Taking apart a PSU.... safety issues.

[Brad]

Volenti, how come 90% of failed psu's we get at work have blown those two big black caps?

[WireX]

bah, psu caps are nothing.

Just get an instant camera with a flash bulb.... (Read: It says in big letters on the side "Do not open for risk of electrical shock" so of couse I did)...

weeeee 110V fully charged... I was doing some testing with it, and my hand slipped. Ouch.

Then, just to see what it would be like, I hooked up a 1.5V led to the cap in series... Quite the burnt black led I have now.

oh yeah, and when I had just opened the camera, I was holding the pcb with pliers (So I didn't shock myself of course) and they were metal, bad idea, the bloody cap arched the contacts (because I was grounded) and it scared the shit outa me (then I shocked myself)

-WireX

[Volenti]

Volenti, how come 90% of failed psu's we get at work have blown those two big black caps?

Really? like exploded? or simply failed? Most of the powersupplies that I've seen dead (including ones that have died on me) have had the mosfet timing circuit fail, causing 2 banks of mosfets to turn on at the same time = magic smoke. Some of the supplies will simply blow the internal fuse but the symptoms are usually terminal anyway.

Hmm, the only 2 things that I can think of that would cause an electrolytic cap to explode are exessive heat (naked flame on the case of the cap, don't ask how I know this ) or overvoltage, it's highly unlikley that a sufficient spike could get through both filters (one on the powersupply board, usually another on a small seperate board attached to the inside of the powersupply case at the mains treminals) to cause the caps to rupture, certainlly in 90% of cases.

Unless the specific caps being used don't have a very large safety margin between their rated and operating voltage, and mains voltage surges(which the filters won't have any effect on, unless you use a UPS) are pushing the caps over edge...

Electrolytic caps can dry out over time and have less storage potential, but that's normally 5-10year process.

I live in australia and we use 50hz 240v mains, and the powersupplies are slightly different to accomadate this, but the overall design principle should be the same...

[Brad]

most of these have burn marks on the psu case next to the caps, and the caps themselves look all rooted.

NZ is on 240v 50hz too

[Volenti]

Ahh, yea forgot you lived in land of the long white cloud

I was having a closer look at one of my dead powersupplies and the main caps have a 200WV rating on them. Now normally caps are rated just in volts (and their capacity), not watt/volts (if that is what WV means) taken at face value that's over 100v short of what they should be (rectified 240v being around 340v) so unless their's a voltage divider or something before them theorectly they should go boom...*frowns, has an even closer look at the board traces...*

Ahha!, their not in paralell like I thought, their in series, that makes more sense.

There are some other components that could cause trouble should they fail , mainly medium and high current diodes that would do nasty things should they decide to pass current in both directions. It can be difficult at times to find the exact fault since a critical component can set off a chain reaction when it fails, taking out other components down the line, collateral damage so to speak.

I'd put my money on either the caps being under spec voltage wise, 200+200=400 with the 340v DC they see, not a huge saftey margin, or another component failing, and causing the caps to see 340 each, not between them, then boom.

Note, electronics is only a hobby of mine, I could, and have been wrong before.

[EMC2]

Volenti - the WV you sometimes see on caps stands for Working Voltage. (I hate caps rated that way as it takes into account derating, and they usually don't spec what derating numbers they used or what circuit configuration. Just give me the true rating)

An additional failure mechanism you have in caps associated with AC voltages is the amount of ripple current versus the ESR value of the cap. If the ripple current is too high, the cap heats up and can indeed go terminal

[Brad]

hmm, I have some here rated at 350V, that are blown