I made up a prototype current regulator for the LED ceiling using a CA3140 and a BUZ11 mosfet, all on a chassis that I once built for a motorcycle voltage regulator.
And it worked, but current was suspiciously high on the upper settings.
Cue the scope and….oh woe…. it was oscillating as the current demand was increased.
As the mosfet’s input capacitance was asking a bit much of the op amp output, rather than try to stabilise it, I ditched the mosfet.
Two 3mm holes later, and a TIP31A (npn) and TIP32A (pnp) had become a Sziklai pair.
The resulting circuit was still oscillating, but much less and at much lower frequency.
From reading an op amp book (did I ever understand Bode plots?) and viewing the Bob Pease/Nat Semi video, I put a 1? + 0.1µF across the output to counter the inductance of 15m cable loop, which made things better.
Then from the book, the video, and the CA3140 data sheet, I put a guessed 1n5 capacitor between pins 8 and 1 on the op amp to slow it down a bit.
Hooray, the result was no oscillation and a nice soft overshoot-less transition between levels.
Actually, it now worked without the output RC, but it is still there as every little helps.
Even the move to a couple of old laptop bricks wired in series instead of the lab PSU failed to upset things.
So pleased was I that I treated the lucky little prototype to a wooden foot so that it will not get knocked over and crush its rather exposed components.
Here it is. Most of the wiring is the rather complicated switch-based variable resistor.
It will get used as it is while, until time is found to make a proper microprocessor controlled version – so March 2013?…
Should you feel the need, DON’T RESPOND BELOW as we are having a comment spam crisis.
Instead, email firstname.lastname@example.org
No email addresses are collected for marketing purposes from responses to this blog. I will keep it that way for as long as possible.