Power supply current regulation oddities

[madworm]

I'm having some newly discovered issues (thanks to my new scope) with my old power supply. I've attached a photo of the analog part of it. It uses a TL084 quad opamp for the voltage and current control loops. The reference voltages are created by 2x 8-bit DACs (not shown), which themselves are controlled by a microcontroller. It takes care of the user-interaction.



Issues discovered so far:

* about 10% voltage overshoot when the output is activated (via relay) - this would kill 3.3V devices.
* massive current overshoot, the current control loop seems deliberately slow.

I've played with the relevant capacitor a bit (C26, top right opamp). If it is removed altogether, the thing never gets out of over-current mode once entered. Replacing it with 22pF makes it unstable. I will have to get a 5pF-ish trimmer cap. to experiment further.

The way the over-current limiting is implemented here seems somewhat flawed to me. The time it takes to kick in is way to long and it also depends on the set-current. For low values (say 20mA) it takes quite a long time on a human timescale until current is finally limited.

See the attached traces. The colors match the dots in the schematic.



I'm not actually thinking about fixing or redoing the control circuitry at all, I'm just playing with it. Maybe there are some simple things to improve its behaviour. I thought about adding an external module to compensate for its flaws, but getting an all new power supply sounds more attractive. This things is about 20 years old btw.