PSU recap that ended badly, ideas?

[chud]

Re: PSU recap that ended badly, ideas?

370: the larger heatsink on the output side sticks out of the back of the case and has a metal guard around it - I haven't touched that at all solder wise (and it isn't live).

I only removed the primary side one (small with no guard as it sticks inside the case) to get to some caps that were between it and other one.
I am pretty sure in the past I have touched that when pc was on, moving cables about inside the case and not got zapped and cant believe you could exposed live metal and it pass any basic safety test.

Why do you say heatsinks are usually live? surely the elastomer pads and little plastic screw insulator things are there to electrically insulate as much as conduct heat?

[chud]

Re: PSU recap that ended badly, ideas?

Can hopefully see I have hot glued plastic film to anything that I thought may be possible to cone into contact with the heatsink

To the left is the bridge wire that was exposed under the sink which joins two legs of the fet's but not the middle legs that are backing onto the sink. ..and you can also see the track I lifted along with the holes little copper tube (oops!) which goes to a little ceramic disc cap, but its joined underneath too, so its ok



These are the fet's using the zalman paste and some elastomer i removed from another old psu (that had no paste on it to begin with)

[chud]

Re: PSU recap that ended badly, ideas?

ok, i blew more money on it and bought some new aerospace grade Bergquist sil-pad 2000 thermal pads, the best Farnell had.
This time I used no paste whatsoever, everything was cleaned in degreaser just for good measure too.
And what is the outcome? well I now get *just* 150-160V between the primary heatsink and earth. These pads are brand new, clean and have 4kV breakdown resistance so there is absolutely no way at all anything is conducting to it - anything remotely able to touch it has plastic hot glued to it.

I took more measurements this time.

#between heatsink and earth with psu on and running a test pc 100% ok;
- uA setting on multimeter says ~3406 uA AC, in mA setting its 1.19mA AC
- varies about 154v to 161v AC and Hz setting says 290Hz although dont really understand that.
- Hz on 5v is 300-320Hz and 12v 140-170Hz, relevant??
- 12v and 5v lines were about exactly that give or take 0.1v

#between heatsink and earth with pc plugged in but not attached to pc;
-current at mA setting is 0.02mA
-sometimes its about 26v AC and can be as much as 95V AC, if i touch negative probe to something else (say other heatsink) and back it changes value sometimes,but not often.
-unlike before and more like other psu's the 3.3, 5 and 12v lines on the ATX connector measure between only 0.1mv and 0.0mv.

The voltage in either case rises almost straight away, say within about a second, to drop off to 0 takes a good few seconds, about in line with the pilot light on the switch going out.


Can anyone understand what on earth is going on, its like its somehow getting statically charged ?


"
1 mA can be felt;
5 mA is painful.
Above 15 mA, a person loses muscle control, and 70 mA can be fatal."

I mean if its really got 1mA across it and its at 190V you will get a shock so it cannot be right at all?? It certainly felt like i got zapped with 5mA previously!!

[stj]

Re: PSU recap that ended badly, ideas?

the sink is held down by solder tabs, do the tracks they solder too go anywhere?

[chud]

Re: PSU recap that ended badly, ideas?

sink is only attached by the FETS legs, it doesnt make contact with anything else.

[Toasty]

Re: PSU recap that ended badly, ideas?

Ah-ha! That's the problem. There are 3 elongated holes in the board that match indentations in the heatsink where mounting lugs are to be mounted. Are you saying that those lugs never did or do not exist?

The one near the single transistor looks to have been formed (stamped) to hold the leg, but it's missing. That is most likely why you are seeing a voltage on the heatsink. No ground connection. As I said way back ^^, many manufacturers use the heatsinks to complete a connection across the board. All 3 of those lugs interconnect the circuit.

Toast

[momaka]

Re: PSU recap that ended badly, ideas?

Ah-ha! That's the problem. There are 3 elongated holes in the board that match indentations in the heatsink where mounting lugs are to be mounted. Are you saying that those lugs never did or do not exist?

The one near the single transistor looks to have been formed (stamped) to hold the leg, but it's missing. That is most likely why you are seeing a voltage on the heatsink. No ground connection.

Out of curiosity, I just tested a Bestec PSU that has its primary heatsink soldered to the negative side of the bridge rectifier. And the result: -165 VDC between primary heatsink and ground with the PSU plugged in. It seems to me that this is exactly how chud's PSU would have been if the mounting lugs were installed. If they weren't installed, that's actually a good thing because the negative side of the bridge rectifier is not a ground (not at 115VAC line input anyways... but I'm guessing it's probably the same thing even with 230VAC line).

chud: with the PSU off and unplugged, try measuring the resistance between the primary heatsink and the negative on the bridge rectifier. If you get very low resistance, like 1 or 2 ohms, then that's definitely where the voltage is coming from. If you don't get any resistance, try measuring the DC voltage between the primary heatsink and ground with the PSU plugged in (keep one hand behind your back and be careful here!). Have the negative probe on the ground/case of the PSU and the positive going to the primary heatsink. Is the voltage positive, negative, or zero?

[Toasty]

Re: PSU recap that ended badly, ideas?

I agree that the mains side heatsinks are typically hot, but from his statement that the heatsink is only attached by the FET legs, something is amiss. I went back to the underside shot on the first page and as chud stated, the heatsink is not connected to the board by the mounting lugs.

So that leaves only the probability that there is a connection being formed by one of the 3 FET mountings. An insulator may be missing somewhere. Either that, or the heatsink is coming in contact with an exposed conductor.

Pull it out and do a resistance check from all of the FET legs to the heatsink.

[stj]

Re: PSU recap that ended badly, ideas?

if there is no reading for resistance it could be capacitively coupled,
in this case you may be able to fit an earth-bond between the heatsink & case.

but *only* if the heatsink is 100% isolated from all the fet pins.

[chud]

Re: PSU recap that ended badly, ideas?

Toasty, no lugs on the heatsink, it was stamped to take some but none there, just the 3 fet's legs connect it to the board. (The small hole next to the fet thing on its own was created by me desoldering the wrong hole - looking at the board side it appears it was to take a diode although I dont think that is relevant to this)
On the front of the psu where the copper fins come out,there are bits of rubber as if to stop any chance the copper fins get bent back abit to make contact with the case and the exit hole in the case has a plastic surround/grommit. Again I dont know if this is relevant.
momaka, will have a go at what you suggest in next day or so, need a few days off from the damn thing!

There is definately no continuity between any fet legs and heatsink,at least with multimeter, I presume the 4kv thermal pads aren't breaking down at 230v. Indeed appart from the very first silver thermal paste and no insulators 'mistake' there has never been continuity between heatsink and legs.

As you can see I have covered most everything vaguely able to touch the heatsink with plastic and hot glue - anyone see anything i couldn't? [caps either side of r17 & r17 itself now also have plastic on them since taking pic]

The wierd part is that the voltage is lower (~160 down from 230v AC) since using these new thermal pads and no white paste. Also there is less mV on the outputs, now it seems to be normal levels when off.

stj; has me replacing the caps somehow created this? i recapped the entire psu with same uF rated caps, just used next voltage up where possible. Because most of them are physically bigger (even same uF & V ones, Panasonic equivalents all larger (often taller) are they too close to other things causing problems? Th two main input caps are same diameter but about 10mm higher (hence had to Dremel sone of heatsink away).