Sign in with
Sign up | Sign in
Soon At Tom's Hardware: Full-Scale Power Supply Testing
By , Patrick Afschar,
1. Test Criteria

Often overlooked, the power supply unit (PSU) plays a crucial role in supplying the PC with stable and sufficient power. Whether you only need a 300W PSU for a modest small form factor box or a 1,200W model for a gaming juggernaut, the quality of your power supply will not only affect the system's performance, but also its longevity.

Bad power can mean early failures...and there are plenty of bad PSUs out there waiting for the unwary. This is why we felt it necessary to start testing PSUs in order for our readers to make better component decisions.

Test equipment plays an important role when evaluating power supplies though, which is why it's such a difficult job to do properly. Only with the help of highly precise measurement instruments can we document the differences between comparable models in an objective and reproducible way. Tom's Hardware performs all of these tests in professional labs, just like the ones found in PSU manufacturing facilities, where we get access to equipment from Chroma. For those new to this space, Chroma is the de facto vendor of PSU testing equipment.

When testing, we measure the following details:

  • Standby power consumption at 230V
  • Inrush current at 115V and 230V
  • Energy efficiency at 20, 50, and 100% of the stated maximum sustained load at 115V and 230V (according to the 80 PLUS consortium's certification)
  • Energy efficiency at 25W, 50W, 85W, 300W, and 500W of power draw at 115V and 230V
  • Hold Up Time after losing power input at 115V and 230V
  • 12V rail overload test at 110% of the specified maximum sustained load
  • Short circuit test at 115V and 230V
  • Measurement of the temperature difference between intake and exhaust air (maximum delta) at 115V and 230V


In addition to the above mentioned values, Tom's Hardware also takes into account the ergonomics of the PSU and its accessories. This includes features like the the length and design of the cables, cable modularity, and any useful accessories. All important technical data will be presented in a table.

2. Test Method: High-Precision Measurement Instruments

The reliability of PSU testing depends on the test setup and the measurement instruments used. Since the significant differences can be small, you need high-precision equipment for them to be documented in a way that's both clear and reproducible.

The cornerstone of all of these measurements is a defined and stable input voltage. Using power straight from the wall socket is out of the question, since this power source is subject to several confounding factors that might lead to significant fluctuations. Instead, we use the high-precision Chroma AC Source 6530, ensuring that all tests are provided with accurate AC voltages of 115V and 230V at a perfect sine curve of 50 or 60 Hz.

We use the precision power meter Yokogawa WT210 for measuring the voltage and current fed by the Chroma. The device attaches directly to the power source and measures voltage, current, power, power factor, and frequency with an accuracy of about 99.9 percent. With a maximum input limit of 26A, it's ideally suited for monitoring the power consumption of high-performance systems.

Clearly-defined DC loads have to be simulated on the different individual rails during all of the measurements. Since this also requires precision instruments, we use a total of eight Chroma DC electronic loads from the 6330-series.

Four of these are capable of generating a maximum load of 600W each to be used as loads for the 12V rails. The remaining four Chroma DC loads (300W maximum each) are responsible for load generation on the 3.3V, 5V, 5V sb, and -12V rails. Three dedicated control units are used for the basic configuration and to recall programmed scenarios. All units come with a feature for generating a controlled short circuit.

We also use a modern oscilloscope, the Tektronix Scope DPO3034, for graphic processing of the measurement results. The four-channel, 300 MHz oscilloscope provides us with readings of voltage, current, frequency, ripple, and rise/fall time. Thanks to its integrated Plug'n'Play PC interface, LAN, and USB connections for external storage, the DPO3034 can be comfortably integrated with the rest of our IT infrastructure.

With the press of a button, a screenshot gets saved as an image file directly to the external storage device connected to the oscilloscope. Not least of all, our Tektronix TCP150 measurement probe is able to handle currents of up to 150A and is an important addition to the oscilloscope for monitoring current flow.

3. Power Consumption In Standby Mode, Inrush Current, Energy Efficiency

Power Consumption in Standby Mode

As long as the power supply's main switch isn't turned off or the computer isn’t physically disconnected from the power grid, the PSU will consume some power even when shut down. This is called standby power and is required to turn on the PC.

Some computers even provide sufficient standby power to charge a cell phone through a USB port. A constant load on the 5V standby rail (5Vsb) has to be defined and used in order to compare the standby power of different power supplies. We chose 0.25A, which equals a power consumption of 1.25W. The test result shows what the PSU actually consumes in order to provide the required 1.25W. The higher the number, the lower the quality of the power supply in this respect.

Inrush Current

Turning on a computer creates a very strong temporary peak in its input current due to the charging of capacitors and so on. If the inrush current is too high, this could negatively impact other devices connected to the same power strip or even the local power grid if they are really sensitive.

Tom's Hardware tests each PSU to make sure its inrush current stays within the ATX specification of 50A at 230V and 100A at 115V. Our 150A Tektronix TCP150 probe enables us to gather helpful screenshots.

Energy Efficiency According to the 80 PLUS Specification

Next to its power output, the most important characteristic of a PSU is efficiency. It's no surprise, then, that almost every PSU manufacturer out there today advertises the compliance of its products with the 80 PLUS certification. A PSU must reach a certain efficiency under given load percentages in order to use this label.

The different levels (standard, Bronze, Silver, and Gold) symbolize what efficiency level for which the PSU is suitable. The next certification, Platinum, is already in preparation. 80 PLUS has its main focus set on 50% loads, which are most common in everyday use. Tom's Hardware tests whether PSUs are labeled correctly or if certified models comply with the requirements of the 80 PLUS initiative. Since the certification enjoys worldwide recognition, all tests are carried out at voltages of 115V, as well as 230V.

The table shows the criteria the individual 80 PLUS levels are based on.

Energy Efficiency20% Load
50% Load
100% Load
80 PLUS 80%80%80%
80 PLUS Bronze 82%85%82%
80 PLUS Silver 85%88%85%
80 PLUS Gold 87%90%87%
80 PLUS Platinum 90%94%91%


Energy Efficiency Under a Range of Defined Loads

In addition to the efficiency testing done to check the 80 PLUS compliance, we also test the efficiency of power supplies under a number of defined loads. This allows the user to get a picture of PSU efficiency under a range of typical loads. To depict the most realistic scenarios possible, the following loads are used:

  • 25W: Compact all-in-one Atom-based PC without a dedicated graphics card
  • 50W: Office PC without a dedicated graphics card
  • 85W: Multimedia PC with a dedicated graphics card
  • 300W: Gaming PC with a powerful CPU and a modern graphics card
  • 500W: Gaming PC with a high-end CPU and multiple graphics cards
4. Hold Up Time, Overload And Short Circuit Test, Temperature

Hold Up Time

Although modern power grids are stable and of fairly high quality, there are always some power fluctuations. Modern power supplies have capacitor banks to handle such power fluctuations for a certain period, called the hold up time. The ATX standard for hold up time is a minimum of 16 milliseconds.

Tom's Hardware measures this by permanently monitoring the PSU's output voltages. After cutting off the power supply, the oscilloscope records the voltage curves in order to accurately measure how much time passes before the voltages drop. These measurements are carried out both at 115V and 230V.

Overload Test

A PSU's power rating normally describes the maximum amount of power it can supply over an extended period of time without being damaged. Here, the total is the sum of the individual 3.3V, 5V, 12V, -12V, and 5V standby rails combined. To check whether a PSU operating at its maximum specified load is truly at its limit or whether the manufacturer has left the PSU with some headroom, we perform an overload test. The load on the 12V rails are set to 110% of its specified maximum. If the PSU can run the test for five minutes without dropping voltage below the ATX specification's 11.4V (a maximum 5% deviation), the test passes.

Short Circuit Test

A critical short circuit doesn't happen very often, but it's not inconceivable, and the result is often a dead PSU. Today, however, many PSUs have circuits to protect them from short circuit damage. To test the protection circuit, Tom's Hardware manually triggers a short circuit on a 12V rail. The power supply should turn off immediately so it doesn't overheat and becomes permanently damaged. If the PSU can be used normally again after deactivating the short circuit, the test passes.

Temperature

Like any electrical appliance, a PSU operates at a certain efficiency. The energy not used for powering the computer and its peripherals is released as heat into the environment. Tom's Hardware continuously monitors the temperatures of the inflowing and exiting air to track the warming effect, and the biggest difference measured between these two values is reported as the test result.

Accessories and Ergonomics

In addition to the results of the numerous performance tests mentioned above, the PSU's accessories and ergonomics are also taken into account in our final results. The testers turn their attention to details, such as included mounting solutions, housing construction quality, cable modularity, and cable lengths. Anything else making the user's life harder or easier gets mentioned as well.