Drive an audio transformer from an AOP. Good or bad idea?

[zamproject]

yep...don't tell me...i can also load a great dish of those

 

[Cqwet Dbdfte]

My younger experienced ears tell me high NFB ckts suppress low level info, reverb tails, room tone, etc. That may be due to the suppression of harmonics mentioned, I don't know. IMHO, distortion down to .0001% from the use of NFB is a rabbit hole.

IM is more relevant than THD. Global feedback could be minimized with local degeneration, in cathode, emitter, or source, leave bypass cap out.
Otala et al had day with this, I think, long time ago. Global feedback to input applied with a time delay thru the stages. Will not show up with static signals, but maybe so small as to not make much difference. Hmm.

 

[abbey road d enfer]

Matti Otala was the guy that said global NFB could not work because the NFB signal would come too late. It's the same train of thought as the Zenon of Elea paradox, which claimed that the arrow could never attain the turtle. It's been debunked a long time ago.

 

[Cqwet Dbdfte]

https://hifisonix.com/wp-content/uploads/2017/10/The-Theory-of-TIM-Matti-Otala.pdf
https://linearaudio.net/sites/linearaudio.net/files/otala low tim amp.pdf
He made some hay with this, fun as long as it lasted

 

[Paul W]

I'd think that within reasonable limits there'd be a fairly big middle ground of "about right feedback" where the feedback linearizes the amplifier's response and there's not much distortion of any kind, because the feedback ensures that the output is very close to a properly scaled-up version of the input signal. (By constantly comparing a scaled-down version of the output signal to the input signal.)

Whatever the necessary level of feedback is for linearization, I'd think there'd be a somewhat lower level of not enough feedback, but it's not clear that there'd be a level of too much feedback, unless it's overcorrecting and oscillating around (hunting back and forth across) the target curve. Is that an issue?

I have no idea how to relate any of that to selectively affecting odd vs. even (or low vs. high) order harmonic distortion

 

[abbey road d enfer]

I don't know why you edited out this part of your post; "Do classic circuits not do that because they wanted transformer distortion, or because they didn't think it was necessary because the transformers were sufficiently linear, or what?", but I have an answer: Yes (particularly "or what?").

 

[Paul W]

I don't know why you edited out this part of your post; "Do classic circuits not do that because they wanted transformer distortion, or because they didn't think it was necessary because the transformers were sufficiently linear, or what?", but I have an answer: Yes (particularly "or what?").

I edited that out because I had an attack of insecurity about my competence to discuss any of this and almost deleted the whole thing. The papers linked by Cqwet Dbfte seem relevant, but I don't understand them, so I thought maybe I should shut up.

The fuller version of the deleted part was something like:

I'd think that using a tertiary winding and doing end-to-end feedback (where the far end is post-transformer) would be a dead obvious way of minimizing overall distortion in transformer-coupled circuits. Do classic circuits not do that because they wanted transformer distortion, or because the transformers were sufficiently linear, or what?

If the answer is "or what" (and not that they wanted the distortion, or that the transformers were sufficiently linear), what's the "what" in that "or what"? Which is to say:

Say What?

 

[abbey road d enfer]

Say What?

When a xfmr is grafted onto an existing design, there is almost no other choice than using the best possible xfmr there, according to the drive capability of the output stage, size and budget constraints.
When implementing a xfmr into a new design, unless a custom xfmr is not permitted, there is no pragmatic reason to not use a tertiary winding. The technique is well-known (after all, the Williamson amplifier dates back from 1947). Compared to the Williamson amplifier, the tertiary winding still permits the secondary to be fully floating.

 

[rh001]

A whole lot of Asberger's has infested this thread, as with so many threads in Group DIY. One can find the ideal value of negative feedback by varying the feedback resistor while monitoring various high quality music recordings of well played good quality instruments that vary a lot from solo instruments, full bands/orchestra of different types and of course voice. Have to have hand on speaker amp voume control to keep monitoring volume the same as varying negative feedback. Once a sweet spot is found one should also vary the feeback shunt resistor, again with hand on speaker volume control. Have to go through several iterations as optimal resistor values are closed in upon. And to make this more time consuming and tedious one also has to take into account amount of gain that is needed and whether or not that gain will be variable in final design. Of course hiss has to be taken into account, both total and spectral. After zeroing in on best sound then measure distortion to see what it is. Might be higher than lowest possible. Of course good quality amp and speakers are needed and doublechecking with good headphones doesn't hurt.

 

[Paul W]

A whole lot of Asberger's has infested this thread, as with so many threads in Group DIY.

So, you think adding some assholery is going to help?

It's Asperger's, BTW, and using it as an insult is an insult to the many fine people on the spectrum, most of whom can read the room and behave better online than you do.

 

[JohnRoberts]

Matti Otala was the guy that said global NFB could not work because the NFB signal would come too late. It's the same train of thought as the Zenon of Elea paradox, which claimed that the arrow could never attain the turtle. It's been debunked a long time ago.

IIRC Otala was trying to push his newly invented form of distortion he called TIM (transient intermodulation distortion). Around the same time (70s?) there was another wannabe distortion SID (slew induced distortion). The common theme for these was slew rate limiting and perturbations caused when NFB circuits recover poorly from slew overload events.

IMO these were pretty much nothing burgers, of course marketers live for stuff like these to spin up.

JR

PS: Everybody be nice its the rules..

 

[Paul W]

When implementing a xfmr into a new design, unless a custom xfmr is not permitted, there is no pragmatic reason to not use a tertiary winding. The technique is well-known (after all, the Williamson amplifier dates back from 1947). Compared to the Williamson amplifier, the tertiary winding still permits the secondary to be fully floating.

Is there usually a need for a custom transformer? I'd think there'd be fine stock transformers with tertiaries, and that the exact ratio of the tertiary wouldn't be critical because you could compensate in the circuit's resistors or whatever that determine how the input is scaled down for use as feedback.

 

[abbey road d enfer]

Is there usually a need for a custom transformer?

In many cases, probably not. However, when I started working on my output stage, there was not much choice of affordable xfmrs. I coud have based my work on Haufe, Sowter or Lundahl, but the cost was prohibitive.
A LL1585 costs about $40 by qty of 100. The transformer I use costs about $6.
Two reasons for that: it uses standard M6 lams and it has no magnetic shield.
Global NFB makes it unnecessary.
I used the same subcontractor as for the small power xfmrs I used thousands of.
Thye didn't charge for the "design" because I did it for them.