Testing PSUs

  • Thread starter Thread starter Greg Lorriman
  • Start date Start date
G

Greg Lorriman

Hello,

I've been aware for a few years that dodgy PSUs are a significant
cause of hardware problems, and further particularly poisonous
problems as they can be intermittent or look like something else
(video/HD etc). I really would like to be able to test PSUs as
although I'm not a pro I am constantly fixing people's machines.

I've trawled around the web and googled the groups and I can't find an
easy reasonably cheap way in which a non-pro and non-electronics bod
can reasonably test a PSU for unstable voltages and other problems. (I
was originally hoping that a cheapo $5 multimeter might be do the
job).

What I am wondering is whether it is the case that 90% of PSU problems
could be diagnosed with a carefully thought out strategy with
inexpensive equipment and non-pros in mind.

Here's a naive illustration of what I mean :

* Use easily-constructed gadget to apply maximum loads (perhaps a few
knobs might be needed for differently rated PSUs).

* test lines (forgive the terminology abuse) with cheapo multimeter,
and a cheapo ampmeter.

*Eventually remove the PSU cover and look for leaking capacitors.

* Perhaps also, if multimeters can do it : set the multimeter to show
the max/min voltages reached to test for unstable voltages without
constant (boring) vigilance (I'm rather thinking of those max/min
thermometers I've seen; don't know if multimeters can do this).

I would love to be the guy to write up a web page dedicated to non-pro
PSU testing, but alas I'm just a humble programmer bod. What I have
seen sure ain't aimed at fiddlers like me. And yet I am salivating at
the prospect as I can clearly see the value of being able to test
PSUs.

The idea of buying an oscilloscope and trying to follow some of the
advice I've seen on the web and in these groups sends shivers up my
spine, but so far that seems to be all there is.

Is this idea realistic? And does anyone have any ideas along these
lines?

thanks,

Greg
 
The cheapo $5 meter is not going to do it. You must splurge
'big time' for the $20 3.5 digit multimeter. No way around
that major expenditure. Power supply must be fully loaded by
computer when testing. Computer with full load (multitasking
so that many peripherals are being used simultaneously) then
has each DC voltage measured with that multimeter. Voltage
must remain in upper 3/4 of limits per chart, provided with
procedure (called '3.5 digit meter and computer load'):
http://www.tinyurl.com\2musa

Alternative is to purchase a large DC load, adjustable so
that every DC voltage is 100% loaded, then monitor each
voltage with an oscilloscope and meter (looking at both ripple
voltage and DC steady state voltage), and then verify power
supply still maintains these output voltages when a variac
lowers AC voltage from 120 VAC to 90 VAC. Easy as hell.
Getting equipment is the complex part.

Power supplies must work even at those extremes
simultaneously. Yet the 3.5 digit multimeter connected to a
fully working computer is a second best alternative.

Still none of this will verify that other essential
functions are installed including EMI/RFI filtering,
overvoltage, and overpower protection. Short out all DC
voltages and start up supply. Power supply must also pass
that test without failure. Intel defines a minimum size wire
for shorting.

Some will recommend a plug-in tester that does not load
supply as effectively as above '3.5 digit meter and computer
load' test. Plug-in tester does even less than a 3.5 digit
multimeter - for almost same money.

This may also provide useful information:
alt.comp.hardware.pc-homebuilt entitled "Power Supply" on 5
Feb 2004 at
http://tinyurl.com/32j4g
 
* Use easily-constructed gadget to apply maximum loads (perhaps a few
knobs might be needed for differently rated PSUs).

* test lines (forgive the terminology abuse) with cheapo multimeter,
and a cheapo ampmeter.

*Eventually remove the PSU cover and look for leaking capacitors.

* Perhaps also, if multimeters can do it : set the multimeter to show
the max/min voltages reached to test for unstable voltages without
constant (boring) vigilance (I'm rather thinking of those max/min
thermometers I've seen; don't know if multimeters can do this).

I would love to be the guy to write up a web page dedicated to non-pro
PSU testing, but alas I'm just a humble programmer bod. What I have
seen sure ain't aimed at fiddlers like me. And yet I am salivating at
the prospect as I can clearly see the value of being able to test
PSUs.

The idea of buying an oscilloscope and trying to follow some of the
advice I've seen on the web and in these groups sends shivers up my
spine, but so far that seems to be all there is.

Is this idea realistic? And does anyone have any ideas along these
lines?
Click to expand...

To give a power supply a test with relative inexpensive equipment is no too
hard to do. You can go to some places like Digi-key and look for some low
value resistors . Get a few and a few switches. You may have beter luck on
e-bay finding some at a more reasonable price. YOu will need them to be in
the low ohm value , around 5 ohms and less. Also high wattage. Set up a
test stand and use the swithes to add more resistors for a bigger load.
Look up Fluke for a multimeter. They are not going to be your run of the
mill $ 10 meter. They make some that will recorde the highest and lowest
voltages. While outside of your price range they also make a solid state
recording scope of sorts. I use one at work from time to time and they are
nice to have if you can get someone else to pay for them..
 
Back
Top