I recently spotted a very clean Leader LPS151 power supply on my way out of an exceptional junk store. I tried in vain to resist–you can never have too many power supplies. We have lots of these around the labs at Linear, and someone always seems to “borrow” the one on my bench. The tag said $75, but the guy let me plug it in to give it a quick check, which revealed that the negative output was dead and the two positive outputs only got to about ¾ of where they should be. With this information I offered him $50 and a promise to not ask any more questions, and he accepted.
Time to dig in. I didn't even have to look for a paper copy of the manual, as BAMA (the Boat Anchor Manual Archive, http://bama.edebris.com/manuals/leader/lps151/) had a copy, refer to page 15 if you're interested enough to keep reading. In addition to the three actual power supply circuits, there is a fourth supply centered around IC101 that powers the control circuitry–let's call it the “Brain Power.” The first step in troubleshooting ANYTHING should be to verify that the raw, unregulated supplies look reasonable, both DC voltage and AC ripple. All was well. However, looking at the Brain Power outputs, the +7.5 and –12.4V supplies were too low. Furthermore, D107 (a 6.2V zener diode used as the reference for everything else) only had about 4V across it, IC101's inverting input was 400mV below its non-inverting input indicating that something in the loop was broken. So after a bit of poking around, I started to suspect U101 itself–either its output was cooked or one of the inputs took a hit somehow (unlikely in a closed metal box, but not impossible... think flying resistor lead entering a ventilation hole...) But I would never find the true cause–I stupidly shorted –12.5V to ground with a meter probe which caused both Q102 and D103 (which provides the +7.5V regulation) to fail as shorts, which in turn blowed the top the top off of one of the other op-amps. Darn. I blindly removed all of the op amps and put in sockets, replaced the dead zener and transistor.
This wasn't supposed to be a formal plug, but the LT1013 is billed as a high performance upgrade for the MC1458, and the LT1012 as an upgrade to the LM301. These are probably overkill for a bench supply but I had samples at my disposal. (The LT1012 is one of my favorites –30V operation with offset not too far removed from that of a chopper but MUCH lower noise. I recently used it as the reference inversion amplifier on the LTC2758 18 bit DAC demo board.)
I had resigned to not make any more mistakes in this project, so I at least wanted to make sure that the Brain Power was up and running somehow before giving the other supplies the chance to destroy themselves. The output supplies work on the principle that the pass transistors are passively turned on, while one of two op amps controlling either the output voltage or current turns the pass transistor off, with the voltage and current “turn off” signals “Ored” together through diodes. Nodes “VA” and “VB” provide the turn-on voltage, and they are only enabled when the Brain Power is up and running. This means you can't just pull the (now socketed) op amps–the outputs would slam on as hard as they could. Lifting one lead of R202, R302, and R403 disables the +6V, +25V, and –25V supplies, respectively.
Plugging in and throwing the switch–no smoke! The Brain Power was up and running, and VA, VB were turned on. I soldered the lifted resistors back in and threw the switch again and it worked! All three supplies could be adjusted to their full output voltage, and they all transition smoothly between voltage control and current limit.
I added one more feature to help this supply stay on my bench and not be mistaken for a company owned one. I popped out the green power LED and installed a tri-color, self blinking one that cycles through color patterns. I'm not 100% sure how these work internally; the ideal circuit would be current sources for each of the individual dice, but checking current with a scope revealed a strong dependence on supply voltage, so it may just use resistors. But the important point is that at some points in the light pattern, the current drops to zero, and the original LED was driven by the –12.4V Brain Power supply through a 1.2k resistor. Thus during times when the tricolor LED was drawing no current, the full 12.4V would be across something inside, probably not good. A 3.9V zener shunt around the LED connector cures this. (There's an error in the schematic I have, it shows the LED cathode going to the –6.2V reference, it actually goes to ground, and the LED is backwards.)
This was completely worth the effort and price. Now I've got a supply that won't walk, and learned a few good circuit nuggets along the way.