How to test a power supply ?

Hello,

Some have suggested to buy a voltmeter and then test the power supply.

Since computers have special plugs and such, how would one go about
testing the power supply ?

Are there special plugs necessary ?

(Not that I would ever try it, way to dangerous !)

Do you have a link to a website with some pictures ?

Bye,
Skybuck.

 

How about buying a power supply tester, they sell them pretty cheap.

 

'Skybuck' wrote:
| Some have suggested to buy a voltmeter and then test the power supply.
|
| Since computers have special plugs and such, how would one go about
| testing the power supply ?
|
| Are there special plugs necessary ?
|
| (Not that I would ever try it, way to dangerous !)
|
| Do you have a link to a website with some pictures ?
_____

All mains voltage is contained within the metal ATX power supply box. Just
follow basic safety proceedure used when working on ANY device connected to
mains voltage (i.e. don't stick screwdrivers into the supply when it is
plugged in.)

Keep in mind that mains voltage is ALWAYS present in the ATX power supply
EVEN when the system front panel switch is turned off AND the power switch
(if any) on the back of the power supply is turned off as long is the power
cord is plugged into the mains.

The System front panel switch just handles logic level voltages. However,
you must supply a POWER ON logic level TO the power supply to get it to turn
on.

Diagrams and specifications for the ATX power supply are at
http://www.formfactors.org/developer/specs/ATX12V_PSDG_2_2_public_br2.pdf .
Included are pinouts for the various power plugs.

A digital voltmeter should be used for measuring voltages.

The voltages under no load will not be representative of the voltages under
load. Voltage measurements when the system is operating are much more
useful. Thus the reason for system monitoring chips.

A good computer repair shop is recommended.

Phil Weldon

| Hello,
|
| Some have suggested to buy a voltmeter and then test the power supply.
|
| Since computers have special plugs and such, how would one go about
| testing the power supply ?
|
| Are there special plugs necessary ?
|
| (Not that I would ever try it, way to dangerous !)
|
| Do you have a link to a website with some pictures ?
|
| Bye,
| Skybuck.
|

 

Hi, stranger!
To _test_ a component to some designed specs is NOT starting by buying
an instrument(s). Maybe first learn what it does, the PSU, and then
build(?) a laboratory to tackle this job.
Then you will be surrounded by voltmeters, ampermeters, passive and
active loads and so on including plugs and wires.

Good hunting

Stanislaw

 

http://www.compusa.com/products/product_info.asp?pfp=cat3&product_code=
332184&Pn=ATX_12V_Version_2_0_Power_Supply_Tester

They have them right in the store if you have a CompUSA local to you.

 

Just buying anything is useless before you know
what you wnat to measure. Are you interested in
the voltage under load? In the temperature
behaviour under load ? In the spikes dependent
on the load ? In the bandwidth under load ?
There is tons to measure.

Rene

 

Make sure you get a voltmeter with a USB connector, in case you have
to reinstall the OS on the voltmeter


martin

 

'martin' wrote:
| Make sure you get a voltmeter with a USB connector, in case you have
| to reinstall the OS on the voltmeter
_____

Actually, you CAN get Digital Multimeters with serial and USB ports - for
logging, and, I suppose, for flash updates for the more sophisticated DMMs.

Phil Weldon

| On Fri, 08 Jun 2007 21:11:17 -0700, in sci.electronics.design Skybuck
|
| >Hello,
| >
| >Some have suggested to buy a voltmeter and then test the power supply.
| >
| Make sure you get a voltmeter with a USB connector, in case you have
| to reinstall the OS on the voltmeter
|
|
| martin

 

Yes I know, but considering all the problems the OP is having with
bios and stuff, I thought I would just add to his confusion


martin

 

They are basically useless. The only way to fully test a PSU is under load.
If has to be plugged in and the computer needs to be running. A voltmeter is
the only way to go. Just because a PSU works when its not under load ie
using a PSU tester does not mean it wont fail under load, once its gotten
good and warm.

Dan

 

You better change your name in stupid american like the rest.
Stop crossposting !
fup set

 

In alt.comp.hardware.pc-homebuilt "Angry_American"

I wouldn't say that.
I've replaced about a half-dozen power supplies by different makers over
the past three to four years; and *every single one* failed to properly
power-on the cheap $20 tester I have. Only GOOD PSUs brought it to
life.

That is true ... but only "to fully test".
A Power-Supply-Tester is a handy way to check a failed supply and
confirm it has failed. Only if it says the supply is *good* and yet the
system shows signs of power-supply failure do you need more extensive
tests. When the Power-Supply-Tester fails to power-on when hooked to a
PSU, taking voltage measurements afterwards is overkill.

Just toss the thing.

And MOST bad supplies won't power up even a cheap tester.
It's quick and dirty; but definitely NOT useless.

But if it fails the PSU tester, then why go to the bother of all the
other shit? And, in most cases of a failed supply, it WILL fail.

A HELL of a lot quicker too, than probing around with a voltmeter.
Just don't trust it to find marginal cases; or especially don't expect
it to find power supplies that were underpowered when bought; unless, of
course, the being underpowered is what caused the PSU to fail.

That's why you keep BOTH in your toolbox.

 

Unless it has a digital readout, it's not much good except for telling
if the PSU is completely dead or not. The one I tried said that my
PSU was fine even though the +12V rail was too low to let HDs spin.

A cheap digital multimeter and a bent paperclip to short the green
Power-On line to either of the black ground lines next to it are
usually good enough, but sometimes a couple of 10 ohm, 8-10 watt
resistors are needed to load down the +5V (connect each between a red
wire and a black wire).

 

If you don't open the case there's nothing that should bite you.

You need a "dummy load" to load down the PS to test it (many won't
even power up without a load of some sort)

None of the commercial units, I've seen, put enough load on a PS to
stress it (pulling 10 watts from a 300+ watt supply isn't a real
test).

There were some dummy load schematics on the web. What would be good
(don't know if anyone is selling it) would be a plug that can be put
between the PS connector and mobo and analyze and monitor the actual
voltages with some lights to tell you if the power is clean.

 

: > How about buying a power supply tester, they sell them pretty cheap.
: >
:
:
: They are basically useless. The only way to fully test a PSU is under load.
: If has to be plugged in and the computer needs to be running. A voltmeter is
: the only way to go. Just because a PSU works when its not under load ie
: using a PSU tester does not mean it wont fail under load, once its gotten
: good and warm.
:

I don't disagree but I do wish I'd had the cheapo $10-$15 ps test
thingie around before I hooked up a RAID array to a brand new Ultra
power supply that shipped from the factory with its +5 swapped with
+12. I didn't bother testing it with a voltmeter ahead of time as I'd
never seen one of these switchers come on without some sort of even
minimal load.

 

SkyTard,

In the motherboard box or at the PC store, there are little splitter
cables that split a supply line into two lines. You cut one end off one
of the splits and plug it in to have four wires available. That is only
two of the supply rails, however. The ATX connector (also needed) is a
product you would buy on digi-key or the like from AMP, but you'll also
have to buy wire and pins and a crimper.

Again, if you do not already know these SIMPLE, BASIC things, you have
no business delving into them.

Since you obviously don't even know how to attach test devices to a PC
supply, you certainly would not have the first clue about testing one
under loaded conditions, whether they (the loads) be fixed or variable,
and that is what is needed to test the supply, not just some lame, simple
voltmeter test. PC supplies also have a nasty habit of shutting down
when loading changes on certain rails, so you also have to have a good
deal of familiarity with the ATX PS Design specification as well, which I
am certain you do not.

You were way over you head when you started messing with your PC, and
you are certainly way over your head on testing the supply for it. You
likely would not even have a clue how to properly load it, much less what
you need to look at or for when it is loaded.

Do you even know what the word "ripple" means?

You cross posting Usenet retard.

 

Not a very bright response. That places you damn near at his level.

 

One cannot properly OR safely test a supply hooked up to the computer
either.

It needs load testing, but ON THE BENCH. It needs to have each rail
examined with a scope to look at the ripple voltages on each rail when
they are all under their full rated loading spec.

That's where one will see the big difference between the cheap shit and
the well made supplies.

It's just like the bad old days of the seventies when car stereo makers
had misleading and erroneous claims about their audio output power
levels. Yeah, it puts out 15 Watts per channel... at ten percent
distortion!

 

The claim that is trademarked is bullshit. What that is is public
domain, dipshit.

The ONLY proper way to test a PC supply is FULLY loaded to the spec of
that particular supply, and with a scope to examine the ripple voltage so
that one can test for compliance to the ATX PS Specification.

It may look fine, it may supply full voltage, but if the ripple is too
high, then it is really FOOL Voltage!

The ATX spec is very stringent on ripple, as that is what causes
failures in logic, and we can't have that on something we would call a
computer.