ATX Power Supply Units Roundup. Part VII

It’s been a while since we tested power supply units, but today we are going to make up for that by offering you a detailed review of 15 power supply units with the wattage below 1000W. You will find the solutions from Antec, Cooler Master, Etasis, Floston Energetix, FSP, Rosewill, Seasonic, Thermaltake, Ultra and Zalman in our new monster roundup.

by Oleg Artamonov
09/04/2007 | 10:40 AM

Our previous article was about the 1000W monsters of the PSU world and was called 1000W Power Supply Unit Roundup. 1000 watts is a pretty round number, yet this wattage is not quite called for yet as we found out in that review. Even top-end gaming stations require much less power.


For this review I took power supplies that should be more appropriate for today’s end-user – a dozen models ranging from 550W to 850W. This is the wattage range the person who’s assembling a top-end gaming PC with one or two graphics cards should be choosing his power supply from.

Antec Neo HE 550 (550W): Error Correction

I reviewed this power supply before and found it to have a large output voltage ripple of unclear origin (for details see our previous article). That was a version A3.1 sample and Antec has been kind to offer us an A4 version for our tests.

The “HE” in the model name stands for High Efficiency. The Neo HE 550 is claimed to be very quiet at work notwithstanding its 80mm cooling fan.

The A4 and A3.1 revisions do not differ much in their internal design. It is a classic design with large T-shaped heatsinks, an active PFC device, and a PCB with connectors for detachable cables in the rear part of the unit.

The quality of assembly is high as you can expect from Seasonic who is the real manufacturer of the Neo HE 550.

The A3.1 and A4 revisions have identical specs. The revision number can be read from the barcode sticker on the side of the PSU – in small print near its left edge.

The PSU offers the following cables and connectors:

Included with the PSU are:

This selection of connectors is quite standard for a modern PSU. One could only wish to have four instead of two graphics card connectors but owners of SLI or CrossFire configurations are going to prefer higher-wattage PSUs whereas for a system with one graphics card, even with a GeForce 8800 GTX or a Radeon X2900 XTX, this PSU will suffice in both load capacity and amount of connectors.

The PSU boasts superb stability of the output voltages. The +12V diagram is a solid green field. The +5V voltage shows the biggest deflection, 4%, from the nominal value (the allowable maximum is 5%).

The most interesting part of the test is about the output voltage ripple. It’s all right with the high-frequency pulsation as it is much weaker than allowable.

The low-frequency ripple diagram recorded at maximum load still shows noise bursts with a duration of about 5 milliseconds each. This time, however, these bursts are within 25 millivolts, which is far below the allowable maximum (which is 50 millivolts for the +5V and +3.3V power rails and 120 millivolts for the +12V rail). So, I just want to congratulate Antec and Seasonic on having found a working solution of the problem!

The PSU uses an Adda AD0812HB-A71GL fan (80x80x25mm, 3010rpm). Its speed is constant at loads below 200W and then begins to grow linearly up to 2600rpm. The PSU is indeed very quiet at work. Its fan sounds like a barely audible whisper at loads below 300W.

The efficiency is 85-86% through the typical load range and lowers to 82% at near-maximum loads. The power factor is above 0.95 at typical loads as you can expect from a PSU with active Power Factor Correction.

So, Antec and Seasonic have successfully got rid of the problems the revision A3.1 Neo HE 550 had. Now it is a high-quality, stable and very quiet power supply capable of supporting a very advanced modern configuration.

Cooler Master Real Power Pro RS-850-EMBA (850W)

We already tested a Real Power Pro series product from Cooler Master in an earlier review. It was a 1000W unit and I was pleased with its good electrical parameters as well as quiet operation. Quietness is a fairly rare thing to find in a PSU with such high wattage. Let’s now see how the less powerful model from the same series behaves.

The PSU is large with a length of 180 millimeters. It is 40 millimeters longer than a standard ATX power supply, but this is not a problem. I doubt anyone would take a small system case to assemble a configuration that requires an 850W power supply.

The external panel of the PSU is perforated, the only blank spot being occupied by the mains connector. Besides everything else, this ensures good thermal conditions because the less resistance to the airflow the PSU elements offer, the more air can be pumped through the PSU case per minute at the same fan speed.

There is a LED indicator next to the mains connector. It is green when the PSU is turned on and red when the PSU is off or has evoked its protection. There is no On/Off switch here while a voltage switch is not required: the RS-850-EMBA is equipped with an active PFC device that supports the full voltage range from 90V to 260V.

The internal design is the same as that of the 1KW model except for the differently shaped heatsinks. The PSUs belong to one product series not only due to marketing reasons. They are indeed based on the same platform. It is quite a common thing, by the way. Having developed the circuit design and PCB of a power supply unit, the manufacturer rolls out several models with varying wattage that only differ internally in the ratings of certain components, in the size of the heatsinks, and in the performance of the cooling fan.

The PSU uses two power transformers. Although many manufacturers claim this to improve the stability of the output voltages, this is not exactly true. A single power transformer, with a twice higher rating and larger, would do just as well, but the dimensions of the PSU electronics are limited, especially vertically (the PSU case must accommodate a fan), so it is often expedient to install two smaller transformers than a single and large one.

The quality of assembly is high. I have no complaints whatsoever.

The PSU offers six 12V output lines but there is only one +12V power rail inside it; it has an allowable current of 60A (720W). There are six 18A current limiters on the PSU output which are set to comply with the safety regulations that require that there be no more than 240VA of power going into the PSU in case of a short circuit.

It’s not quite clear why the PSU label shows only the peak currents. It is more common to specify sustained currents and, additionally, peak ones.

The PSU offers the following cables and connectors:

Thus, this PSU can power up almost any modern PC configuration imaginable without adapters, even if it is a SLI or CrossFire system with two top-end graphics cards each of which has two external power connectors. The PSU delivers more than enough power for that. Our measurements show that such systems don’t even come close to the 850W mark.

The output voltages are very stable except that the +3.3V deflects by 4% from the nominal value under maximum load (the allowable deflection is 5%). The +12V voltage remains very stable at a point slightly higher than the nominal value.

The output voltage ripple is large at max load, but always remains within the acceptable range. The pulsation amplitude is going to decrease at lower loads.

The PSU uses a 135x135x25mm fan from Young Lin Tech Co. (the DFS132512H model). Some say that a larger fan is always a quieter fan. This is not true. The amount of noise depends on the fan speed, on the quality of the bearing, on the design of the PSU, etc. From a PSU developer’s point of view, a large fan may look better since it stretches the entire width of the PSU case and thus cools all of the case volume, leaving no “dead zones”.

The fan speed is kept at about 820rpm up to a load of 600W and then begins to grow up to 1050rpm. As a result, this PSU is very quiet but also rather hot. The difference between the air temperature on its input and output was over 15°C during this test.

The PSU boasts over 85% efficiency through most of the load range, from 150W to the maximum.

Thus, the Cooler Master Real Power Pro RS-850-EMBA leaves a very nice impression. It features a neat assembly, a full set of connectors that is going to be sufficient for any modern system, a very quiet operation, and good electrical parameters. This PSU will be an excellent buy if your PC needs such a high-wattage power source.

Etasis ET850 (850W)

We have already tested products from Etasis in an unobvious way. This firm develops and manufactures Zeus series units for SilverStone. This time Etasis plays under its own brand in our roundup. The sample for our review was provided by Sundial Micro.

It’s impossible not to register at least an external resemblance of this PSU to the Zeus series. It is the same case design with the cover opening backwards. The fan is in the back part of the case again, making you wonder at first whether the PSU is cooled actively at all. The On/Off switch, mains connector and operation mode indicator are all placed in the same way, too. The only difference is the color of the case. Zeus PSUs are black while the ET850 is painted a somewhat unusual matte dark-gray color.

Internally, the PSU consists of two full-size PCBs facing one another. This helps reduce the component density of each separate PCB, but also limits the size of the heatsinks and makes the PSU more crowded overall, which hinders proper cooling.

The top PCB carries most of the high-voltage components: an active PFC device (its choke is in the top left of the photo and its controller chip resides on the small upright PCB in the center) and a rectifier with two parallel-connected 390µF electrolytic capacitors on the output. The standby +5Vsb source is placed here as well (in the bottom right of the PCB in the photo above).

An interesting thing in this particular PSU, there is a seat for a relay on the PCB (a little below its center) which is, however, not soldered in. The purpose of this relay would be to close thermistors that would limit the current when the PSU is started up. In its cold state, the thermistor (it is the large green disc standing near the fuse and the 6-pin connector) has a rather high resistance. So when the PSU is plugged into the wall socket, this resistor limits the inrush of current, protecting the PSU’s input circuitry. This current heats the resistor up and its resistance lowers as the consequence. After that, it has no effect on power supply operation. The relay must have been meant to close the thermistors so that after they have done their job in the first second, their heating up wouldn’t contribute to increasing the PSU temperature and their resistance wouldn’t affect the PSU’s input resistance.

Well anyway, the developer chose not to install the relay at all.

The other PCB carries the power transformer with transistors as well as the low-voltage section of the PSU. The latter is interesting since the ET850 is one of the few PSUs with truly independent regulation of voltages. The two small additional cards you can see in the photo above to the left of the output cables are switching regulators based on an L6730 chip with synchronous rectifiers. These regulators are compact thanks to their high operating frequency (400kHz) and synchronous rectification. Each of them resides on a card the size of a matchbox and does with a tiny heatsink for cooling.

It’s hard to tell if this design is better than the regulator with a saturable choke that is traditionally employed in PSUs. The latter has an excellent efficiency and a good regulation coefficient, but is not exactly independent. It can only work in pair with a “senior” full-featured regulator. Perhaps the reason for choosing this design is that Etasis drew from their earlier engineering solutions when developing this PSU.

The PSU can yield almost all of its output power (840W out of 850W) via the +12V rail which is split into four outputs by means of 18A current limiters.

The PSU offers the following cables and connectors:

The output voltages are very stable. The +3.3V voltage is the only one to get red at maximum load. None of the three tracked voltages goes out of the allowable range in any point of the diagram.

The output voltage ripple is about 45 millivolts on each output line at max load. This is within the allowable limit.

The ET850 employs a Sanyo Denki fan (San Cooler 80 9A0182S402). Its rated speed is 3400rpm. Located at the back of the PSU, this fan is going to be inside the system case in an assembled PC. This should make the whole system less noisy. The efficiency of cooling is not affected: the PSU has vent holes in its front and rear panels only, and the direction of airflow doesn’t depend on the position of the fan.

The PSU proved to have an extremely high operating temperature in my tests, however. The difference between the input and output air temperature was over 25°C while the fan only sped up to 2700rpm. Moreover, I tested the PSU on the next day when it was rather hot in our test lab – about 26°C – and the PSU shut down when working at full output power and could only start up again in ten minutes.

I don’t understand this behavior. The fan controller should have made the fan work at the maximum possible speed at such high temperatures or the manufacturer should have installed a more powerful fan. I can only hope I tested a defective sample.

The efficiency is 83% at the maximum and generally higher than 80%. These numbers are good, but not impressive anymore against modern models with an efficiency of 85% and higher. The power factor is near 1.0 – the active PFC device does its job well.

So, the Etasis ET850 is a good power supply with an original circuit design solution and superb electrical parameters but with one drawback. Its fan controller did not make the fan rotate at full speed and the PSU would overheat under max load. An interesting fact, being very similar to the ET850 in its internal design, the SilverStone Zeus SST-ST85ZF didn’t have that problem. Its fan would speed up to 3400rpm at a load of 700W.

Floston Energetix ENFP-750W (750W)

The name of Floston Electronic Enterprise may not ring any bells in a mass user’s ears although the company has been on the market for a while. Its products are mostly meant for modders, including highlight lamps, various fans, etc. Floston offers PSUs, too.

Floston’s Energetix PSUs look really exciting. While most other PSUs come in cardboard boxes, these are shipped in neat silvery cases that resemble a toolbox. Well, you can use the case as a toolbox indeed because it only has a piece of foam-rubber inside to hold the PSU firmly in place.

The PSU has a dark glossy case (which is easily scratchable, by the way). Its dimensions are standard, but it’s got a fashionable 135mm fan. There is a slit for a not-installed mains voltage switch on the left of the front panel, which is the consequence of using the same case for different PSU models: the Energetix is equipped with active PFC and doesn’t need such a switch.

A certain drawback, the corner of the case with the mains connector and the On/Off switch is blank without vent holes.

The PSU is assembled neatly. I have no complaints about the quality of manufacture. The heatsinks are medium-sized with large lengthwise ribs in their top.

On the other hand, the ENFP-750W does not have any exceptional features, unlike the two above-described models for example. It doesn’t have two transformers or three regulators. It is quite an ordinary high-wattage PSU with active Power Factor Correction.

The combined allowable load on the +12V rail, which is split into four output lines, is 50A. The total allowable load on all the main rails is 730W. The remaining 20W is due to the -12V rail and to the standby source.

I want to note that the label is not quite correct joining the two latter rails together with a combined load limit of 20W. These rails are in fact absolutely independent from each other and if there’s a full 15W on the standby source, the -12V rail can be loaded by more than 5W. However, the -12V rail bears little load in a modern PC (it may power COM ports controllers and onboard sound but hardly anything else), so my remark is largely for the sake of truth.

The PSU offers the following cables and connectors:

This is an ordinary selection of connectors but the cables are about 10cm shorter than a regular PSU offers. It would also be good to have SATA power connectors on two different cables because you can have SATA DVD-burners and SATA HDDs installed in different areas of your system case so that it’s not convenient to connect them all to the same power cable.

The graphics card power connectors are designed in a peculiar way. They are ordinary 6-pin connectors with an additional 2-pin piece hanging about to transform this connector into an 8-pin one.

One of the two additional “ground” pins in the new 8-pin connector is meant for a more accurate correlation between the graphics card’s signal ground and the PSU’s ground. There’s current flowing through the “ground” wires of the cable at work, and so there is a voltage drop on them, resulting in a divergence between the two mentioned levels. In the composite connector of the ENFP-750W the additional piece is connected to the main connector rather than to the PSU. This design will only compensate the change in the ground level due to the voltage drop on the connector pins only rather than on the entire cable length. Anyway, this is good, too. So if you’ve got a graphics card with an 8-pin power connector, don’t be afraid to plug the 2-pin piece in!

The cross-load characteristics of this PSU look good. The voltages go out of the allowable ranges only when there is a great load misbalance that just cannot occur in a real PC. The +3.3V voltage is ideal. The +12V voltage is within a 2% deflection through over a half of the diagram. The +5V voltage, the least stable of the three, is going to fluctuate within 1-2% above the nominal value in a real PC.

I want to remind you that the typical power consumption of modern PCs corresponds to the bottom third of the diagram, up to 40-60W on the Y-axis, because all the heavy consumers are powered by the +12V rail, which corresponds to the X-axis.

The output voltage ripple is normal at full load. In the oscillogram above one vertical square corresponds to 50 millivolts whereas the industry standard allows an amplitude of 50 millivolts on the +5V and +3.3V rails (the top and bottom, red and orange, lines) and of 120 millivolts on the +12V rail (the middle, green, line).

The PSU employs a Young Lin Tech DFB132512H fan. Fans from this maker are often used in PSUs that are cooled by a 135mm fan.

The fan speed is constant up to a load of 400W and then begins to grow up in a linear manner. The speed is average, with a minimum of 1230rpm, so the ENFP-750W is average or somewhat worse than average in terms of noise. It is going to satisfy many users, yet you can compare its results with those of the Cooler Master RS-850-EMBA for example. The latter’s fan is only half as fast at low loads.

The efficiency is good even as today’s PSUs go. It is 85% at loads from 350 to 400W. The PSU overcame the 80% mark at a load of about 100W, i.e. at less than 15% from the maximum.

Thus, the Floston Energetix ENFP-750W is a good modern PSU that delivers conscientiously what is written in its specs. It offers stable voltages, a full selection of connectors (but again, I would want it to have a second cable with a couple of SATA plugs), and average noisiness. Without a doubt, many people will also be attracted by its original packaging, a neat aluminum box.

FSP Epsilon 700W, 800W, and 900W

We have regularly reviewed products from FSP Group, including the high-wattage Epsilon series in which we tested the 600W and 700W models. The company has recently extended its model range towards higher wattage, up to 1010W. So, I am going to test new Epsilons and take the Epsilon 700 for the comparison’s sake.

All Epsilon series products have attractive violet-blue cases. The coating is high quality and durable. It is difficult to scratch it as opposed to the dark glossy metallization of most other PSUs.

The On/Off switch is highlighted with blue at work but the fan lacks any highlighting.

The internal design of FSP Group’s power supplies has long been a shocking view for reviewers and buyers due to the size of the heatsinks. Two out of the three heatsinks are simple aluminum bars without any ribbing. The third heatsink has but very superficial ribbing. Why? Because the developer preferred to reduce the heat generation of the transistors and diode packs rather than to increase the size of the heatsinks. There are much more elements in this PSU than usual. Most of them are connected in parallel. In an ordinary PSU, the 30A power rail has one diode pack rated for a current of 30A, but there are two such packs here, each working at half the rated load. Why? The correlation between heat generation and current is non-linear with diodes. When a 30A current is flowing through one 30A pack, the latter is generating more heat than two 30A packs, each with a current of 15A, would.

Of course, semiconductor components cost money and two packs are more expensive than one, but the manufacturer saves on the heatsinks and on the overall weight of the PSU. The latter thing makes the transportation cheaper and also helps save on the European electronics recycling tax that depends on the weight of the device.

I can’t catch any difference in the circuit design of the two senior models (800W and 900W) except for the different color of the PCB and the ribbing on the heatsinks. The marking on the PCB reads “FSP850-80GLN REV.: 1” and there is a list of possible PSU wattages on the left, from 500W to 900W.

It was to be expected that the heatsinks would grow larger. Whatever tricks the manufacturer may resort to, there is ever more heat generated inside the PSU as its wattage increases. So, the heatsinks in the Epsilon 800 and 900 are larger than in the Epsilon 700, yet still much smaller than in other PSUs of their wattage.

In an earlier review I mentioned a small card in Epsilon PSUs that created a load on the PSU when it was started up. Otherwise, some mainboards could not start up because they did not provide the load necessary for the PSU to function normally at first. This load used to be provided by ordinary resistors that were turned on all the time, but this affected the efficiency at small loads. Moreover, the resistors get hot at work.

The Epsilon 700 has this card (it is covered with heat-shrinkable film and looks like a black square in the photographs) but the Epsilon 800 and 900 have not. They only have an empty slot the card used to be plugged into. We can’t check the PSUs out with many mainboards to see if the compatibility problem has vanished, but I suppose components that may need as much as 800W of power do consume a high enough current from the PSU at the start anyway.

Each of the three PSU models has four output +12V lines, but there is only one +12V power rail inside the PSU. The purpose of each line is marked on the label and the cables of the lines are color-coded (yellow, yellow-green, yellow-blue, yellow-black).

The more powerful units have a proportionally higher load capacity of the +5V and +3.3V rails as well as of the +12V rail.

Although the label seems to suggest that the max combined output power of all the +12V lines equals the PSU’s total output power, it is not exactly so. There is a limitation written in small print below: the load on the +12V must not exceed 70A for the 900W model. This means 840W rather than 875W as you might have thought.

The FSP Epsilon 700 offers the following cables and connectors:

All the cables are sleeved.

The senior models, Epsilon 800 and Epsilon 900, have more connectors:

So, there’s everything necessary here and the cables are long enough for any system case.

Alas, the cross-load diagram for the 700W model looks bad. The +5V voltage drops quickly as the load on the corresponding rail is increasing. As a result, it goes below the allowable minimum at a load much smaller than declared (about 130W for the +5V and +3.3V combined, the declared maximum being 155W).

It’s even worse with the higher-wattage models. The 800W model has a too high +12V voltage (see the red color in the left of the diagram). Both PSUs went out of the allowable range for the +5V voltage even sooner than the Epsilon 700 did. Of course, modern PCs do not load the +5V rail much, but when the manufacturer promises up to 175W, you expect to have 175W, not only half that wattage, even if you’ll never need that much.

FSP Epsilon 700

At full load the output voltage ripple was at the limit (50 millivolts) on the 700W model’s +5V and +3.3V rails.

FSP Epsilon 800

This limit is exceeded by the Epsilon 800: 60 millivolts on the +5V rail and 100 millivolts on the +3.3V rail.

FSP Epsilon 900

The picture is different with the senior model: the pulsation is a little higher than the allowable 50 millivolts on the +5V and +3.3V rails but there are short spikes on the +12V rail. With those spikes the pulsation is as high as 120-150 millivolts on the +12V rail, the allowable maximum being 120 millivolts.

Thus, the Epsilon 700 is the only PSU out of the three that complies with the requirements of the Power Supply Design Guide in terms of output voltage ripple at full load.

There are ordinary 120mm fans from Protechnic Electric in each PSU. The junior PSU has a MGA12012HS-O25 fan with a rated speed of 2500rpm.

The medium model uses a MGA12012HB-O25 fan with the same performance and speed but running on a ball bearing instead of the HS model’s slide bearing. Both fans have translucent blue impellers, but lack any highlighting.

The senior PSU model employs a MGA12012YB-O25 fan. Unfortunately, the product catalogue didn’t work on the Protechnic Electric website so I couldn’t find its exact characteristics. But judging by the consumption current indicated on the label (half as much again as that of the MGA12012HB-O25), this is a high-performance model with a rated speed of about 3000rpm.

The fan speed diagrams are very similar between the three PSUs. This is normal because if the PSU electronics is guided by the temperature of the heatsinks, it is going to maintain the same fan speed under identical conditions even with different fans.

The speed is changing linearly, from 1150rpm at min load (50W) to 2300-2500rpm in the Epsilon 700 and 800 to 2800rpm in the Epsilon 900. The initial speed isn’t low (as opposed to the early Epsilon PSUs whose fan would start out at a speed of 825rpm) and the linear regulation is no good in terms of noise. Ideally, the fan speed should not change at all until a certain temperature. After that, it should be increasing to protect the PSU from overheat.

So, today’s FSP Epsilon PSUs are noisier than average. They are not silent at min load and become louder at higher loads.

The PSUs all have an efficiency of over 80%, up to 87% in some point of the diagram, which is an excellent result. But as I wrote above, the Epsilon architecture is conceived to achieve a high efficiency by means of reducing the loss on the rectifiers’ diodes. The power factor is over 97% through most of the load range as it typical of PSUs with active Power Factor Correction.

Alas, the new FSP Epsilon PSUs leave an ambiguous impression. On one hand, these PSUs are neatly assembled and offer all the necessary connectors. But on the other hand, the stability of the output voltages and the voltage ripple are poor and their noise can be characterized as “worse than average” whereas I know that Epsilons can be very quiet. I hope FSP Group will be working to correct the mentioned problems and reestablish its reputation.

Rosewill Turbo Series RT550-135-BK (550W)

Unlike the previous famous brand, Rosewill has never taken part in our tests although offers quite a lot of PSU models. Rosewill doesn’t have its own production facilities, though. For example, the reviewed model is manufactured by the Hong-Kong-headquartered Wintech Electronics (we have reviewed its products before) as is indicated by the UL certificate number on the PSU label.

The PSU has a larger case than standard (with a length of 155mm instead of the standard 140mm) and presents a odd mix of a fashionable 135mm fan with an outdated design without active PFC (the mains voltage switch is a clear indication of that; a PSU with active PFC wouldn’t need one). Today, only low-end PSU models are made without active PFC. As a matter of fact, the RT550-135-BK is the only power supply without PFC in this review.

Internally, it is a typical PSU from Wintech. It is a classic design with a half-bridge inverter, once very popular but now becoming obsolete due to the arrival of more advanced controllers. The assembly is not very neat as you can see by the capacitor in the top right corner. The developer could not fit it into the PCB and the capacitor hovers above the surrounding elements on its legs. The drop of glue it was fixed with on the nearby choke had already cracked when I opened the PSU up and the capacitor was loose.

Note that the PSU has no PFC whatsoever, passive or active. This indicates its targeting at the American market because a PSU can’t sell in Europe without being equipped with at least passive PFC. In the United States the version 4 and higher Energy Star certification can only be obtained with a power factor of 0.9 and higher (such a power factor can only be achieved with active PFC), but this certification is not obligatory.

By the way, Wintech products are easily identifiable, even without the UL certificate number, by the characteristic coloring of the components: yellow capacitors, blue transformers, and orange heatsinks. Compare with this or this power supply for example.

Having a total wattage of 550W, the PSU offers only two “virtual” +12V lines with a combined current of somewhat higher than 30A. This is not much, just a little higher than the requirements of the version 2.2 Power Supply Design Guide to 450W power supplies. For example, the Thermaltake Purepower RX (see below) with a rated wattage of 550W offers a max load of 41A on the +12V rail and the 500W Zalman ZM500-HP provides a load up to 34A.

The PSU offers the following cables and connectors:

The cables are sleeved.

The PSU has a normal selection of connectors but I guess there should be more SATA and PATA power plugs. The cables are somewhat too short. They are 10-15cm longer in most other PSUs.

The RT550-135-BK doesn’t yield very stable voltages. The +5V voltage is higher than necessary in the bottom part of the diagram and the +12V is not too stable, either. Anyway, the PSU copes with every load combination successfully, leaving the allowable limits only when there is a strong load misbalance that can hardly occur in a real PC.

At full load there is a weak high-frequency pulsation: about 20 millivolts on the +5V rail, 45 millivolts on the +12V, and small spikes on the +3.3V.

Alas, the picture grew worse when I switched the oscilloscope’s resolution from 10 microseconds/div to 2 milliseconds/div. Besides high-frequency pulsations, there is a strong low-frequency ripple on the +5V and +12V rails (at the doubled frequency of the mains, i.e. 100Hz). The combined ripple amounts to 50 millivolts on the +5V rail (i.e. exactly equal to the allowable maximum) and to 80 millivolts on the +12V rail (the allowable maximum is 120 millivolts).

I couldn’t identify the manufacturer of the fan. It bears Rosewill stickers on both sides without any additional information. I can only tell you it’s an 11-blade 135mm fan with blue highlighting.

The fan speed is adjusted linearly from 1080 to 1650rpm. Thus, the Rosewill power supply is somewhat noisier than average (it would be average if the fan speed did not increase at all up to a load of 300-400W).

The efficiency of this PSU is rather low by today’s standards. It hardly reached 80% and fell back to 76% at full load. The power factor is 0.65 on average just as you can expect from a PSU with any PFC (compare this to the power factor of 0.7 and higher of PSUs with passive PFC and 0.95 and higher of PSUs with active PFC).

Thus, the Rosewill RT550-135-BK is a mainstream power supply, unexceptional in any of its parameters. It has average voltage stability, average noisiness, an acceptable selection of connectors, etc. Its 135mm fan distinguishes it from previous-generation models, yet this fan doesn’t guarantee silence. The buzz of its impeller and the noise from the airflow is quite audible at speeds higher than 1000rpm.

Seasonic M12 SS-700HM (700W)

Seasonic is a recognized manufacturer of high-quality and quiet power supplies. I once reviewed the S12 SS-500HT model and was very pleased with it. Let’s see if the SS-700HM is as good as its predecessor.

The PSU has a black matte case. Its paint is powder-sprayed and durable. It’s hard to scratch it off to the metal even when you are installing the PSU into your system case.

A special feature of the M12 is the 60mm fan on its rear panel in addition to the main 120mm fan. The small fan is turned on only when the PSU temperature reaches a certain threshold value; it then blows at the power transformer and the heatsink with switching transistors. The hottest section – the heatsink with the output diode packs – is not in the way of the airflow from this fan, however, and the additional fan doesn’t help much in cooling it.

Besides that, there are connectors on the rear panel for the detachable cables of graphics cards and peripherals.

The internal design is typical of Seasonic. It is almost the same as in S12 series models except for the heatsinks. They used to have twice the amount of ribs, placed in two tiers. From an engineer’s point of view these are PSUs with a single-ended converter and active PFC.

I have no complaints about the quality of assembly, which is high.

The PSU has four “virtual” +12V lines with a maximum current of 56A which corresponds to 672W. In other words, the PSU can yield almost all of its total output power via the +12V rail.

It offers the following cables and connectors:

Included with the PSU are:

Each cable is sleeved in this PSU.

The +12V voltage remains exemplarily stable except for the very end of the diagram where it is more than 1% above the nominal value at max load. The +3.3V voltage looks good, too, but the +5V bottoms out under load. Anyway, this is not even half as bad as with the above-described FSP Epsilon PSUs, so the cross-load characteristics of this PSU look good overall.

The output pulsation isn’t strong. It is barely noticeable on the +5V and +3.3V rails and does not reach 50 millivolts on the +12V rail, the allowable maximum being 120 millivolts. That’s an excellent result.

The main fan is a 120mm AD1212HB-A71GL model running on two ball bearings and having a rated speed of 2200rpm. It is made by Adda.

The auxiliary fan is a 60mm AD0612HB-D71GL with a rated speed of 4500rpm.

The auxiliary fan was indeed motionless at loads below 260W. It started up only when the PSU had heated up to a certain temperature. It is about at the same moment that the speed of the main fan, which had remained constant at 810rpm, began to grow. The auxiliary fan didn’t stop when the load was reduced back to 250W but kept on rotating at a very low speed at which it did not affect the cooling or noise of the PSU.

At maximum load the speed of both fans is just short of their possible maximums: 2000rpm for the main one and 4200rpm for the auxiliary fan. Of course, this is not silent, but the M12 is generally quieter than most of its opponents. From the PSUs reviewed in this article it can only be challenged by the Cooler Master RS-850-EMBA in terms of noise. The Zalman ZM600-HP is about as quiet as this Seasonic.

The PSU has an excellent efficiency as you can expect from a product with an 80+Plus certification that requires 80% and higher efficiency at 20% from full load in 110V power grids. The maximum efficiency is 86%. The PSU overcomes the 80% barrier at a load less than 100W, but I perform my tests in a 220V power grid. In a 110V grid the efficiency is going to be lower by a couple percent.

So, the Seasonic M12 SS-700HM is a high-quality and quiet PSU that offers good electric parameters and all the connectors sufficient even for a serious gaming station. It is somewhat noisier than the less powerful Seasonic S12 SS-500HT and Zalman ZM460B-APS we reviewed earlier, so you shouldn’t hunt for the SS-700HM specifically unless you need its impressive 700W output power. But if you do need a 700W power source, this PSU is going to be a good choice. Don’t also forget about the detachable cables of the M12. The S12 series doesn’t have this feature.

Thermaltake Purepower RX 550 AP (W0150, 550W)

Thermaltake was the first manufacturer of coolers that took to selling power supplies. It was later joined by Zalman and, recently, by GlacialTech. The company’s model range is impressively large, including several lines, ranging from inexpensive solutions to 1000W monsters. The Purepower series belongs to the mainstream. It is higher than the low-end TR2 series but lower than the expensive Toughpower series.

The PSU has is painted a matte dark-bronze color that distinguishes Thermaltake’s products from glossy PSUs.

The fan grid is designed in a peculiar way. PSUs from this price category normally have a wire grid but this one is punched out and neatly shaped with curved-in edges. Looking prettily enough, this grid is, however, no match to the ordinary wire with a minimum surface area when it comes to minimizing the resistance to airflow.

Internally this PSU seems to resemble PSUs from FSP Group with its three parallel heatsinks. But the actual manufacturer is Channel Well (CWT) and the only common trait with the FSP Epsilon series is the number of heatsinks. The circuit design of the Purepower RX is closer to the classic examples without the doubling of power semiconductor components to reduce heat dissipation as is typical of the Epsilon series.

The PSU case is designed in such a way that one side panel belongs to the bottom while the other moves together with the cover. There is quite a lot of free room behind the PCB. This is due to the size of the fan – it wouldn’t have fitted into the case otherwise.

This PSU has three “virtual” +12V output lines with a combined allowable current of 41A. Interestingly, the minimum load currents are specified on the label as well. The PSU won’t fail under lower currents (the Power Supply Design Guide is explicit on that point), but is not guaranteed to yield stable voltages under such conditions.

The PSU offers the following cables and connectors:

All the cables are sleeved.

Each of the three voltages is perfectly stable. There is no orange or red in the diagram while yellow (denoting a deflection within 2-3% from the nominal value) can only be seen in the area of minimum or maximum loads. This is a sure indication of independent voltage regulation although it is not listed among the official features of the Purepower series at the Thermaltake website.

The output voltage ripple is within the norm: about 35 millivolts on the +5V and +3.3V rails (the allowable maximum is 50 millivolts) and about 50 millivolts on the +12V rail (the allowable maximum is 120 millivolts).

The PSU employs a Yate Loon D14BM-12 fan remarked as Thermaltake TT-1425B. This 140x140x25mm fan has a rated speed of 1400rpm, according to the Yate Loon website, but I’ve got doubts about that. Describing the PSU at its own website, Thermaltake says “1900rpm Fan Inside.” Anyway, it’s not the rated speed proper but the efficient regulation depending on load or temperature that is important.

The fan is half-covered by a piece of celluloid film that drives airflow back into the rear part of the case. This is a questionable solution since it reduces fan performance almost by half. The film may also rattle irritatingly if not fastened properly (I had this problem in one PSU where the screws were not tightened to the end).

The fan speed remained constant at 1000rpm until a 200W load. Then it began to grow up quickly until reached 1800rpm at a load of 475W. This must be the rated speed of the fan because it didn’t change after that while the temperature of the exhausted air was growing up. I don’t quite grasp the idea of this regulation. Why make the fan work at maximum speed when the load is not yet the highest? Perhaps the developer hopes that no one will use this PSU at its max load.

So, the Purepower RX 550 AP is an average model in terms of noise. It is not silent, but quite acceptable for a majority of users.

The PSU is 85% efficient under typical loads and its efficiency goes down to 83% at full load. The power factor is over 0.95 as you can expect from a PSU with active PFC.

The Purepower RX 550 AP (W0150) is a high-wattage power supply suitable for SLI or CrossFire systems, let alone systems with a single graphics card. It has good electrical parameters and works rather quietly. It is not exactly quiet despite its large 14cm fan and is inferior to the PSUs from Zalman, Seasonic and Cooler Master in this respect.

Thermaltake Toughpower 600 AP (W0103, 600W)

Funnily enough, I found the Purepower RX model (discussed above) to have independent regulation of voltages although such regulation is touted as the key feature of the Toughpower series. So, the division of PSUs into the Purepower and Toughpower series is largely based on marketing than on technical reasons. The former series includes models with a wattage rating of below 600W and the latter, from 600W and higher.

This PSU has got the same case as the Purepower RX but with a black, not bronze, glossy paint finish. Anyway, the characteristic fan grid is instantly recognizable.

The internal design is a copy of the Purepower RX, confirming my words about marketing, not technical, reasons for the division between the two PSU series from Thermaltake.

The PSU is manufactured by CWT. It has an active PFC device, three parallel heatsinks and a non-removable side panel. All of this is already familiar to us.

The PSU is declared to have four +12V output lines with an 18A current limiter on each, but the maximum combined current is not indicated. I only found it at the manufacturer’s website. It is 48A, which means 576W. This is close to the PSU’s total output power.

The PSU offers the following cables and connectors:

So, the only difference from the Purepower RX is the six SATA power connectors instead of four. Every cable is sleeved.

The PSU’s cross-load characteristics are superb. A 2% deflection is exceeded at minimum and maximum loads only. The diagram for the +12V voltage is green (meaning that the voltage deflects by less than 1% from the nominal value) through the entire load range.

The output voltage ripple isn’t strong. It is about 25 millivolts on the +5V and +3.3V rails and within 60 millivolts on the +12V rail. This is about half the allowable maximum.

The PSU uses a Yate Loon D14BM-12 fan, too (it seems to have a rated speed of 1900rpm). Half of it is covered with a piece of celluloid film to drive the airflow to the back of the case. This film may rattle if fastened loosely.

And again, the fan speed is kept at 1000rpm until a load of 200W with something but grows up to the maximum during the next 200W. At loads from 450W to 600W its speed doesn’t change but the temperature began to grow up quickly. Moreover, I smelled overheat plastic at full load although the PSU went through the test unharmed.

The meaning of this regulation is still unclear to me. Perhaps the developer wanted to ensure an optimal thermal mode for the PSU in a load range from 300 to 400W as the most probable for a modern top-end system. This makes the PSU cool, but not quiet. Having a fan speed of over 1500rpm at a 350W load, it is only average in terms of noisiness.

The PSU has a good efficiency – up to 86% at the maximum with a smooth reduction to 83% towards higher loads.

Thus, Thermaltake’s Toughpower 600 AP and Purepower RX 550 AP are in fact twin products. They have almost identical specs, identical electronics, and very similar effective parameters. They differ in the amount of HDD power connectors, but this difference is insignificant for a majority of users. Moreover, the strange regulation of the fan that reaches its max speed at a load of 450-475W makes me think that the specified output power of the Toughpower 600 AP is somewhat overstated while that of the Purepower RX 550AP is closer to reality. Speaking in general, these PSUs are going to be interesting for owners of rather advanced gaming systems as they yield stable power at a reasonable level of noise.

Thermaltake Toughpower 850 AP (W0131, 850W)

You may disagree with me, yet I think that an 800W power supply is the reasonable maximum the owner of a top-end gaming station with two graphics cards should consider as a possible buy. Our measurements show that a PC with a Core 2 Quad and a pair of GeForce 8800 GTX does not consume even 500W under full load. So, an 850W is going to provide a reserve for now as well as for the future. Higher-wattage PSUs represent the manufacturer’s desire to shock the customer with big numbers and to appear in computer-related news but they are not in fact necessary for today’s PCs.

This PSU looks exactly like the Purepower RX 550 AP and has a bronze-colored case with a punched-out fan grid.

But when you take a look at the back, you see rows of connectors for detachable cables. The mainboard cable and two graphics cards cables are the only non-detachable ones in the W0131. I wonder why the CPU cable is not non-detachable, either. Have you ever seen a PC without a CPU? The Purepower RX series includes PSUs with detachable cables too, but we haven’t reviewed them yet.

The internal design isn’t different from the two previous PSUs. They are all based on the same platform and only differ in the ratings of some components and in the size of the heatsinks. Particularly, the Toughpower 850 AP has larger heatsinks.

The card with output connectors is located at the rear panel of the case in the previously empty space.

The PSU has four “virtual” +12V lines with a maximum combined current of 62A. The current limit is raised almost twofold on two of the output lines, to 30A. This is the consequence of the use of a unified PCB. This PCB offers room for four shunts only and the required power has to be delivered by relaxing the overcurrent limitations.

But opening the PSU up I found a reinforced shunt on the +12V3 line only whereas the +12V4 line had an ordinary, 18A shunt. Three 18A lines plus one 30A line do provide the necessary current, yet I’d like to see the PSU being more true to its own label!

And my second complaint is that I could find no information about how those lines were wired. This is important information considering the different current limiters. I had to find that out by myself: the +12V1 line powers the CPU (the first half of the 8-pin connector); the +12V2 is for the CPU (the second half of the connector) and for one detachable graphics card cable; the +12V3 line powers the mainboard and the non-detachable graphics card cables; and the +12V4 line powers the second detachable graphics card cable and all the peripherals.

The PSU offers the following cables and connectors:

Included with the PSU are:

All of the cables are sleeved.

The cross-load diagram is almost ideal with this PSU series as we have had a chance to see with the two previous models. The maximum deflection from the nominal value is 3% while the +12V voltage diagram is a solid green field.

The output voltage ripple is stronger now, yet within the norm: up to 30 milliseconds on the +5V and +3.3V rails and up to 110 millivolts on the +12V rail.

The PSU uses a Yate Loon D14BH-12 fan remarked as TT-1425B (the D14BM-12 fans of the two previous PSUs were renamed likewise). Yate Loon declares a rated speed of 2800rpm for this fan while Thermaltake says 2300rpm. Judging by my measurements, Thermaltake is nearer to the truth.

And again, the fan speed remains at about 1200rpm under loads below 300W, then grows up quickly up to 2100rpm under loads up to 600W (the PSU temperature gets stabilized at that as the diagram shows), and then it is the PSU temperature that begins to grow. The latter amounts to an alarming 19°C (to remind you, this is the difference of the air temperatures at the input and output of the PSU) and the meaning of this strange regulation remains a mystery.

The Toughpower 850 AP is somewhat noisier than its lower-wattage predecessors. It is especially noticeable under low loads because the min fan speed has increased from 1000 to 1200rpm. It is only in certain parts of the diagram that it surpasses its predecessors by beginning to increase the speed later, but anyway, you shouldn’t expect the more powerful PSU to be quieter than its lower-wattage mates in this particular case.

The PSU notches an efficiency of 87% at a load of over 400W. That’s an excellent result. The power factor is 0.95 on average, increasing a little towards higher loads.

So, the Toughpower 850 AP is a good high-wattage PSU with superb electric parameters and capable of powering any modern configuration except for exotic components like thermoelectric Peltier elements. It’s got two drawbacks. First, the manufacturer should have been more attentive to the PSU specs and specify which line powers which connectors since the different +12V lines are not equivalent to each other in terms of allowable load. Second, this PSU is not quiet. It is average or even somewhat worse than average in this respect and is noisier than its lower-wattage mates due to the use of a faster fan.

Ultra Products X-Pro ULT-XF800S (800W)

Ultra Products has already been represented in our earlier reviews. For this review it offers an 800W PSU with a rather queer exterior:

Its case is made from aluminum and it is cooled with two 80mm fans. It is not the only PSU to have this type of case – the aluminum boom has affected many industries, including power supplies – but such models are still rare in our reviews.

The PSU features a push-pull cooling design with one fan sucking air into the case and another exhausting it. There are vent holes near the first fan for some air to pass through because the performance of the intake fan is lower here than that of the exhaust fan.

The heatsinks are impressively large, yet their design is somewhat disappointing. I had expected the aluminum case to be used to dissipate heat as well (thanks to its excellent heat conductivity) but it proves to be a purely decorative element. None of the PSU components has thermal contact with the case. The PSU is cooled only by the stream of air passing through it.

The PSU has a dual-transformer design. This is about all you can see behind the massive heatsinks that occupy almost all of the top part of the case.

With a max output power of 780W, the allowable load on the +12V rail, divided into four “virtual” output lines, is 660W or 55A.

The PSU offers the following cables and connectors:

The idea to connect the CPU fan via the power supply may look odd, yet it is not without a reason. Having shut down, the PSU is still powering the attached fans from the standby source for a few minutes more to additionally cool your system components. On the other hand, it is better to control the CPU fan with the mainboard’s tools, so you may want to limit yourself to connecting only your system fans to the PSU (up to two fans).

Each cable is sleeved.

The output voltages are stable, but not ideally so. The +5V voltage deflects most from the nominal value, but goes beyond the allowable range only at near-maximum loads. The +12V and +3.3V voltages do not leave the “yellow zone”.

The output voltage ripple is within the norm. I can only note the odd shape of the impulses on the +12V rail, perhaps due to the non-symmetrical dual-transformer design (it’s clear that the transformers are different – they have different markings and even different colors).

The first, external, fan is a Young Lin Tech DFS801512H (it is marked as “Ultra” but the original name and the UL certificate number help determine the actual manufacturer). It is an 80x80x15, 3200rpm fan.

The second, internal, fan is a Young Lin Tech DFC801012H (80x80x10mm, 3500rpm).

I should note that the use of “thin” fans involves one unobvious problem. The rotating speed being the same, a typical 80x80x25mm fan creates half as much static pressure as a typical 80x80x15mm fan. It is the static pressure that largely determines what volume of air the fan can pump through a dense arrangement with high aerodynamic resistance, so one 80x80x25mm fan can prove more efficient than a couple of “thin” fans.

Here, the fans start up at a speed of 1300-1400rpm and accelerate linearly to 2800-2900rpm after a load of 200W. As a result, the PSU is of average noisiness. It is not noisy, which is even a surprise considering its wattage and the cooling system consisting of two weak and thin fans.

The efficiency is 85% at the maximum and above 80% on average. The manufacturer’s website specifies a typical efficiency of 78%, but this must be just a reprint from the respective recommendations of the Power Supply Design Guide rather than a real number because the real efficiency of the PSU won’t drop so heavily even in an 110V power grid.

Thus, the Ultra X-Pro ULT-XF800S is a midrange high-wattage power supply. It doesn’t offer exceptional electric parameters and doesn’t work noiselessly, but it is capable of powering up an advanced PC configuration without problems. The aluminum case should be viewed as a design element because the material has no effect on the PSU, particularly on its cooling.

Zalman ZM500-HP (500W) and ZM600-HP (600W)

Products from Zalman have been tested in our lab regularly, receiving good reports from us. Like Thermaltake, this company began as a cooler maker, but it doesn’t hurry to produce super-powerful PSUs. It focuses on quiet operation instead. That’s why the 500W and 600W models presented here are the senior ones in the company’s product range although Zalman is planning to release an 850W model.

The ZM500-HP and ZM600-HP are built on the same platform, so I’ll be discussing them both together.

The PSU shows unmistakable traits of a product by FSP Group. It has a black matte case of a standard size and is cooled by a single 120mm fan.

The internal design is similar to the above-described FSP Epsilons except for two things: the card with connectors for detachable cables on the rear panel and the additional heatsink that is connected to the heatsink with diode packs by means of a heat pipe.

The ZM500-HP and ZM600-HP do not differ in their internal design.

The mentioned heatsink is near the external panel, under the vent grid, and the air that is being exhausted from the case is passing through that heatsink. This ensures enough airflow for the heatsink but the dense ribbing increases the aerodynamic resistance of the whole thing, thus reducing the amount of air that is being pumped through the PSU per minute.

You may worry that the air will flow around the heatsink, but it won’t. It is clear that the heatsink occupies most of the vent grid and is covered from above by the PSU fan. So, it will be cooled indeed and the only arguable question is how this additional heatsink affects the aerodynamic resistance and noisiness.

As I said, there is a card with connectors for the detachable cables at the rear panel of the PSU.

The PSU has half a dozen connectors. The cables for a second CPU (if you need one) and for a second graphics card are joined into one.

The junior model has two +12V lines with an allowable current of 34A. The first line has a current limit of 25A rather than 18A as usual.

The other model has four output lines with a combined load capacity of 42A. The label of each PSU shows what connectors are powered by what line.

The PSUs are equipped the following cables and connectors:

Included with the PSUs are:

Thus, the PSUs have all the necessary connectors. It’s only for a SLI or CrossFire configuration that you’ll need adapters since such a system requires four power connectors (two for each graphics card).

The cross-load diagram of the junior model resembles the FSP Epsilon. The +12V voltage is higher than necessary and not very stable while the +5V and +3.3V voltages go below the acceptable limit when the load grows on the respective power rails. This PSU is going to cope with a modern PC that puts but a small load on the latter two rails, but it does not comply with its own specifications.

The 600W model is somewhat better, but not ideal, either. The +5V voltage bottoms out heavily when there is a high load on the appropriate rail while the +12V voltage is much higher than the nominal value when the load on the +12V rail is low. The PSU is going to power the PC all right, yet diagrams with more green, rather than with yellow or red, are much more agreeable to my eye.

The output voltage ripple isn’t strong. It is much lower than allowable on each of the three tracked power rails even at maximum load.

Both PSUs make use of Adda AD1212MB-A71GL fans (120x120x25mm, 2050rpm). The rated speed is rather low, promising low noise.

In the junior model the fan started from 940rpm and was accelerating steadily and linearly at higher loads. It did so not too fast, however. The maximum of speed is only 1350rpm. This indicates a quiet PSU although I guess the fan might have started to accelerate from a certain load only to make the PSU even quieter.

In the ZM600-HP the fan started out from 1000rpm and stopped at 1600rpm. This is somewhat worse than in the ZM500-HP but still good. The PSU is not silent at low loads but the fan speed is being increased less rapidly at higher loads than in many other PSUs.

Each model has a good efficiency that quickly reached 86% as the load grew and then kept around that value until a load of 400W and higher when it dropped to 83%.

Thus, the high-wattage power supplies from Zalman are not good in only one parameter, in the stability of the output voltages. They share this problem with the FSP Epsilon PSUs but they are not as noisy as the products from FSP Group, having less powerful and high-quality fans. They also feature detachable cables. I can’t recommend these PSUs as the best buy due to the mentioned problem, yet you may want to consider them if silence is your priority. And if your system consumes no more than 350-400W at full load, you may even want to consider the lower-wattage ZM460B-APS that we tested earlier.


I think the leaders of this review are the Antec Neo HE 550, Cooler Master Real Power Pro RS-850-EMBA, Seasonic M12 SS-700HM, and the three models from Thermaltake.

The first three of them are high-quality and exceptionally quiet power supplies. If you want not only top-performance but also a quiet PC, you should consider them. These models are each nearly silent under low loads and do not become irritating at high loads.

Thermaltake’s PSUs feature superb electrical characteristics but are not quiet. They are rather average in terms of noise. Oddly enough, the 850W Toughpower is generally louder than the 550W and 600W models from the same brand, which refutes the widespread opinion that a PSU with a higher wattage rating is going to be quieter than a lower-wattage model under the same load.

The Zalman PSUs would fit into the quiet models category well enough if it were not for the poor stability of their output voltages.

The FSP Epsilons have a couple other drawbacks besides the mentioned instability (the Zalman models are actually based on this platform from FSP). First, they do not fit within the allowable maximum of the output voltage ripple. And second, they are rather noisy at work. So, although I was very pleased with first models from the Epsilon series, the development of the series has taken a wrong way since then.

The Etasis PSU had serious problems with overheat. The fact that it endured a temperature increase of 25°C indicates a large margin of safety, yet a more powerful fan would do it good.

The PSUs from Floston and Ultra are sturdy midrange products. They don’t set any records, but deliver their specified parameters. If you don’t have any specific requirements, these models will be a reasonable choice.

The Rosewill power supply is somewhat odd due its combination of an outdated circuit design (particularly, it is the single model in this review to lack Power Factor Correction) with a fashionable 135mm fan which does not give this PSU any advantages. This model is average even in terms of noise.