Test MethodologiesThe power supply has finally earned its rightful place among system components and with this new found popularity more review sites have embarked on testing the device. As I read PSU reviews I find there are just about as many test methods as there are websites testing. Ever since I reviewed my first
PCPower & Cooling PSU and shortly after OCZ allowed me to review their very first pre-production
PowerStream 420, my interest in power supplies piqued.
Unfortunately without the proper test equipment and having to learn electrical theory of switching power supplies from scratch the road has been strewn with the potholes of inexperience. I have tried to compensate for this lack of test equipment drawing from my background in Audiophile hardware and immediately began immersing myself in related technical material. My desire to expand my knowledge along with my studies in Neurophilosophy, Philosophy of Science and Neuroscience of which Computer Science is an integral discipline, all have been invaluable to my technical writing.
So what does all this have to do with Power Supplies and Test Methods? Many Reviewers today have become complacent in their testing. We often find a redundant review formula with little derivation or creativity. I can empathize with those sites whom cannot afford appropriate test equipment such as a Chroma ATE PSU tester; however, it's these sites which need to put that much more effort into their reviews so they serve the reader, even if only to provide photos of circuit topology, describe the design and provide pricing as well as availability. Even among those sites that can afford a Chroma (for example) I fear the best we are doing is to design tests which do little more then replicate manufacturer product specifications to validate their accuracy.
After sifting through hundreds of reviews over the years, I have found just a handful of sites that have the appropriate test setup and the wherewithal to design test criteria which transcend the typical formulae. One site I hold in high esteem is XBitLabs and as an example of their technical prowess one need only look to their 2004 article: “
Power Consumption of Contemporary Graphics Cards....Part II”. In that article XBitLabs is the first and only test site I know of to isolate and monitor graphic card power consumption by measuring both the current draw of the PCIe slot and external connector simultaneously. With the modern graphic card approaching unprecedented power demands on the system power supply this has become a valuable resource. My only criticism would be they also measure for amperes. The topic photo exemplifies XBitLabs modified external connector and below the original shunts wired into the AGP slot (this has now been upgraded at XBitLabs for PCIe).
Recently I came across an article on the subject of PSU test methods I consider required reading for anyone who measures, or reads PSU reviews. From Hardware Secrets;
Why 99% of Power Supply Reviews Are Wrong. I'm sure that article will have a substantial impact on most PSU reviews you have read. I have compiled a list below of hardware review sites which I feel have made a contribution to the accuracy of PSU testing and testing in general;
XBitLabs,
ExtremeOverclocking,
SPCR,
Hexus,
PCPerspective and
JonnyGuru. On the subject of my own test methods I have simplified the process by building two test systems from each camp, Intel/AMD and creating an environment which will task the PSU to its limit. This was done by calculating ampere draw from each component running at stock settings and maximum overclock, while under IDLE then full LOAD. Insofar as using a multimeter to measure Load fluctuation, I refer you to the following quote from the Hardware Secrets article above;
Power supplies used on the PC use switching-mode technology. They are closed-loop systems, meaning that the power supply measures its output voltages and corrects them if there is any fluctuation. This is done by the PWM circuit, which is in charge of switching the primary transistors. In other words, if there is any fluctuation on the output voltages, the PWM circuit will know it right away, increasing or decreasing the duty cycle of the signal applied to the switching transistors in order to correct this.
Since the frequency of the signal applied to the transistors are in the range of KHz, it would take only a few microseconds to the power supply to correct any fluctuation found on its outputs. And no multimeter would be capable of measuring the power supply fluctuation, if any....since the power supply found on the PC have five different outputs (+12 V, +5 V, + 5 VSB, +3.3 V and –12 V) you would need to connect five multimeters to the power supply at the same time...Even if you connected five multimeters, you would need to read them at the same time.
You begin to see the conundrum for those of us ill-equipped (using multimeters). Temp measurements can be taken and perhaps the only way a multi-meter can be of any value would be to monitor the motherboard directly, as I have done for Vcore. Still the question remains, with most sites reproducing data in watts rather then amperes is this the best measurement? I have been adamant about measuring power demands in amperes rather then watts which I can best explain with the following analogy;
Measuring video card current requirements using wattage, is analogous to determining the brightness of a candle by measuring the heat of the flame. The following passage may clarify the relationship between amperes, watts and how watts are a concomitant event;
the SI unit of power. Power is the rate at which work is done, or (equivalently) the rate at which energy is expended. One watt is equal to a power rate of one joule of work per second of time. This unit is used both in mechanics and in electricity, so it links the mechanical and electrical units to one another. In mechanical terms, one watt equals about 0.001 341 02 horsepower (hp) or 0.737 562 foot-pound per second (lbf/s). In electrical terms, one watt is the power produced by a current of one ampere flowing through an electric potential of one volt.Try finding one article, review or table which provides a measurement in amperes for any given video card or PSU (if you find it please write me). I specify video cards because these will demand more current from our power supplies then any other component in the modern PC. This especially true with the emphasis on Gaming, the impact of SLI/Crossfire and the transistor count increase in recent GPU’s. Even with our newfound evidence devaluating the multi-meter alone as a tool to evaluate power supplies, there is still no defined and universally accepted test methods.
Insofar as our test methods, as I've stated generating a "real world" load on the power supply does have merit, at least as long as the greatest draw from the VGA cards are those which demand some of if not the largest number of amperes. While employing a digital multimeter on the 6-pin PCIe connectors looking for voltage fluctuations on a switching power supply does seem fruitless, we did nonetheless see fluctuations. Using two Extech
MiniTec MN-26 which were both calibrated just before this test, made my readings even harder to simply dismiss. Might this have something to do with the ST85ZF's "true" independent voltage regulation? In the interim measuring Vcore was accomplished at the motherboard circuit level thanks to information concerning volt mods for both the Asus M2N32-SLI Dlx and Asus P5W DH Deluxe. For the Asus P5W DH example I would like to thank Next Hardware / Memory Xtreme for the forum topic
ASUS P5W DH Mods from which the photo below was borrowed and modified.
Onto our test results -->