X-bit labs

Roundup: 14 Power Supply Units from Thermaltake (page 9)

Category: Other

by Oleg Artamonov

[ 07/28/2006 | 08:34 AM ]

So, there are two different and, what’s important, two greatly different voltage regulators in this PSU (one follows the group voltage regulation design whereas the other lives by itself and doesn’t depend on anything else). This made me make two cross-load diagrams, in one of which the +12V line was provided by the main regulator (12V1) and in another by the additional regulator (12V2). Our testbed has two independent +12V load channels, so I could perform the measurements simultaneously, but it would have been hard to show the results in a single diagram.

This is the +12V1 channel, i.e. the main regulator designed in the classic way with group voltage regulation. To remind you, it powers the mainboard, graphics cards, and some of the hard drives.

Unfortunately, the +12V voltage bottoms out heavily and is below the lowest permissible limit (11.4V) under loads of 150-160W. Is it a high load or not? For example, one Radeon X1900 XTX graphics card consumes about 120W; two GeForce 7900 GTX cards in a SLI subsystem have a power draw of about 160W. The voltage on the W0049’s output is going to be 11.3-11.4V under this load, which is likely to lead to system malfunctioning.

The picture is merrier if you use the additional regulator (12V2). This regulator works by itself and doesn’t depend on how the other power rails are loaded. It can yield the promised 23A without violating the acceptable limits.

On the other hand, the +3.3V voltage lost stability when the 12V1 line was under-loaded. But this problem occurs if there is a high load on the latter rail, which just cannot happen in a modern computer system.

Alas, the poor stability of the output voltages is not the only drawback of this power supply. At a load of 580W (i.e. 100W short of the maximum!) the voltage ripple on the PSU output was 56 millivolts on the +5V rail (the allowable maximum is 50 millivolts), 87 millivolts on the 12V1 line, and 20 millivolts on the +3.3 rail. On the 12V2 and 12V3 lines there were high spikes up to 300 millivolts (the allowable maximum is 120 millivolts).

So, I have to say that the attempt to adapt an old power supply to new operating conditions ended in a total failure. The poor stability of the 12V1 voltage and the strong high-frequency pulsation on the output make this power supply absolutely unsuitable for a more or less power-hungry computer (with a SLI or CrossFire graphics subsystem or even with a single high-consumption graphics card). What is the funniest thing, a high-quality 400W power supply can easily cope with a typical SLI system whereas the declared wattage of the W0049 is higher by 300W!

If you’ve already bought a W0049 and found your system unstable, I advise that you try to power your graphics card through an adapter from a Molex connector on the cable that belongs to the 12V3 line (choose one of the cables with floppy mini-plugs on their ends).

After so disappointing results there is not much sense in testing the fan speed and efficiency factor – what’s the use of this knowledge if the power supply is absolutely trash? Well, anyway…

Both fans are usually connected to the same speed controller in such power supplies, but here only the speed of the external fan changes considerably. The PSU is quiet under low loads; you will hear its noise from loads of about 250W and higher. Two identical fans from Space Fan (the B802512BH model) are employed here.

The efficiency of this power supply is average, and the power factor is rather low for a model with active PFC (only 0.95 at the maximum).

So, the first-revision W0049 model is not just a poor product. It is practically unfit for work and is not worth even half its price. The attempt to modernize an old power supply by putting an additional regulator into it only created more problems, partially coming from the section of the PSU that was left unchanged (hence the low stability of the +12V1 line – its load capacity didn’t matter much at those times when computers were mostly powered from the +5V source) and partially from the imperfections of the new additional regulator (hence the very high voltage ripple on the +12V2 and +12V3 lines). If you want to buy a W0049, I strongly recommend that you buy its second revision (discussed in the next section of this review). If you’ve already bought the first revision and encountered all the problems mentioned above, you should follow my advice and power the graphics card through an adapter or, which is better, replace the PSU with a better one (even a lower-wattage but high-quality new PSU will prove a better choice).

The irony of this situation is that all manufacturers that produce ATX12V 2.0 power supplies appeal to the customer with the two +12V lines. This is usually featured in large print on the box and is then followed, in smaller letters, with an explanation that this “technology” ensures unprecedented power, stability, reliability, etc. Our readers already know that there is usually only one +12V power rail inside the PSU and it is only divided in two by means of current limiters which cannot affect the stability of the output voltages even theoretically. The W0049 is in fact the first power supply tested in our labs that really has two separate +12V power rails, but its power, stability and reliability are far beneath criticism as you have just seen.