PS-36KX fixed!

KF7ZIM came through! At $120-200 for a replacement supply, a $10 gamble on a few components seemed like a good deal. I bought the parts and Alexander put in the work. After warming up the soldering iron and replacing both semiconductors that we suspected of failing (the voltage regulator is now in an IC socket; he didn't want to risk directly soldering a new chip and overheating it, and now it'll be easier to test/replace next time), it still didn't make any difference. At this point, he clipped the wires connecting one of the 4 big pass transistors mounted to the rear heatsink, and tested it, finding that the emitter-collector junction showed a short (i.e. it was conducting without anything connected to the base). Scratching our heads over the unexpected combination of that with the base-emitter & base-collector readings, I had him clip another set. Same thing. Finally we determined that all 4 transistors were blown.Alexander remembered that he had gotten a set of power transistors at SEA-PAC, and went looking in his junk collection. He came back, we looked them up online, and started comparing specs. They were rated for more than twice the current, but had fairly close characteristics in most things that mattered. Since they were just used for a linear power supply, we weren't as concerned about some of the specs as we might have been had they been used for something else. (We really didn't care if they worked at 2MHz; this was DC...) I approved them as being close enough. Before committing to a bunch of work with pliers, solder and heatsink grease, we tested one of the transistors by sticking its pins into the end of the wires (the pins slipped in nicely between the stranded wire and the insulation) and used a wire with alligator clips at both ends to connect the case. W turned the power supply on, and voila! No alarm, and everything worked the way it was supposed to. Open circuit voltage varied as it should with the control knob on the front panel between 12-15V. He shut things down and went to work. Less than an hour later, it was put back together with 8 wirenuts to make it easy to disconnect the transistors should the need arise to test them again. The supply works great!

The shouldn't have been enough current flowing for long enough to kill the transistors (and the SCR "crowbar" circuit is supposed to prevent that!) and the most likely theory on the failure came to us after pondering the situation and the discovery that both of the fuses on the connected battery were also violently blown. Apparently there was indeed a short, and both bench supply and battery took on a sudden heavy drain. I believe that when the fuses blew (with a nice arc) the sudden disconnection did what a mere short wouldn't: it caused a self-oscillation of high current in the inductive components of the system (including the wires outside the power supply) and generated a high voltage at just enough current to punch through the transistor junctions. In other words, it wasn't that the max current was exceeded, but rather the peak inverse voltage rating. It didn't damage the external diode that I added, nor did it harm the smaller transistor that the VR IC used to drive the larger transistors. I'm glad for that. Too bad we didn't go ahead and clip the wires and test those diodes to begin with; we would've saved $10 and the wait on those components. As it was, Alexander's SEA=PAC find saved us $20 and another wait -- and we got beefier transistors to boot.

There are advantages to having an electronic enthusiast for a son! -- esp. when his hands and eyes are better than yours....