Power supply problems often masquerade as device or software problems. In this Daily Drill Down, Rodrigo Zamora gives you a troubleshooting process to isolate these power-related problems.
Isolating PC power supply problems can be either very straightforward or extremely difficult, depending largely on the type of problem, the symptoms that you observe, and when the symptoms occur. Although a completely dead PC is almost always due to a faulty supply, other types of problems are not nearly as obvious. In fact, there are some power supply problems that can be easily mistaken for a problem with another component or with the configuration of the system. Since every device or component attached to your PC depends either directly or indirectly on the power from the system’s supply, developing a methodology for isolating power problems can be quite useful.
Why not try to isolate the problem by immediately replacing the supply? Simply replacing a power supply is common practice when a supply problem is suspected. Although this can be a very effective troubleshooting method (especially for dead systems), there may be situations where doing so is just not an immediate option. For example, if you are working in the field or a system uses a proprietary supply, you may not have a proper replacement readily available. In some situations, it may be ineffective and inconvenient to replace a supply as a first option. Even if you do have a replacement available, you may find that the power supply is not the source of the problem. So to save time, you should do some simple troubleshooting first.
ATX power supplies
Although the general principles discussed in this article can be adapted for AT types of PC power supplies, I will primarily discuss the ATX supplies since they are currently the most common. For basic information on power supplies, see my Daily Drill Down, “Choosing the right PC Power supply.” Unlike AT supplies, ATX supplies are connected to a PC’s motherboard at a main 20-pin connector. The newer ATX 12V supplies have an additional 4-pin +12V connector, and the ATX standard also allows for an auxiliary and optional connector. For this article, however, I am only concerned with the main 20-pin connector whose pin-out is shown in Figure A. The color at each pin represents the color of the wire on the supply cable that will be connected to the pin. Most, but not all, ATX supplies use this color scheme. Notice that ATX supplies use five different voltage rails of which the +3.3V, +5V, and +12V are the most important. The GND pins are at common ground potential that on PCs, like on most other electronic devices, is also the potential of the system chassis.
|This shows the ATX main 20-pin connector and color scheme. The key keeps the supply cable from being connected backwards.|
When you plug in an ATX supply—and the rocker switch on the supply, if there is one, is on—there will always be a nominal +5V at the 5VSB pin (pin 9) regardless of whether the computer is turned on or not. This standby voltage is used to power the push-button circuitry that actually turns on the machine. It is also used in conjunction with the PS_ON pin (pin 14) to allow software such as the operating system to control the power to the system. Normally, the PS_ON pin will be at a nominal +5V when the system is off. When you press the case switch to power on the system, the voltage at the PS_ON pin will drop to ground potential (0V), and the ATX supply will be powered on. A short time (a few hundred milliseconds) later, the power supply will send a Power OK signal to the motherboard via the POWER_OK pin (pin 8), and the system will start its boot procedure.
Proprietary power supplies may not use the standard 20-pin ATX connector. However, these power supplies will still provide the ATX voltages and signals albeit in a different configuration. You should still be able to troubleshoot these supplies in a similar fashion as standard ATX supplies.
The following are some basic and relatively inexpensive tools for troubleshooting power supplies that you are likely to have on hand.
The multimeter is the single most important troubleshooting tool for an electronics technician. Multimeters come in two flavors: analog and digital (DMM). Both types serve the same purpose, but I will concentrate on DMMs since they are easier to read and use. Figure B shows a DMM.
|Here, a DMM measures an AC line voltage.|
Although I won’t go into detail on how to use one, you can find a good tutorial on the subject here or in any good basic electronics book. An important consideration when using a DMM to test PC power supplies is to set the meter for the type of voltage you will be reading (AC or DC), and if your meter is not auto ranging, set it to a range higher than the voltage you will be reading. All modern PCs use low DC voltages inside the case so set your meter to DC when reading these voltages. The exception to this rule is when measuring voltage inside the actual power supply housing, which contains high voltages and should never be opened by someone not qualified to do so. Line voltage (i.e., voltage from a wall outlet) will be about 115 VAC RMS (U.S.) or 230 VAC RMS so remember to set your meter to AC and use caution when measuring it.
ATX power supply tester
This very simple device (Figure C) consists of two power resistors connected to +5V and GND through an ATX 20-pin connector. An LED is provided to show POWER_OK, and a jumper wire drives PS_ON low so that the supply can turn on.
|An ATX power supply tester is a simple device.|
This tester provides an inexpensive and quick good/no-good test that does not require any devices to be attached (not even the motherboard), as shown in Figure D. Notice that the power supply is not connected to any devices other than the tester. Although hard to notice in this picture, the LED is lit indicating a good supply.
|This image shows testing a power supply with an ATX power supply tester.|
This simple test allows you to quickly determine if the source of a power problem is the power supply or another PC component. To use an ATX power tester, you first unplug the power cord at the PC end. Then, you disconnect all devices from the power supply including drive bay devices, motherboard, and any fans that are directly attached to a power supply connector. Next, attach the power supply’s main connector to the tester’s 20-pin connector. Reconnect the power cord to your PC and check to see if the tester’s LED lights up and that the fan is working. If the LED does not light up, you can be quite certain that your supply is bad. If it does light up, you can be reasonably sure that it’s good. However, as Figure E shows, these testers are not extremely thorough so use them as a general test only. Remember to unplug the cord before removing the tester.
|ATX Power Testers are not necessarily thorough.|
The same power supply tested in the image above has had its +12V line (yellow wire) cut. The ATX tester is still indicating a good supply because it only checks for Power OK by loading the +5V rail. Use the ATX tester as a general test only.
BIOS setup and sensor software
Most motherboards manufactured after 1997 come with onboard sensor circuitry that measures system temperatures and voltages. This can be helpful when a supply problem is suspected. Most of these boards will provide the readings in a section within the BIOS setup, but they can also be accessed through software usually provided by the chipset or motherboard manufacturer. An excellent third-party sensor program for Windows is Motherboard Monitor. It’s shown in Figure F. For Linux, you can find support for many motherboards here.
Isolating the supply in a dead system
A completely dead PC is one where no activity or power can be detected when you try to turn it on. Its symptoms are very easy to spot. There will be no lit LEDs, no spinning fans, no video, and no beep codes. Faulty supplies are the main cause of completely dead systems.
Many times, a power supply will not fail immediately but will instead fail gradually. Before complete supply failure actually occurs, the system may lose power during operation and then power up again only after an extended period. Eventually, you may be able to turn on the power supply (i.e., the fans spin and the power LED lights up), but there will be nothing on the display and no BIOS boot activity. You can treat this situation in the same manner as a dead system problem and troubleshoot it accordingly.
Another important cause of completely dead systems is using an ATX power supply that does not provide enough current on its +5VSB rail to power the Wake On LAN (WOL) feature of newer motherboards. This usually involves older ATX supplies that are not compliant with the newer ATX standards. Removing the cable that connects the NIC to the WOL header on the motherboard will resolve the issue but will leave the system incapable of using the WOL feature.
You could use the following steps to quickly isolate a power supply as the source of a dead system’s problem.
Check the obvious
Start by looking for problems that may be easily overlooked. This can save you from wasting precious time and also helps to keep your professional pride intact. Make sure that you are getting power at the receptacle by checking to see if other devices plugged in to the same outlet have power. You could use a DMM to check the actual AC voltage at the receptacle and make sure that it is at or close to nominal. Determine if the cord has continuity either by using a meter, swapping the cord with a known good one, or plugging it in to the outlet and measuring the AC output at the connector end (as shown in Figure B). After establishing that the cord is okay, make sure it is firmly connected to both the receptacle and the supply. Check that the 115/230 VAC selector switch (if the supply uses one) on the back of the supply is set to the proper voltage. Checking the selector switch is especially important if you have just installed a supply or built the system. Try powering the system to see if the problem has been resolved.
Disconnect external peripherals
To isolate the system from all external peripherals, disconnect them and leave only the power cord connected to the system case. If you can then turn on the PC, the problem is due to an external peripheral. Reconnect them one at a time to determine which peripheral is at fault. Remember to turn off the system before reinstalling each device.
Check internal power
Remove the cord from the supply and open the case. Make sure that the supply’s connectors to the motherboard and those being used with devices are firmly inserted.
If you have an ATX tester, you can save time by using it with the method discussed earlier to immediately isolate the supply from the rest of the system. If the tester indicates that the supply is bad, you should replace the supply. Otherwise, the problem may be related to a failed internal device.
If you are using a multimeter instead, set it to measure DC voltage and connect the black probe to any convenient place on the chassis. Reconnect the power cord. Before proceeding, determine whether or not the ATX supply is actually turning on when the system power switch is pressed. Do this by first placing the red probe on the PS_ON pin (14) that usually has a green wire attached. This should give you a reading of about +5V on your meter. After pressing the power button, the signal should swing close to 0V (a few millivolts), indicating that the supply has been switched on. Next, determine whether or not the supply is providing stable power to the system. To do this, move the red probe to the Power OK or Power Good pin (8) that usually has a gray wire connected. Your meter should show approximately 5V, but a reading between 3 to 6 volts is acceptable. If this value is close to 0V or out of range, your supply is probably bad. Replacing the supply at this point will often fix the problem. However, there are situations in which a bad internal device can result in the supply not producing the proper output voltages. To be thorough, you should isolate the supply from internal devices.
Disconnect internal devices
Since PC power supplies must have loads connected to operate and provide proper regulation, you should not operate them when completely disconnected. Test one internal device at a time by using the following technique:
- Disconnect a device from the supply.
- Try to power the system.
- If the PC does not turn on, reconnect the device and test the next device.
Keep in mind that the 20-pin connector from the supply must remain connected to the motherboard in order for the supply to turn on and provide a POWER_OK signal. If the system does not turn on after all internal devices directly connected to the supply have been checked, you can then unplug the system, remove all adapter cards from the motherboard, reconnect the cord, and test for power. If the PC still does not turn on, the most practical option left is replacing the supply.
Lockup and stability problems
If the system you are troubleshooting is suffering from random lockups and reboots, it could be that the power supply is to blame. Unfortunately, there are many other culprits that can cause the same type of symptoms. Lockups and sporadic reboots may also be related to software, hardware, power supply problems, external power problems, system configuration, etc. One of the difficulties in determining that a power supply is the source of lockup and stability problems is knowing to suspect the supply in the first place.
Although it is hard to establish a troubleshooting methodology for isolating a supply as the cause of lockup and stability problems, there are a few symptoms and situations that may lead you to suspect the supply:
- Problems began after upgrading a system with a power-hungry device.
- You’ve experienced sporadic memory errors that do not occur at the same memory location.
- Lockups and/or reboots are not related to any single program and do not occur at any specific moment.
- You have a power supply that feels very hot when operating.
- You experience poor hard-disk performance, even after reformatting the drive.
Some power supply problems that may cause boot, lockup, and stability problems include:
- Overloaded power supplies—loads that exceed either the maximum output of the supply or of a particular voltage rail.
- Dirty power or poor regulation.
- Malfunctioning supply fans.
- Malfunctioning internal supply components.
- Supplies with poor Power Good timing.
You can begin isolating a power supply as a source of lockup and stability problems by examining the most common causes of these problems. Use a DMM, BIOS sensor info, or a program like Motherboard Monitor to determine how close output voltages are to nominal (Figure F). Voltages that deviate greatly from nominal (especially at lower voltages) can indicate supply problems.
|Motherboard Monitor can display a variety of motherboard sensor data and can be set to trigger an alarm or send notification if a value is not within the limits you specify.|
Notice that in the image above, all the voltages shown are within one percent of their nominal values. This indicates that the supply is providing excellent regulation and should be working well.
Determine if your power supply’s output power specifications are adequate to support all of your system’s components. If your system tends to lock up during bootup, you may be overloading the +12V rail.
To continue isolating the supply, try to eliminate as many other sources of lockups and stability problems as possible. Some of the most common include:
- Bad or incorrectly configured RAM.
- Heat-related problems. (Check your fans and airflow.)
- Hardware conflicts. (Even on PCI, some devices do not like to share IRQs.)
- Software conflicts. (Keep your firmware and drivers up to date.)
- Corrupt system or program files.
- Noisy and dirty outlet power. (Make sure the power received by the PC is clean, properly grounded, and contains no ground loops.)
These are the most common causes of system lockups and reboots but are by no means the only possibilities. After you have exhausted your options, try replacing your supply with a high-quality unit. Remember that cheap power supplies that operate very closely to or at their specified maximum ratings are the cause of many problems.
Replacing a suspect power supply can be a quick and effective solution. However, it can also be a hit or miss proposition that can result in wasted time. To ensure thoroughness and as a requirement for some types of problems, you should try to isolate a power supply as the source of a problem. This is not always easy and can require a considerable amount of patience before the actual problem is found, but by recognizing symptoms, analyzing operating information, and understanding system requirements, you will be able to establish system stability.
The authors and editors have taken care in preparation of the content contained herein but make no expressed or implied warranty of any kind and assume no responsibility for errors or omissions. No liability is assumed for any damages. Always have a verified backup before making any changes.