A [H]ard Look at Power Supplies

HardOCP’s [H] Enthusiast finally spins up to correctly cover computer power supplies. We show you how we will be doing it, what equipment we will be using and the methods to our madness.


Our Review Setup

The most important facet for testing a power supply is to be able to isolate the power supply and our results from environmental factors as much as possible. This allows us to be able to both replicate our testing on each unit and be sure there are no confounding factors being introduced by variability in the test environment.

To that end, we have developed and constructed the following test setup outlined below. The block diagram outlines exactly how the flow of the equipment works and below we will give you an explanation of what is taking place in the testing cycle.

Article Image

Power In

In our power supply reviews everything begins with one of the most important factors to control, the incoming power. In our reviews we will be using a Tripp-Lite LCR2400 Line Conditioner to isolate our test setup from outside electrical fluctuations as much as possible. The Tripp-Lite LCR2400 provides Automatic Voltage Regulation (AVR), AC surge suppression, and EMI/RFI filtering. The LCR2400 has a rated AC surge suppression of 1440 joules and EMI/RFI AC noise suppression rating of 75 dB. The AVR settings on the LCR2400 allow for nominal operational output of 120v at an input voltage range between 89volts to 147 by trimming output voltage by 9.6% when the voltage is between 128v and 147v, boosting output voltage by 8.9% when the input voltage is between 103 and 110v, and boosting output voltage by input voltages by 18.8% when between 89v and 103v. Following the Tripp-Lite LCR2400 in our test setup there are two distinct paths to follow.

The SM-8800 Load Tester

Path one first takes us to the SunMoon SM-8800 ATE. Here we have our programmable load tester which allows us to put a set load on each of its ten available fields. For each testing cycle we will preprogram a battery of loads which will be applied directly to our PSU of interest through the panel on the lower left, unlike some other test equipment the SM-8800 is specifically designed to interface with ATX PSUs. Our readers will notice in various reviews that not all ten of the available fields will be used with all PSUs as 12v rail counts have gone from one to two to multiple and now there seems to be some shifting back to one. In general, our reviews will report the amperage load and voltage output on the 3.3v, 5v, +5vsb, and however many 12v’s values the PSU has. We will not be testing or reporting the -5v as it has been dropped from the ATX specification as of ATX12v2.0 revision, however if we do encounter a PSU that still has the -5v we will report it’s inclusion for those people who still need that output. Also at this stage we are able to connect our USB Instruments DSM1M12 Digital Oscilloscope to the SM-8800’s waveform output. The USB Instruments DSM1M12 Digital Oscilloscope provides all the functionality of older CRT/LCD oscilloscopes in a newer much smaller and cheaper PC-based Oscilloscope. The DSM1M12 has its own analogue-to-digital converter that directly interfaces via a USB connection with a PC, in our case an IBM xSeries x330, in order to read DC output with the EasyScope II software bundle. Here we will be logging the DC ripple, the superimposed AC waveform on the DC output measured as voltage change over time, of each power supply tested.

Also at the SunMoon SM-8800 load tester we have the Nautilus Integration PSTT-1 Transient Load Tester. The PSTT-1 is a stand alone unit that is responsible for selectively and transiently loading power supplies with an addition of 9.25A on the 12v rail and an additional 3.75A on the 5v rail when triggered. The transient load duration on the unit is selectable between 10ms, 100ms, and 1s. The unit also has a set of BNC connectors which will pass the transient load profile off to the USB Instruments DS1M12 Digital Oscilloscope which is connected to our IBM x330 for data logging. The Transient Load Test is meant to simulate what occurs in a system during load changes as the typical loading of a system is not perfectly static as a load tester at a set load DC load is but rather is constantly changing as a user changes their usage pattern. For instance if a user is starting up a PC that has a large RAID array there is a heavy load initially during spin up that declines after spin up, or when a user goes from idling to loading up their favorite 3D intensive application or vice versa. These are the types of events are a more normal usage pattern for a modern PC than static loads and are what we are attempting to model with this equipment.

Variable Voltage Input & Extech Power Analyzer

Path two takes us from the LCR2400 to the Staco 3PN2210B Variable AC Transformer (Variac). Here we are able to specify the input voltage that our PSU is going to see anywhere from 0v to 130v. Since we have isolated ourselves from the electrical grid’s voltage fluctuations and are running at a 120v nominal input to the Variac, the Variac allows us to simulate various PSU input voltage scenarios over the realistic operating range of a PSU in the US (we will not be doing 220v input as most PSU are moving to full range input) in a controlled and reproducible fashion. Immediately following the Variac is the Extech Power Analyzer, which is used to read our power draw on the AC side to give us half of the measurements we need in order to determine efficiency and Power Factor (PF). Efficiency is calculated by the input wattage read at the Extech Power Analyzer divided by the output wattage of the PSU and is recorded across all of our tests. Power factor is the ratio of real power (kW) to apparent power (kVA). For our readers, the main concern with power factor is proper utilization and loading of a UPS, which is specified in kVA.

The Incubator & Temperature

As our paths converge once more, we are brought to the test chamber. Each power supply being tested is housed in our Quincy Labs 10-140 incubator. Here we are using a slightly modified incubator in our testing to allow us to address one of our primary concerns in power supply testing, temperature control. Unlike using a case the incubator gives us finer temperature control and allows us to adjust our temperature test parameters in a reliable and reproducible manner from test to test. The incubator has a set temperature range of 15c to 62c (with a minimum temperature of 1c above ambient temperature) and will hold that temperature +/- 1c indefinitely. The incubator has been modified to vent the hot air being expelled by the PSU out the back of the incubator so as to not endlessly build heat inside the incubator. To monitor these parameters we also have our Sperry DT-506 4 Channel Digital Thermometer. The Sperry DT-506 uses type K thermocouples to log temperatures between -50c and 1000c at +/-(0.3% +1c). Having a 4 channel digital thermometer allows us to monitor both intake temperatures and exhaust temperatures on the PSU and up to two other temperature points simultaneously.