What follows is a tale of carelessness, stupidity and laziness. So far, I haven't found an excuse to add ignorance to the list. ;) As you may recall, I was testing an 8" floppy drive that was reading inconsistently on an Altos 8000 system, when, while testing with a replacement drive, the 24V power supply gave out.
First I acquired a replacement series pass (2N3055) from Radio Shack. That did not seem to fix the problem. Further, due to *carelessness* (#1) during the replacement and testing of the 2N3055, I managed to break the center lead off of the TIP31A which drives it. So, I ordered a replacement for the TIP31A, and, while I was at it, ordered several LM723 Voltage Regulators, just in case (good idea, as it turned out). Then, I replaced the TIP31A, put it back together and --- nothing. No output. Oh well, I guess the LM723 must be bad. Then, in the process of checking that out, I tilted the board over on its back and, due to my *carelessness* (#2) and *stupidity* (#3)) FLASH!!! BANG!!!! POP!!!! Though switched off, the power supply was still plugged in - I had neglected to unplug it. The board came in contact with what SHOULD HAVE BEEN neutral, which likely would have caused no harm. I didn't actually realize that until I was on my way back home with the parts to replace what had blown out. [Also a design note: Altos did not cover up the input terminals on the transformer or use quick disconnect plugs - either of which would have saved the day.] After some thought (and a test), I realized that, due to prior laziness / neglect (#4 - about 10 years+ worth), and a dose of stupidity, the mains hot and neutral were reversed on my bench. Thus the power switch was on the neutral side, and the mains to the power supply was "hot" even when not turned on. At least this was not *originally* my fault. That circuit came that way when we bought the house. But, out of laziness, neglect and stupidity, I had not fixed it, even though I had known about it for years. So, this blew up the -12V supply (quite literally) -- cross out a pair of rectifiers, an LM723 voltage regulator (spectacularly blown up, with a hole in middle and two leads blow away), and a 2N2905 transistor, a couple of resistors and a trace on the circuit board. (At the time, the 2N3055 series pass transistor was not connected in the circuit). The fuse didn't blow - it wasn't part of the circuit, being on the neutral side of things because of the miss-wired mains in the house - the short was from hot to ground. (A GFCI, had I had one, would undoubtedly have tripped). Next, I pulled out all the blown components, and tested what was left. Being at least halfway clever, I proceeded to test the +24V and +12V supplies first. Once I replaced the original LM723 on the +24V supply, which was already undoubtedly bad, the +24V supply was fine - the short to the mains had not affected it. But the +12V supply's LM723 had apparently been collateral damage - the -12V supply uses the +12V supply as part of its LM723 circuit, so the two are connected. (I knew something was wrong, because even though I could set the supply +12V, the adjustment worked backwards, and under load the voltage dropped too much). So, I replaced the +12V supply LM723, and the +12V supply came back to life. Both tested fine under load, no ripple, only 20mv of noise. During these tests, I moved the same 2N3055 from circuit to circuit to avoid having to completely reassemble the supply. That saved time. But, do you think I was done being careless? Noooooooooooooope. Next, I soldered in the components for the -12V supply, including its 2N3055, and turned it on. KA-BANG!!! FLASH!!! DEBRIS CLOUD !!! Out of carelessness (#5) I had managed to put the electrolytic input capacitors in BACWARDS. I now had a nice coil of cardboard and aluminum sticking up out of one of them. Then the next day (today) I found the can itself lying on the floor - this power supply predates the modern practice of scored break lines in electrolytic capacitors. It could easily have hit me in the eye. So, I took both of those capacitors out, and tested the one that had not blown up. It seemed to be OK - no excessive leakage. So, I put in a replacement and... 0V output. Blown fuse. What the? Ahhhhh - forgot to check the rectifiers - and the capacitor reversal took both of those with it. The other semiconductors were fine. Replaced the rectifiers, put in the capacitors correctly and, finally, the -12V supply came to life. Adjusted, load tested, checked ripple. (Still need to replace the surviving capacitor. I don't trust it. And the smaller signal transistor I used is a bit underrated compared to the original (650mw vs. 800mw), so I expect I will replace both in a week or so, but for now the supply is quite usable. Ah yes - one other thing. The circuit (which the bench shares with a wall-wart powered water softener) that had hot and neutral reversed is reversed no longer, as of this afternoon. It now also has a GFCI, for good measure - a good idea on a bench circuit, methinks. So, maybe tomorrow, I can put this machine back together and get back to testing the floppy drive. So, SAFETY FIRST. Learn from my experience - it could have been deadly or resulted in serious injury (say, if a shock or reaction to the capacitor explosion had caused me to tip over on my stool an hit my head, or that capacitor can projectile had hit me in the eye). To paraphrase from the Direct TV commercials: DON'T BE LIKE THE LAZY ME. DON'T BE LIKE THE CARELESS ME. Be like the smarter, conscientious me who knows better. UNPLUG stuff before you work on it - DOUBLE CHECK. HAVE A GFCI for your workbench. Wear some kind of safety goggles or at least hardened corrective lenses when soldering and testing power supplies, especially. When replacing filter capacitors in a power supply (or anywhere else, for that matter), take an extra second to double check their polarity (I usually do - but "spaced" it this time). [Which reminds me of the IBM 1414 I/O synchronizer that had been at the U. Wisconsin Registrar and then the U. W. school of business, that had easily more than 10 smallish filter capacitors on the backplane that were backwards for years (these were hooked up with little sockets that slipped right over the pins on the backplane). Each day it took 10 minutes for things to settle down and the printer to stop showing its red Sync Check light because of the noise in the circuit. But they never blew up. The FE's finally got tired of seeing that anomaly and spent an entire day hunting it down.] I wish I could say I was in a hurry. That excuse wouldn't cut it anyway - but I was actually taking my time. Just careless. Sigh. JRJ
