by Oleg Artamonov
07/28/2006 | 08:34 AM
Started as cooling systems manufacture, Thermaltake Technology has long overgrown these limits and is currently offering, among everything else, a very extensive line of PC power supplies. There are three lines, to be exact: TR2, PurePower and ToughPower.
The company positions its power supply units (PSUs) as high-quality products intended for computer enthusiasts and advanced users and prices them accordingly. However, there is no unanimous opinion about Thermaltake’s PSUs on Web forums where they are often a subject of hot argument. So, I got curious about all this and tried to clear things out for myself and for you, too.
The entire model range of Thermaltake power supplies is covered in this article, from the obsolete but still selling W0009 (it was the model Thermaltake began shipping PC power supplies with) to a model from the newest ToughPower series. Perhaps some of the products to be discussed are not at all interesting as a purchase, yet their results in tests help understand the overall structure of Thermaltake’s model range better and find the reason why the company’s power supplies provoke such contradictory reports from users.
Before getting to the PSUs proper, I’d like to remind you in brief our power supply testing methodology.
One of the crucial parameters of any power supply is the stability of the voltages it yields. In our tests we examine the three main power rails (+5V, +12V, and +3.3V) by putting varying loads on them. As you may know, the so-called group voltage regulation is used in quite a lot of PC power supplies to simplify design. The downside of this is that the voltages outputted by the PSU depend on each other. So, it is important to check out how all the three voltages change in the entire range of permissible loads. As the result of this test a cross-load diagram is constructed in which the X-axis shows the load on the +12V rail and the Y-axis shows the combined load on the +5V and +3.3V rails. The diagram shows a colored area inside which the PSU’s output voltages do not deflect by more than 5% from their nominal values. The exact value of the deflection is marked with color from green (a deflection of less than 1%) to red (a deflection of over 4% but less than 5%).
The cross-load diagram gives you an understanding of how the particular power supply behaves in general as well as in a specific computer system. It also reveals certain problems that may occur. As opposed to this method, many reviewers limit themselves to measuring the output voltages under a few loads or even in a particular computer, but such tests only give an answer to the question, “Will this power supply work in the given computer?” and leave the rest of the issues not attended to. Particularly, problems that may arise if this PSU is installed into a differently configured computer are not identified, which is wrong.
The second step in our tests is to measure the amplitude of pulsation of the power supply’s output voltages. This pulsation may occur at two frequencies: at the frequency of the PWM-regulator of the PSU (from 70 to 130kHz, but usually not higher than 75kHz) and at the double frequency of the electricity network (100kHz in our area). This much simpler test produces a few numbers that should be within the limits described in the industry standard.
The third step is to measure the speed of the PSU fan depending on the load. The load on the tested power supply is being increased in a step-like manner from the minimum point of 50W to the maximum permissible load. The power supply works for a while under each load until its temperature stabilizes; the fan speed is measured after that. Besides, we also check if the PSU can work normally under different loads, including the maximum load it is rated for – the PSU must be able to work for 30-40 minutes under it. I know from my experience that power supplies with overstated wattage either fail in the first 10-15 minutes or show unmistakable signs of being about to fail, e.g. begin to smell of burnt insulation; that’s why I don’t think it’s necessary to test a power supply under its maximum specified load for too many hours. But again, each PSU is tested for at least half an hour in our labs for its ability to output the full power declared by the manufacturer.
The last item in our test procedure is about the efficiency and power factors. You should not confuse these two parameters – they have nothing in common at all. The efficiency is the ratio of output power to input power. The difference between the input and output power transforms into heat you have to pay for in your electricity bill. This heat also makes it harder to cool the PSU and makes a higher fan speed necessary.
The power factor has nothing to do with the processes taking place inside the PSU or on its output. It is a ratio of active input power to full input power. The difference between these two powers is called reactive power and it presents an interest for power engineering specialists rather than for computer users: reactive power does not do any work in the load (and it is not dissipated in it in any way, heat or anything), but is just flowing forwards and backwards in the circuit, purposelessly loading the wires, sockets, substations and other equipment, even including the generators at the power station. That’s why it is mostly the electricity suppliers’ rather than the consumers’ trouble to try to reduce the amount of reactive power, i.e. to increase the power factor (well, this indirectly affects the consumers, too, but it would be too great a digression to dwell more upon this topic).
I also check the power supply visually for any assembly defects and for extra functionality, count up its cables and connectors, and give my opinion about it from an aesthetic point of view. After all, the exterior of the power supply matters for quite a lot of users.
This is one of the oldest models in Thermaltake’s PSU line-up, and it has already taken part in our tests a number of times and under different brands because its actual manufacturer Sirtec supplies it to different companies. The HPC-420-102 DF is still available in shops, so I included it into this review to give you a fuller overall picture.
The suffix “102 DF” in the model name denotes a version of the power supply without power factor correction (“202” stands for models with passive PFC, “302” means models with active PFC) and with two cooling fans (“Dual Fan”).
This power supply has a plain gray case. One of its fans is located at the back panel, the other is placed on the bottom panel (I’m referring to the panels’ positions as they would be if the power supply were installed in the system case). It has a Turn on/off switch and an input voltage switch.
The internal design of this power supply is absolutely ordinary. It is assembled neatly, so I can find nothing to complain about. The PSU has an input socket with an integrated filter in addition to the filter on the main PCB.
The W0009 is designed up to the already out-dated ATX12V 1.3 standard (the current on the +12V rail can be up to 18A). This version of the standard was introduced as transitional from version 1.2 to 2.0 and became obsolete after the arrival of version 2.0-compliant units. So, in a modern computer, which mostly loads the +12V rail, the real load capacity of the HPC-420-102 DF won’t be higher than 250W: the maximum of 216W on the +12V rail plus 25-40W on the +5V and +3.3V rails (modern systems just don’t need any more power from these low-voltage lines).
The PSU is equipped with the following cables and connectors:
The mainboard’s connector has 20 pins, of course. The 24 connector was not yet described in the ATX12V 1.3 standard and was not actually required in power supplies for which the maximum allowable current on the +12V line was rather low. It’s nice to have two SATA power connectors here – you won’t have to use adapters.
The cross-load diagram for this PSU doesn’t look nice to my eye. The +5V voltage is so high that it varies within a range of 5.2-5.25V through almost half the diagram. The +12V voltage is not very stable, either. It fluctuates rather much and is going to get the nominal value in older computers in which most of the total load was on the +5V line, contrary to modern systems.
At a load of 400W (12V/210W, 5V/156W) the voltage ripple on the +5V and +12V rails is 25 and 57 millivolts, respectively. Both low- and high-frequency pulsations are present.
The PSU is cooled with two SuperRed fans (CHA8012BBS-M and CHA8010CBS-A models). The speeds of the fans are quite efficiently controlled so that the PSU is quiet, although not exactly silent, under loads below 150W. The fans become audible at higher loads, the speed of the faster of them getting as high as 3000rpm and more at the maximum.
The PSU’s efficiency factor is surprisingly good – above 80% in most of the load range (I say surprisingly because it’s mostly modern power supplies the W0009 can hardly be counted among that are usually highly efficient). The power factor is just what you can expect from a PFC-less unit: not higher than 0.7 at the maximum (for comparison: units with passive PFC and active PFC usually have a power factor of 0.75 and of over 0.95, respectively).
So, even if you meet a W0009 in shops, you will hardly be seriously interested in it. This power supply was designed to comply with the now-obsolete ATX12V 1.3 standard, so its real load capacity is going to be low in modern systems (well, a majority of PCs will be satisfied even with that – you will only have problems if you’re using a top-end central processor or a SLI/CrossFire graphics subsystem). Moreover, the tested unit’s +5V voltage is a too high and +12V voltage is unstable. Considering that the price of the W0009 doesn’t differ much from the prices of new ATX12V 2.0 products, I see no sense in purchasing it.
This is yet another power supply whose actual manufacturer is Sirtec. It is newer than the one described in the previous section but belongs to the version 1.3 standard, too, with all the ensuing consequences.
This PSU resembles the previous model externally, except that the fan on the top has moved to a side panel and the input voltage switch is now missing. The W0020 features an active PFC device with the full range of input voltages, so it just doesn’t need that switch.
A big letter X with the Xaser brand logo in the middle is fastened to the side panel of the PSU with four screws. If you take it off, you’ll see this plastic window:
Of course, they don’t make windows in power supplies for no reason. There is a LED behind it. The LED has three colors and is blinking at work.
You can see in this snapshot that the LED is shining in green and red at the same time. A number of blinking algorithms is written into the LED and they replace each other automatically. Another such LED is fastened on the case of the fan in the PSU’s top panel:
As you see, this is not a highlighted fan proper, but a fan with a separate LED.
This is a multi-colored LED, too (its red light is shining right into the camera in the snapshot above and there is also a green reflection on one side). It is blinking with all its colors following a few algorithms that change automatically.
Apart from its fans, this power supply is designed in quite an ordinary manner and differs but little from its predecessor. The additional active PFC card on the left makes a difference, though. The main regulator is based on a SG6105D chip; the PFC device is based on a UCC3818N.
By the way, the first of the blinking LEDs resides on the card with the active PFC device.
Besides some screws and a power cord, the box with the power supply also contains a device to control the speed of the PSU fans. This device is to be installed into a 3.5” bay of your system case.
It can change the speed of the back-panel fan only. The other knob is meant to control the speed of one of the system fans; the system fan must be made by Thermaltake and must support such speed management.
Realizing that users may not like the necessity to look for for a specific fan to use with the speed management device, Thermaltake includes one with the PSU. This is not a regular fan, but an X-Blower that occupies a standard 80x80mm seat.
As I said above, the W0020 complies with the ATX12V 1.3 standard and differs little from the W0009 model. Its maximum load on the +12V rail is only 18A.
The PSU is equipped with the following cables and connectors:
Alas, the cross-load diagram looks bad again. The +5V voltage is set too high and the +12V voltage is not very stable.
At a load of 450W the voltage ripple on the PSU’s outputs was 31 millivolts on the +5V rail and 71 millivolts on the +12V rail.
I measured the speed of the PSU fans at the two extreme positions of the manual control. This only affects the speed of the back-panel fan, but the automatically adjusted speed of the second fan also changes because the PSU temperature depends on the airflow intensity.
This power supply is rather quiet at the min fan speed. The second fan is faster than 2000rpm, but it is hidden deep inside in an assembled computer and is not very audible.
When the manual control is set at its maximum position, the fan speed exceeds 4000rpm. There’s no talking about silence now, the PSU just roars loudly. The minor decrease in the speed of the second fan is but a poor compensation for that.
The PSU’s efficiency is a little lower than that of the previous model. The power factor is 0.98 thanks to active power factor correction.
Generally speaking, there are only external differences between the Butterfly W0020 and the above-described W0009. I mean the exterior design and the accessories. The Butterfly comes with a blower and a fan speed controller (for two fans which must come from Thermaltake and must explicitly support this controller because it is attached via a separate connector) and is highlighted with blinking three-colored LEDs. From the technical viewpoint, it is an obsolete ATX12V 1.3 unit a modern computer can only get about half the declared wattage from – after that the PSU’s +12V power rail will get overloaded. This lack of power is also aggravated by the lack of stability of the output voltages. So, you should better turn to other PSU models, even to other models from Thermaltake. The system blower and the multicolored LEDs can be purchased separately whenever you want them.
The W0023 differs dramatically from the previous model. It has a black case (the paint isn’t glossy, but matte; it even feels coarse to the touch). There are no LEDs here, no transparent windows, no decorations, only a large sticker Xaser on the side panel. You should be aware that the protruding grid of the fan on the top of the case may prove to be a problem when you try to install this PSU into certain system cases.
The internal design of this power supply has nothing to do with the W0020, either. The actual manufacturer of the PurePower 560APD is Seventeam Electronics as is indicated by the marking on the filter card (soldered to the mains connector) and on the main PCB of the power supply. The original name of this model is ST-522HLP.
The most unusual thing about this PSU is its heatsinks which are copper bars with aluminum ribs. Otherwise, we’ve got an ordinary enough design with active PFC and group voltage regulation. An ML4800CP chip is used as the main controller here (it combines a PFC controller with a switching power controller proper). An additional PS222S chip protects the output voltages and currents from overshoot.
The PSU label claims a peak output power of 600 watts but doesn’t specify the conditions under which it is achievable. The name and the official specification of the product imply a maximum output power of 560W (not peak, but sustained). The marking on the PCB reads “Seventeam ST-522HLP”, thus suggesting an output power of only 520W. Anyway, I checked this unit out under a load of 525W and it easily passed through the tests.
The PurePower 560APD surpasses the previous two models in the declared load currents, but doesn’t go too far beyond the ATX12V 1.3 specification with its allowable load of only 22A (264W) on the +12V rail, which is far lower than ATX12V 2.0 units offer.
The PurePower 560APD has the following cables and connectors:
All the cables, except for the latter, are sleeved and painted different colors: black for the main mainboard power cable, red for the AUX and ATX12V cables, blue for the PATA cable, and green for the SATA cable.
The cross-load diagram of this PSU is not ideal, but acceptable. It’s only at high loads on the +5V and +3.3V rails, i.e. in a very unlikely situation for a modern computer system, that the voltages violate the limits described in the ATX standard as acceptable. If the load distribution is typical (a total of 25-50W on the +5V and +3.3, and 50W and higher on the +12V rail), the voltages are all rather close to their nominal values.
At a load of 525W the high-frequency voltage ripple is 57 millivolts on the +5V rail and 42 millivolts on the +12V rail. If you sum this up with the low-frequency pulsation, the voltage ripple is 71 millivolts on the +5V and 54 millivolts on the +12V rail. I want to remind you that the maximum allowable voltage ripple is 50 millivolts on the +5V and +3.3V rails and 120 millivolts on the +12V rail. It means the PSU doesn’t comply with the requirements of the industry standard when working at full load.
But if the load is reduced to 100W on the +5V and to 20W on the +3.3V rail (this gives you a total load on the PSU of a little lower than 400W), the ripple diminishes to 25 millivolts on the +5V and to 15 millivolts on the +12V (although the load on this rail remains the same). I want to again recall the fact that modern computers do not load the low-voltage rails much, so the problem with the strong pulsation of the output voltages isn’t going to show up in practice, but this is just an excuse, of course. The PurePower 560APD was developed to comply with the ATX12V 1.3 standard and must provide the required parameters without any allowances for the changes in power consumption of computers that have occurred after the standard was accepted.
A Sunon KD1209PTB1 fan is installed on the top panel of the PSU, and a Sunon KD1208PTB1 is placed at the back panel. The speed of the fans is varied depending on the temperature. There’s no means of manual control – you can only use the mainboard’s tools to monitor the speed. The PSU is not silent, but rather quiet. The speed of the fans doesn’t change at all at loads below 200W, but then begins to grow up linearly. The sound of the fans becomes annoying only when the load approaches 400W.
The PSU efficiency is average at 80%. The power factor is 0.99 at the maximum as is normal for power supplies with active power factor correction.
The W0023 doesn’t look an appealing buy to me. This power supply exceeds by just a little the requirements of the obsolete ATX12V 1.3 standard, and in such cases you should consider the allowable current on the +12V rail rather than the total wattage of a unit. Here, we’ve got 22A or 264W. It means that in a modern computer that consumes most of its power from the +12V rail the maximum allowable load on the PSU will be not 560W but 264W plus 30-40W more on the low-voltage rails. That is, about 300 watts in total. So, if you are comparing the W0023 with newer power supplies, you should regard it as a 300W unit. Well, an absolute majority of home computers will be quite satisfied with 300 watts of power, but the price of a power supply depends directly on its specified wattage. It means the W0023 is rather expensive in comparison with new ATX12V 2.0 units, yet there is no sense in this overpayment.
Running a little ahead, I want to inform you that I tested two W0049 units that differed greatly in their design as well as quality. For some unclear reason the manufacturer didn’t change the number of the newer model (the marks “revision 1” and “revision 2” are put by myself – the units are both marked as just “W0049”), so you have to distinguish between them by their declared characteristics (which somewhat differ). And I say it will pay to differentiate between the two versions of this PSU model!
The PSU has a dark glossy case and is cooled by two 80mm fans. One fan is in the front panel, the other is at the back. This placement of fans is rare and is most often seen in low-end PSUs. The actual manufacturer of this model is Sirtec.
The internal design of this PSU proved to be very unusual. This is so far the only model I know of that has two independent +12V regulators. In other PSUs there is only one regulator and the splitting into the two outputs is made by limiting the maximum allowable current on each of the lines. The regulator is placed on a separate large PCB (it’s on the right in the snapshot; you can see two parallel-connected diode packs of the rectifier).
But to all appearances this design was employed out of necessity rather than to ensure higher stability of the output voltages (it’s this stability that the manufacturers of PSUs with “virtually divided” +12V lines promise you, although the division of a single internal power rail into several physical outputs doesn’t make anything more stable). They just had to adapt an old high-power server-oriented PSU to today’s realities, i.e. to the ATX12V and EPS version 2.0 standards. This is obvious even from the specification: the power supply can yield a huge current on the +5V rail, up to 50A, but the +12V rail of the main regulator is only rated for 15A. The additional regulator adds two more +12V lines (these are separated “virtually”) with a combined current of up to 23A.
Since the PSU has two independent +12V regulator, it’s interesting to know what connectors each is responsible for. Alas, the enclosed manual makes things only more confusing. It seems that its authors had a very vague notion of the PSU they were writing about, and the designations “+12V1”, “+12V2” and “+12V3” are used in the manual at random or are sometimes replaced with a general designation “+12V” without any number.
So, I opened the PSU and found the following:
The PSU offers the following cables and connectors:
The cables are all sleeved in multicolored braided tubes.
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.
When I was testing the first revision of the W0049 power supply I found that its parameters were quite different to what the Thermaltake website said, although it didn’t mention a release of a new version of the PSU. Moreover, we took the sample from a retail shop, so the explanation that I was dealing with an engineering sample didn’t work. So, we contacted Thermaltake and they were kind to offer us a sample of the W0049 model whose characteristics coincided with the specification you can find at their website.
The PSUs are very much alike on the outside: two fans and a shiny dark case. The single difference is that the newer version has no vent holes in the side panel (it is the panel the label is attached to).
The internal design of this PSU is dramatically different from the previous revision, but it is made by Sirtec, too.
There is a PCB on the heatsinks here, too, but it now carries the low-power section of the PSU electronics (a regulator controller and fan speed controller) rather than an additional regulator. Such PCBs are often installed in perpendicular to the main PCB, but it would collide with the second fan here, so they had to place it at the top. You can see it even in the snapshot that there are no power elements on it (remember the couple of diode packs in the previous PSU?).
The additional PCB is connected to the main one with two groups of contacts.
The PSU follows the classic circuit design with group voltage regulation. Although it has three +12V lines, like the first revision, there is only one +12V source inside it. Then, there are also three shunts and a circuit that controls the passing current. As a result, we’ve got three “virtual” +12V lines that only differ in the over-current protection thresholds.
Some readers have inquired if it’s necessary to load all the three +12V lines for the PSU to work normally. Well, it’s clear that these lines converge into one inside the PSU, so there’s absolutely no difference between putting a load of 4A on both +12V1 and +12V2 lines and putting a 8A load on only one of them and not using the other at all.
In comparison with the previous revision, the allowable load on the +5V line is reduced to 30A (because this PSU was designed from scratch rather than redesigned out of an older model; today, there is just no sense in developing a PSU that allows high consumption from the +5V output), but the total load on the +12V lines can be as high as 52A (624W, which is just a little short of the PSU’s full output power). These changes in the specification are the most noticeable external difference between the first and second revision of the W0049.
The PSU offers the following cables and connectors:
The connectors for additional fans serve an obvious purpose: they are connected to the PSU’s fan speed controller and thus give you an option of automatic adjustment of the speed of all the system fans. What is not quite clear to me is why they have 4-pin Molex connectors. The user will have to search for adapters because an absolute majority of fans have small 3-pin connectors.
The +12V1 line of the PSU powers the mainboard and SATA drives, the +12V2 line is for the graphics cards, the +12V3 line serves the CPU and PATA drives.
Unlike the previous revision’s, the cross-load diagram of this revision of the PSU looks very good. The +12V voltage exceeds the allowable limit in the top part of the diagram – it is 12.6V there – but this is not a big problem since modern computers have low consumption on the +5V and +3.3V rails (within half a hundred watts), so it is the bottom part of the diagram that presents a more practical interest to us.
And it’s all well at the bottom: the +12V voltage is very close to the nominal value (the wide green band is exactly where it should be), the +5V voltage is a little higher than necessary but doesn’t go beyond the acceptable limits, the +3.3V voltage deflects by about 2% from the nominal throughout this entire range (but as you remember, the deflection of 5% is allowable).
At a load of 650W the voltage ripple was 26 millivolts on the +5V rail, 48 millivolts on the +12V rail, 12 millivolts on the +3.3V rail. It’s all normal, the numbers are nearly two times lower the allowable peaks, so the new W0049 is head above its ill-fated predecessor from this point of view.
A T&T 8020H12C NF2 fan is installed on the internal panel of the PSU; a Thermaltake TT-8025A is on the external panel (the real manufacturer of this fan is not disclosed). Although the fans have different sizes (the former is 20mm and the latter is 25mm thick), their speeds coincide with high precision. At loads below 350W the fans are both rotating at about 1300rpm. They are not audible and the PSU is overall very quiet then. When the load is increased further, the speed of the fans grows up linearly to about 3000rpm, but considering that even SLI systems keep within a total power consumption of about 400W, you’ll have to do something unusual to fully load your W0049.
The PSU is 85% efficient at the maximum which is a very good result (but note that it is about 82% at full output power). The power factor isn’t as high as PSUs with active power factor correction usually deliver, but it is acceptable – the difference of 2% (the W0049’s 0.97 against the other units’ typical 0.99) is really negligible.
I tested the PSU with an APC SmartUPS SC 630, and the maximum load on the PSU was 380W when powered from the electricity mains and 325W when powered by the UPS battery. These are normal numbers. The PSU shouldn’t have any problems with UPSes as some other models with active power factor correction have.
And so you can see that the second revision of the W0049 has nothing to do with the first one, except for the name (which is of course strange – why didn’t Thermaltake differentiate between these power supplies more sharply?) This is a high-quality and high-power product with excellent parameters and quiet operation under a wide range of loads.
The second-revision W0049 can be a good choice even for a very powerful computer system, but do not confuse it with the first revision! You can tell them apart by the specified parameters: the newer unit has a smaller load capacity on the +5V rail and a bigger load capacity on the +12V rail.
The TWV abbreviation in the name of this power supply stands for Total Watt Viewer, i.e. indication of the amount of power consumed. What’s curious, the detachable cables of this PSU seem to be a more interesting feature than the power indicator (yes, it looks pretty but has no practical purpose), yet this is not shown in the model name. Moreover, this PSU has independent voltage regulation, which is not emphasized in any way in its characteristics. Thermaltake officially declares this feature for its newer ToughPower series, but not for PurePower.
The PSU has a black matte case and is cooled with one 12cm fan (highlighted with blue LEDs). The actual developer and manufacturer of the unit is Sirtec.
Prominent inside the PSU case are the composite heatsinks which are thick aluminum bars with slanting ribs riveted to their sides. The next thing that may catch your eyes is the three toroidal chokes on the PSU output instead of the usual two. This means independent voltage regulation. There are usually two chokes in ordinary power supplies. One is necessary for group voltage regulation (it helps reduce the interdependence of the output voltages), and the other in the so-called magnetic amplifier circuit regulates the +3.3V voltage (group regulation isn’t sufficient for it). This solution simplifies the overall design of the power supply, but makes the +5V and +12V voltages dependent on each other: the PSU’s regulator aims at a certain arithmetic mean between them.
The addition of one more choke solves the problem altogether: the PSU’s main regulator is now only responsible for the +12V voltage, instead of trying to find the optimal values of the +12V and +5V voltages simultaneously. In an ordinary unit this would result in poor stability of the +5V voltage (the output voltage on this rail would depend greatly not only on its own load, but also on the load on the +12V rail), but there is an additional regulator on the +5V in this power supply, the same in fact as on the +3.3V rail. It’s only the low-power -12V rail that is left unregulated, but the low currents make it possible to put an integrated linear regulator like the LM7912 on it. So in the end the power supply becomes more expensive (the extra choke costs some money by itself and also makes the assembly of the PSU more difficult by occupying quite a lot of room in its case), but its output voltages become much more stable. I don’t claim this is a crucial advantage because a majority of high-quality power supplies yield stable enough voltages throughout all the range of loads, but it’s good to have this extra regulation anyway. Of course, if the circuit is assembled carefully and works well.
The non-detachable cables on the PSU have connectors for the manual fan speed controller and for the indicator of consumed power. The rest of the cables are detachable, although I don’t quite understand why the mainboard and CPU power cables are made such. You always have to use them while the extra pair of contacts reduces the overall reliability of a circuit.
The longest connector has 24 pins, it’s for the mainboard power cable. Next to it, there are two 6-pin connectors to power your graphics cards, and an 8-pin one for the CPU. Then, there are five 4-pin connectors for PATA drives – these have only two voltages, +5V and +12V. The cables for SATA drives are complete, with +3.3V voltage besides the rest. The SATA power cables are attached to the PSU with two connectors, with a main 4-pin one and a separate 2-pin one (there are two such sockets below) through which the +3.3V voltage is supplied.
The box with the PSU contains:
Besides the cables, there is a module to control the speeds of two fans and to display the amount of power the PSU consumes. This module is inserted into a 3.5” bay of your system case and it resembles the module included with the W0023. Unlike Zalman’s controllers, for example, this one is only meant for work with fans that have a dedicated connector for speed management. To put it plain, it only supports fans from Thermaltake, and one such fan is included with the power supply to make your life easier. It is a 12cm TT-1225A (Everflow R121225SU, 2900rpm) without highlighting. I tried it out and found that its speed was varying from 1250 to 2850rpm. This is quite a high speed for a quiet computer. Enthusiasts usually slow the fans of their systems down to 800-900rpm. At this speed the fans are practically silent but produce enough airflow to cool the system case. At speeds above 2000rpm any 12cm fan is noisy just because the strong airflow they create produces a hissing sound. So, if you want to assemble a really quiet computer, you are unlikely to be satisfied with the speed controller included with the W0057. It’s impossible to make the fan you receive with it quiet (this is the most powerful model in Everflow’s 12cm fan line-up), while replacing it with another fan is going to be problematic because the controller supports only a few select fan models.
The 7-segment display of the module shows you the amount of power consumed. Such devices generally lack precision, but this sample did quite well. In a range of loads from 50 to 300W its showings were indeed accurate. At higher loads the indicator would show a 30-40W higher value, while at lower loads its showing would drop not to 0, but to 10W.
This power supply complies with the ATX12V 2.0 standard. The maximum load on the +12V rail can be as high as 36A (432W). The +12V2 line is a separate item in the last line of the label where the combined allowable loads are specified, although the +12V power rail is split in the W0057 in the traditional “virtual” way, so it doesn’t matter for the PSU which line, +12V1 or +12V2, is loaded if the combined current in them is below 36A. These lines aren’t connected with the +5V and +3.3V lines, so I don’t grasp the purpose of specifying a combined load for the +3.3V, +5V and +12V1 lines.
The power supply has no problems outputting voltages under any permissible load, but unfortunately its cross-load characteristic isn’t perfect despite the independent regulators. The +12V voltage is set so high that it is close to the acceptable limit of 12.6V under zero load. At the maximum load it is still 3% above the nominal value. The +5V voltage is very high, too, but is also not very stable: the graph changes from a precise coincidence with the nominal to a 5% deflection which is a rather poor result for a power supply with independent voltage regulation. Of course, the W0057 looks good against many other models, but I had hoped to see some better results from it.
At max load the output voltage ripple was 35 millivolts on the +5V rail and about 65 millivolts on the +12V rail (the allowable maximums being 50 and 120 millivolts, respectively).
I disabled the manual control unit before measuring the fan speed. Alas, the PSU is far from quiet. The fan starts off at a speed of 1200rpm and quickly accelerates as the load grows, reaching 2000rpm under a load of less than 200W. You can compare this result with those of really quiet power supplies like the Zalman ZM460-APS (460W) or the Seasonic S12 SS-500HT (500W) where the fan speed is a little over 700rpm initially and is not higher than 1500rpm even at the max load.
The efficiency of this PSU is low by today’s standards, reaching only 77% at the maximum. It’s no better with the power factor which barely reaches 0.95 while PSUs with active power factor correction typically have a power factor of 0.98-0.99.
I can’t say the W0057 is bad, but it is rather expensive for ordinary users (it is selling in retail stores for about $100) and unexciting for PC enthusiasts with the parameters it has. The PSU is rather noisy at work and this cannot be cured by the manual controller – it doesn’t allow reducing the speed of the fan below 1200rpm which actually seems to be very close to the bottom limit of stable operation of the powerful fan employed in this PSU. This speed controller is also useless for a second fan because a rather powerful model is included with the PSU and you cannot slow it down to below 1350rpm. It can only be replaced with Thermaltake’s fans explicitly supporting such speed controllers. The selection of such models isn’t large and includes only powerful models with a high maximum speed, and you can’t slow them down to silent level, either. So, if you need a quiet power supply, you should consider other models. Even lacking a manual fan speed controller, many of them are quieter than the W0057 but cost the same or lesser money.
Resembling the previous model from the outside, the W0063 has a slightly different design of the case. It is made by HEC/Compucase whereas the W0057 is manufactured by Sirtec.
The PSU is very neatly assembled. It has an active PFC device and a classic circuit design with group voltage regulation. The vertically positioned card in the middle of the case carries the regulator and PFC controllers. A little lower, on the main card, there resides a standby source on a special-purpose TNY267P chip. There is yet another card near the right panel of the PSU, which carries an output voltage supervisor (on a Weltrend chip) and a fan speed controller.
The declared characteristics comply with the ATX12V 2.0 standard: the combined current on the +12V rail is as high as 30A whereas the +5V and +3.3V have an even higher load capacity than required (the max load on these rails was greatly lowered in the latest versions of the ATX power supply specification).
Although the input voltage is given with a slash as “115/230V”, the PSU doesn’t have an input voltage switch, just like any other power supply with active PFC: such PSUs can support either one input voltage or a whole range of voltages from 90 to 265V without manual switching (support for only one voltage is implemented as a means of making the PFC device cheaper; this voltage is always 230V because it’s low input voltages that put the PFC circuit under greater load).
The PSU offers the following cables and connectors:
The cross-load diagram for this power supply looks rather good, but the instability of the +5V voltage raises my apprehensions. This voltage is too high in the area the power consumption of modern computers corresponds to (the bottom right of the diagram). The +12V voltage, on the contrary, deflects no farther than by 2% from the nominal value in this area.
Unfortunately, the output voltage ripple is high at maximum load: it is a little higher than the acceptable 50 millivolts on the +5V rail. On the +12V rail the voltage ripple was 70 millivolts, which is below the acceptable limit of 120 millivolts. Take note of the oscillation period: its frequency is about 100kHz which is thrice higher than in typical power supplies. The higher operating frequency allows using smaller ferrites (transformer and chokes), thus simplifying the PSU’s component layout and improving its cooling, but the downside is that the higher the frequency, the stricter the requirements to the components are. That’s why such power supplies are rather rare. I can only recall the new series of PSUs from FSP Group that we have tested earlier in our labs - they had an operating frequency of 110kHz.
A Thermaltake TT-1225A fan (12V, 2.8W) is installed in this power supply. Its real manufacturer is Young Lin Tech., Ltd.
The PSU is not quiet; unlike in the W0057, its fan speed doesn’t change much as the temperature grows up, but the average speed is high enough for you to hear the hiss of the air. Well, the PSU might be very, very quiet even in comparison with non-Thermaltake models if the fan speed was growing up in a linear manner from somewhere like 700-800rpm (obviously, if the fan rotating at 1300rpm is capable of cooling the PSU at a load of almost 400W, its speed can be greatly reduced for a 150W load; even considering the lower efficiency under low loads, there’ll be over two times less heat dissipated in the case at 150W than at 400W).
The efficiency of this PSU is 81% at the maximum and below 80% on average. This is an average result. The power factor is good at 0.99.
Thus, the W0063 is a rather typical midrange power supply that doesn’t stand out among its competitors with any of its parameters. It offers a standard selection of connectors, produces an average amount of noise, and yields sufficiently stable voltages. On the other hand, there’s nothing I could find fault with. The high output voltage ripple will hardly be a problem in practice because it exceeds the required limits only by a little at full load and becomes weaker at low loads.
This PSU will fully satisfy many users who need just a well-made power supply without any extras. If you need maximum stability or silence, you may want to look at other PSU models (a manual modification of this PSU to have the fan speed controlled more efficiently is beyond the scope of this review).
It’s difficult to tell this power supply from the previous model visually as well as by the model name or even by the parameters. However, they are two different products. Particularly, the W0068 has an input voltage switch which means that it lacks active PFC (running a little ahead, I want to inform you that it lacks passive PFC, too). Another difference – you can see it when you turn the PSU on – is the blue highlighting of the fan.
The actual manufacturer of the PSU is again HEC/Compucase.
The internal design differs from the W0063 more. The different heatsinks should catch your eyes in the first place, but on closer inspection you can also find that this power supply is based on a different PCB (which is logical since the active PFC device was located on the main PCB in the W0063, while the W0068 lacks any power factor correction).
However, the W0068 employs in fact the same components and it shouldn’t matter much for the end-user how these components are placed on the PCB. The circuit design is unchanged, too. The PSU is assembled in an ordinary way with group voltage regulation.
What’s more important, the declared output parameters of the PSU are unchanged – it has exactly the same load capacity as the previous model. Of course, it now makes sense to write the output voltages with a slash since manual switching between them is required, but this is hardly of any interest for the user because the input voltage is usually selected once and for the entire lifetime of a power supply (and if it is set wrongly – the lifetime proves to be just too short :)).
The PSU offers the following cables and connectors:
So, the only difference from the W0063 is the configuration of the power connectors for hard and optical drives, but their total number is the same. Most users won’t even notice this because two or three such connectors are usually quite enough for a regular PC system.
The cross-load diagram for the W0068 looks good except that the +5V voltage might be more stable. Well, that’s nothing but fault-finding because there are going to be no problems with the voltages in practice.
Both low- and high-frequency pulsation can be observed on the PSU’s output. At a load of 450 watts the voltage ripple is 25 millivolts on the +5V rail, 105 millivolts on the +12V rail (over half of this pulsation is at the double mains frequency, i.e. 100Hz), and 35 millivolts on the +3.3V rail.
The PSU uses a Young Lin Tech fan, the DFS122512M model. Its speed is controlled in an linear-like manner, but the PSU is not quiet. The fan speed is about 1100rpm at min load and then grows up to 1800rpm at high loads. The W0068 will suit you if you are not very fastidious about noise, but people who seek complete silence may not like it.
The efficiency of this power supply is very close to the 80% mark, but doesn’t go beyond it. The power factor is quite typical (for a PFC-less unit) at 0.68. I want to note that this parameter is only important for power companies and large organizations with a lot of computers. A low power factor leads to extra loss in the power system, i.e. in the wires, and may even lead to problems associated with distortion of the shape of the line voltage in some cases. This is not a problem for home users. PFC will only make a difference if the line voltage is unstable in your area, but you need an active PFC device with a full range of input voltages, from 90V to 265V, then.
So, the W0068 doesn’t differ in anything important from the above-described W0063. The same design, the same load capacity, and rather close parameters… This PSU is a mainstream model. Its parameters are not exceptional, but it is neatly designed and provides good enough parameters as to satisfy a majority of users. But like with the W0063, if you need as quiet a PSU as possible, you should consider other models. Despite the efficient fan speed management, the W0068 is not quiet.
The choice between the W0063 and the W0068 depends on whether you need active PFC. On one hand, most users don’t actually need it, but on the other hand, the price difference between these two models is only about $3. Of course, the W0068 has slightly better cross-load characteristics than the W0063, but the difference is small and may vary in this range from one sample of the same PSU to another.
This model doesn’t have a serial number or any marketing name since it ships only bundled with Thermaltake’s system cases (Tsunami Dream, Soprano, etc). However, I have a reason to place it in exactly this place of the review, before the TR2 XP550 (W0070) model. They are obviously much alike to each other in their internal design as well as in their model names. So, the XP480 can be viewed as a junior model in the low-end TR2 series (but speaking formally, it lacks “TR2” in its name and is not described in the corresponding section of the Thermaltake website).
This PSU has a simple gray case with two fans. Its actual manufacturer is HEC/Compucase.
The PSU offers the following cables and connectors:
It’s roomy inside this power supply: the components are all of modest size, the heatsinks are simple bars with punched-out “fingers” on top. The assembly quality is high, however, just as you can expect from HEC, so I can’t have any complaints here. The circuit design is classic, with group voltage regulation and without PFC. The main regulator is based on a UC3843B chip; the output voltages supervisor is based on a HY510N chip. The marking on the PCB – a table that tells which fuses to use for units with different wattage – set me aback as I couldn’t find wattages higher than 300W in it. The PSU label has a somewhat different number in large letters, hasn’t it?
At about the same time as I was testing the XP480 I bought a cheap microATX system case Ascot 6KR2 for my home computer (this is an ordinary HEC/Compucase shipped under the Ascot brand for a Russian distributor). It came with a 200W HEC 250AR-TF power supply and I was greatly surprised to find that the circuit design and the components used in that PSU were an exact copy of the Thermaltake XP480! The only difference between the two was their cooling: the 250AR-TF is cooled with one fan while the XP480 with two fans. Well, the extra cooling may help the transformer and the choke coils in the XP480, but its transistors and diode packs won’t feel any difference. They had to make the heatsinks smaller in the XP480 in comparison with the 250AR-TF to put in the second fan.
So, what is the real wattage of the XP480?
The label promises a sustained output power of 400W and a peak load of as much as 500W, but the currents don’t reach the level of the standard 300W unit as described in the ATX12V 1.3 specification – it must have a load capacity of 18A on the +12V rail whereas this PSU offers one ampere less there. On the other hand, the typical 250W power supply of the ATX12V 1.3 standard has a +12V load capacity of 17A, i.e. exactly as in the XP480. That’s another meaningful coincidence after the circuit design being identical to the 250W HEC.
The cross-load diagram of this PSU looks well, except that the +5V voltage is a bit too high when there is a low load on this rail. It doesn’t go beyond the acceptable limits, though, deflecting by no more than 4% from the nominal value (a 5% deflection is considered acceptable). It’s only in the top right of the diagram – where the load power is 310-320W – that the power supply behaves in an odd way. Do you see those multicolored patterns?
After the building of the cross-load diagram, the next text in our PSU-related reviews is the measurement of the output voltage ripple. These measurements are in fact made as the final step in the test procedure because I measure the voltage ripple under the maximum load on the PSU. Before doing that I am increasing the load in steps of 50-100W and leave the PSU to heat up for 20-30 minutes under each load. After that I measure the power factor, efficiency, and fan speed (the heating up is actually necessary for the fan speed to become constant). And it’s only at the maximum load power that I attach our oscilloscope to the PSU’s output.
Alas, the XP480 just didn’t make it to that point. When the load exceeded 310W, the output voltages suddenly dropped by 10-20% (I took a final look at the multicolored patterns in the cross-load diagram just in this area) and the overload protection was triggered in the PSU at a load of 320W (I say, where is the promised peak of 500W?!) My further attempts to find a maximum load under which the PSU would be stable ended in a loud plop inside the PSU case and the power supply shut down. For ever. The autopsy showed that the switching transistor of the high-voltage inverter had actually exploded in two.
So, my apprehensions were well grounded and I think it’s a proven fact that the XP480 is a remarked 250W unit from HEC. Contrary to the impressive specification, the XP480 just cannot yield more than 300W of power. 250W is the maximum load the XP480 is sure to sustain for some long time, not for just five minutes. But let’s get back to the tests. Perhaps it will do as a 250W unit?
The PSU has two cooling fans; their speed is adjusted in an almost linear manner until about 250W. After that the fans work at their max speed. By the way, this is indirect evidence that the PSU was initially meant for a peak load of 250W.
The efficiency of this PSU isn’t high, being only 75% at the maximum and closer to 72% on average. This is higher than the standard demands (ATX12V 1.3 power supplies must have an efficiency of no less than 70% at full load), but isn’t impressive at all in comparison with modern models whose efficiency is closer to 80%.
The power factor is low at 0.68, but this is quite a typical number for a model without any kind of power factor correction.
I’d like to divide my impressions about this power supply in two parts. One part is concerned with the 250W unit from HEC and the other with the 400W label from Thermaltake.
As an inexpensive 250W power supply for office computers and even midrange home systems the XP480 looks quite good. It has the necessary connectors (even too many of Molex ones); its parameters are good and it is quiet (particularly, much quieter than the original HEC 250AR-TF equipped with one cooling fan). So, there is nothing to cavil at. It doesn’t look impressive against today’s monsters with twice the output power, but do you really need hundreds of watts if there is no top-end graphics card and high-consumption central processor in your computer? Not to run long for examples, I am now writing these words on a computer with an Athlon 64 3800+, a Radeon X800 XL graphics card, one gigabyte of system memory, one hard drive and one DVD-burner, and all this stuff works without problems in a barebone system case with a 250W power supply from Enhance with a maximum current of only 16A on the +12V rail. I guess this configuration is the home mainstream for today, let alone office machines that often lack even expansion cards because everything is integrated into the mainboard. The XP480 suits well exactly for such computers.
However, Thermaltake thought it necessary to stick a label on this power supply that declares an output power of 400W and even promising a peak output power of as much as 500W! But as I made sure in my tests, even 300W is too high a load for this power supply. Depending on the particular conditions, either its overload protection reacts, or the attached system hangs up due to a sudden drop of the output voltages, or the unit just burns down. And the XP480 comes in system cases that can hardly be called cheap: the retail price of a Tsunami Dream is about $140. This case is unlikely to be bought for a low-power office machine I guess.
For me, the problem is not even that the power supply isn’t up to the class of the system case it comes with, but that the customer is overtly lied to. The wattage specified on the PSU misleads the user as to how much power his/her system needs. Imagine a person that buys a Tsunami Dream and assembles a gaming system in it with two graphics cards and a top-end processor. He hears a plop on first turning the system on, opens the case, looks at the label that reads “400W” and comes to the conclusion that this is too low wattage for his computer! And this user then goes to buy something like a 680W unit, for example the not-very-cheap W0049. And this user will be lucky if he gets the second revision of this W0049 because otherwise he may find that 680W is not enough for his system, too! So now you see where those stories that a power supply of 700W and no less wattage is required for a gaming computer come from. And all these watts cost you money whereas a power supply with an “honest” 400 watts of output power would be more than enough.
I mentioned this model explaining the position of the XP480 in this article. The XP480 doesn’t have an official model number whereas the XP500 belongs to the TR2 series and has a model number of W0070. The TR2 series comprises inexpensive PSU models from Thermaltake that are not worthy of the PurePower brand.
The XP480 and X550 have a remote semblance. You can guess common design features, but the cooling fans are placed in a different way in the XP550. There are two of them here, but they are now located on the opposite sides of the case (that’s why you can’t see the second fan in the snapshot above).
The PSU is manufactured by HEC/Compucase.
Everything is easily recognizable on the inside, though. The internals of this PSU are the same as were installed in the XP480, except that the heatsinks are larger here (they don’t have to be adapted for the second fan that used to hang from above) and the input capacitors are now 680µF rather than 470µF. Otherwise the circuit design is identical. To all appearances, this is a version of the 300W HEC unit re-marked by Thermaltake as having 430W wattage. Here, the second fan may help the PSU a little – they had to make the heatsinks smaller in the XP480 to find place for the second fan, but the heatsinks are full size here and should cool the semiconductor elements better.
The PSU offers the following cables and connectors:
There is again an abundance of Molex connectors for PATA drives, but a second SATA connector and a graphics card power connector have been added.
480W wattage is declared for this power supply, and there’s no mention of the peak output load on the label. The load currents aren’t much higher, though. The most important +12V line is now declared to provide a 1A higher current.
The cross-load characteristics of this PSU are similar to the XP480’s, which is logical since they have the same circuit design. The +12V voltage is stable, the +3.3V is very good, but the +5V voltage is a little higher than necessary. And once again I want to draw your attention to the rainbow patterns in the top right of the diagram where the PSU worked at about 380W – the +5V voltage is suddenly very low there.
I couldn’t measure the output voltage ripple for this PSU due to the same reason as with the XP480: half a hundred watts short of the declared maximum output power, at a load of about 380W, the PSU’s switching transistor overheat and burned down in the power supply. It was so hot that the insulating plastic spacer under the screw the transistor was fastened with melted.
The PSU has two 80mm fans connected to the same speed controller. So, their speeds coincide with a pretty high precision. The speed is adjusted in a linear manner, with a bend in the middle of the graph. The PSU is not silent – and you can’t expect a low-end model to be such – but it is quiet enough at everyday work as to provoke no complaints from most of users.
This PSU is more efficient than the XP480, but not too much – the graph touched the 80% line only once in the diagram. The power factor is low due to the lack of PFC and is about 0.65 on average. The PSU allows to install a passive PFC device and it is indeed installed in the W0069 model whose power factor should be higher at about 0.7-0.72.
So, like its predecessor, the XP550 isn’t true to the declared wattage. This is in fact an ATX12V 1.3-compliant 300W unit from HEC reinforced with a second fan. This may make it a 330W unit, but not better. The PSU just burns down at a load of 380W due to overheat of the switching transistor.
So, the W0070 can be viewed as a good low-end 300W power supply with accurate assembly, good parameters, and quiet operation… But on the other hand, Thermaltake put a label that reads “430W” on this product and this number has nothing to do with reality. This PSU cannot yield this output power for more than a couple of minutes. What’s the most disappointing thing, Thermaltake prices this PSU by the declared rather than real wattage and it costs now $45 or even more (unlike the XP480, the XP550 comes to retail shops apart from system cases). I think it’s unfair to put so high a price on a mediocre power supply built to comply with an obsolete standard.
Next to the low-end W0070 there stands the expensive W0073 which belongs to an opposite market segment. It is an enthusiasts-targeted product.
You can correctly guess the real manufacturer of this power supply – HEC/Compucase – even by its appearance. The PSU has a black case and a single 12cm fan with blue highlighting. Well, it’s not only the fan that’s highlighted here as you’ll see soon.
Yes, we’ve seen this before! This is a copy of the W0063, at least the stuffing is absolutely identical to it: the same heatsinks, the same electronics, active PFC, and group voltage regulation…
You will find the difference in the rear part of the PSU where a separate card with a few connectors resides under a small plastic casing. Is it yet another PSU with detachable cables? Not quite.
There are only three connectors at the back. This is too few even if the mainboard and CPU power cables are not detachable. Two of these connectors are 6-pin for graphics cards, while the third and most interesting one has 8 pins.
You get a so-called Power Station with this power supply. This station goes into a 5.25” bay of your system case and is attached via a special cable to the PSU’s 8-pin connector.
It’s at the rear panel of the Power Station that the connectors for peripherals are placed: 6 connectors for PATA drives, 4 connectors for SATA drives, two for floppy drives. This solution – one cable goes from the PSU to the 5.25” bay and then the power cables for the drives spread out from this bay – is meant to simplify system integration, improve airflows in the system case, etc.
The Power Station itself is an empty box without any electronics. There’s a card with the connectors at its rear panel; on the front panel there are three multicolored LEDs that indicate voltages on the PSU’s output: green is for +3.3, red is for +5V and yellow is for +12V.
If the power supply works normally, it always yields all these voltages at once, so you’ll have all the three LEDs of the Power Station shining when your computer is turned on. It means that this is just a piece of pure decoration without any functional load. Enthusiasts may find some other use for the neat 5” box provided by Thermaltake and put something more useful in there, like a manual fan speed controller, for example.
At the back panel the Station’s connectors are highlighted with a couple of blue LEDs. One might say this helps not miss the necessary connector, but the LEDs don’t shine when the computer is turned off. And when the computer is working, I wouldn’t recommend touching those connectors at all.
The connectors on the power supply case are highlighted in a similar fashion.
Besides the big Power Station, the PSU comes with a smaller Power Station Mini. This small box is meant for powering additional fans (it can be bought separately, by the way). This box is connected to a Molex connector and can supply power to four fans with Molex connectors and two fans with the more widespread 3-pin plugs. Alas, no fan speed control is provided; there is always full +12V voltage on all the connectors.
The Power Station Mini doesn’t have any set place to be fastened at; it can be fixed anywhere in the system case by means of dual-sided scotch tape. Note that you should attach only fans to it. The Power Station Mini is not meant for such a load as hard/optical drives, let alone graphics cards.
And here’s a list of cables you receive with the PSU to connect all this stuff together:
The cables have identical connectors on both of their ends, so it doesn’t matter with which end you connect them to the Power Station and with which to the powered devices.
Besides that, the PSU itself has two non-detachable cables:
The electrical parameters of the PSU coincide with those of the W0063, except for the increased total wattage. The maximum load capacity of its +12V rail is 30A or 360W. It means you won’t be able to put a higher load on the W0073 than on the W0063 in practice because modern computers, with their low consumption from the low-voltage rails, will be limited by the load capacity of the +12V rail which is identical in these two power supplies.
The cross-load diagram for this power supply has quite an ordinary look: the +12V and +3.3 voltages are close to their nominal values, and the +5V is a little too high. It’s all normal here, without any serious problems.
At a load of 500W the voltage ripple was 25 millivolts on the +5V rail, 44 millivolts on the +12V rail, and 19 millivolts on the +3.3V rail.
A DFS122512M fan from Young Lin Tech with blue highlighting is installed in this PSU. Its speed is nearly constant at a little over 1200rpm at loads up to 250W and grows up linearly at higher loads. This resembles the behavior of the W0063’s fan, but the linear growth of speed began later in the W0063 and only a small stretch of it fitted into the diagram (I guess this is due to variations in the parameters of the employed components since the circuit design is identical in these two PSUs).
The W0073 is not very loud, at least it won’t be perceived as irritating by most of users. It is however not exactly silent. There is an audible hiss of air at the min fan speed which is over 1200rpm. So, if you do care about the noise produced by your computer, you may want to consider other models. Other users are likely to be quite satisfied with the noise parameters of the W0073.
The efficiency factor graph is similar as the one of the W0063 with a maximum of 80%. The average efficiency is about 78-79%. The power factor is high, as is expectable from a power supply with active power factor correction.
Generally speaking, the W0073 is a full analog of the W0063 plus the Power Station. These two models have the same circuit design and nearly coinciding real parameters, while the higher specified wattage of the W0073 isn’t crucial because it offer the same allowable load on the +12V as the W0063 – the rest of the lines do not matter much for modern computers. However, the W0073 is considerably more expensive at about $135 against the W0063’s $75. And this $60 difference buys you only the two Power Station boxes and the detachable cables. It’s up to you to decide whether they are really worth it!
After a few models manufactured by HEC/Compucase, here is yet another Sirtec. This power supply has a dark shiny case and is cooled with a single 12cm fan.
This PSU doesn’t seem to resemble any of the earlier tested ones at first sight, but on closer inspection you can notice that its circuit design and components are the same as in the above-discussed second-revision W0049. You say there’s no second card on the heatsinks? Well, it just now stands upright along the rear panel and is secured on its own connector which was straight and long in the W0049, but is now short and L-shaped. Like the W0049, this power supply is based on a CM6800G chip, has an active PFC device, and supports group voltage regulation.
The declared load currents are the same as the W0049’s, which is somewhat odd. Calculate the total allowable load on the +12V rail. It is 52A or 642W, i.e. higher than the full output power of the PSU. Well, let’s assume that there’s 50A (50A * 12V = 600W) there, and they just forgot to write the combined limitation for all the three +12V lines on the label (I said it a number of times in my earlier reviews that it’s the combined output power of all the +12V outputs that matters in a power supply with “virtual” division of the +12V power rail into several outputs; the current limits on each of these outputs is not important at all).
More about limitations. The division into several output lines was introduced to comply with the safety standard that demands that there be no more than 240VA on user-accessible lines. In other words, the current shouldn’t exceed 20A for a voltage of 12V. They usually limit the voltage at 18A to have some safety margin, but here the +12V2 line has a limit of 23A, so the power supply doesn’t formally comply with the mentioned safety standard.
The PSU has the following cables and connectors:
The cables are sleeved.
The cross-load diagram looks good. The W0083 easily copes with any permissible load. In the most important section of the diagram (the bottommost third of it), the +12V voltage is close to the nominal value while the +5V voltage is slightly higher than necessary, but doesn’t violate the allowable limits.
At a load of 550W the voltage ripple is 33 millivolts on the +5V line, 50 millivolts on the +12V line and 30 millivolts on the +3.3V line. Most of this ripple is due to 100Hz pulsation (the doubled frequency of the electricity network in our area) whereas high-frequency pulsation is not very intensive.
The PSU employs a Hong Sheng A1225S12D fan. Its speed depends but slightly on the temperature at loads below 200W, remaining always at about 1250rpm. The speed grows up rapidly at higher loads, getting as high as 2500rpm. So, the noise parameters of this power supply are average, just like those of many other models from Thermaltake. The W0083 is not exactly quiet or loud. Its fan is audible, but most users won’t mind that. The fan is fast under high loads, but this usually means that the owner of the computer plays some heavy 3D game and is not likely to bother about the noise from the power supply.
The efficiency of this PSU is expectably good (the W0049 was efficient, too) at about 83% on average. The power factor is not very high at 0.95 in comparison with other models with active power factor correction, but any PSU with active PFC is much better in this respect than PFC-less units and units with passive PFC that have a power factor of 0.65-0.68 and 0.7-0.75, respectively.
So, the W0083 is a variation of the second-revision W0049 with slightly lower wattage and a 12cm cooling fan. It costs $5-10 less than the W0049 but doesn’t differ from it in the actual parameters, especially considering that most computers just don’t need the extra wattage of the W0049. What’s interesting, the W0049 with its two small fans proved to be quieter at work than the W0083 with a single large fan. I think the fan speed is managed not optimally in the latter model. They could have set the speed lower without worsening the cooling of the PSU.
If you haven’t grasped the idea that model numbers don’t mean much in Thermaltake’s power supply nomenclature, you may be a little surprised to see that after discussing two high-wattage and expensive products we now return to the low-end TR2 series. This time it is a newer, ATX12V 2.0-compliant model.
There’s no use to try to guess the actual manufacturer of this power supply by comparing it externally with the above-tested models. The W0093 is made by Channel Well, the supplier of PSUs for Foxconn and Antec. So, this is a fourth Thermaltake’s PSU contractor after Sirtec, HEC and Seventeam that we know about.
The PSU looks quite ordinarily for an inexpensive model (relatively inexpensive I should say; yes, the W0093 belongs to a low-end series, but its price is $70-80 due to its high wattage). I don’t have any complaints about the quality of this modest design, though. The PSU is equipped with passive PFC (the large coil in the top left of the photograph).
Quite a lot of the PSU’s electronics reside on a separate card that is placed near the side panel. The case has an L-shaped rather than U-shaped cover, i.e. one side of the case is always closed.
There is a two-color LED next to the Turn on/off switch and the input voltage switch. The LED shows the operation mode of the PSU as “turned on” and “standby”. Make sure the input voltage switch is set at the correct position when you’re installing this power supply because the switch isn’t protected against accidental presses and it’s just too easy to set it wrong with fatal consequences for the PSU.
There are additional vent holes in the side panel. A rather hot stream of air comes through it when the PSU is working – the output diode packs and the group regulation choke are located nearby. Some users hold an opinion that the power supply should never blow hot air back into the system case, but you shouldn’t block these vent holes. This would greatly worsen the thermal conditions inside the power supply while the computer doesn’t feel much worse for a weak stream of hot air.
The PSU is equipped with the following cables and connectors:
The cables are all sleeved.
The PSU offers two +12V output lines (“virtual”, of course) with a combined current of 29A (348W). What’s interesting, the original model name by the CWT catalogue is mentioned – PSF450S-30 – and the number 450 in it gives me some suspicions about the real wattage of this power supply.
The cross-load characteristic of this PSU is very good expect for the +3.3V line whose voltage changed rather too much depending on the load, although it always remained within acceptable limits. The +5V and +12V voltages are very stable. Moreover, none of the voltages deflected by more than 5% in the whole range of loads allowable for this PSU. As a result, the cross-load diagram is shaped like a regular rectangular which is usually observed with power supplies that have independent voltage regulation.
Alas, the PSU couldn’t yield its promised 500W of output power – its protection was triggered at a load of about 475W. However, it escaped the ill fate of the XP480 and the XP550 models and worked normally under any loads below 475W. I guess the number 450 in the name of the original CWT model is indeed true and Thermaltake shows some cunning here, declaring a pretty-looking but wrong number in the specs.
At a load of 475W (the highest load the PSU was stable at), the voltage ripple was 45 millivolts on the +5V rail, 73 millivolts on the +12V rail, and 26 millivolts on the +3.3V rail. So, the ripple was close to the allowable maximum of 50 millivolts in the first case; in the other two cases there is a solid reserve yet. Only high-frequency pulsation was observed on the +3.3V rail; both low- and high-frequency pulsations were on the remaining two rails, the low-frequency constituent prevailing.
The PSU has a TT-1225A fan manufactured by Adda (the AD1212US-A71GL model). A transparent piece of plastic covers one half of the fan. One of its ends isn’t fastened and contributes to the PSU’s noise. As a result, the PSU produces a quiet but distinct humming at any load.
The fan speed is controlled in Thermaltake’s traditional way. It’s even funny that the fan speed management doesn’t vary much between the different actual manufacturers of Thermaltake’s PSUs. The fan rotates at about 1150rpm under loads below 200W, but then its speed begins to grow up linearly to 2500rpm. So, its noise parameters are average. It is not exactly quiet, but will satisfy a majority of users nonetheless, especially if you take care to secure the mentioned piece of plastic well…
The efficiency of this power supply is unexpectedly low at 68% under maximum load and 76% at best. The power factor is never below 0.7 thanks to passive power factor correction, but a high efficiency would be of much more use for home users than a high power factor.
Summing it up, I’d say the W0093 is a well-made midrange product, but its real wattage is 450W rather than 500W. It delivers stable voltages, works quietly and offers all the necessary connectors. On the downside are its low efficiency and rather strong low-frequency pulsation on the output, but this will hardly be a problem for ordinary users. Thermaltake has also greatly improved since the XP480 and the XP550, overstating the wattage of this PSU by only 50W rather than by 50%! :)
The Toughpower series is new in Thermaltake’s product nomenclature and includes expensive power supplies whose independent voltage regulation as the emphasized feature (before this, top-end and midrange power supplies from Thermaltake belonged to the PurePower series; low-end models to the TR2 series). But we’ve already found out that the PurePower W0057 has independent voltage regulation, too. It’s just not touted there.
Although the PSU doesn’t look like the low-end W0093, it is manufactured by Channel Well, too. The UL certificate numbers – it’s the easiest way to determine the real manufacturer of a power supply without opening it up – are different for the W0093 and W0097 but are both registered by CWT. The PSU has a black matte case with a large aluminum label on a side that reads “Toughpower” (Thermaltake seems to have a fancy for this kind of labels, putting them on many of its power supplies).
The internal design of this power supply is going to look familiar to you if you’ve read our reviews of Antec or Foxconn PSUs also manufactured by CWT: the characteristically shaped heatsinks with that large rectangular white resistor on one of them (this resistor works as a load for the +12V channel so that you could start the PSU up without external load), and there is a sticker “550A” on the transformer that provokes associations with the Foxconn FA-550A model.
The PSU doesn’t support power factor correction at all. Thermaltake also offers the Toughpower W0096 model which is equipped with an active PFC device.
This power supply is declared to have an output power of 550W (20W of which fall on the -12V output and the standby regulator). The maximum combined current on the +12V power rail is 36A (or 432W – take note that this is smaller than the sum of the limits for the separate +12V lines which is 38A). What’s interesting, the +5V rail allows a load current of up to 40A, which is generally untypical for modern ATX12V 2.0 power supplies. The requirements to the load capacity of their low-voltage rails are not strict in the new standard, so the allowable current on the +5V rail is rarely even 30A.
The PSU is equipped with the following cables and connectors:
A curious fact, the mainboard power cable has not one sense wire as required by the standard, but three, on all the highly loaded lines (you can see two sense wires in the snapshot above, the third one, for +3V, is on the reverse side of the connector). This sensing is necessary to compensate the voltage drop on the cable wires which may be considerable at high load currents.
The cross-load diagram for this power supply looks nearly perfect. It is predominantly green; the voltages never go beyond the acceptable limits in any point of the diagram, so the boundary of the diagram is determined by the specified load capabilities of the PSU (for example, the slanting top right of the diagram is the limitation of the maximum output power at 530W, this line is wavy due to bilinear interpolation of the graph from separate dots into a solid fill).
It’s no worse with the output voltage ripple. There is almost no low- or high-frequency pulsation. At a load of 530W the voltage ripple is a mere 15 millivolts on the +5V rail, 30 millivolts on the +12V rail and 10 millivolts on the +3.3V rail.
The PSU is equipped with a TT-1225A fan (it is actually manufactured by Adda, the AD1212UB-A73GL model; note that the W0093 employs a fan marked as TT-1225A, too, but it is quite a different model!). The dependence of the fan speed on the load is not linear, and resembles an exponent, rather. The speed grows up quickly as the load is getting higher. The speed ceiling is low, however, at less than 2000rpm, so the PSU is not noisy even under maximum load. On the other hand, it is not absolutely noiseless. The fan rotates at about 1050rpm at the minimum. This is good, but you can meet better (for example, the minimum fan speed is below 750rpm in the Zalman ZM460-APS, the FSP Epsilon, and in the Seasonic S12).
The efficiency of this PSU is average, reaching 80% at the maximum. The power factor is low, about 0.65 on average, due to the lack of any kind of power factor correction.
The Toughpower left a good overall impression on me. This is a high-quality product with very good parameters, maybe except for its power factor, but this parameter is not very important for ordinary users. It has all the connectors you may need, delivers stable voltages with low pulsations, and is quiet at work. I should note, however, that the W0097 is not at all cheap. Its price of $120 is quite a lot for a 550W power supply.
Thermaltake’s power supplies leave one with a whole spectrum of emotions. Well, there could hardly be one emotion and one opinion because the company’s model range includes products of different classes and manufactured by four different brands. Generally speaking, Thermaltake’s power supply nomenclature lacks order, including completely different models with unsuggestive names. For example, the W0063 is a high-quality midrange product, but the W0070 is a low-end model that is not worth half the money asked for it. The W0073 is the same W0063, but with a couple of extra accessories in the box. So I can’t find any logic in the model names. It’s better with the marketing names of the products: the letter index indicates the type of the power factor correction device (“AP” stands for active PFC, “PP” stands for passive PFC, and “NP” means lack of any power factor correction), but you have to come to this arcane knowledge by yourself because there are no such suffixes on the Thermaltake website, just model numbers like “W00xx”.
This nomenclature rises to a level of absurdity with the two revisions of the PurePower 680APD model, one of which is almost completely unfit for work whereas the other is a very good power supply. Both have the same model number of W0049.
Of course, the older power supplies from the TR2 series – the XP480 and the XP550 (W0070) – look least attractive of all. They are not so very bad by themselves, but their original wattage ratings were 250W and 300W, respectively, and were transformed into 400W and 430W after they were branded as Thermaltake’s products. Quite naturally, none of these power supplies can give you the specified output power. They burn out at much lower loads, although they would have received better feedback from me if Thermaltake had been more modest. As you may guess, I’m disappointed when a power supply doesn’t perform that well in my tests, but I’m just raging when I find that a power supply’s characteristics differ radically from what is specified by the manufacturer!
Thermaltake made some amends to the users with its newer TR2 series units. The specified wattage of the W0093 is overstated by only 50W, and it passed through my tests alive. When I tried to squeeze the specified output power out of it, its protection worked sooner than its transistors got burned.
Most of the PurePower series units are midrange models. They don’t have exceptional parameters, but are free from obvious defects, if you don’t count in the fantastically ill-designed first revision of the W0049 model. However, it’s no use to make general statements about the whole PurePower series because it includes a lot of very diverse models, some of which differ from each other not only in the circuit design but also in their actual manufacturer, while others are only externally different (for example, the W0063 and the W0083).
And finally the Toughpower series (to be exact, its single representative I tested in this review) left a very nice impression. Well, it was expectable because this power supply betrayed its actual manufacturer at first glance and we had already tested CWT’s power supplies of this class under the brands of Antec and Foxconn. So again, the good performance of this model was not a surprise to me.
I also want to add a few words about the noise characteristics of Thermaltake’s power supplies because it’s one of the focus points for the discussions at Web forums. The models with manual speed controllers are discussed in the first place. Of course, it’s a very subjective thing to evaluate the noisiness of a power supply in terms of “noisy” or “quiet”, but Thermaltake’s models are mostly right in the middle between these two categories. For example, my colleagues from Silentpcreview.com would most likely have condemned all the models in this review as noisy, but their resource is oriented at those users who set silence as the main priority and who select other fans in the system in such a way as to not hear them at all. As for ordinary users, a majority of them will be quite satisfied with the noise characteristics of the power supplies from Thermaltake, I think.
I also want to note that manual control over the fan speed doesn’t necessarily mean that the power supply works quietly. For example, the W0057 comes with a control panel, but you cannot slow the fan down to below 1200rpm. Moreover, this panel cannot silence the second fan enclosed with the power supply, either. And you cannot easily replace the fan with a quieter one because the panel only supports some special fans from Thermaltake. So, if you pay more for such an enhanced-functionality power supply, think if you really need this extra functionality and if you will be able to use it at all.
Click here for the cross-load data for the tested power supplies.
Click here to download the viewer for the cross-load data.