Non-constructive comment:
Oscillation on a solderless breadboard ground? Say it ain't so!

Hopefully more constructive comments:
Simulation shows the EL2030 being very ringy, perhaps that's the culprit. That's a really old opamp and is obsolete, perhaps for a reason! Have you tested the buffer on it's own to see if it's ringing? I spent 0 time chasing down that model BTW - it was in the datasheet. I know little about CFA's, I'm not sure including a resistor makes capacitive load driving more stable or not. I'd perhaps research that a bit more. I'm also surprised the input to output offset is so large after having matched them, but if you're happy, fine.

I have no problems running your lib files - I've attached a file with them both working.

 

Non-constructive comment:
Oscillation on a solderless breadboard ground? Say it ain't so!

I have done nothing to beef up the ground. On one _very_ high frequency oscillator just about every hole on the ground bus was used -- either by a component lead or an added ground jumper if no component needed that hole. Note that it oscillated at about 350MHz. and other frequencies. All at the same time.

Hopefully more constructive comments:
Simulation shows the EL2030 being very ringy, perhaps that's the culprit. That's a really old opamp and is obsolete, perhaps for a reason! Have you tested the buffer on it's own to see if it's ringing? I spent 0 time chasing down that model BTW - it was in the datasheet. I know little about CFA's, I'm not sure including a resistor makes capacitive load driving more stable or not. I'd perhaps research that a bit more. I'm also surprised the input to output offset is so large after having matched them, but if you're happy, fine.

I raised the feedback resistor (in simulation) to 820 ohms and most of the overshoot is gone. I will be trying that on the real hardware today, I think.
I was surprised at the offset. It may be because I had a resistive load on the output of the FET's when I matched them. The first criteria was the Idss so I settled for what offset I got. The offset in in the feedback loop of the function generator. The only effect is that if it gets too large the current source starts to saturate.

I have no problems running your lib files - I've attached a file with them both working.

Thanks for doing that. To be honest, I did not even try them. They looked different than I expected so I just assumed that it would take me some time getting them to work.

The delay of the OPA659 sim's as being about 575ps. The OPA653, with its internal gain setting resistors, is less than 500ps. I can live with that.

By the way, you may not have noticed but the OPA653, OPA659 and the ADA4817 are all FET input voltage feedback op-amps. Pretty amazing parts. The ADA4817 has a gain-bandwidth product of over 1GHz. The OPA653 has a bandwidth of 500MHZ at a gain of +2.

I am adding the J-FET's to the EL2030 just for the breadboard testing. (And the challenge ).

 

@tindel
I have added the J-FET's to the SIP holding the EL2030. I feel better now.

I am still not happy with the operation in the function generator. I decided to clean up a bunch of things in the SBB prototype.

I think that subject belongs in this thread.
https://forum.allaboutcircuits.com/threads/thoughts-on-a-high-frequency-function-generator.142253/
See you there...

@OBW0549 I have missed you. See you there as well, I hope.

 

@OBW0549 I have missed you. See you there as well, I hope.

Oops. Sorry 'bout that, been on Amazon.com buying Christmas presents for the grandkids. I'll pay more attention now...