So I accidentally blew up my trusty TTi CPX400A PSU (2x 20A/60V limited to 420W). Was playing with a DIY 3-phase motor drive, and ended up regeneratively braking into the PSU. With a motor that - when connected to a 240V SLA bank - regens at around 1 kW. Ended up with an overvoltage trip, and upon restarting, it failed to regulate at anything over 15-20V at zero current, and substantially lower if there was a load connected.
Oops.
So off I went trying to find the fault. Surprisingly the pass transistor was fine; the topology tripped me up for a good while. It's a bit unusual (even for a SMPSU with a linear post-regulator):
It consists of a PFC stage with a 390V output, which is fed into two forward converters (for isolation) running at a fixed duty cycle. These do not regulate based on the set output voltage; unloaded they output 350+ volts. That is then fed into a magamp.
"What's a magamp?" I said.
Turns out it's a magic inductor that essentially acts as a buck regulator (it's the large toroidal inductor w/ stranded wire in the pictures). There are no switches driving it; it takes the raw switching signal from the forward converter, and by adjusting the reset current, regulates it to whatever you end up setting the PSU to (plus a volt overhead for the linear post-reg). Hadn't even heard of one before this; it seems like a nifty technology, but it's
weird.
Anyway, it's then rectified, filtered, and the final regulation is done by a beefy NMOS. Nothing unusual here.
Once I stopped looking for a broken feedback path, I figured that the magamp wasn't being driven correctly. Turns out the control transistor - a MJE350 - was toast, as was most of the overvoltage shutdown components. Ended up replacing Q8, D22, R27, R28 and R34. Works perfectly well now; total cost was something like $5/channel to repair it.
Anyhow, attached photos because yay teardowns, right? Also,
here's the service manual (PDF, 5.8 MB).