Power Designs Model 2005 Precision Power Source - Teardown, Repair (?), Measure

[ChristianConkle]

I picked up a Power Designs model 2005 precision power source that was described as working but with a ~10mV error that the seller couldn't calibrate out. Based on step b.4 of the adjustment and calibration procedure I figured this would be an easy repair: there's a series of resistors that are supposed to be jumpered out in order to bring the zero voltage within range of the calibration pot.

Opening up the unit revealed no obvious failures. There is some corrosion on the main amplifier board, but I don't see any blown caps or other visibly failed components. None of the capacitors are shorted. And sure enough, rearranging the jumpers allowed me to zero out the adjustment.

Alas, not done! Measuring the no-load voltage of the supply set to 20V showed an instability of a bit less then 10 mV, with an odd profile: periods of stability within 1mV or so interspersed with occasional excursions about 10mV lower. Unfortunately, until I receive and set up the GPIB port on my HP 3478A meter, I don't have any logging capability. Suffice it to say, the supply is not within spec (ripple and noise < 100 uV peak to peak, stability better than 0.001%+100 uV per 8 hours, and calibration accuracy better than 0.1%+1 mV).

For this failure mode, the manual's troubleshooting section is less than helpful: "No specific check can be suggested since failure to regulate within specifications may be caused by any of the components in the supply." Cue an hour of poking around with the meter and generally familiarizing myself with the schematic. I pulled out a few of the transistors and didn't find any failures, although the reported gain values did seem all over the map. (I couldn't find specifications, much less data sheets, for any of the key transistors.) Several operating voltages were way off from the schematic: Vce of Q1 and Q2 should be 9V and 8V respectively when outputting 20V (at 500mA); I was seeing values of 20-30V. (I don't have a suitable load, so all measurements are at whatever current flows through the 3478A in 30V range.)

Testing the ovenized voltage regulator, I saw a steady 4.987V across the main regulator diode (across pins 3 and 8 on the oven interface). But CR10, which is described as "provid[ing] a constant collector voltage for the input section of the differential amplifier," showed a varying voltage of around 3.8V, rather than the specified 5.4V. Oh well, looks like the oven has to come out!

Removing the oven board assembly is straightforward but fiddly. There are three screws around the circumference of the oven mount connector (on what the schematic calls the "front view" of the amplifier board, i.e. the bottom of the unit). These need to be loosened to remove the oven cover. Unfortunately they're all slotted screws pointed straight into the edge of the case or into other components, so I found it very difficult to get a screwdriver on them. I ended up removing the two screws that mount the whole oven assembly to the board (accessible from the "rear," i.e. top, of the amplifier board) in order to wiggle the thing around a bit to get a better angle. Thankfully the builders left a bit of slack in all of the connecting wires, so I didn't have to desolder anything.

Once the oven cover screws are removed, the other steps go much as described in the manual. When it says to "extract the oven board," it means to just pull straight up: there is a connector that will come free. I found a pair of parallel-jaw pliers helpful.

I desoldered the dodgy CR10 and tested its Zener voltage. 5.29V; below spec, but not by much. What happened to the dodgy 3.8V reading from before? Maybe it wasn’t being provided the expected current? But the upstream components, CR22 and Q10, were within spec. At any rate, after spot-testing the other bits of the oven board I reassembled everything and turned it back on. Lo and behold, the voltage between oven assembly pins 3 and 5 was now 5.35V, still slightly below spec but probably good enough. Maybe a dodgy connection on the oven board plug connection?

Warming the unit back up, I saw improved stability, although still way out of spec: fluctuations of a few mV off the 20V setpoint. Turning off the 10V gain and dialing the output down to 1V gave me a pretty stable output. The calibration was now off, though, and I needed to re-do the jumper wires to get it within the 2.5mV range of the main calibration pot.

Unfortunately, while messing with the jumpers I managed to drop a bare wire into the running unit, resulting in a bang and a blown AC fuse. Further testing is on hold until new fuses arrive.

I'm not actually sure what I shorted. Something near the oven assembly connector, I think. Fortunately, there was no sign of blue smoke, and nothing looks blown up.

[donlisms]

No followup?  Just when it was getting good!