of amplifier. PSU also. All in "black box".
Somehow I don’t think you understood my message. Nevertheless, I would appreciate if you could explain how does, in your view, the input filter affect the amplifier stability.
In my view, the main role of a compensation method is to provide the stability of the feedback loop, and not to “handle” (whatever that means) the amplifier bandwidth.
I understood you message. I posted up screen shots. You know exactly what I’m showing.
You’ve conflated my comments about limiting BW to ensure good amplifier behavior specifically with loop stability which is not what I said.
I consider making sure an amplifier is well behaved with any kind of input signal or load part of the compensation design. Hence, limiting the input signal BW is part of the job - I posted a screen shot a few days ago of the sx-amp comp design as it was evolved as well.
But, there is not any point in having a reasoned discussion with you Syn08. This will quickly degenerate into a slinging match and I am tired of it.
I’ll continue to do it my way. You do it your way.
You’ve conflated my comments about limiting BW to ensure good amplifier behavior specifically with loop stability which is not what I said.
I consider making sure an amplifier is well behaved with any kind of input signal or load part of the compensation design. Hence, limiting the input signal BW is part of the job - I posted a screen shot a few days ago of the sx-amp comp design as it was evolved as well.
But, there is not any point in having a reasoned discussion with you Syn08. This will quickly degenerate into a slinging match and I am tired of it.
I’ll continue to do it my way. You do it your way.
For the purpose of DIY, you are entitled to you own definition of "compensation". Wrong, but it is yours and nobody can take it away from you.
P.S. I was trying to be polite and considerate - unfortunately this compensation method did not make you stable enough to have an educated discussion.