Complementary symetrical jfet opamp

I started to play with the idea of designing a complementary symetrical jfet opamp and at this stage I tried some simulations and it works. For this simulation the closed loop gain is aproximately 100 [because of the feedback resistors 100 ohm and 10kohm] and as you can see the output voltage is 100mV for 1mV input.
I would like to know what the big proaudio gurus think about it.
I was interested in this topology because it has the advantage to cancel some nonlinearities because of the symetrical structure compared to the usual designs.
For the input jfets I think about the 2SK389 and 2SJ109 and for the rest the 2SJ74 and the 2SK170. I supose if the input fets are paired the 2SJ74 and the 2SK170 will work.
Hope soon to make one opamp on the breadboard and tweak it.
So, Fred Forssell, John Hinson, PRR what do you think about it and how it can be made to work in the real world.
Here is the screen capture.


chrissugar

"Aren't the latest generation Forssell op-amps differential too..? Anyhow, looks interesting!"
Are they? I didn't know about it.

I have to say a big thank you to Fred Forssell, Erno Borbely and Nelson Pass because their white papers and schematics opened my appetite for the discrete opamps. I learned a lot from them.

Please, would you share your thoughts about this comp sym opamp.
Thanks

chrissugar

Chris,

I am not an expert so you can take this FWIW.

In my humble opinion, unless you use a current mirror on the differential pairs you defeat the purpose of the symmetrical design. Your design goal seems to be minimizing distortion, but the mismatch in the differential pair due to unmatched currents will cause a different type of distortion.
I know the simulator might say that you have exactly the same amount of current flowing in the input devices. That is very misleading and it never happens in real life unless you force them with a CM.

Are you detecting blobs of distortion in the ones you have built?

If I changed something about the Forssell opamp it would be a current mirror on the input pair and an output stage that is easier to get parts for. The V parts seem to be going out of existence slowly.

I am breadboarding my fifth opamp design this week and the only thing the simulator seems to be good for is setting up bias points. Even that is only within 10%.

Cheers,
Tamas