Thank you!
Any time that I read "distorsion thermique" and "cascode" in the same sentence, I'm interested.
Member
Joined 2009
Paid Member
I am reading Peufeu's articles and I am enjoying them but cannnot but wonder: why did he stopped at the VAS? Or is he attacking the output section later???
I think the output is the biggest concern for memory effect...am i wrong?
The big take-away for me, from Peufeu, was that the solution to each stage drastically improved the performance of the stage. In each case the linearity of the stage was significantly improved. How then, can it ever be possible to assess whether the improvement in 'sound' was due to the alleviation of so-called memory distortion, or from the clear and unambiguous improvement in linearity ?
I don't understand...do you mean a trick that potentially improves two aspects of the problem and that also sounds good (to the author) is not a good solution???
In my view whenever you pass from "hearing an electronic device" more to "listening some musicians in the room" most probably you are experimenting the improvement of a basic downfall of the electronics' physics rather than an improvement in "classical" parameters...for example, I consider I have very good hearing but I cannot differentiate amps from its different THD values, and I seem to be inmune to mild crossover distortion
but I can tell wether the amp is single ended or push-pull...sort of...
I'm with you on this. Moreover, not only did linearity increase, often so did the open loop gain. Also what I appreciated from Peufeu was that often what he did was to create a constant power transistor (I don't recall if he explicitly stated this).
This aspect is also an important result from Arto Kolinummi's study. After attacking each stage separately and greatly improving its inherent linearity, he came to the point where he no longer needed global feedback for ppm distortion. And attention to the thermal memory issues was an important part.
Jan
Drastically ? significantly ?
What a load of rubbish.
If these things made such a big difference then it would have been fixed decades ago !
He is right, Nigel. It's just we never learned to think in terms of basic linearity and how to fix it. The book I linked to above is pretty much unique, nobody did such an in-depth study, and it is almost unknown to audio designers.
And to be honest, we get by pretty well with the usual circuits and gobs of feedback. The 'alternative thinkers' normally don't get any further than cascodes and bootstrapping. But some are interested in knowing all the gory details and Arto was one of them, and he wrote it down.
Edit: if you just want to start thinking about it in an easy way, this is of interest: Linear Audio | your tech audio resource
Jan
I would suggest, to differentiate: amplification in "audio" and amplification of something else. The demands and results are very very different. Good or better solutions in one section may be a desaster in the other section often.
AUDIO-PHYSICS is much much complexer than a bit e-technique.
AUDIO-PHYSICS is much much complexer than a bit e-technique.
There is guy that is obsessed with cascodying all his (BJT and other...) outputs
until he understands (all) the reasons why it sounds better...
Maybe it is time to try cascoded-CFP outputs.
Several authors, including our Peufeu, if I remember correctly, praised the cascode assembly for limiting thermal distortion, or "tail" which I like better, but it is interesting to analyze how this is...any thoughts?
If I understood well, it is related to lower reverse Cbe of the arrangement but also maybe to the physical separation of the two elements, one who cares about "voltage gain" and the other which handles "current gain". Please enlighten me about this intriguing problem.
Apart using electric strategies to deal with MD, I have not read extensive strategies to deal with the problem from a thermal point of view...meaning, for example, the be junction temperature has a non-linear dependence on current (or is it the other way around? I imagine it is both ways...) then, for example, why don't we use the transistors where this relationship is flatter? I mean, apart class A, maybe class B (the other extreme) is also less influenced by thermal changes? I have to see T° curves of the be union.
Maybe it is better to use BIG power units for this reason...
Cheers,
M.
Jan, I think JC would love the man. You should let him know. He is infatuated with open loop.
Member
Joined 2009
Paid Member
Yes, but he also wrote that it sounded really good and we have not the constraints that commercial enterprises have...
Remember Sony and Sansui have reputed amps with it.
Sorry, I probably meant cascoded-Szicklai pair. More material for study...
Or better, I build it and then I study it
as I do generally, hehe.A good point to start exploring cascodying the output would be a one stage single ended common source amp. I have four
better start with the mosfet...yeah, the putative improvement in sound could be from increased linearity and not from better MD...Any thoughts about "partaged task" of cascode being good for MD?
Meaning partaging thermal tails, probably of different kind...could this exist?
Cheers,
M.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.