what size oscliliscope do you need to test a pc power supply

After a lengthy hiatus, nosotros're back with a new PSU and case reviewer. Every bit we boot off our revised power supply testing and reviews, we wanted to cover the fundamentals of how we test and what to expect. Some of this is still a work in progress, every bit we have not gathered all of the equipment nosotros would like to have, and every bit we move frontward we will periodically provide updates to our PSU testing procedures. And with that out of the fashion, let's discuss how nosotros're going to go well-nigh testing ability supplies.

Constructive testing of a ability supply requires far more than merely connecting it to a PC and using a $x multimeter to check the voltage rails. At the very least, it requires specialized (and very expensive) equipment. At this point, most people that really know a few things about PSUs would say, "Yes, OK, y'all need an adjustable load and an oscilloscope." While information technology's true you lot need those items, you lot can't simply grab whatsoever old adjustable load and oscilloscope. What you really need is very precise, programmable electronic loads with transient testing built-in and an oscilloscope that should comply with exact specifications, amidst other meters and equipment. So of course you demand to know what you are doing, every bit it'southward non simply a matter of connecting a PSU to the equipment and pressing a few buttons; at that place are exact loading and testing procedures, described in technical papers and guides, that need to exist followed.

Programmable DC loads are an absolute necessity if you lot want to test a power supply. To that end, we acquired two loftier precision Maynuo M9714 1200 Watt and two Maynuo M9711 150 Watt electronic loads, which will allow us to describe up to 2400 Watts from 12 Volt lines and up to 150 Watts from each of the 3.iii Volt and 5 Volt lines. Equally these are quick-response programmable models, they will too permit us to perform transient tests in the futurity.

When testing a ability supply, using fifty-fifty the all-time of multimeters are entirely useless. An oscilloscope is an absolute necessity and not simply any oscilloscope. Intel's ATX pattern guide denotes that the oscilloscope should have a bandwidth of 20MHz; notwithstanding, things are non near as uncomplicated as that. Even if you do want to purchase a proper oscilloscope, buying a 20MHz oscilloscope is a mistake. Digital oscilloscopes need to be capable of acquiring samples at least ten times faster than the frequency they are required to resolve. Then, you need a 20MHz oscilloscope with a sampling rate of at least 200 MSa/s, and low range or USB continued devices cannot get anywhere shut to that number.

There are of course many other minor details but we will not bore y'all with those. It should suffice to say that for the time being nosotros are using a Rigol DS5042M oscilloscope, which has a bandwidth of 40MHz and a real fourth dimension sampling rate of 500 MSa/south. Although that sounds impressive, really even this device is non good enough if y'all want to perform transient tests properly and it cannot resolve noise out of the ripple of a indicate; these are tests we plan to add together in the future.

Compared to the above items, testing the efficiency of a PSU is relatively elementary, once you know exactly how much power you are cartoon from it. Our electronic loads tell us exactly how much ability is being drawn at a given time; therefore, we merely need a good Air conditioning power analyzer to tell us how much ability the unit of measurement is cartoon from the AC outlet. Note the "good" part, as yous demand a power analyzer capable of displaying truthful RMS values, as PSUs can generate a great deal of harmonics.

Our Extech 380803 power analyzer does a very good task at reporting the level of power that our PSU requires at any given time. We should note that all testing is being performed with a 230V/50Hz input, delivered by a 3000VA VARIAC for the perfect adjustment of the input voltage. Unfortunately, we cannot perform tests at 110V/60Hz at the moment, every bit that requires a high output, programmable AC power source. As a rough estimate, conversion efficiency drops by one% to i.5% when the input voltage is lowered to 110V/ 60Hz.

Thermal and noise testing are another complex procedure. Thermal testing is relatively simple; we only had to acquire two loftier precision UNI-T UT-325 digital thermometers. With four temperature probes, we can monitor the ambient temperature, the frazzle temperature of the PSU, as well as the temperature of its master and secondary heatsinks. Noise testing still cannot be performed while the unit is existence tested, equally the very equipment that is used to test it generates a lot of dissonance. Everyone says that it is incommunicable to keep the unit loaded with the equipment far apart in order to perform dissonance testing and aye, that truly is incommunicable. Then, information technology cannot be done, right? Wrong.

One of the basics of the scientific method is that y'all isolate the problem from a system and resolve information technology on its own. In other words, instead of trying to practise the impossible and measure the noise of a power supply while we are testing it, there is nothing keeping us from using a non-intrusive light amplification by stimulated emission of radiation tachometer to tape the speed of the fan instead. Then, we can simply test the unit of measurement on its own, with the fan hotwired to a minor fanless, adjustable DC PSU that we made, taking noise readings with our Extech HD600 for the RPM range of the fan and cross-referencing the two tables. Not quite that difficult, was it? In that location is a catch withal; as the unit of measurement volition non be powered at the fourth dimension of audio level testing, the meter cannot tape whatever scroll whine noise. Coil whine is clearly audible during testing though and we will make certain to written report it if (when) we encounter a PSU whose coils could have used a little bit more lacquer. The background noise of our testing environment is about 30.4 dB(A), which figure resembles a quiet room at night. Equipment noise usually becomes audible when our instrumentation reads in a higher place 33.5 dB(A).

In order to facilitate testing ability supplies more effectively, nosotros created a test fixture for the connection betwixt the PSU and the testing equipment, also every bit a proprietary hot box. The hot box is not much more than a closed case with an air-heating device, which is controlled via a DAQ and our software. It is imperative to heat the air inside the box, not the box itself, in order to create adept testing conditions. Admittedly, this self-fabricated contraption is not perfect as information technology is modest and has a very slow reaction rate, only it does piece of work well for the means of simulating the environment inside a estimator case. Therefore, testing will exist performed at room temperature (maintained at 25 °C) and inside the hotbox (at 45-fifty °C). Remember that efficiency certifications are performed at room temperature (25 °C) and a ability supply can easily fail to meet its efficiency certification standards inside the hotbox!

As for the testing procedure, there are specific, detailed guidelines on how to perform it. All testing is done in accordance with Intel's Ability Supply Design Guide for Desktop Class Factors and with the Generalized Examination Protocol for Calculating the Energy Efficiency of Internal AC-DC and DC-DC Power Supplies. These ii documents describe in item how the equipment should be interconnected, how loading should exist performed (yep, yous do not but load the power lines randomly), and the basic methodology for the acquisition of each data fix. However, not all of our testing is covered and/or endorsed past these guidelines.

(Update, April 20th, 2015, Click for the detailed post)

Even though these documents are just a few years old, their methods neglect to account for mod "enthusiast course" figurer SMPS. The industry has been making leaps on the creation of more than energy-efficient devices, continuously lowering their ability requirements. Nowadays, the vast majority of computers that require very powerful PSUs merely employ multiple components, such as numerous graphics cards. Equally the majority of energy-consuming components crave a 12 Five source, PSU manufacturers have been continuously driving the 12 Five output of their units upward, while the 3.3V/5V outputs remained inert or are getting weaker. In that location are many pattern rules that modernistic "enthusiast-grade" PSUs do not adhere to nowadays, such as the current safely limits and the maximum size of the chassis, but this particular change creates a trouble with the generalized examination protocol.

According to the generalized test protocol, the derating cistron D of the iii.3V/5V lines should exist:

Just put, the formula is maximum rated power output of the unit divided past the sum of the power output ratings of each individual electric line.

Yet, this formula ofttimes leads to the overloading of the 3.3V/5V lines with >i kW PSUs. The effect is specially severe in some loftier efficiency units, in which the designers moved the 3.3V/5V DC-to-DC conversion circuits on the connectors PCB, reducing their maximum power output significantly. Although some PSUs would operate normally fifty-fifty if their 3.3V/5V lines were overloaded, the continuous degradation of the 3.3V/5V lines in comparing to the 12 V line resulted to PSUs actualization in our labs that could non operate under such atmospheric condition. Therefore, we had to alter the derating factor formula in gild to compensate for real world testing. Without at to the lowest degree 2 meaning energy consumers, no modern system requires > 500 Watts. Greater power demand suggests the presence of devices that load only the 12 Five line (i.e. GPUs, CPUs, liquid cooling pumps, Peltier outcome coolers, etc.). Subsequently sure calculations and research, for units with a rated power output over 400 Watts, we volition be using the following formula:

Which effectively half the impact of the 3.3V/5V lines on the calculation of the derating factor, imposing the divergence on the 12V line. Furthermore, the loading benchmark of the iii.3V/5V lines for a load rating X (in % of the unit of measurement's maximum output) is now changed to:

For the 12 V line(due south), the loading criterion remains unchanged.

This formula results to the more realistic representation of the requirements that actual systems have, at least up to a power output realizable today.

(End of update)

Furthermore, there are no guidelines on how transient tests should be performed and the momentary power-up cross load testing that Intel recommends is far also lenient. Intel recommends that the 12V line should be loaded to < 0.1A and the 3.3V/5V lines up to just 5A. Nosotros also perform two cross load tests of our own design. In test CL1, we load the 12V line up to 80% of its maximum capacity and the 3.3V/5V lines with 2A each. In examination CL2, we load the 12V line with 2A and the iii.3V/5V lines up to lxxx% of their maximum combined capacity.

Furthermore, it has been suggested that efficiency testing needs to be performed at specific load intervals (20% - l% - 100%), which is considered to be the normal operating range of a PSU. However, modern systems can easily have their energy demand drop dramatically while idling, which is why we will be testing power supplies starting at five% of their rated capacity, not twenty%. Note that the conversion efficiency of all switching PSUs literally takes a dive when the load is very low, so large drops of >10% are expected and natural.

Any questions or comments on our PSU testing procedures are welcome, and as noted earlier we programme to add and/or improve some of the testing over the coming months with some additional hardware. We will provide an updated commodity when/if such changes are required.

vincentungs1936.blogspot.com

Source: https://www.anandtech.com/show/7820/how-we-test-psus-2014

Share this post