I have a pair of old power supplies here, and one of them has a failed SCR. It's a 2N6509.
However, when I went and replaced it, I found that the new device did not work. I put a scope on the SCR, and found that it's being turned on and off very very quickly. It's only on for about 6 microseconds, then left off for about 200us. So I think on the working one, with the old SCR, it's not being given enough time to latch in. To verify, I removed the working old scr and put it in the bad power supply. It worked just fine.
I tried 2N6509s from two different sources, but they both latch up, like one would expect an SCR to. My guess is that this power supply (It is old, like 20 years old) relied on the imperfections in the SCRs to work properly. The new ones seem to have no gate turn off action. Any thoughts on what I should do? The triggering voltage is 15V, so I tried an IGBT in series with a diode, but that didn't work. I just blew a fuse. Putting the scope on it, I found that my diode wasn't chopping off all of the negative voltage. It seriously reduced the negative portion of the sine wave (From -170V to about -50V), but this is still well over the Vec breakdown for the IGBT. That part confuses me, too. It was just a standard 1N400x diode. I tried two in series and it didn't get any better.
I looked for some GTO SCRs, but I couldn't find anything. I couldn't find any reverse blocking IGBTs on mouser/digikey, either.