Mixed feedback

[Bazukaz]

Hi,
Recently i found the following article for improving output tranformer's performance.Maybe it would be useful for ESLs , especially lower quality trannies ?

http://www.lundahl.se/pdfs/papers/feedbck.pdf

Regards,
Lukas.

[bobo1on1]

Looks like a negative resistance to me.

It might not be suitable for driving capacitive loads like esl's.

[orjan]

The negativ fedback is only set to be enough to nearly cancel the serial resistans in the primary.

[maudio]

Yes, and the winding resistance of the primaire is the least of your worries when designing an ESL-transformer

What makes a good esl transformer so difficult to design is the combination of required output voltage (together with the v/turn ratio determining # of sec turns) and the large frequency range: the low frequency limit requires a large # of prim turns but the high limit requires low internal capacitances and low leak, what in turn requires as little turns as possible. In other words, the problems are at the secundairy side.

I am currently trying to design an amplifier/transformer combination that includes the transformer in the feedback loop. But it proves very difficult to obtain enough bandwidth

[jcx]

the point is that you're missing is that cancelling the pri resistive current causes a very large reduction in low frequency distortion

http://www.analog.com/library/analog..._handbook.html

chapter 6 ~ 6.70+

the question is could you usefully rely on this to extend esl transformer usable bandwidth lower limit

[maudio]

Originally posted by jcx
the question is could you usefully rely on this to extend esl transformer usable bandwidth lower limit

Interesting paper, thanks for the link!

I believe this method could to some extend be useful to lower low frequency distortion by the transfo, if it could be made to work properly in capacitive loads. But than again, I don't believe low frequency distortion is something to worry about. Transformer distortion will be overshadowed by other esl-related problems anyway (resonance, non-uniform membrame motion etc) in the low freq range.

But to answer your question, it will not extend the low frequency limit. The lower bandwidth limit of a practical ESL transformer is not set by distortion problems but by core saturation limits that impose a hard limit on the voltage the transfo can handle at a certain frequency, derating linear with frequency. In addition primairy inductance sets another limit (Zprim must not drop below what the amp can drive @ 20 hz). Whatever comes first sets the limit.

[orjan]

Well, the primary inductance is only a small signal aproximation. At large signals, close to core saturation, the value decreases.

And, yes, it works. I used it with line level transformers. It extends the usfull, as in some set distorsion value , low frequncy limit.

With transformers you could do as in quad 63s, use 2 for each esl. Then you can have half the voltage ratio.

[maudio]

Originally posted by orjan

With transformers you could do as in quad 63s, use 2 for each esl. Then you can have half the voltage ratio.

That's a good way of doing things but the price you pay is lower primaire inductance as both primaire coils are in parallel and they don't share the same magnetic circuit.

Even better solution is therefore to place the two coil-former from those two tranformers on one core, which can be done by using two coil-formers on a double C-core. To my knowledge this is the best way to make an ESL-transformer.
At least I hope so, I am trying it right now

[frank ziel]

Yes, yes, yes -
and we are all waiting for you, to have grate success -
and tell us all about...
Regards,
Frank