Test Power Supply Unit (PSU)

Information:
The following steps are completely safe if the following steps are done exactly as shown.
warnsmall.png
Warning

Do not under any circumstances open the case on any PSU! The capacitors in it will hold enough current to kill or severely burn you for several days even after powered off and unplugged.Unplug the PSU until instructed to plug it in.



Note: This is meant to determine if a suspect PSU is bad or not and not meant to be a definitive PSU test tutorial.

Method One

Step One
1. Switch off the PSU if it has a switch and unplug it from the wall power or UPS. This must be done because even with computer powered off, the PSU is still on and has voltage present in its connectors.
2. Open the computer and locate the large power connector. It is normally on the right side of the motherboard.
3. Locate the locking tab on the connector, press it in to release and wiggle the connector and pull it out of the socket.
This is what it looks like and will have 20 or 24 pins and may be a white color.



Step Two
1. Bend a paper clip as shown in the illustration below.
2. Insert one end in the connector hole with the green wire and the other end to any one with a black wire. We have jumpered the green and a black conductor now.
3. See illustration above to locate the proper connectors.
warnsmall.png
Warning

Do not jumper ANY conductors other than GREEN and BLACK.



4. After jumpering the green and black conductors as shown above, plug in and switch on (if previously switched off) the PSU.
5. Check the PSU fan, it should be running now, if so the PSU is working. The case fans may run at this time which is normal. They may not run too which is also normal.
5. At this point if the PSU fan runs but the PC still won’t power on or has other issues, continue to method two below.

Method Two
1. Unplug the PSU. This method will use a Volt/Ohm meter (VOM) if available, to check the voltage out put. Leave the jumper connected. See illustration below.
2. Adjust the VOM to so it will read at least 20VDC, volts DC, direct current. Put the black VOM lead on any black conductor and the red on another color.
Yellow = 12VDC (11.4 – 12.6)
Red = 5VDC (4.75 – 5.25)
Orange = 3.3VDC (3.135 – 3.465)
These color conductors will have the same voltage on any of the PSU connectors.
3. Plug in the PSU and switch it on. Readings should be ± 5% (in red above) of the nominal voltages. In this picture the VOM is reading a 12VDC lead and shows 11.77 which is within specs and ok. If outside if these ranges, the unit is bad.
4. If the volts are within range but there are still problems, go to method three below.



Method Four
1. A better way to test a PSU is with a dedicated PSU tester like below. It will put a load on the PSU and test volts under load. This shows this PSU faulty due to 0VDC on the 12V 2 connector.
2. A PSU tester will automatically let you know if voltages are where they should be. Many will have an LCD display turn red for a bad PSU and emit a loud series of beeps.



To show a summary of test steps and options of methodology.



Common PSU Connectors



SATA drive power for hard drives and CD/DVD/BluRay drives.



EPS 12V which is usually plugged in near the CPU socket for CPU power.



PCIe for graphics cards. These may be 2×6, 1×6 + 1×8 and other configurations.



Common peripheral, Molex, connector for fans, lights and other case accessories. Adapters are available to use this for SATA, PCIe, and other connectors as well.



Floppy power connector, it is still on many PSUs.
You are done! This should determine whether this PSU is good or bad.
smile.gif

Reference: Power supply unit (computer – Search results – Wikipedia, the free encyclopedia

Also posted at 2xg's blog site http://miscblogscontributors.wordpress.com/2012/10/11/how-to-test-a-power-supply-psu/
 
Last edited by a moderator:
Click to expand...
Re: PSU how to test one

That is a really nicely produced tutorial!

But I do have a couple comments, if you don't mind. There is no mention of any ESD precautions. While I understand this tutorial is focused on testing a power supply outside the PC case, it is very likely users will be removing the PSU from the PC case, and/or disconnecting power connectors to test the PSU with the PSU still installed inside the PC case. In either event, if ESD precautions are not taken PRIOR to sticking their hands inside the case, the user may very well destroy an ESD sensitive device, such as the CPU or RAM module without even being aware any static discharge occurred. Therefore, I recommend some warning in addition to unplugging from the wall to touch bare metal of the case interior (or use a anti-static wrist-strap) before reaching in to put user and computer/PSU "at the same potential" thus ensuring no discharge can occur.

It it very rare for capacitors to hold on to charges as you describe in your warning because even budget supplies bleed that off in seconds when power is removed. HOWEVER, if there is damage, a faulty component, or poor assembly techniques uses - all bets are off so the important part of your message (Do not under any circumstances open the case on any PSU!) still certainly applies.

Sadly, there are many who feel since it is (relatively) safe to remove the side panel of a "running" PC and dig around and not risk electrocution, they feel they can do the same with a power supply that contains lethal voltages inside - perhaps to dig around inside the PSU with a multimeter! That is not recommended unless you are a formally trained electronics technician.

I note you say,
This is meant to determine if a suspect PSU is bad or not and not meant to be a definitive PSU test tutorial.
Click to expand...
That is great. But note those methods can confirm a PSU is bad (if a voltage is missing, for example), but not necessarily that a PSU is good. And I note you also said at the bottom,
You are done! This should determine whether this PSU is good or bad.
Click to expand...
You cannot conclusively test a PSU with a multimeter - especially using the paper clip method. To properly test any power supply (whether it be a computer PSU, battery, or car engine) the supply MUST be under a proper, realistic load - preferably tested over a wide range of "expected loads" (a requirement for 80 PLUS certification). A failing (or very cheap) PSU can measure within tolerances with no load (idle), then fail when put under a greater (but still realistic) load. So to measure a PSU properly with a multimeter, the PSU must be connected to the motherboard (realistic load) and powered up. That means jamming (with considerable force!) hardened steel, highly conductive, sharp, pointed probes into the main power connector deep in the heart of a live computer. That can take some significant physical dexterity (not to mention strategic and correct placement of your tongue) where one tiny slip can cut through several tiny circuit traces. Disciplined ESD prevention is essential! Therefore again, IMO testing with a multimeter should be left to a professional, or at least someone with this type experience.

Also, most multimeters are incapable of measuring ripple and other anomalies that "ride" on DC voltages and affect computer stability. Your +5VDC can measure +5.00VDC but if the PSU is letting too much ripple through, freezes, system crashes, reboots and other problems can occur.

The plug-in testers are good, especially those with a digital readout, IF they go out at least 1 decimal point like this PSU Tester (not sure if the Rosewill does for the +12V - I keep a PSU Tester in my truck tool bag). These testers typically have a little 10W resistor (well, 10W is big, for a resistor) for a small load, but not a realistic one. And none of those plug in testers measure ripple either.

Therefore, the only conclusive way to accurately test a PSU is with a qualified technician using an oscilloscope or a power supply analyzer - sophisticated (and expensive) electronic test equipment requiring special training to operate, and a basic knowledge of electronics theory to understand the results. Therefore, conclusively testing a power supply is done in properly equipped electronics repair facilities.

This is why swapping in a known good supply is often the best way to test a supply.

In my canned text for testing PSUs, I include a link to the ATX Form Factor PSU Design Guide and reference “Table 2. DC Output Voltage Regulation” on Page 13. That way, I don't have to worry about me typing wrong information in my canned text, like your tutorial saying the 3.3VDC high-side tolerance is only 3.165 volts! ;)

One last thing, I promise. There is nothing in the ATX PSU Design Guide, or the main ATX Form Factor Standard that "requires" a standard wire color code scheme (except for the newest standard, SATA cables). They are only recommended color coding schemes and PSU makers do not have to comply - though most do. Still, there are some, typically older PSUs from off the wall manufacturers that did not use standard color codings. Therefore, it is always best to go by, or at least verify by pin number - which is standardized (unless modified like older Dell and HP PSUs and motherboards).

Sorry if that was more than you asked for.
 
Re: PSU Test

Not buying that for a PSU test. To properly test a PSU, it needs to be under a variety of realistic loads, then measured with sophisticated test equipment such as an oscilloscope and power supply analyzer to view the output waveforms for stability, voltage tolerances and [very importantly] ripple. Other testing criteria include hot-box tests (stability in 40 - 50°C operating environments), in-rush current, and hold-up times, as well as efficiency - all things that OCCT test cannot do.

The description of that PSU test, "...to get the most out of your Power Supply", makes no sense. You want out of your power supply exactly and ONLY what the connected devices demand at any given point in time. No more, no less. For those reasons, a proper PSU test can only be done in a properly equipped testing facility by properly trained personal who can understand and interpret the results. Therefore, all "normal" users can do is make sure they have a supply with enough horsepower and if the quality/reliability of the output is in doubt, swap in a known good PSU to see if problems remain or go away.

To make sure the size is adequate, for experts and the less experienced alike, I recommend the use of a good PSU calculator and IMO, the best (by a long shot) is the eXtreme Power Supply Calculator. The developers are constantly updating their databases as new CPUs and graphics solutions hit the market ensuring this calculator remains current and accurate. And it offers users extensive flexibility for factoring in abundant combinations of RAM, fans, drives, alternative cooling solutions, and more. And then it also recommends a suitable UPS size as well.

Because this calculator is so flexible, it does not grossly overestimate recommendations as other calculators do.
 
Re: PSU Test

I tend to agree, but also think that the best way to test a PSU is to swap it out for another one. No tool can be as accurate as the good old PSU swap. This software is just a first test to see what it says and is by no means conclusive, although, I have found it pretty reliable in most cases.
And I agree with the site recommendation, have been using it for years.
 
Re: PSU Test

Digerati said:
Not buying that for a PSU test. To properly test a PSU, it needs to be under a variety of realistic loads, then measured with sophisticated test equipment such as an oscilloscope and power supply analyzer to view the output waveforms for stability, voltage tolerances and [very importantly] ripple. Other testing criteria include hot-box tests (stability in 40 - 50°C operating environments), in-rush current, and hold-up times, as well as efficiency - all things that OCCT test cannot do.

The description of that PSU test, "...to get the most out of your Power Supply", makes no sense. You want out of your power supply exactly and ONLY what the connected devices demand at any given point in time. No more, no less. For those reasons, a proper PSU test can only be done in a properly equipped testing facility by properly trained personal who can understand and interpret the results. Therefore, all "normal" users can do is make sure they have a supply with enough horsepower and if the quality/reliability of the output is in doubt, swap in a known good PSU to see if problems remain or go away.

To make sure the size is adequate, for experts and the less experienced alike, I recommend the use of a good PSU calculator and IMO, the best (by a long shot) is the eXtreme Power Supply Calculator. The developers are constantly updating their databases as new CPUs and graphics solutions hit the market ensuring this calculator remains current and accurate. And it offers users extensive flexibility for factoring in abundant combinations of RAM, fans, drives, alternative cooling solutions, and more. And then it also recommends a suitable UPS size as well.

Because this calculator is so flexible, it does not grossly overestimate recommendations as other calculators do.
Click to expand...

Wrench97 said:
Yea I've never seen any value to that test software.
Click to expand...

It is not meant to be a true test in the way you two are viewing it. It is just meant to see how the PSU handles two key components under full load: the CPU and GPU. It doesn't guarantee the PSU is fine if it passes, but it seems like it would be a good tool to test scenarios where users are seeing their system shut down when running high resource intensive tasks.
 
Re: PSU Test

It is not meant to be a true test in the way you two are viewing it. It is just meant to see how the PSU handles two key components under full load: the CPU and GPU. It doesn't guarantee the PSU is fine if it passes, but it seems like it would be a good tool to test scenarios where users are seeing their system shut down when running high resource intensive tasks.
Click to expand...
I just don't see it as a test of the PSU. There are many things that can cause a system to shutdown unexpectedly. There are no monitored sensors in power supplies or on motherboards that monitor voltages for tolerance adherence. So if there is a sudden shutdown caused by the power supply, there is no logging to point fingers at the power supply.
 
Back
Top