BJT Control Analysis

No.10827018 ViewReplyOriginalReport
So I've whipped up this design for a constant current/voltage power supply. The pic is a simplified version but contains most relevant details. I already have it up and running on a proto board and it seems to work rather well up to 30-40W.

The only way I could stabilize it was by trying different compensation caps/'impedances', and given all of the different active devices I wasn't able to apply my somewhat limited control theory knowledge to stabilize it. However, the constant current, while stable, will produce some odd waveforms when the load is pulsed.

So my main question is how can I model a BJT for "control block diagram" purposes? I have the pole-zero eqns mapped out for the op-amps, but can't figure out how to model a BJT as a large-signal freq dependent block.
Alternatively, anyone have advice for how this type of circuit should be analyzed/designed if it were say commercially produced? From what I've seen/heard, actual engineers wouldn't be using this type of approach but I don't see how they could use intuition to know that this type of circuit would be stable..