All FET folded cascode driver for "JLH"-buffer with no GNFB

[revintage]

This idea is with some unusal takes. Feedback is local and only over the cascode as the buffer is enough feedbacked in itself. Also the locally feedbacked cascode will act very much like a triode.

Output stage is a JLH with a PNP driver that makes it a pushpull follower with unity gain, low distortion and low Zout.

RDC is for balancing initial offset of the buffer and only works if Q4 has lower Vbe than than Q1.

The DC-servo is integrated with the voltage-controlled gyrator( a CCS will not work against a JFET) at the bottom.

With Q1, Q3, Q4, Q5 and Q6 on the heatsink thermal-drift will be quite low even without the servo.

The Q5/Q6 CCS is fed by 20V to allow Q1 to swing up to 15V-Vbe.

Tried sims with a BJT in place of J1 but got indications of a less pleasant overtone structure.

[Wavebourn]

Hi Lars; is your servo stable enough?

[revintage]

Hey Anatoliy,
Don´t know, just drew a std configuration. Any suggestions how to do a stable one? Will anyway try the circuit without servo at first.

[Wavebourn]

You have 3 poles, and at least 2 of RC networks have the same time constants.

I'm afraid it will oscillate.

[revintage]

OK, then we skip R13/C3 to get only two poles. Also we make C1 bigger, like 2,2u. Will that be good enough?

Or let´s go crazy and dump the servo.

[Wavebourn]

I would make it smaller

[revintage]

Yeah, but at some point it begins to impair lowend response/phase. No problem going to .47u or .33u anyway.

[Bigun]

Would there be some benefit if impedance of feedback could be lower, say 2k series and 220R shunt ?

[juma]

Lars, that looks like a nice gain stage
I see you love the gyrator - I never tested one (except long time ago in parametric EQ) in preamp/gain stage - what sonic benefits you experinced with it?

[Wavebourn]

It may sound gorgeous, when properly built.