PSU recap that ended badly, ideas?

[chud]

hi,
quick story recap -
pc failed to switch on again after poweroff, was intermittant then every time failed. Full power off and on would get it POSTing ok.
Replaced all obviously failed caps on output filter side and then almost always posted first time but sometimes not.

Decided to recap entire thing (further 12 caps including primary side, with best Panasonic ones I could get FM, TSED series etc.) and after connected to my pc (which nicely blew my USB controller after powering on) I thought I best test it some more before making same mistake again....

Situation: PSU loose, connect AC cord only, these are voltage readings, the 5v standby is ok but voltage is getting to 3.3, 15 and 12v rails somehow? (Voltages measured with negative lead on pin3)

pin | V | my readings
1 | 3.3V | 1.67
2 | 3.3V | 1.67
3 | COM |
4 | 5V | 2.31
5 | COM |
6 | 5V | 2.31
7 | COM |
8 | PWR_OK | 33mV
9 | 5VSB | 5.05
10 | 12V | 5.59
11 | 3.3V | 1.67
12 | -12V | -5.3
13 | COM |
14 | PS_ON | 5.05
15 | COM |
16 | COM |
17 | COM |
18 | -5V | -3.4
19 | 5V | 2.31
20 | 5V | 2.31

Everything seems ok, I cannot for the life of me see anything wrong but something obviously is, has anyone encountered this or electronically minded suggest what may be wrong?

(The only other thing I removed was the 3 FETs? on the primary side that were on the heatsink to be able to get to some of the small caps, but seem to be all back on ok.) 2 were identical and large, one was smaller. The 2 larger ones had same writing on and may have gone back reversed, but assume as same item that is irrelevant.

If anyone could help that would be great!
cheers

[stj]

Re: PSU recap that ended badly, ideas?

did you miss any caps hidden inside black shrink-sleeve?

[Toasty]

Re: PSU recap that ended badly, ideas?

Either the 115/220 switch is in the wrong position or 1/2 of the input circuit is dead. You're getting near 1/2 voltages. The input mains is "stepped-up" 350-400v to operate the PSU. Check the mains caps and surrounding circuitry.

[chud]

Re: PSU recap that ended badly, ideas?

stj: 99.9% sure nothing missed, it was working ok prior to replacing the last 12 caps, so pretty sure its something wrong relating to them or removal of fet's to get to them.

Toasty: switch deffo at 230V - without it being powered on with green-ground short, do you know why would there be any voltage at all on anything other than 5v standby (which is only correct voltage?)

I don't want to risk another session in a pc to get 'on' voltages after it blew the USB controller out on my main computer. (Incidently it did POST, then said keyboard wasn't present (PS2), then rebooted) - then I yanked power quick as realised wasn't right. It had also reset my CMOS and blown the usb.

I will have to buy a desoldering unit to remove the fet's again as it took hours with wick!

[Toasty]

Re: PSU recap that ended badly, ideas?

230v mains? Then you should have around 300-325vdc across the mains filter caps. Measure the DC output of the bridge rectifier.

What brand and model # PSU?
Any pictures of the top and bottom of board may help us.

Make sure you didn't get a cap in backwards.

The 5vsb is independent of the 3.3/5/12 outputs. It supplies standby and startup voltages to those other circuits. As I said before, it seems that the output from the mains caps is only running at 1/2 voltage. Check that bridge rectifier.

If this is a PFC PSU, could be a failure in the PFC circuit. Diode, cap, FET...

Keep hunting. Go back over solder work and make sure:
a) you didn't jump a trace where you -shouldn't- have
b) make sure you crossed a trace where you -should- have.

Heavily reflow and connections to heat sinks as some manufacturers use them as conductors/circuit jumpers.

Happy Hunting!

[stj]

Re: PSU recap that ended badly, ideas?

check you didnt leave out any insulators used to mount stuff to heatsinks.

[chud]

Re: PSU recap that ended badly, ideas?

Toasty: thanks ill try and get some pics up tomorrow - what I was getting at is why would there be anything other than the 5v standby current if its not powered on?

stj: i'd removed the thermal rubber pads the fets were mounted on to the heatsink in favour of decent thermal paste, id assumed they were just that and not performing any insulation? the metal backs of the fets arent conductive right?

[370forlife]

Re: PSU recap that ended badly, ideas?

Think we may have found your problem, the thermal pads do also insulate it, as the back is almost always connected to the center leg. Hope you didn't throw out the little plastic washers that went in the holes with the screws too.

[chud]

Re: PSU recap that ended badly, ideas?

hmm there were definitely no washers of any kind on any of them, literally a bolt going through the larger pair of fets? and the sink, with a nut on the end, on the single fet? was a bolt with a threaded hole in the sink.

I'll have to wait for a couple of days until I get a desoldering unit to get them off and test, I cant face another hours battle with bloody wick!

[370forlife]

Re: PSU recap that ended badly, ideas?

Well, there won't always be a plastic washer, if there is it would be between the head of the screw and the part. But, if the part(s) are to-247 or to-3p, they don't need the washers as the holes don't ever make contact with the metal on the back.

TO-247 and TO-3P both still need the insulating pads.

http://media.digikey.com/photos/Fair...261-TO-247.jpg (TO-247)

http://media.digikey.com/photos/Fair...TO-3P-5(L).jpg (TO-3P, most have 3 legs though, this is just an IC in a to-3p form)

http://media.digikey.com/photos/STMi...tos/TO-220.jpg (TO-220, to-220's with the steel tab need the plastic washer,)

[stj]

Re: PSU recap that ended badly, ideas?

hate to break this to you,
the thermal characteristics of the elastomer pads is better than silicone grease.

[chud]

Re: PSU recap that ended badly, ideas?

Here are some pics - desoldering unit ordered , will be about 2 days before I can get sink block off.


<BR>

<BR>

<BR>

<BR>

<BR>

<BR>

<BR>

[chud]

Re: PSU recap that ended badly, ideas?

ok, desolderer came, removed the heatsink, replaced the rubber pads and reassembled.
With it powered on, but not switched on, I am getting mV readings on the 3.3, 15 and 12v lines rather than 1/2 full voltage readings.
I checked this against 2 other psu's and found they do give some mV readings on some but 0 on others - the ones with mV readings are order of magnitude less, my max was 220mV on 12v, others were like 10mV.

Anyway I just had it pluged into a test pc and it fired up ok, bios voltage readings seemed fairly normal. Had it running for a few minutes and it seemed ok, no funny smells or lots of heat on either copper heatsink (that stick out).

Then after about 10 minutes i did a third round of touching it to get idea of temp and got a huge shock from the copper heatsink on the primary side (previous two touches were ok) - anyone have any ideas??

[chud]

Re: PSU recap that ended badly, ideas?

...the thermal paste i was using was 'Antec Formula 5' which contains some silver (I left some between sink and pad , and pad and FET, as there was white paste there previously), apparently although it is excellent thermal conductor it is meant to have negligable electrical conductivy..... Anyone have any real experience about this?

I also find it rather dangerous the metal PSU case is earthed, yet the heatsinks that stick out of the case are not and can get mains voltage on them easily, the only protection being very thin rubber pads??

[stj]

Re: PSU recap that ended badly, ideas?

a resistance will reduce the current, not the voltage.
use a meter on "resistance" to check between the heatsink & leads on each part to find the problem.

btw, that must have been a bitch - 380v roughly!!!

[chud]

Re: PSU recap that ended badly, ideas?

i'd very carefully checked for any shorts with multimeter before soldering it back in and there was nothing - just wondering if at high voltage the paste would allow leakage? I read up on it and there are odd articles saying its a bit conductive, but not at what voltage and how conductive it became.

Running under the metal of the sink at one end is a bridge wire that can be in contact with it and a couple of resistors the other end it sits on, either of which could possibly contact it if there ceramic is imperfect. There is also a vertical resistor that sits extremely close to it on one side that has an uninsulated leg that is possibly able to touch it.
I can put a rubber jacket thing on the large resistor and maybe put hot glue on the other exposed resistors and bridge wire under it? Seems oddly designed so many things are able to make it live especially as it is not earthed and easily touched?

[stj]

Re: PSU recap that ended badly, ideas?

silver past was intended for cpu's - i doubt they expect more than 5v near it.

if it really is silver then a good question is what voltage the stuff between the silver will breakdown at.
because then the power can arc through the past.

[chud]

Re: PSU recap that ended badly, ideas?

Dismantled it all yet again, used new elastomer pads from an old psu that hadn't touched any silver based paste, and re thermal pasted it vary sparingly with white Zalman (presumably zinc oxide paste). I also covered anything near it that was exposed with that plastic sheeting stuff used in the rest of the psu stuck down with hot glue.

The result - plugged in, but not powered on the mV readings were lower on 3.3/5/12 in line with other psu's I tested but still some were like 0.8mV instead of 0.

Pc again worked fine - but the exposed metal fins were again at 230v !!
The only wierd thing I observed was it took several seconds to reach that voltage, so it wasn't instant. Anyone got an ideas how on earth this is getting live???

[stj]

Re: PSU recap that ended badly, ideas?

no idea, btw, the reason you get a reading on the outputs is residual power in the output-caps.

[370forlife]

Re: PSU recap that ended badly, ideas?

Hmm, typically heatsinks are live in psu's, but one with giant copper fins sticking out of it probably shouldn't.

Did you desolder the primary heatsink and put rubber pads on that too?