Sign in with
Sign up | Sign in
Your question

Testing power supplies

Last response: in Components
Share
July 12, 2006 11:38:03 AM

Can someone give me some info on how to test a Power Supply to see if it is good before I try to install it. Thanks

More about : testing power supplies

July 12, 2006 12:33:47 PM

Assuming its an atx,,,, you will need to jumper (MB 20 pin connector) the green wire to any black/ground. That provides the PSU start signal.

Make sure this 20 pin connector is not connected to the MB when you are going to jumper the green & black pins & make sure you have at least 1 HD or CD/DVD drive attached to the PSU, all you really need here is a load for the +5v & +12 Rails.

Once you plug it in & jumper the green to black, the PSU fan & the attached drive should start spinning.

NB. If you suspect that you might be COLOUR BLIND do not attempt this.
There no hazardous voltages on any external wires on the ATX PSUs but if you are colour blind you might be testing the wrong wires.

You can now measure the voltages if you have a DC volt meter.

The three primary rails to check are:
+3.3v which should be at any of the orange wires.
+5v which should be at any of the red wires.
+12v which should be at any of the yellow wires.

If within range of +or - 10% for the rails you can assume its good.

This should give you a fairly good indication oh the health of any atx PSU.

I can provide additional info on the other wires or rails & psu mods if you need.
July 12, 2006 12:35:34 PM

There will be no current until you put it under load.

The answer is there is no easy way for the average consumer to verify if the PS is good, short of just seeing if it will power on and the voltages are good, which is better than nothing. The only thing you can do really to verify it is actually good under full load is to put it in and hold your breath.
Related resources
July 12, 2006 12:41:01 PM

Stop scaring the person...

There many good ways to test a PSU.
You can even get a PSU tester & just plug it into the 20 pin ATX connector.
It provides the power up jumper & load resistors for all the rails & lights to indicate a propper working PSU,,, & they are cheap as little as 8.00 U$.

eg. http://www.dansdata.com/quickshot018.htm
July 12, 2006 12:45:51 PM

Quote:
Stop scaring the person...



I'm not. The average consumer can't do it. If they want to buy the test equipment, have at it, but it won't put it under full load and one shouldn't assume it will. Is it a good test? As good as any that most people can do, and if you want to do something rather than nothing, then do it.

Quote:
When you connect the multimeter to ground and a live wire, is that is series or parallel :?:


It could be both. If the meter is the only current path, then it's series. If you're measureing across the load, it's parallel. The only way to measure the current supplied by the PS is to put it under load and then measure it in series....or, with a current probe which would be better/easier. But, then you don't know if you are at maximum load. Short answer, it's not really an undertaking for the typical consumer.

If one wants to measure the voltages, go for it. It may not tell you what you want to know, but it will probably tell you what you want to hear. :wink:
July 12, 2006 12:50:18 PM

Strangely,,, thats both. If your meter is really good & on the right setting you will get a reading of sorts, & if it does not have overload protection then you meter might smoke.... no kidding here...

It would be ideal to measure both voltage & current but for most basic testing you are interested in voltages only & you should measure from:
BLACK to Orange for +3.3v
BLACK to Red for +5v
& BLACK to Yelloy for +12 v.
July 12, 2006 2:13:23 PM

hey man

i came across this power supply tester which i think it's better than the one suggested previously

Link: Tester produced by Frozen CPU


Features:

Tests 20-pin and 24-pin power supplies

Tests SATA power

Tests Pentium 4 power connector

Tests PCI-Express power connectors

Tests Xeon power connectors

Tests Floppy drive connectors

Tests standard 4-pin power supply connectors

Tests for +3.3V, -12V, PG, +5VSB, +12V, -5V, +5V outputs

cheers!!!
July 14, 2006 3:21:21 AM

I have two PS testers. One just has the green light to tell if it powers up and the other checks all the rails. I have had them test out good and still not fire up a system. Bruce
July 14, 2006 4:35:16 AM

are there programs that test out psus??

PSUz instead of CPUz etc etc

why dont some smart dude make a program like CPUz for PSUs????
July 14, 2006 5:36:08 AM

Quote:
You need to get yourself a multilmeter and measure the current (measured in amps) on the wires to ensure they are within the sticker specs.
That is outright wrong advice in this case.

Measure voltages, not amps. Have the meter set to a DC volts scale and its leads connected to measure volts (black lead in the meter's COM socket, its read lead in the socket labelled V, Volts, VDC, or V with a straight line over it, either solid or dashed, which also indicates DC volts).
July 14, 2006 6:45:39 AM

Quote:
hey man

i came across this power supply tester which i think it's better than the one suggested previously

Link: Tester produced by Frozen CPU


Features:

Tests 20-pin and 24-pin power supplies

Tests SATA power

Tests Pentium 4 power connector

Tests PCI-Express power connectors

Tests Xeon power connectors

Tests Floppy drive connectors

Tests standard 4-pin power supply connectors

Tests for +3.3V, -12V, PG, +5VSB, +12V, -5V, +5V outputs

cheers!!!



FrozenCPU.com is a great place to get stuff like that.


http://www.frozencpu.com/psu-165.html
July 14, 2006 8:01:49 AM

The Psu Tester Isn't expensive. I think the tester is better then shorting the power supply
July 14, 2006 12:53:24 PM

The PSU tester does the same thing, so whats your point?
I bet you dont overclock either?
July 14, 2006 1:17:28 PM

It's possible the tester uses a ttl signal to turn it on, which is how it's suppose to be turned on, not shorted. I don't necessarily see how shorting it to ground will do any damage, but you aren't suppose to do that. Maybe the tester uses a pull down resistor if it doesn't use a reall ttl, who knows.
July 14, 2006 1:28:05 PM

The very same green wire is whats shorted on the motherboard by your START or your ON button.
All the switch does is short the green to ground.
You can leave off the switch connector & short the 2 pins on the mother board to power up the pc as well which is how many people power up boards outside a pc case.
July 14, 2006 2:03:31 PM

Quote:
You need to get yourself a multilmeter and measure the current (measured in amps) on the wires to ensure they are within the sticker specs. Black wires on the power supply are ground. Red are +5V DC and yellow are +12V DC. As for the other colors.. you'll need to look that up or go by the diagram on the link below.

Make sure the multimeter is set to measure current (I) and set to DC power (direct current). The black lead connects to ground and the red lead connects to a voltage rail (red or yellow wire)


For Heaven's sake don't do this! At best it will just overload the PSU and cause it to turn off, at worst something will go on fire.

When measuring current, multimeters are designed to be connected in series, and therefore to present as low a resistance as possible, in order to be transparent to the circuit they're measuring (and conversely, when measuring voltage, they provide a very high resistance as they're meant to be connected in parallel). So if you set a multimeter to measure current and connect it across an output rail of a PSU, you're basically just shorting the PSU.

The other posters are right: find a PSU tester and use that. It's still not as good a test as plugging the PSU into the system it's intended for, though - a PSU might run fine with a tester but struggle with a system which has odd power usage, such as a server with 12 hard drives which all try to draw 3A on spinup. Although this is obviously a pretty rare example.

If buying a new PSU, I try to get assurance that it's OK by looking at reviews and finding a reviewer who has connected the PSU to a beefy system or tester and checked that it really does work as it's meant to.
July 14, 2006 2:55:24 PM

Quote:
The very same green wire is whats shorted on the motherboard by your START or your ON button.
All the switch does is short the green to ground.
You can leave off the switch connector & short the 2 pins on the mother board to power up the pc as well which is how many people power up boards outside a pc case.


Wrong. The on button is a momentary switch. The green wire is held low by a ttl level from the motherboard. That is one reason why the PS must supply power to the MB all the times, so the MB can generate the on signal when the button is pushed. The other reason is of course so the MB can power up from LAN, etc.

ADDED: Page 19, 3.3.2 PS_ON#
July 14, 2006 6:56:28 PM

I have tested PS with all kinds of testers and still ended up changing the PS to get the system to run. I find that also is the best test the MOBO and all the hardware that surround it. Bruce
July 14, 2006 7:28:42 PM

Quote:
Can someone give me some info on how to test a Power Supply to see if it is good before I try to install it. Thanks


You can jump the green and any black wire to see if it tuns on.
You can use a multi-meter to test volts.
You can use a cheap "tester" that will do the same as both methods above...but none of these will show if the power holds up under load or if the power is "clean".

To do that you will need a loading unit,meters and a scope.

You could use what a real psu company does:

http://www.pcpowercooling.com/technology/Chroma-8000-AT...

That testing unit runs all the tests and while the psu is in a 50C state rather than the 22-25C all other psu makers use to inflate the real "clean" output of thier products.
July 14, 2006 7:59:19 PM

Quote:
You could use what a real psu company does:

http://www.pcpowercooling.com/technology/Chroma-8000-AT...

That testing unit runs all the tests and while the psu is in a 50C state rather than the 22-25C all other psu makers use to inflate the real "clean" output of thier products.


I've been scanning this whole thread to see if someone would give this very answer. We have a winner! Wize reply, ZODude!
July 14, 2006 8:47:17 PM

a power supply normally is good no matter what the watts are. But if you want a really good one (dont know why you are wanting to test the PSU, the really only way to test it is to test it under a load). What i would recoomend is looking at what the specs are for the power supply and there are power supplies that have dual rails. Those are the better ones, they distrubute voltage and power to devices better becuase instead of using one rail for all your devices, you have two. Testing it for current and the volts is a good idea, only when you suspect your PSU unit is going bad. Otherwise, why fool with that?...
July 14, 2006 9:03:40 PM

Dual rails aren't actually better, and in fact they probably don't really exist, but that's a whole other flamefest, er, I mean topic of discussion. :wink: In a nut shell you have 3 options. Single rail. Single rail with dual current limiting, or actually dual rail. The middle one is what the vast majority of dual rail supplies really are. The last one is virtually non existant and would be hugely expensive, and the first one would probably be the all out best choice if you could find one that would supply the current required.

ADDED: There are also tri or quad rail supplies, but again, that's another discussion.
July 14, 2006 9:06:15 PM

Quote:
a power supply normally is good no matter what the watts are.


That comment is about as useless as tits on a bull.

Quote:
But if you want a really good one (dont know why you are wanting to test the PSU, the really only way to test it is to test it under a load). What i would recoomend is looking at what the specs are for the power supply and there are power supplies that have dual rails. Those are the better ones, they distrubute voltage and power to devices better becuase instead of using one rail for all your devices, you have two. Testing it for current and the volts is a good idea, only when you suspect your PSU unit is going bad. Otherwise, why fool with that?...


Maybe go back and read a bit and figure out what this thread is really about?
July 15, 2006 4:24:05 AM

& you dont see your contradiction in your statement?
Think about it for a bit.
July 15, 2006 12:31:39 PM

Instead of being so coy, why don't you just come out and tell me. I'm not so bright so you're going to have to hold my hand and lead me to the right answer. :roll:
July 15, 2006 3:34:25 PM

The mb can provide the trigger for the PSU start via any method.
All thats required is a logic low signal be it a "short circuit to ground" via the momentary switch or a logic low from the MB which is activated via the same momentary switch or some other wake up signal from any feature card or some other embeded/onboard feature.

So the PSU does NOT provide any start signal,,, it is dependent on one.
July 15, 2006 3:41:48 PM

Did I say the PSU supplied any start signal?

Quote:

Wrong. The on button is a momentary switch. The green wire is held low by a ttl level from the motherboard. That is one reason why the PS must supply power to the MB all the times, so the MB can generate the on signal when the button is pushed. The other reason is of course so the MB can power up from LAN, etc.



I said, the PS supplies some power to the MB all the time (unlike the old AT supplies) and the motherboard sends a TTL signal to the PS to turn on fully, which is initiated by the momentary push button switch. The PS is not suppose to be turned on with a short, and as I said prior, I don't see how that would cause any problems, but one should probaby refrain from doing it long term.

You said this:

Quote:
The PSU tester does the same thing, so whats your point?
I bet you dont overclock either?



And in reply, I said this:


Quote:
It's possible the tester uses a ttl signal to turn it on, which is how it's suppose to be turned on, not shorted. I don't necessarily see how shorting it to ground will do any damage, but you aren't suppose to do that. Maybe the tester uses a pull down resistor if it doesn't use a reall ttl, who knows.


What is contradictory?
!