Testing my power supply after lightning strike (is it dead?)

Joined
Nov 23, 2007
Messages
22
So after a thunderstorm, my server stopped powering on. I didn't observe the thing breaking since I wasn't at home when it happened. It was hooked up to a surge protector, but hey this is lightning we're talking about -- not your every day power surge. The PSU is an Antec TruePower 550 W unit. I thought it could either be the power supply or the board. I took out the PSU, and plugged it into the wall to test it with a multimeter. Here are my observations:

1.) Voltages dissappear after the power supply has been running for a while. They seem to be fine immediately after I turn the power supply on. I don't know if this is a measurement error or correct behavior or what.

2.) Testing the purple wire with ground with a multimeter says no voltage when the switch on the back is ON. This is the +5 V diagnostic wire... it should say +5 V even if the fan is not spinning.

3.) Switch on back of power supply is OFF and shorting the purple wire and the ground still causes the fan to spin. The purple wire should only have an effect IF the switch on the back is ON. There is a visible spark when I do this and the wire gets really hot. Why does this happen with no voltage when I test it with the multimeter? Also, if the switch on the back is OFF, why is this happening? There should be no power at all going through anything.

4.) Shorting the green and the black wires causes the power supply fan to start spinning.

5.) Shorting an orange (+3.3) wire with a black wire causes a spark, and the power supply shuts down completely (have to unplug and replug the AC power plug to get it back running again). Weird.

6.) If the power supply fan is running (green and black shorted) and power supply is ON, the Molex drive power connectors say no voltages. There are no voltages when the power supply fan is not running or when the power supply is OFF either. In other words, I have never been able to get the drive connectors to read any non-zero voltage with either +V terminals (yellow+black or red+black) on the plugs. Weird.

Does this mean the power supply is shot? I hope it is the PSU, and not the board -- otherwise, this could get pretty expensive.
 
Shorting out any wire besides a green and black causes a large amount of current to flow through it, which of course heats up the wire and causes the power supply to be overloaded. Of course it will shut off, that is its overcurrent protection.

You should NEVER short out any wire besides the green and the black one. Shorting out the 5v and 3.3v wires was frankly a very stupid thing to do, and could potentially have had disastrous results. It would have been far more dangerous if you had attempted to do that with a 12v line, which could have possibly electrocuted you or caused the power supply to catch fire or blow up.

If the molex doesn't have any power, it most likely means your PSU is shot, although some components are apparently still working on it. Connect a fan to the molex connectors and see if it spins up while shorting the green and black.
 
So the fan started spinning which is bad because it suggests that the PSU is all right and that the board is dead. This is not good. Also, I'm extremely lucky that nothing bad happened.
 
So the fan started spinning which is bad because it suggests that the PSU is all right and that the board is dead. This is not good. Also, I'm extremely lucky that nothing bad happened.

That's not necessarily the case. Test out the power supply on a cheap spare system that you don't really care about. Power supplies generally are good about knocking themselves out before knocking other components out, unless they are faulty or have a bad design.
 
The Truepower 550W is an old and not very good quality PSU anyway, so I suggest you just spend sixty bucks and replace it with a more modern quality unit. One way or another, you'll find the source of your problem, and you'll end up with a much more reliable PSU.
 
1.) Voltages dissappear after the power supply has been running for a while. They seem to be fine immediately after I turn the power supply on. I don't know if this is a measurement error or correct behavior or what.

Probably some capacitors or the voltage regulators are screwed.

2.) Testing the purple wire with ground with a multimeter says no voltage when the switch on the back is ON. This is the +5 V diagnostic wire... it should say +5 V even if the fan is not spinning.

Purple is 5V standby, without that getting power, you can't even turn on the motherboard. That's very bad.

3.) Switch on back of power supply is OFF and shorting the purple wire and the ground still causes the fan to spin. The purple wire should only have an effect IF the switch on the back is ON. There is a visible spark when I do this and the wire gets really hot. Why does this happen with no voltage when I test it with the multimeter? Also, if the switch on the back is OFF, why is this happening? There should be no power at all going through anything.

Shorting purple to ground should NOT cause the PSU to start. Green to black should do that, but CERTAINLY NOT with the switch flipped to off. If there's a spark and the wire is hot it means that there's a short and it's going out to ground to keep you from getting electrocuted or killing other hardware.

4.) Shorting the green and the black wires causes the power supply fan to start spinning.

Good.

5.) Shorting an orange (+3.3) wire with a black wire causes a spark, and the power supply shuts down completely (have to unplug and replug the AC power plug to get it back running again). Weird.

That should do nothing. That means that ground is LIVE and thus there's a short inside the power supply, so it's shutting itself off when it notices the 3.3V going weirdo as a result of you shorting it with ground, in order to protect itself and anything connected.

6.) If the power supply fan is running (green and black shorted) and power supply is ON, the Molex drive power connectors say no voltages.

Cables disconnected, broken, or VRMs gone, capacitors popped, etc.

Does this mean the power supply is shot? I hope it is the PSU, and not the board -- otherwise, this could get pretty expensive.

Yes, the power supply has a massive failure inside, but it's very good that the protection circuits still work, otherwise you could have started a fire, or worse, zapped yourself. You must replace it.
 
The Truepower 550W is an old and not very good quality PSU anyway, so I suggest you just spend sixty bucks and replace it with a more modern quality unit. One way or another, you'll find the source of your problem, and you'll end up with a much more reliable PSU.

What do you recommend?
 
btw- the answer to #3 is because even though you turn the switch off, the input capacitors still hold a charge (these are the big AC caps that filter some of the supply line noise and voltage variations). when they are charged and you try to turn the PSU on, the energy in them will begin to power the rest of the system. usually there is enough power to spin the fan for a half second or so.
 
Back
Top