Introduction

How many reviews of power supplies have you read? I bet you that in the majority of those the reviewer will mention something along the line of this:

Power supplies are generally not thought of as the most interesting PC components, but they can often be the most vital, especially for system stability when you're pushing your hardware to its limits

While we assume that this statement is true in most cases, how accurate is it? To find out we went out and bought two cheap no-name power supply units and put them through a series of in-depth load tests.

The No-Name Products

We visited two local PC stores with a small amount of cash stuffed in our wallet with the aim of walking out of them with a working, brand new PSU.

The first shop we visited was Forcom where we bought the cheapest ATX unit available: a no-name 450W ATX PSU with 120mm fan and P4/AMD connector, it even features SATA power connectors!



This is what we got for €23, a Stability Power unit, Forcom sells it as a 450W unit, while a sticker on the PSU labels it as 500W.

Next we sought out a higher rated model at Alternate, a €49 Sweex Low Noise 650W PSU.



While not exactly a no-name brand, this Sweex unit was the cheapest 650W unit we could find, you can find a bit more info here about this PSU.

Time to start our tests to see how if these units can perform as advertised ->

Stability Power 450~500W

This power supply is as generic as can be, dull grey color, one specification label for all different models, ranging from 300 to 550W.

There are not a lot of connectors on this PSU, hardly enough to power a very low end system, don’t try to use for a new Intel or AMD system though, there’s no 8-pin power connector or 6-pin PCIe. This unit is for Pentium 4 and Athlon XP…



The all in one label shows maximum loads for each volt rail:

+3.3v: 28A

+5v: 38A

+12v: 20A

-12v: 0.8A

+5VSB: 2A

Older generation system did get CPU power from the 5v line, this explains the overpowered +5v and the underpowered +12v.



Our first test is loading the PSU up to almost 50% of rated capacity:



The efficiency is respectable at 77%, the voltage lines remain quite stable too, not too much deviation seen. The power factor is low, 0.57 (57%), most brand name units hit 90%+ numbers with ease.

Let’s try for close to 75% load:



Efficiency drops noticeably; to be able to output 361W we are drawing 533W at the wall outlet. The 12v rail drops quite low, while the -12v is up, clear signs of instability.

Do we dare to increase further? We didn't quite reach 100% load.



We increase the load on the 3.3v and 5v rails, the outcome is bad, quite bad, efficiency drops to 66% and we’re at the maximum of what this unit can take, after 5min we heard a buzzing noise followed by a loud pop and visible smoke, the Stability Power was dead. The numbers you see above are far from encouraging, the -12v line is way up, and while this unit is rated for 450~500W it doesn’t even come close to reaching 400W.



One dead, one to go, will the Sweex survive the tests ?

Sweex Low Noise 650W

Sweex makes a multitude of products, ranging from KVM switches, LAN cables, mice, routers, webcams, and also power supplies, we got their highest end 650W Low Noise unit for €49 at Alternate.

Sweex (sweex.com) manufactures computer-related products and has grown into a leading, mature player in the European market for consumer IT. From its head office in Delft (the Netherlands) and the nearby logistical centre in Hook of Holland, an intricate distribution network attends to Sweex partners in 45 countries in Europe, the Middle East and Northern Africa. The Sweex office in Taipei (Taiwan) guarantees the best possible time to market for each product. The 50 Sweex employees possess clear-cut ‘can-do’ mentalities, with teamwork being one of the major pillars for the company’s success.

In the looks department the Sweex doesn’t disappoint, it has a nice finishing and a hidden fan inside, there are also enough connectors to power modern day AMD/Intel systems.

Looking closer at the specifications label we can see that this unit has 2 12v rails;

3.3v: 32A

5v: 30A

+12v1: 17A

+12v2: 16A

-12v : 0.3A

+5 VSB : 2A



Promising numbers here, so let’s start the first test, loading the PSU close to 50% of rated capacity:



At half load the Sweex delivers close to 80% efficiency, the volt lines remain stable with minimal fluctuation. Power factor however is noticeably low at 0.51 (51%)

Let’s try with a slightly higher load, closing in on 75% of rated capacity



Efficiency remains high, this unit is very close to a 80Plus rating indeed. Again voltage rails are well within ATX specs.

Let us try for 100%? We started increasing the load to obtain that goal the PSU wouldn't take it, here's the maximum it would do:



This was the highest we could get before the PSU would auto-shutdown due to overload, it does reach 663W at the wall outlet, but with lower efficiency at 76% the output is a far cry from the rated 650W. Voltage lines remain stable though throughout the test.



Overall the Sweex was not half bad for the price; it did not break down when overloaded, got stable voltage rails, better than expected efficiency. It doesn’t deliver the rated wattage though which makes this unit incorrectly advertised for sure.

So how exactly did we test these power supplies? Geoffrey our local PSU guru will go into that on the next page ->

PSU Testing Station at Cooler Master Venlo, NL







CoolerMaster Europe kindly invited us to have a sneak preview of their newly composed power supply testing laboratory, the ideal opportunity for us to write an interesting article for our readers.

Test Equipment Detailed

A power supply is not 'just' a power supply, there are some variables which can differ depending on which route there has been taken during the stage of concept and design. Yes, there are ATX specifications, but these are just rough guidelines, they tell you nothing about the PSU itself. And off course, there is always the difference in the nations electrical source, either being it 220V 50Hz or 110V 60Hz. Therefore you'd better carefully choose your testing equipment, these professional tools cost quite a lot and it would be a real bummer if you noticed afterwards that some tests could not be completed because none of your equipment supports it. The setup at CM Venlo comes close to €50.000.

Chroma 6500 series programmable AC source>
To start with, CoolerMaster made sure to have US and European standards covered, meaning they have both 220V and 110V AC sources available. Therefore they have chosen the Chroma 6500 series, a programmable Alternating Current (AC) power source. This is essentially a power source which can be programmed to have a output power anywhere between 0 and 300 Volt. There is the ability to alter the AC frequency and phase shift plus you have the option to choose between 30 pre-programmed waveforms, and if that isn’t enough you can always program your own waveform. There is a lot more then meets the eye, in the end I guess the optional graphical display is a great addition to this AC source.



Yokogawa WT210 digital power analyzer

The WT210 is a high-end digital power analyzer. It can correctly determine the power usage, current, voltage, frequency, power factor and many more with up to 0,1% accuracy. And with up to 26A we have enough headroom to go readout multiple sources at once, your house breakers probable won't even support that much. The WT210 is an excellent way to have an exact readout of how much current the whole setup is drawing, power supply inclusive.

Chroma DC programmable loads

In order to properly stress the to be tested power supply's Cooler Master bought 3 Chroma 6330 series programmable direct current (DC) loads. 2 types: 63303 and 63306, where the last number stands for the maximum power it can load, i.e. 63303 = 300 Watt. These DC loads come with a fast responding microprocessor regulating the load; on front you have an easy turn-knob which allows precise adjusting. The inside electronics allow for a readout of the current drawn and also shows you the live voltage. The load is left floating and the device is also equipped with OCP, OVP, OTP and OPP protection circuitry: shortage will not damage any inside electronics. Better: the Chroma 6330 comes with a pre-programmed short circuit test button on front! Besides that you have the ability to use multiple units in parallel, you can even have them working together synchronously.



Tektronix DPO3000 Digital Scope

Tektronix is a well know brand in 'scope-land', their DPO3000 series have a bandwidth support up to 500MHz 4 channels. It has all the basic measuring options, next to that there is triggering supported for analysis of I2C, CAN, RS-232 and many more. The scope has a plug-and-play computer interface and comes with front and rear USB ports(for mass storage devices) as well as a LAN port. On front you'll find a 800 x 480 high resolution (WVGA) color display, the bigger the better right... More info can be found here



Flukes, thermal probes, handscopes...

For single line analysis a Fluke 88845A hand scope is present. This device is not exactly a normal multi-meter; it's very high precision and has frequency and wave analysis options. Not comparable with the Tektronix DPO3000 off course but very complete in its own way. And it doesn't stop there; we also found thermal probes attached to the inlet and outlet of Cooler Master Test subject. Both these probes were connected to the Center digital thermometer which comes with a computer interface with software included. Results can be logged and are being directly visualized in graphs.

Testing Cooler Master Power Supplies

Now that we have tested two cheap PSU units it is time to see what you get if you decide to pay a bit more. Cooler Master provided us with three different units to be put through their paces.

• Real Power M 620W
  
• Silent Pro M 500W
  
• UCP 900W

We put got some help of Cooler Master employees turning the 1001 knobs on the load testing gear, here you can see Ron the local tech guru doing his thing:



The first of the Cooler Master PSU we tested was the 620W Real Power M, we hooked it up to the test station and were ready to go with 50% load:



Stable volt lines, high power factor, efficiency at 85%. Impressive numbers.



Efficiency takes a small hit at 75% load, but remains above 80+. Volt lines are again very stable. On to 100% load:



To be able to get 80Plus certified you need to have 80% efficiency at 20/50/100% load, the Real Power M 620W passes that test.

Next we wanted to see what it takes for this PSU to automatically shut down, we started the overload test, when the PSU failed we reset the system and measured the values at the edge of stability… here are the results:



This 620W unit was able to deliver 856W of power, and although efficiency was no longer above 80% you have to take into the account the fact that this unit running 38% above the rated wattage, which is nothing less than stellar!

More CM PSU and Results Table



The Chroma DC-loads come with an ATX compatible external printed circuit board where you plug all the different voltage rails of your PSU onto. The thick white wires on the left connect the DC-load with the printed circuit board, the black and red wires are used to read the voltages on the printed circuit board as if it where your computer mainboard. What you can't see are the small pins specially added for scope monitoring, but you can spot the probes right. As you may have noticed, the alternating current power supply cable has been dismantled; the reason behind it is to read out the current flowing through it. An AC clamp-on probe will react on the current flowing through one of the conductors of the cable via an electromagnetism principle, but you may measure only one conductor as once so that's why you have to cut the cable. On the intake and outtake ventilation holes we've also added temp monitoring probes but since our testing schedule was really tight we did not take note of these results.


Busy bees: Geoffrey and Cooler Master’s Technical Marketing Manager René Grau at work

For our tests we tried different load conditions, roughly 50%, 75% and 100% load. We then measured the voltage and current for each volt rail and then calculated the PSU's total output power. Via the Yokogawa WT210 digital power analyzer we had the ability to read out the power factor and the total input power of the power supply, the result of the difference between the output and input power is what we call efficiency. The AC clamp was used to analyze the AC inrush current which can be quite high with high-end PSU's, we also had a look at how fast the PSU boots and how the volt rails are pulled "high" in a stable condition.

Overview of all test results



Focused on efficiency we saw the high-end Cooler Master UCP 900W come out as best. Not a surprise since high-end PSU's really have to be efficient in order to keep temperatures reasonable and noise levels acceptable. With the load at only 500 Watt the PSU scored 90% efficiency which is the best score we could obtain that day. Runner up is the Silent Pro M500 from Cooler Master which had a efficiency score of somewhere in the mid-high 80's, followed by Cooler Master's Real Power 620 which also scored 80% + efficiency.
The Sweex 650W unit had lower efficiency scores then the Cooler Master units we tested, at 76%~79% its not that much worse then the Cooler Master quality PSU's but what was really unsatisfying is that we could not even scale up to the maximum rated output power: at roughly 500W this 650W model shuts down. The story gets even worse for the Stability Power 500W unit, at roughly 400W load this unit began to make a weird noises followed by a PSU shutdown, bad smells and a small wisp of smoke coming out of the blower hole. Just before the shutdown we could measure efficiency as low as 66% which is terrible low and may cause the PSU to burn down at higher loads because of problematic cooling performance (if it did not gave up to you already that is).

Another very unsatisfying result is the power factor of the budget power supplies. The ideal number is 1 (100%), Cooler Master stays very close to that number with their PSU's, the low-end no-name brands however don't seem to have any kind of power factor correction circuit. These numbers are not dramatic since we're talking about a single unit and low current streams, not like heavy machinery and stuff like that, but you should not neglect these results either.

We also performed a short circuit test on each PSU, luckily the OCP was working well enough and none of the test samples died in the process.



The above charts show you the output power and PSU's efficiency, with each of the PSU's rated for at least 500 Watt or more it is remarkable how the Cooler Master units have no problem to keep the efficiency up while increasing the output power. Over 400W load the no-name power supply's have either stopped working or have significant lowered efficiency, again I remember you that you should not take into account the scores of the Real Power 620 loaded with 850Watt because we were using the power supply way out of specs.

Focusing on the 50/75/100% load and efficiency we get this:



But there is more, let us have a look at some of the extra tweaks Cooler Master has added just to make sure they deliver a quality product ->

Rear on/off button missing?



The higher end Cooler Master units seem to have the On/Off button missing, but what up what that? Did they simply forget? Off course not, let us first have a look at the following scope image:


Sweex 650W power supply boot

This is the booting process of the Sweex 650W power supply with 500W load. The thing with computer power supply is that they have a large amount of capacitors which draw a lot of current when the power comes on. In the above picture can you clearly see the peak current (inrush current) which happens at the moment the power is switched on. With this unit we have a peak value of 33 Amps. It's not that hard to imagine that a high loaded 900W unit will draw even more current... The thing is that these power-on/off buttons also have to deal with the AC current flows and in some cases the button just isn’t good enough. That's where Cooler Master came up with the innovative idea to have a high power relays instead to deal with the large current flows, they have also developed a circuitry which spreads out the power-on current drawn which makes the inrush current a lot smaller. Here is what we measured with the Cooler Master UCP 900W 100% loaded:


CoolerMaster UCP 900W 100% loaded booting process

As you can see the Cooler Master unit has no problem to score lower then the Sweex unit. Another interesting thing we saw during the day was the booting process of the Stability Power 500W unit, have a look at how it struggles to get its voltage rails flat with a 380W load applied:


Stability Power 500W booting process

Thermal Imaging Tests

An ideal conductor has zero resistance, unfortunately that is still not the case with semiconductors and so there always some heat that you will have to get rid off when using a AC-DC switching power supply like those we use in our computers. The more power you draw from your PSU, the more current will flow through the high power diodes or transistors/fet's, knowing that they are not ideal conductors but always have a small amount of resistance it not that hard to imagine that the more power is needed the more the PSU's will heat up at the inside. Thermal tests are not to be neglicted, unfortunately we did not have the time to stress test each unit over larger periods of time, non the less did we get to see some very exciting media which show some of the idea's CoolerMaster has been thinking of when developing their units.

In the two videos below you can see with thermal imaging camera where the heats is expelled from the power supply. The first one is a generic unit:



And this one is from a Cooler Master UCP 700W PSU:



With the UCP 900 you can see how the PSU is designed to have only one airflow direction, meaning that heated air can have only one way out and that's at the outside of the case. With generic units that is not always the case, more ventilation holes is not always better because the pressure of the fan will blow the air in different directions and if there are two ways to exit the PSU housing then the air will take both exits: some heated air will travel back into the pc housing. For some this may seem only details of course but you can not debate about the fact that CoolerMasters design will not be the best way to get the heated air out of your computer as fast as possible. But to be honest, it are these details which define a better product from a lower quality one, and it are these details which are quickly looked over when reviewing these power supply's.

Conclusive thoughts

So what have we learned here today? Well, much more than with the usual power supply tests we hope. You see, out there you have all these reviews where just a few real world load test are being done by using a high-end computer system. In their conclusions they are indeed correct with the statement that power supply X can keep a high-end system running stable for a period of Y hours, at sound level Z. We have had such articles too in the past but we quit doing them because we know there is much more to testing a PSU than seeing how the 3/5/12V rails behave under load.

For starters, define a high-end system, define mainstream, and how can you possible be sure that such a system can really push your PSU? Without clear numbers you are only guessing. Today we had the ability to program the load per volt rail at very high precision; we could read the current flows and calculate the power usage and efficiency of the tested units. We could honestly and accurately state that a power supply advertised for X Watt is able to actually do it (or fail to do so in some cases), and what efficiency it maintains.

There is much more to reviewing PSU's, we had the ability to analyze the inrush current, the booting process, transient response if we wanted, heat, voltage levels, power factor, ... the list goes on, however time was running out and we chose to limit the amounts of tests to be able to hook more units to the load tester and have more data to show you from different PSUs.



Coming back to our test results, we were astonished by some of them. For example we could properly stress the Cooler Master Real Power 620 up to 850W without it shutting down or burning down or anything. Completely different from what we saw with the low quality no-name units, these didn't even deliver the stated currents and stopped working at roughly 80% of their maximum rated power. They also lacked proper power factor correction and compared to quality builds they were simply not good enough. The Stability Power unit even scored as low as 66% efficiency in one of our tests. In contrast, the Cooler Master UCP 900W scored fabulous, we saw 90% efficiency which certainly deserves a high-end recommendation:


CoolerMaster UCP 900W

Both Real Power M 620W and Silent Pro M 500W performed beyond their specifications, with the Silent Pro having a small lead in efficiency, the price difference between these two is small, Real Power M 620W: €79, Silent Pro M 500W: €89. Both can easily make up part of a quality budget build and deliver enough juice to today’s high end gear.


Cooler Master Silent Pro M 500W & Real Power M 620W

We are sure to return to Venlo with a new batch of power supplies to torture, we were promised some alone time with the load test gear at our next visit so we could really delve into the details, with any unit we bring along, so you can expect more PSU reviews coming your way here at Madshrimps.

We hope you liked this article and how we tested, before you go let us thank Cooler Master for letting us turn their knobs and overload their products. Special thanks go out to Marloes for the tour around the facility, René for all his help during the test sessions, and Ron for his technical insight behind the Cooler Master design philosophy. Until next time, take care!