Non-energic is not the same as non-ergodic, the confusion is not helped by the fact that Birkhoff invented the first term, in 1927, and then did fundamental work on the second a few years later. I should have included a date reference.
Nonenergic feedback is feedback implemented with lossless components only, transformers, inductors and capacitors and so on.
That means you have conservation laws and thus can use the whole arsenal of 300 years of mathematical techniques for conservative systems like mechanics.
I have just touched the start of this, it has an obvious theoretical appeal and should provide limits to what is possible.
Not sure if that is worth the effort, especially when in practice, of course, I don't want to have feedback transformers.
They would probably appeal to a certain niche.😉
Thanks Jan, sorry I didn't have it ready in time for September edition.
Best wishes
David
Last edited:
Douglas,
I might get lambasted for this, but do you accept that something that may not be measurable may in fact sound different or better?. I recall Nelson Pass talking about the global feedback being taken off of the Stasis, placed from the Vas to inverting input, measuring the same yet sounding "better".
Colin
I might get lambasted for this, but do you accept that something that may not be measurable may in fact sound different or better?. I recall Nelson Pass talking about the global feedback being taken off of the Stasis, placed from the Vas to inverting input, measuring the same yet sounding "better".
Colin
I've always been intrigued by such reports. Surely we can agree that ANY audible difference results from different electrical waves at the speaker terminals; I hope we will exclude extra-sensory perception!
So we would have an audible difference between two wave shapes that measure the same. The only possible conclusions* are 1) we hear more sensitive than we can measure or b) we're measuring the wrong (or at least not relevant) things.
Considering the huge evidence we can set 1) aside as very unlikely, so the remaining option is we measure the wrong thing.
One thing I never see measured is amplifier performance right after a large burst or a strong passage. I have read the reports by the usual suspects (DS, BC come to mind) on thermal transients in output stages, for instance, but my own research indicates that within the stable bias condition over a longer period, sub-second variations in bias current can exists that are 'smoothed out' in the average bias measurements. What if a soft passage is reproduced just after a strong passage that shifted the bias, before the bias loop has returned to the rest position?
Anybody has any experience with such things?
Jan
*I am assuming for the moment that the reported audible difference is objectively, repeatably there, which of course is another oil drum of worms, but for another occasion perhaps...
Last edited:
It's actually quite well-known in the tube world, though the mechanisms are different. Crowhurst wrote extensively about it in the 1950s, and both Morgan and I have designed amps (Crystal Palace and Red Light District, respectively) taking this explicitly into account.
Coupling and bypass caps, especially at cathodes- like I said, different mechanism, same effect.
Please define "TMC can work better". If you mean TMC can provide higher loop gains, that's incorrect in any apple to apple comparison. The FDNR is there, but it can be safely ignored in any non-pathological, practical, case. Caring for that FDNR sounds like Mr. Zan caring about audio amplifiers being non-minimum phase systems, because of the bipolar beta RHP beyond Ft, at some 500 MHz. Come on...
Anyway, I know somebody in Canada that will get pretty angry if you don't properly cite prior art in the TMC vs. TPC equivalence 😛. Expect a letter to the Editor 😀.
TPC "lossy" and TMC "lossless"? You lost me here.
No, it's not those "loop gain probe" thesis guys I'm working for. I built the probe, using a commercially available pulse transformer, for measuring the loop gain and stability margin of a mixed signal IC library, bicmos wideband, two pole compensated, opamp. Works fine for 1KHz up to 6-7MHz.
I would say that my Fig 1C is attempting to be output inclusive compensation- but it will almost certainly be unstable because of the extra phase-shifts in the output stage.
Looks good/familiar 🙂, manually it is simpler to put the conclusion ahead (2 pole equivalence), calculate the impedances for TPC, then identify the coefficients. Then see what's left (like the FDNR, etc...), analyze the physical interpretation (if any) and check if their magnitude and phase are anywhere relevant in the particular circuit topology. That's how I did it, and for common 3 stage amplifier topologies (either Blameless or the derived symmetrical) the equivalence is in the % range.
I know Edmond has some special designed topologies where the TMC can have a significant advantage over TPC. Those look interesting, but still have to be experimented and evaluated on a board. My experience with Edmond's auto bias was a disaster (in terms of stability), so I take all his extreme designs with a big grain of salt.
Ah yes, I am very familiar with that review; many thanks. However the Halcro is not much help to us as it is a highly unconventional (and very complicated) design. I should have said that I was thinking of more conventional designs, and especially the differences in distortion performance between single-ended and push-pull VAS's.
BTW, anyone know precisely what happened to Halcro? The "latest news" on their website is dated 2010, so it doesn't look good.
Last edited:
For high power (= high and symmetrical slew rate) I prefer complementary differential input and push-pull VAS, rather than SE VAS + CCS.
If I understand correctly, this delta-wye transform business is crucial to the "Output-Inclusive Compensation is a form of TPC" battle cry. But if it yields non-physical components it seems to me of little use in understanding things. I'm not afraid of FDNR's (or worse) but I would hesitate to design them into an amplifier.
I was hoping that someone could tell us exactly how this theory helps with practical amplifier design.
- Status
- Not open for further replies.