Taking apart a PSU.... safety issues.
 

I intend to mod my enermax so that I control the fan speed.
I've read that PSUs are not to be messed with unless you know what you're doing, so enlighten me.. :)

How can you ensure that it is fully discharged?

pull the power cable and hit the atx power switch, that should take care of it.

--Matt

there are huge capacitors in the PSU that can give you a severe shock; you will want to discharge them (pushing the power button when the PSU is unplugged may or may not do it).

or just do what i do, DONT TOUCH THEM LOL!! ..and you'll be fine

Yeh one thing NOT to do is have it plugged in and start licking the PCB.. thats a bad bad thing :)

Carefull modding enermax. when you put the 5V line way up your 12V shoots up past 13V just a warning you might fry some stuff. or your psu could start on fire like pH's

What are you talking about... "putting the 5V line way up"...

I just want to add fan control.

ooh i see didnt see it in the post :)

To discharge capacitors, you really should use a resistor to do it.
Basically, you should short each of the big capacitors through the resistor until it measures zero voltage with a multimeter.

Don't ask me what Ohms or Voltage because it depends upon what you are discharging and how fast you want it to discharge. The voltage that the resistor can handle should be more than the voltage on the capacitor. The resistance (in Ohms) should be higher if you don't mind waiting or lower if you want to dump the power in a hurry and don't mind dumping some wattage through the resistor. I think that there are some web pages that talk about it in more detail.

I just use a fan to discharge it. I haven't been shocked yet.

how, exactly, do you discharge the capacitors w/ a fan?

Also, how long after you unplug a PSU will they stay charged up?

can be a very long time - weeks, months...

High voltage capacitors can stay dangerous for quite some time. I don't know if it is years, since that depends upon the leakage characteristics and how big the capacitors are. I think the "big" capacitors in PC PSU's are only 35V, but they may be big enough to deliver dangerous (to your heart) amounts of amperage. I'd be more worried about the capacitors in a monitor or TV, since they are much higher voltage and/or current. There's nothing in a PC that uses 10,000V.

To use a fan, simply touch the two lead wires from the fan to the two wires from the capacitor. Assuming you didn't just fry your fan with a 35V or more shock from a large capacitor... the fan will bleed off the electrical energy stored in the capacitor.

But they taste soooooooooo good - and the current kinda tickles :p

The capacitors within your psu will drop fairly quickly. You have to remember, in an ATX supply, there is a 5v constant, always draining current from your house. So, when you unplug it, that 5v regulator is gonna burn off those big caps fairly quickly. If you are THAT worried about it, take a meter and test the 5vsb lead against a ground or the casing of the PSU. If it reads 0, then they are discharged. Plus, even if they are fully charged, its 35v DC current, which wont shock you, it will just tingle A LOT and kinda burn your skin. Not that I'd know or anything. Yeah, and dont lick em either :D

they taste sooooo goood!!!

What about the caps on the AC coming in? I don't know much about supplies, but won't those be charged to ~170v from the AC line?

OK, I don't design PSU's, so check for yourself if you really want to be sure. However... there shouldn't be any capacitors on the incoming AC. What would be the point? Capacitors smooth out changes in voltage, and the transformers need that smooth sine-wave signal coming in from normal AC.

I've never seen a capacitor attached to the incoming AC voltage lines.

Thats a good point . . .

I was thinking the AC would run through a diode and cap before the transformer, which would make no sense.

Carry on :)

Capacitors are used quite often in AC, infact every single mains fluroscent light has a capacitor in it.

A basic description of how a computer powersupply works;

For north america, you have 110v, 60hz AC coming in, this goes through a small filter ( essentially a low pass filter,which includes small capacitors across the active and neutral) this is then fed through a diode array which converts it to around 155v DC (the difference in voltage is due to how AC is rated in RMS rather than peak voltage)

This is then filtered by the 2 large high voltage capacitors in question, and then "chopped up" to produce a very high fequency square wave AC.

The square wave signal is fed into a small multi tap transformer, which creates the various voltages, which are converted back to DC again, filtered, regulated, and ready to power the components in a PC.

The reason for all this hassle, (ie, why don't they simply feed mains into a normal transformer) is size, cost and efficiency, the relatively low frequency of 60 hz requires large, expensive and inefficient transformers, the higher the frequency (up to a point) the more efficient and smaller the transformer can be.

Great answer Volenti! Thanks! Now you see why I prefaced my remark with: I don't design PSU's. My power supply education is based upon a hobby book on PSU design (using those heavy 60Hz single or dual-tap transformers). Those simple designs didn't prefilter the AC or convert the AC into a higher frequency.

The one thing I'm vague on is the "chopped up" phase of the PSU circuit you describe. Perhaps you can expand on that Volenti?

But that gets us back to the first point, Is it safe? Give it 10 minutes and yes, because the AC caps are filter caps, they arent designed to store much of a charge like the big DC caps he's refering to. The DC caps wont be up for very long, because they continue to feed the squarewave circuit to produce the 5vsb.

Lets simplify more...

110@60hz in. Sinewave current
-Converted to ~150 DC and filtered by BIG caps to reduce ripple
-Re-converted to AC power via FET's with a switching rate much higher. 300hz? beats me. (Chopped up phase)
*newly formed high frequency power*]
-Into little transformers
-out to some more filter caps
-out to 5, gnd, 12, -5, 3.3, etc

I'm not sure what is the exact frequency my self, but the switchmode powersupplies in car audio amplifiers commonly use around 32khz, PC powersupplies would use something similar I suspect.

As meltman said, the high voltage DC is "chopped up" by the high frequency fet's (fast switching transistor) by rapidly turning it on and off, like flicking a switch really fast, thus you get a rough AC current, I can't remember how the circuit attains the +,0,- cycle,(needed for the transformer to work) since the switched DC is initally just + and 0. (from what I recall)

Thanks for the great replies! This is what makes this forum good: actual technical replies instead of noise.

As for creating negative voltage, they could just connect the center tap of a transformer to ground to get symmetric positive and negative voltages.

Exactly. There are a couple of small transformers in most supplies that could very well be center-tapped. OR... they could be using a negative 5v DC-DC converter. There are ways of generating -5v without using a transformer. Actually, come to think about it, I would assume that there is a DC-DC converter because of the amperage rating of the -5v line. It is very low. In a symmetric transformer setup, it would be equal to the +5v line in amps.

later

Just FYI's ;)

Most recent switchers operate in the 100Khz to 1Mhz range (some of the newer PWM converters operate up to 2Mhz). Higher frequencies are used for both better effeciencies and so that smaller inductors may be used.

Regarding generating negative voltages - a center tap is one way. More common is an inverting switcher design (which requires no transformers).
A good example of one is On Semi's NCV33063A controller which can be configured for step up, step down, or as an inverting switching supply.