Compaq Deskpro 386s PSU

[the-compaq-guy]

So after 30 years, it finally happened and the capacitors leaked. I've tried a recap with Kemet's as the primaries and a mix of nichis and Wurths everywhere else, and it won't boot. No fans spinning. No LED activity. I've double and triple checked the soldering and negi-lead orientation. No catastrphic trace destruction and instead more minor blistering at some points. Thus I can only assume its ESR.

Now the main issue here is these are 30+ year old capacitors. The internet wasn't in the best condition in 1989 so I'm really struggling to find any mention of most of these caps at all so I need help with what capacitors would make sufficient replacements please.

Aerovox

M LPH 677 8937 C 250V/300V surge 680u
pos C7, C8

Marcon

CE-US 8945 BYM 25V 220u
pos C2

CE-US 8949 9ZM 35V 100u
pos C24

CE-US 8940 9xb 25V 1000u
pos C21, C22

CE-US 8945 9ym 25V 470u
pos C31

Vishay

511D 8943H USA 5 35V 68u
pos C9

I'm listing this one for anyone else trying to recap one of these PSUs. This family is still available, with data sheet here:
https://cdn.badcaps-static.com/pdfs/...c8b55a8f11.pdf


Nippon Chemi-con

9(2) OT SXF 10V 1000u
pos C27, C28, C44

This family is not available anymore, but this document from Nippon lists the LXY family as a suitable replacement.
https://cdn.badcaps-static.com/pdfs/...21ec008c0f.pdf

[sam_sam_sam]

Re: Compaq Deskpro 386s PSU

Could use any of the following brands

United Chemi-Con ( KY ) ( KZE ) ( LXY ) LXZ ) [ the last two are automobile grade ]
Panasonic Electronic Components ( FR ) FC )
Rubycon [ use low ESR capacitors ) just like KY KZE LXZ LXZ FR and FC
Nichicon ( same as above )

I personally use mostly United Chemi-Con ( KY ) ( KZE ) ( LXY ) LXZ ) unless I can not find the right size then I will look at the other brands mentioned

I hope this helps

[the-compaq-guy]

Re: Compaq Deskpro 386s PSU

I've seen those families floated a few times for a few different things, are they generally suitable replacements for most things that age? If that's the case my life just got a bit simpler thanks.

Any families in particular for the Rubys or Nichis? I have some Nichi PW(M) in there already. And for the primaries, does the family still matter there? I see Chemi-Con KMQ series on RS.

Edit: just saw your edit, thanks for the extra detail!

[sam_sam_sam]

Re: Compaq Deskpro 386s PSU

I've seen those families floated a few times for a few different things, are they generally suitable replacements for most things that age? If that's the case my life just got a bit simpler thanks.

Any families in particular for the Rubys or Nichis? I have some Nichi PW(M) in there already. And for the primaries, does the family still matter there? I see Chemi-Con KMQ series on RS.

Edit: just saw your edit, thanks for the extra detail!

I done quite a few switching power supplies before and have not had any problems using those brands and series

I hope that helps you

[stj]

Re: Compaq Deskpro 386s PSU

last AT class psu i recapped, i just filled it the panasonic FC and FK - no problem.
maybe it's not the caps - maybe the bad caps killed a semiconductor.

[the-compaq-guy]

Re: Compaq Deskpro 386s PSU

I done quite a few switching power supplies before and have not had any problems using those brands and series

I hope that helps you

It does a lot thank you. I have got the caps on order, I took your advice and ordered those chemi-con families (serendipity meant I ordered the LXY series for the cap in which the data sheet requested that series as a replacement. I'd somehow managed to blank that). And I went for Nichicon GX series for the primary smoothing caps as I made the bold assumption (as I'm more solder monkey than electronic designer) that ripple only really matters on the secondary side.

last AT class psu i recapped, i just filled it the panasonic FC and FK - no problem.
maybe it's not the caps - maybe the bad caps killed a semiconductor.

I hope not. It's hard enough finding adequate cap replacements, without trying to dig out datasheets for that lot I'll pop the main FETs out and try them in my multimeter if this second recap doesn't work.

[sam_sam_sam]

Re: Compaq Deskpro 386s PSU

PW is an older series but I have used them and there has not been an issue with them I did not list because it is an older series

[the-compaq-guy]

Re: Compaq Deskpro 386s PSU

So it's now had it's second recap with the above suggested caps and still nada so I fear this may be a silicon problem.

I'm not sure if this happened before the second recap, but when it's turned on, there are no signs of life except for the fan trying to spin up. It spins for about a second then stops for a couple. Rinse and repeat ad infinitum. I don't own an oscilloscope to go deep probing, and I don't know the pinout of the custom board connector. Next step is to trace every line out and work out where they're all going. It's a single-sided PCB so it shouldn't take TOO long...

Any suggestions welcome.

[stj]

Re: Compaq Deskpro 386s PSU

check the output diodes

[the-compaq-guy]

Re: Compaq Deskpro 386s PSU

check the output diodes

You've done this before! There is a Schottky barrier rectifier (SBL3040PT) in which all 3 pins are essentially shorts to GND. Will keep probing but that defo screams of being dead.

[kaboom]

Re: Compaq Deskpro 386s PSU

You'll have continuity thru the transformer secondary to GND on the "outer" pins of the rectifier; don't confuse this with shorts.

But as the center pin of the SBL3040 shows a short, you're reading thru the bad diode to GND via the secondary.

Measure from each output to GND to find (other?) shorts.
It'll be obvious and not mistaken for the tens to several thousand ohms of either load resistors or feedback dividers.

PWs or FCs should be fine for it.


Pull the caps on +5 and look for solder splash. Did the short clear? If not, replace the SBL3040.
Also check for electrolyte leakage from the old caps- it makes electrical leakage and eats traces. Ohm out/jumper anything questionable.
I forget if this power supply has riser boards, but check/replace those caps if present.

Don't run it too long w/o being installed in the case w/ cover on.
The system fan pulls air thru the PSU.
Make sure the case feet are present!

[the-compaq-guy]

Re: Compaq Deskpro 386s PSU

You'll have continuity thru the transformer secondary to GND on the "outer" pins of the rectifier; don't confuse this with shorts.

But as the center pin of the SBL3040 shows a short, you're reading thru the bad diode to GND via the secondary.

Measure from each output to GND to find (other?) shorts.
It'll be obvious and not mistaken for the tens to several thousand ohms of either load resistors or feedback dividers.

PWs or FCs should be fine for it.


Pull the caps on +5 and look for solder splash. Did the short clear? If not, replace the SBL3040.
Also check for electrolyte leakage from the old caps- it makes electrical leakage and eats traces. Ohm out/jumper anything questionable.
I forget if this power supply has riser boards, but check/replace those caps if present.

Don't run it too long w/o being installed in the case w/ cover on.
The system fan pulls air thru the PSU.
Make sure the case feet are present!

You were onto something here, once I popped the 3040 out I tested it on diode mode of my ol multimeter. It was fine. So I decided if it ain't broke don't fix it and put it back in.

I resoldered every connection around where there was an electrolyte leak of one of the old caps (and what prompted me to recap in the first place) and even through my respirator I noticed that wonderful fishy smell of burnt electrolyte. As a side note, I guess I was wearing it wrong. But that told me there was still electrolyte around making my life a misery so I popped out the only diode near where the spill occurred and ... yup. It's failed open circuit. Res of >2M in one direction, and about 1M in the other. There was also a small pool of electrolyte under where it was which managed to evade me blasting the board with circuit cleaner.

Need to work out what a "N4934 g946" is now. Gallium based Schottky diode?

Edit: I assume one of these

[PeteS in CA]

Re: Compaq Deskpro 386s PSU

That it was a 1N4934 fast recovery rectifier was my guess as well. On Semi makes the part as well.

[stj]

Re: Compaq Deskpro 386s PSU

if it's this then use an FR102 or a UF4007
https://uk.farnell.com/on-semiconduc...00v/dp/1017584

[PeteS in CA]

Re: Compaq Deskpro 386s PSU

Maybe UF4002, a 100V part like 1N4934 is, though it's a 50nS recovery time part instead of 150nS recovery time. FR102 is a 100V and 150nS part, a bit more similar to 1N4934.

[stj]

Re: Compaq Deskpro 386s PSU

i always buy 1000v parts,
they make you buy 10 or 25 so atleast the spares your stuck with are more usefull

[the-compaq-guy]

Re: Compaq Deskpro 386s PSU

SHE'S ALIVE

Thanks everyone for the tips. I bought a pack of 10 on-semi from ebay (i think just 100V ones sorry stj) and replaced the diode.

I also went around and reflowed, with loads of flux, every solder joint around the leaked area. Then doused the whole thing rather liberally in PCB cleaner and let it dry. Will probably pop it back out and give it a proper deep clean with distilled water/detergent, IPA soak and cleaning with a soft toothbrush at some point, but it's alive!

Thanks again everyone!