Why you shouldn't open power supplies (w/video)

[Bobnova]

So, it's (constantly, and correctly) said that you should not open computer power supplies due to high voltage.
Being bored, I decided to do two things:
1) Play with high voltage.
2) Explain a bit more.
3) Make a video.


First up is the explanation! There are three main sections to a PSU that contain capacitors that can hold a charge after the unit is turned off and unplugged.
You really shouldn't open a unit that is plugged in, there will by live mains voltage inside, and it has a good chance of killing you if you touch the wrong thing.

In order:
1) The transient filter, this is charge to line voltage (peak) at most, and has fairly minimal capacitance. Worst case, it hurts. You'll probably live. It comes with a discharge circuit that should render it safe within a few minutes.

2) The APFC section and storage capacitor. Usually has a discharge circuit of some sort, as well as bleeding out to an extent powering the 5vSB circuit while unplugged. It is charged to ~380v during operation and has a significant amount of capacitance. It takes a significant amount of time to bleed out under the best of circumstances (minutes), and takes a very long time in some designs. We're talking days, in theory. (Older passive PFC and non-PFC units may hold it even longer)

3) The output side, which has a lot of capacitance, but is only charge to 12v/5v/5v/3.3v/-12v. You aren't going to do anything meaningful to yourself with this. The 12v will drain out rapidly through the fan, anyway.


#1 isn't overly much of an issue. #3 isn't one at all.
#2 on the other hand can do real damage.

I built a circuit to charge a small (330uf) APFC capacitor, charged it to ~75v and discharged it via an alligator clip.


Note the alligator clip welding itself to the lead it touches! Not subtle, this.
It's far from the most impressive youtube capacitor discharge clip, but keep in mind that in service the voltage (and hence, energy) is about five times higher.

The circuit is simple, but I am not going to share it as you can do Bad Things to yourself with it if you're not careful.
If you don't know how to build it, the odds are good you shouldn't

 

[Mr.Guvernment]

i agree "most" people should not open one, unless you know what your doing, just as most people should not build computers... most people should not drive a vehicle.. and yet they do it!

nice video.. myself, soon i will be building an SFF system and had considered i may need to re-cable a PSU to save room..

 

[Ivy]

It was a small CAP. Imagine what can happen when one of the biggest is used, it can truly instantly kill someone. Because it can discharge itself in a fraction of a second. But as long as PSU is not opened there should be no danger. The stuff on the MB is usualy to small to be dangerous, but i still would not suggest to touch the stuff. The biggest and dangerous CAP is inside a PSU.

 

[Bobnova]

The stuff on the mobo is all too low voltage.
You'd have to really work hard to kill yourself with 12v.
380v it's a bit easier

This cap is on the small end of APFC caps too.


EDIT:
It's really too bad I don't have slow motion, or better sound capturing. This thing sounds mean.
Might swap the 330uf cap out for a 680uf.

 

[nightelph]

This deserves a sticky I think, even if not a lot of people read it, if it prevents .01% of people who see the headline from shocking (or worse) themselves, I think it deserves placement.

Bob, I kinda skimmed it but adding something about discharging the caps with a big screwdriver might be a worthwhile addition.

I liked that you pointed out that not only the caps contain power; I've touched live mains accidentally (15years ago ) due to this ignorance. Nasty shock.

most people should not drive a vehicle.. and yet they do it!

LOL!

 

[Lady Fitzgerald]

You should see the damage done by a large power line cap used to adjust power factor. You could weld from one of those. Striking an arc would be easy.

 

[jade]

If I'm scraping computer parts and take apart a PSU...
how long do I have to let the PSU sit before taking it apart. I don't want to get zapped. I saw that the APFC section can take a day to drain, so is one day good?

 

[Bobnova]

If the PSU is an APFC unit (no switch on the rear), a day should render it safe, more or less.
The most I've seen in an APFC cap after that is ~18v. Enough to suck if you have wet/sweaty hands and/or thin skin, but not enough to lay you out.
A passive PFC system I don't know, I haven't taken one apart after less than a month. Ditto for non-PFC.

What I recommend follows. Do me a favor though and don't open them in the first place. It's a bad idea.
If you do decide to take your life into your hands, this is what *I* do:

1) Pop the fan/top housing off. Leave the PCB in place. Don't touch anything shiny.
2) Find an alligator clip type lead and a thin, plastic handled, screwdriver.
3) Clip one end of the lead to the ground lead onto the neutral wire on the receptacle (follow the two wires to the PCB, the one that doesn't lead to a fuse is neutral. The wire should be blue).
4) Clip the other end of the alligator lead to the screwdriver's metal shaft. Hold only the plastic handle.
5) Poke everything shiny you can find, especially the leads of the MOSFETs near the Great Big Cap. If something goes POW you just saved yourself some pain.
6) Pull the PCB out, don't touch anything shiny!
7) Take a voltmeter, set it to HV (something >380v), check the voltage between the two legs of the Big Fat Capacitor where they poke through the PCB. If you don't read voltage, drop the scale to 200v. If you still don't, drop it to 20v. If you still don't, it's safe and off you go.
8) If the capacitor leads have voltage, use the screwdriver (hold only the plastic handle!!!!) to short the two legs together. Expect a pretty solid pop. Alternatively use a 1k-ohm (ish, 470-2.2k works) to connect the two posts, slower but safer.
9) re-check voltage.

If that seems excessive, it's because it's as safe as it gets.

 

[jazon1]

most capacitors you can discharge by touching a ground to the cap on the top of the capacitor if finding the leads on the other side of the board is difficult/imposable.
but defiantly agree be very careful with any high voltage capacitors and always treat them like they are charged until you know for sure there not with, multimeter etc.

 

[Bobnova]

The APFC cap is isolated from ground, you're stuck with using the neutral for that sucker.

 

[SolidxSnake]

If the PSU is an APFC unit (no switch on the rear), a day should render it safe, more or less.
The most I've seen in an APFC cap after that is ~18v. Enough to suck if you have wet/sweaty hands and/or thin skin, but not enough to lay you out.
A passive PFC system I don't know, I haven't taken one apart after less than a month. Ditto for non-PFC.

What I recommend follows. Do me a favor though and don't open them in the first place. It's a bad idea.
If you do decide to take your life into your hands, this is what *I* do:

1) Pop the fan/top housing off. Leave the PCB in place. Don't touch anything shiny.
2) Find an alligator clip type lead and a thin, plastic handled, screwdriver.
3) Clip one end of the lead to the ground lead onto the neutral wire on the receptacle (follow the two wires to the PCB, the one that doesn't lead to a fuse is neutral. The wire should be blue).
4) Clip the other end of the alligator lead to the screwdriver's metal shaft. Hold only the plastic handle.
5) Poke everything shiny you can find, especially the leads of the MOSFETs near the Great Big Cap. If something goes POW you just saved yourself some pain.
6) Pull the PCB out, don't touch anything shiny!
7) Take a voltmeter, set it to HV (something >380v), check the voltage between the two legs of the Big Fat Capacitor where they poke through the PCB. If you don't read voltage, drop the scale to 200v. If you still don't, drop it to 20v. If you still don't, it's safe and off you go.
8) If the capacitor leads have voltage, use the screwdriver (hold only the plastic handle!!!!) to short the two legs together. Expect a pretty solid pop. Alternatively use a 1k-ohm (ish, 470-2.2k works) to connect the two posts, slower but safer.
9) re-check voltage.

If that seems excessive, it's because it's as safe as it gets.

I would refrain from discharging a capacitor by shorting it's leads. One should, instead, use a large, high-power resistor (100k-1M Ohm, and 5W or more for overkill safety). If you're getting arcing, you're not treating the screwdriver/alligator-clips/PCB very well.

Also, keep in mind that rapidly discharging a capacitor (ie, a short) will _not_ always discharge it. Sometimes, after shorting a capacitor for a second and then removing the short will not discharge the capacitor due to a memory effect, and metering a capacitor after a short will confirm this.

I've gotten zapped many a time by touching live 120VAC mains. The classic basic (but not foolproof, by any means) is the one-hand rule: keep one hand in your pocket (and isolated from anything metal or any type of ground or other charge sink), and poke around only with the other hand. If you happen to touch something with a dangerous potential, the current will not flow through your body (or, in particular, your heart), and you'll walk away with a sore/numb arm.

 

[Bobnova]

It's going to take a while to discharge an APFC cap through a 100k-1m ohm resistor.
Having been playing with this APFC cap I have noticed the self-charge bit, most I've seen after charging it to 160v and shorting it is 1.1v though.

 

[SolidxSnake]

It's going to take a while to discharge an APFC cap through a 100k-1m ohm resistor.
Having been playing with this APFC cap I have noticed the self-charge bit, most I've seen after charging it to 160v and shorting it is 1.1v though.

I will agree to that, however, I'd imagine arcing with a 470-2.2kohm resistor, which can result in pitting in the screwdriver/alligator-clip/PCB.

Though, of course, the best way to figure out what sized resistor to use is to calculate it. Calculate the RC constant and then check how long it will take for the cap to reach safe levels, then adjust R accordingly. RC = Resistance * Capacitance in Ohms and Farads, respectively. The voltage across the capacitor is given by V*exp(t / RC), where V is the fully charged voltage of the capacitor (assume the capacitor's maximum rating if the circuit's usage of the cap is unknown), and t is the time, in seconds, after connecting the resistor across the terminals.

 

[Bobnova]

No disagreement here on that front. Calculating is best.
Really, the way I do things probably isn't the best way (the best way is to not open it in the first place...). It's just how I do it.

In my testing I've been unable to get any kind of arc through a 470ohm resistor with the cap charged to ~130v. At 380v I expect things would be different, I need more than 5v input to get that high though and I'm unwilling to hook it up to a PSU as then it wouldn't be an isolated charge like it is running on batteries.
I'm not overly interested in toasting myself, after all.

 

[dejo]

many years ago I had a friend that decided to install a 2farad cap to his car stereo. He had used 0 guage wiring for everything and when he went to power the capacitor, he found out why the sent a resistor with it- when the connection was made it literally blew the tip of the screwdriver off.

never underestimate the power of the magic electricity

 

[wagex]

many years ago I had a friend that decided to install a 2farad cap to his car stereo. He had used 0 guage wiring for everything and when he went to power the capacitor, he found out why the sent a resistor with it- when the connection was made it literally blew the tip of the screwdriver off.

never underestimate the power of the magic electricity

hahahaha ive seen this happen a time or two, actuallly happened to me once... i had charged it, i use a test light instead of a resistor so its easier to tell when its charged, no more light but then i took it out to do some work and appearently i left it out too long (few days) and didnt recharge it hooked up the positive and pshew!!!!!! sparks everwhere and melted half the eye connector off some 4 gauge i think it was a 2 farad as well stopped using them caps once i learned moar about systems though. now i have a battery bank

 

[Ivy]

The stuff on the mobo is all too low voltage.
You'd have to really work hard to kill yourself with 12v.
380v it's a bit easier

This cap is on the small end of APFC caps too.

The 850W PSU im using got 2 big 420 V CAP connected in parallel = 780 uF charge, and those are able to kill.

Those on the board are to small and to low voltage thats right, so there is no danger from a board.

 

[nightelph]

Battery bank Wagex!? Dang, my system is weak. I remember charging the 1 farad cap through a light bulb, acting as a resistor.

 

[wagex]

Battery bank Wagex!? Dang, my system is weak. I remember charging the 1 farad cap through a light bulb, acting as a resistor.

yea i had a bank of semi batteries in my toolbox in my truck i miss that system :'( now its in my little brothers truck lol had to grow up hes rockin some XS batteries though not that i dont want it i plan on doing another build later on once babeh grows up a lil.. she will be here in nov

 

[JigPu]

I've gotten zapped many a time by touching live 120VAC mains. The classic basic (but not foolproof, by any means) is the one-hand rule: keep one hand in your pocket (and isolated from anything metal or any type of ground or other charge sink), and poke around only with the other hand. If you happen to touch something with a dangerous potential, the current will not flow through your body (or, in particular, your heart), and you'll walk away with a sore/numb arm.

So very much this. If you're gonna get zapped, best get zapped "only" in the hand.

JigPu