JohnW
pfm member
I keep thinking about minimum phase margin sounding best.
It doesn't innately feel right.
Artificially "Peaking" the signal seems like compensating for something which may be deficient elsewhere.
I don't dispute what you've observed.
I assume you're referring to phase margin in the output amplifier stage?
I guess though it also depends on what phase margin at what frequency, and also on the overall loop bandwidths, open and closed
I do agree, if only because its technically wrong (and feels like cheating), during simulations I target over 75Deg to approach critical damping - the only mention I could find about PM and sound quality was a rather Cryptic quote along the lines:-
"Everyone's aware that 82Degs sounds optimal..." Well I certainly was not, and considering I could find no other reference to PM and its relationship to SQ I'd be keen to hear from other designers with there own experiences.
Interestingly, 'Unity Gain stable Op-amps' seem to target 45Deg PM at Unity Gain, and this is too tight IMO - 45Deg PM results in well over 20% overshoot.
Every designer is aware that over compensating an amplifier kills the sound, but what's 'Optimal' from the perspective of SQ? some would say the least you can get away with - which makes me wonder if a little "peaking" is preferred as I would theories it subjectively enhances detail. (I'm not suggesting this is a valid approach, but its an interesting observation if true).
You are correct that I could be compensating for the MDAC's design or my systems defiances... which is one of the reason I've built a listening room above the lab where I can install my ESLs + designing the VFET reference amplifiers + MDAC2
Once the labs installed I've got a HP / Agilent network analyser (and small signal Current probes) and plan to break the MDAC2's FB loop to see how close the hardware matches the simulation results - and how stable the PM is between boards.
Yes - I'm referring to the MDAC's gain stage PM. The CL phase response within the audio BW is practically flat.
The MDAC2's Bode plot is rather odd as its a 2nd order multiple nested feedback design - which I could only consider designing (and optimising) with the aid of computer simulation.
Maybe on the MDAC2 we can adjust the Q of the Digital filters to see its effect on SQ - also the Q of the NS modulators at a later date...