IBM 5170 power supply repair, 3Hz voltage spikes on all rails

[dmuntz]

I have a couple of IBM 5170 power supplies that are exhibiting similar symptoms. The DC voltages all have a lot of noise, varying by 0.1 - 0.2V according to my BM786 EEVBLOG(tm) meter, also the negative rails are "high" (-3V on the -5V rail, and -10.5V on the -12V rail). The interesting bit is that I took a look at the DC output with an oscilloscope, and all the rails have a very regular ~3Hz voltage spike that seems to account for the variability seen with the meter. I thought someone might recognize this symptom and be able to point me in the right direction.

In one of the supplies, I've replaced 3 blown WIMA caps (that was all of the WIMA caps), and am mid-recapping a bunch of 1200uF and 820uF caps that were way out of spec. By lucky coincidence, both of the supplies are almost identical to the one described in the SAMS for the 5170 (http://vtda.org/books/Computing/Hardware/Sams/Sams%20-%20IBM%205170%20-%20with%20type%201%20motherboard.pdf). These PSs use 3 LM339N voltage regulators, an LM358N op amp, a D83-004K diodes (mentioned because they're attached to huge heatsinks), a pair of 2SC3320 power transistors (also heatsinked), and two additional transistors: 2N2222 and 2N5682.

[RoGeorge]

With old electronics, the electrolytic capacitors might have dried.  I would change them all anyways, but I'm not familiar with that model so this is only generic advice. 

0.1-0.2 V ripple might not be that much, but the absolute voltage seems way off. Some power supplies can not stabilize well without a load.  Was there any load attached to the power supply while measuring it? 

[dmuntz]

I knew I forgot to mention something. It was tested with load (25r 50w on the 5v rail). These PSs will shut down w/o load. I don't know what IBM was thinking, but these have an absolutely stupid number of sizeable electrolytic caps. I think it was 17 large-ish ones in the 820uF-1200uF range (~1cm diameter, ~5cm high), 2 large ones 1000uF 200V, and another 10 or so small ones.