Indirect feedback with FETs

[x-pro]

I've uploaded a small article about an old but somewhat forgotten technique of distortion reduction applied to a single-ended FET amplifiers (could be used with valves too).

http://www.ant-audio.co.uk/Theory/Ap...e_Feedback.pdf

Perhaps Nelson will make another ZEN ? That approach can be used in many applications. I attach a picture from the article with a (somewhat) simplified circuit.

Cheers

Alex

[lineup]

Looks like the upper FET is run in
Shunt Regulated Push Pull, SRPP

This is used with Tubeamplifiers.
The second harmonic is very suppressed giving lower overall THD

[x-pro]

Originally Posted by lineup

Looks like the upper FET is run in
Shunt Regulated Push Pull, SRPP

This is used with Tubeamplifiers.
The second harmonic is very suppressed giving lower overall THD

It would be good to see some examples. However with the indirect NFB all distortion components may be reduced, not just the second harmonic.

Cheers

Alex

[Elvee]

The effect is so marginal it looks practically worthless.

Are there actual circuits where it would be of some value?

[x-pro]

Originally Posted by Elvee

The effect is so marginal it looks practically worthless.

Are there actual circuits where it would be of some value?

Try it at least with 2SK170 . The FETs you've used have very low gain without the feedback, so the effect is nil.

Cheers

Alex

[x-pro]

Originally Posted by lineup

Looks like the upper FET is run in
Shunt Regulated Push Pull, SRPP

This is used with Tubeamplifiers.
The second harmonic is very suppressed giving lower overall THD

It is not SRPP (where is the shunt?!) .

Cheers

Alex

[MarcelvdG]

Originally Posted by x-pro

Try it at least with 2SK170 . The FETs you've used have very low gain without the feedback, so the effect is nil.

Cheers

Alex

I guess the criteria are:

voltage gain factor of the FET >> gain with indirect feedback

transconductance of the FET times resistance of the two feedback resistors in series >> gain with indirect feedback

Is that correct?

By the way, I like the circuit. It is a variant of indirect feedback I hadn't seen before.

[x-pro]

Originally Posted by MarcelvdG

I guess the criteria are:

voltage gain factor of the FET >> gain with indirect feedback

transconductance of the FET times resistance of the two feedback resistors in series >> gain with indirect feedback

Is that correct?

Yes, you are absolutely correct.

Originally Posted by MarcelvdG

By the way, I like the circuit. It is a variant of indirect feedback I hadn't seen before.

Thank you! I've designed this circuit about 7 years ago and used it with very good results. I won't give away exact circuits used in my commercial projects, however I can give you some ideas how a well polished circuit of this kind can work - by uploading a couple of graphs with spectrum of the output signal from a preamp built using that idea for the VAS. The gain is 20 dB (x10), the first graph is for the output voltage 1V RMS at 1 kHz, the second - for the output 0.5V RMS at 19+20kHz , both on 5 kOhm load.

Cheers

Alex

[DF96]

Maybe I have missed something, but to me the first circuit looks like it has a drain load consisting of an FET wired as a CCS, in parallel with a 1k6 resistor. The second circuit has a slightly modified CCS in parallel with a 50K resistor. I would expect the first circuit to give high distortion. Nothing to do with indirect negative feedback, just a matter of changing the drain load for a triode-like device. Elvee gets much better results for the first circuit because he uses a much higher resistor value.

The 'explanation' also mentions identical FETs (OK in simulation, but where do you buy them?) and FETs whose drain current is independent of drain-source voltage (same question!). The second circuit might work OK, but not for the claimed reasons.

[x-pro]

Originally Posted by DF96

Maybe I have missed something, but to me the first circuit looks like it has a drain load consisting of an FET wired as a CCS, in parallel with a 1k6 resistor. The second circuit has a slightly modified CCS in parallel with a 50K resistor. I would expect the first circuit to give high distortion. Nothing to do with indirect negative feedback, just a matter of changing the drain load for a triode-like device. Elvee gets much better results for the first circuit because he uses a much higher resistor value.

The 'explanation' also mentions identical FETs (OK in simulation, but where do you buy them?) and FETs whose drain current is independent of drain-source voltage (same question!). The second circuit might work OK, but not for the claimed reasons.

You did miss something. The modulation of the input JFET drain current is exactly the same for both circuits, so is the gain. If your explanation was right, the gain in the second circuit should be 30 times higher, but it is not . The trick is that in the second circuit the top FET is not just a current source any more, but rather a non-linear load which compensates for the non-linearity of the input FET. The indirect feedback loop takes care of that, making the resulting voltage transfer characteristic much more linear.

Regarding the second point - for JFETs of the same type the transfer function for the same drain current would be identical enough for a good distortion compensation even when the devices are not matched perfectly (or at all) . The only problem would be a DC offset which would have to be corrected in either circuit by changing the bias voltage on V5 and V2 respectively.

And the third point -- even 2SK170 have enough voltage gain for this circuit to be useful and if required you can cascode two different JFETs in a usual manner - for example, as on the picture attached.

Cheers

Alex