ATX Power Supply Units Roundup. Part IV

We continue testing powerful power supply units available in the today’s market with power capacity ranging between 500W and 600W. Today we are going to introduce to you 7 new models from A.C.Ryan, CoolerMaster, Sunbeamtech, Ultra and Zippy.

by Oleg Artamonov
04/10/2006 | 09:02 AM

Like the last article in our series of PSU reviews, this one is about high-wattage models (500 to 600W) that are selling at retail apart of a system case (for details you can see our previous article called ATX Power Supply Units Roundup: 9 Powerful Models Tested ).


Three out of the five companies whose products I’m going to present to you take part in our tests for the first time. They are Sunbeamtech, Ultra Products and Zippy. The last name is especially thrilling as it belongs to a well-known company that has made a reputation making power supplies for server systems, but now also offers a home user-oriented series under the telling name of Gaming Power.

I want to answer a question from our reader first. It’s not a secret there are many more brands PC power supplies are selling under than actual manufacturers of power supplies, so why don’t I name the real manufacturer for every tested product? Well, I only do so if I am absolutely sure about the name. For example, there can be the manufacturer’s native marking on the power supply’s PCB, or the UL certificate is issued in the manufacturer’s name, or the PSU looks to me exactly like some model from a company which is known to be a PSU manufacturer (you can open up power supplies from Zalman and Nexus to see they are identical to certain models from FSP Group; and no one can have any doubt that FSP is the real manufacturer of its power supplies). But there may be no marking on the PCB; the UL certificate number is not always printed on the label; and I haven’t yet seen all the PSU models in this world – although I’m trying – as to be able to identify one PSU as identical to another. In this case I either do not touch upon the problem of the actual manufacturer of the particular power supply (and it may well be that the company whose brand the PSU comes out under is indeed its real manufacturer) or share with you my suspicions if the unit seems to me similar to a product from a well-known manufacturer. To cut it short, I consider the company whose brand is on the PSU’s label as the real manufacturer of the PSU unless there’s a proof of the opposite.

Now that we’ve cleared this matter out, let’s move on to the power supplies to be tested.

A.C. Ryan Ryanpower2 CableFree ACR-PS2100 (550W)

A power supply from A.C.Ryan’s Ryanpower2 CableFree series once took part in our tests already (for details see our article called ATX Power Supply Units Roundup: Part II). That was a 450W ACR-PS2094 model, and now I am about to discuss the 500W ACR-PS2100.

The two units are identical on the outside: a dark-lacquered and scratches-susceptible case, two cooling fans, and an input voltage of 220V sharp. This unit cannot work in 110V power networks notwithstanding its active PFC.

Here are the PSU’s cable connectors. And it’s all the same as with the ACR-PS2094 model: the cables are all detachable (many such PSUs come with a dead-soldered mainboard cable – have you ever seen a computer without a mainboard?); there are still no separate SATA power cables with +3.3V voltage (SATA power cables are connected to the PATA power connectors); there is still a 4-pin connector for the CPU. The latter is a drawback even if you’ve got a mainboard with a 4-pin connector. Contemporary processors consume high enough currents and any extra connector in the circuit only results in useless heating, unreasonable wattage loss and lower stability of the supply voltage. I think if the manufacturer decides to make a detachable CPU cable, they should make it an 8-pin one. Doubling the number of pins reduces the connector resistance in two times and thus reduces its effect on the whole CPU power circuit in two times, too.

Most of the user manual is dedicated to a harangue on how this power supply suits the modding community and how its detachable cables free you from the necessity to follow the ever-changing industry standards. I don’t quite grasp the author’s idea here because this power supply doesn’t meet even today’s requirements even at the time of its release (it lacks such presently common things as +3.3V power for SATA drives, an 8-pin connector for the CPU, and a power cable for the graphics card).

Inside the PSU case the connectors are fastened on a small card which is then connected to the main PCB. I had some complaints about the soldering quality of the ACR-PS2094. It is much neater here – at least the ends of the wires are tightened with a shrinkable pipe.

The internal design of this PSU is analogous to the junior model and also resembles power supplies from Channel Well (CWT). I do not claim, however, that CWT is the actual manufacturer of A.C.Ryan’s power supplies as there is no evidence of that (the manufacturer’s marking on the PCB, an UL certificate number, etc.) apart of my subjective impressions. I even can’t really name a CWT model that is absolutely identical to the Ryanpower2. There is some similarity, and that’s all.

The PSU features independent voltage regulation (the three choke coils in the left of the snapshot) and active PFC (the separate card with a heatsink in the right of the snapshot).

The PSU complies with the ATX12V 2.0 standard, but the allowable load on the +12V rail is but slightly above the requirements to 400W models (29A as opposed to the ACR-PS2100’s 30A). On the other hand, the Ryanpower2 has a good load capacity of the +5V and +3.3V rails, but the question is if you really need it in practice? It may come in handy in old computers, but such systems can hardly need a 500W PSU.

The following cables are supplied with this power supply:

As I mentioned above, it’s impossible to connect normal SATA connectors with 3-volt power (the enclosed connectors just lack this wire) or a separate graphics card cable to this power supply. You will also have to purchase another mainboard cable if your mainboard has a 24-pin power connector; the enclosed cable has a 20-pin connector only. To be exact, the cable has a 20-pin plug on one end and a 24-pin plug on the other, but if you attach the latter to your mainboard, you’ll have to use the 20-pin plug for the power supply, transferring the problem of connector overload under high currents from the mainboard to the PSU (and this is the very problem the new 24-pin connector has been introduced to solve). The 24-pin connector is necessary: many manufacturers of mainstream PCI Express graphics cards with a peak consumption of 50-60W do not install additional power connectors on their devices, so the mainboard’s power connector has to bear the whole load.

The power cables for hard and optical drives are rather short and may give some trouble to owners of big system cases.

The cross-load characteristic of this power supply looks as superb as can be expected from a model with independent voltage regulation. On the other hand, the unit only reaches its maximum output power when the +5V and +12V rails are both under max load which is a virtually impossible situation in real life. In other words, if this PSU proves insufficient for a modern computer, it will be due to an insufficient load capacity of its +12V rail rather than due to its low overall wattage.

Well, I never happened to get the full declared output power from this power supply. In order to measure the speed of the fans and to check the PSU’s operability under full load, I was steadily increasing the load from 50W to the allowable maximum, stepping 50-100W. The PSU has to work for 30-40 minutes at each step. The ACR-PS2100 passed the step of 450W successfully, but then burned down at a load of about 510W. It put on an impressive show at that, with a series of plopping sounds and sparkles flying out of the fan grids… The subsequent autopsy revealed that the high-voltage section was all burned out (not the active PFC, but the PWM controller itself, farther along the circuit); the heatsink and surviving components were all sooty.

This is why I couldn’t measure the voltage ripple on the PSU’s output.

The power supply has two fans, an 80mm fan from an unidentified manufacturer and a 92mm JDDA SDF9225S. Its speeds are varied depending on the temperature in a linear-like manner. The PSU is quiet at low loads, but at high loads the fans become audible, even though not actually loud.

The efficiency of this power supply is rather low as today’s PSUs go – a mere 74% at the maximum. Of course, this meets the requirements of the standard, but you should be aware that many other PSUs are 80% and more efficient.

Thus, the Ryanpower2 ACR-PS2100 hasn’t any big advantages over its predecessor, the ACR-PS2094 model. The unit couldn’t work under the maximum specified load and it is also not free from the drawbacks of the junior model like too few cables or a paltry selection of connectors (you can’t attach a graphics card power cable; there are no connectors for SATA drives with +3.3V power). Contrary to the boastful epithets in the manual and on the box, this power supply is not better, but inferior to many competing products. The only definite plus of this PSU is the excellent stability of its output voltages.

CoolerMaster RealPower RS-550-ACLY (550W)

I can’t but draw a parallel here. Almost one year ago we published a review in which a 450W power supply model from CoolerMaster was tested second after an A.C.Ryan Ryanpower2 ACR-PS2094 – the junior brother of the above-described ACR-PS2100! So, once again we’ve got RealPower and Ryanpower2 power supplies in the same review, but they have grown 100 watts more powerful now. This extra wattage doesn’t help the Ryanpower2 much as you’ve just seen. What about the RealPower?

The power supply is in a steel case, painted a lusterless black color. It is cooled with a single 120mm fan and lacks a 110/230V switch – just like the Ryanpower2, but for another reason. The RS-550-ACLY features an active PFC device that supports a full range of input AC voltages from 90 to 265V.

The real manufacturer of this PSU is AcBel Polytech Inc., which is indicated by the UL certificate number on the label: E131875.

The PSU doesn’t have detachable cables and there is no protective plastic ring where they go out of the case, just a hole with a rolled-in edge. Some users have reported their concern that the cable insulation may wear off against the edge of the case with time, that’s why I mention this at all. Personally I do not think the cables can be damaged under normal use and don’t see any big difference between a plastic ring and a rolled-in edge of the case.

The unit is analogous to its 450W predecessor on the inside, with active PFC (a Fairchild ML4800CP chip is employed which combines PFC and main regulator controllers; it is the chip with the blue paper label in the snapshot) and group voltage regulation. So, this is quite a regular power supply as they are today. A curious thing is that the PFC choke is wound on an E-type rather than on a toroidal core (in the bottom left corner of the first snapshot). A non-typical thing too is the ferrite ring with a brown-wire loop inside that you can see in the top left corner of the snapshot. This is a current transformer the power consumption indicator, which is to be installed in the front panel of your system case, is attached to.

Don’t be misled by those two tall high-voltage capacitors into thinking that the RS-550ACLY is a classic-design unit with two capacitors connected in series. Both the capacitors are for 450V voltage (the operating voltage on the active PFC’s output is about 400W irrespective of the input voltage) and are connected in parallel, yielding a combined capacitance of 2*150=300µF.

You receive a power consumption indicator along with the power supply. It is inserted into a 3.5” bay of your system case. The box with the indicator contains a micro-ammeter and a blue highlighting LED; it works only with this particular power supply. Owners of light-colored system cases can replace the indicator’s faceplate with a silvery one, which is also included. The showings aren’t very accurate since the indicator has a rude scale and is also not a very high-precision measuring instrument, either. So, it should be rather viewed as decoration. I must acknowledge, however, that the indicator did report the power consumption of the PSU correctly at loads of about 300W. It would yield exaggerated numbers at higher loads and understated ones at lower loads.

CoolerMaster cunningly declares peak currents far above typical currents for the three 12V lines. What’s the hitch? Like with any other ATX12V 2.0 power supply, you should rather look at the combined power of the 12V power rail. It is 360W here, so the combined current in all the three 12V lines cannot be higher than 30A whether it is a peak or not.

As I have written in my other reviews, power supplies with multiple +12V outputs have in fact only one +12V power rail inside which is split in several lines to comply with the requirements of a safety standard. The standard demands that the current in each line was not higher than 20A (the manufacturers even safeguard themselves setting the limit at 18A). Thus, none of the individual lines can be said to have any peak current at all since any current below 18A is normal for it, any current over 18A triggers the protection. They can put the limit of 18A on all the +12V output lines and the power supply will still be compliant with the standard. It wouldn’t require any work on the developer’s part: the PSU’s internals remain the same, and only the protection-triggering thresholds are changed. And the power supply wouldn’t become more powerful for that, as you see.

Thus, a peak current can only be specified for the common +12V rail, which is inside the PSU before it is divided into the separate outputs. But it is for this rail that there is no peak load specified here.

For the same very reason, there is no technical meaning in the words of PSU manufacturers about extra stability, extra wattage, and any other extras that multiple +12V lines bring about. You’d better just filter out this white noise you hear from the marketing departments. Stability, wattage and other characteristics all come from the “basic” +12V power rail and do not depend at all on how many lines this rail is split into on the PSU’s output. In other words, a PSU with a single 12V/36A line is not any worse than a PSU with two 12V/18A lines.

The RealPower RS-550-ACLY has the following cables and connectors:

There’s nothing I could cavil at. The PSU offers all connectors necessary for a modern computer, even a separate 8-pin CPU connector and as many as two power connectors for graphics cards, for SLI or CrossFire configurations. Of course, the SATA power connectors have a +3.3V voltage unlike the deficient connectors of the above-described Ryanpower2. I only think that perhaps it would be simpler for the user to have splittable CPU and mainboard connectors (one 4+4 connector instead of the two 4- and 8-pin ones and one 20+4 connector instead of the 24-pin one plus the adapter).

Lacking independent voltage regulation, the PSU still yields very stable voltages. The output voltages remain within acceptable limits almost across the entire diagram. But like with the Ryanpower2, the maximum output power can only be achieved here if all the rails are fully loaded, which is virtually impossible in a real computer, as opposed to our specialized testbed.

The load being 525W, the output ripple was 23, 54 and 14 millivolts on the +5V, +12V and +3.3V rail, respectively. There is almost no low-frequency pulsation (at the double mains frequency or 100Hz in our case).

The RS-550ACLY uses a blue-highlighted CoolerMaster A12025-25BB-2AN-PI fan (which, however, bears its native marking “Protechnic Electric MGA12012HB-O25”). The graph I got is an exact copy of the one you will find in the user manual. The fan speed remains constant at about 1200rpm under loads of below 200W. At higher loads the speed grows up linearly until it reaches 2400rpm.

The graph is shaped in the same way as the fan speed graph of the RS-450-ACLY model, but is shifted upwards by 400rpm (the speed was varied within a range of 800-2000rpm in the RS-450-ACLY), so the new model is noisier than its predecessor under the same load. Still the noise characteristics of the new PSU are acceptable; I guess you will be satisfied with it, if you do not seek for absolute silence. If you do, consider other models, perhaps even the mentioned junior model from CoolerMaster.

The unit boasts good efficiency (over 80% on average and 84% at the maximum) and power factors.

My impressions about the RealPower RS-550-ACLY are generally on the positive side. This is not an exceptional product and, despite its high total output power, corresponds rather to 400-450W models from other manufacturers in its effective load capacity, i.e. considering the way the load is distributed along the different power rails in a modern computer. But it has good electrical characteristics (excellent stability of output voltages; low level of pulsations; high efficiency) and high quality of manufacture. The CoolerMaster RS-550-ACLY is going to be a good choice for many users, except for those who value silence above everything else. Its noise characteristics are rather average and it is a little louder than the junior model from the same manufacturer (but otherwise the RS-550-ACLY has done better than the RS-450-ACLY in my tests).

Sunbeamtech Nuuo Series SUNNU550-EUAP (550W)

It’s the first time a product from Sunbeamtech comes into our power supply reviews, so I can’t tell you anything about the company except that its Tuniq Tower fans are reported to do well in tests. My search for the real manufacturer of the PSU produced the name of Andyson International Co., Ltd., which I cannot tell you much about, either.

The PSU came to our lab in a cute-looking cardboard box which contained, besides the PSU proper, a very detailed user manual (my thanks go to the manufacturer for it as most other PSU manuals merely contain a couple of tables with specs and a couple of pictures to illustrate the connections), a set of detachable cables, and a fan control panel to be inserted into a 3.5” bay of your system case.

The dark lacquered steel case of the PSU is cooled with two fans, 120mm and 80mm. It’s not very usual to have different-size fans in a power supply; I’ve only seen this in products from Wintech (that are selling under a number of trademarks). The manufacturer claims it to be more efficient than the classic cooling solution with a single 120mm fan.

Although with active PFC, the power supply supports an input AC voltage of 220V only.

On the other side of the unit, there is an abundance of connectors: four connectors for PATA drives, two more for SATA ones (the narrow 5-pin connectors in the bottom right), an 8-pin CPU power connector, a 6-pin graphics card connector, and a group of 2- and 3-pin connectors that I’ll discuss later on. A.C.Ryan would do well to follow this example.

The power connectors for your drives are plugged up with protective rubber caps. There’s a reminder on the caps on the PATA power connectors for you to be careful to connect the cables correctly. Yes, the connector has two keys (cut-off corners) to prevent incorrect orientation, but it is made of soft plastic, so the keys won’t help much if you apply some effort. I hope the manufacturer will see to this problem in the future. Until then, you just have to be careful. Note that there is no such text on the SATA power connectors’ caps because the different shape of the connector makes it impossible that you position it in the wrong direction.

It’s all plain inside this PSU: group voltage regulation, active PFC on a separate card, two 820µF capacitors on the output (the effective capacitance is 410µF since they are connected in series). The small green card on the heatsink (there are in fact two of them there, but the other hides under the mains connector and is barely visible in the snapshots) carries fan control circuitry – not the ordinary thermal control, which is located elsewhere, but an auxiliary circuit for the external fan management unit that I’ll describe below.

The output connectors are fastened on the common PCB the wires are soldered to. This looks neat enough, but I’ve seen prettier solutions in other PSUs.

The only stationary cable here is the mainboard’s one, with a 24-pin connector (a 4-pin part of it is detachable); its length is 47 centimeters.

And here’s what you’ll find included with the PSU:

The two screened cables with built-in filters look interesting. They are designed similar to cables from OCZ that you have seen in our earlier reviews: there is a ferrite ring and a couple of capacitors at the end of a cable which form a simple LC filter to suppress high-frequency interference. The efficiency of the filter is dubious, but why not if it doesn’t make things worse?

The characteristics of this model are like those of the previous one: the combined current on the +12V rail is limited to 30A (with short-term peaks up to 35A) while the load capacity of the +5V and +3.3V rails is high and their combined allowable power is an impressive 280W (it is usually no higher than 200-220W and even 150W in the latest PSU models). So this product looks a hybrid between versions 1.2 and 2.0 of the ATX12V standard.

The most thrilling accessory included with the power supply is the panel to display temperatures and fan speeds.

The panel shines in blue, showing one temperature and indicating the fact that one fan is rotating (I can’t say it shows the speed of the fan since there is not a number here but an animated icon). The panel looks very elegant.

Among the three fan connectors of this power supply, two are meant to control the PSU’s own fans and the third to connect an external fan that will get the same voltage as is supplied to the PSU’s 120mm fan. The two 2-pin connectors below the fans’ ones are the outputs of temperature sensors, one of which is located on the heatsink with the diode packs and the other measures the air temperature inside the PSU case.

The panel allows controlling only one fan and only one temperature – you choose them by attaching the panel to the necessary connectors. The PSU manufacturer, however, does not recommend to adjust manually the speed of the 120mm fan. They say it is already quiet enough, and you may overheat the PSU by setting the fan speed too low. This is why if you remove the plug from the 120mm control connector (it is sealed with a warranty sticker), the PSU warranty is automatically shortened from two years to one.

Many other power supplies combine manual adjustment with automatic thermal control, but the SUNNU550 divides them sharply: the fan is managed automatically when the panel’s knob is in the extreme position (or if the panel is not attached at all) and manually at any other position of the knob.

Besides the fan control function, the panel also features a Smart Protection output. It is attached to the appropriate PSU connector and switches the PSU fans to automatic control (irrespective of the position of the panel’s knob) if the heatsink temperature is over 70°C. Moreover, if the temperature is higher than 90°C or the fan speed is below 400rpm, the panel emits a warning squeak and blinks a warning symbol at you.

I must also say that the PSU uses standard thermo-sensors, standard fans and standard connectors, so it can be attached not only to its own panel, but also to fan & temperature control panels from other manufacturers. There is another collection of cables included with the PSU for that purpose so that you could connect both the sensors and both the fans at once. It’s nice to see such care about the user when many other manufacturers deliberately make their designs incompatible with others’ (using non-standard connectors, for instance) for the user to buy accessories from that manufacturer only.

The unit yields rather stable voltages, especially considering the allowable load range for the +5V and +3.3V rails, but I wish the +5V were lower in the bottom right of the diagram since it is there that most modern computers belong in.

At a load of 540W the voltage ripple was 20, 23 and 28 millivolts on the +5V, +12V and +3.3V rail, respectively.

The fans installed in the PSU are labeled as “Sunbeamtech”, so I can’t tell you the name of their actual manufacturer. Both fans are black, with shiny blades and without any highlighting. The number of blades is somewhat unusual: the smaller fan has 11 and the bigger has 9.

As promised by the manufacturer, the 120mm fan is really very quiet. Its speed isn’t higher than 1500rpm even at full load (the speed was managed automatically during the tests). It’s worse with the 80mm fan which can spin up to 3000rpm and become rather noisy. The control panel allows you to find an optimal balance of the speeds of the two fans, though.

The efficiency of this PSU isn’t high, barely notching 80%. The power factor is quite typical for a unit with active power factor correction.

So, Sunbeamtech has made a successful enough debut in our reviews. The SUNNU550 features good electrical characteristics and excellent functionality too. It has detachable cables (and unlike the cable-free Ryanpower2, the SUNNU550 is up to the latest requirements), allows to control the speed of the PSU fans and even connect an external fan, and has two temperature sensors accessible from the outside. What’s important, the additional features do not look like useless trumpery, but are really well thought out from a technical standpoint. So, that’s indeed as fine a debut as you may wish.

Ultra Products X-Finity ULT-XF500 (500W)

Ultra Products is a new name in our reviews, but my search for the actual manufacturer of the power supply yielded immediate results. It is the Taiwan-based Wintech who is already well known to us by power supplies of the MGE and SinTek brands. The unit can be identified as Wintech visually, too. No other company uses bright yellow capacitors and blue transformer cores.

I noticed a funny thing about the box. If you put it on a side, the big pretty slogan that reads “Limited Lifetime Warranty” transforms into an alarming “Limited Lifetime”. :)

The PSU has a dark lacquered case and is cooled with a single 120mm fan. It can work in 110 and 220V electric networks, but with a manual switch only. The switch is to be found in quite an odd location: it is inside the computer when you install the power supply. This makes sense, though, because you won’t need that switch after you’ve assembled the computer. It’s even right to hide the switch from accidental presses since the PSU would just go kaput immediately if the input voltage switch were set to a wrong position.

The internal design of the PSU is quite typical, without any extras. There is neither active nor passive PFC here (so, this version of the unit cannot be supplied into Europe); the regulator is based on a KA7500 chip (it is an exact analog of the TL494, one of the oldest PWM controllers). There are two 1000µF capacitors on the PSU’s input. The bright colors of the different components are typical of Wintech’s products.

There is a 25mm gap between the PCB and the front panel. Our colleagues from Silent PC Review suppose it has something to do with efficient cooling, but I’m inclined to take a simpler view of the things. I think this gap is the consequence of Wintech making all its power supplies using the same barebone. For example, the power supplies from MGE Vortec and SinTek (both are manufactured at Wintech facilities) have one 80mm fan besides the 120mm one, and the ULT-XF500 uses the same PCB whose dimensions are meant for an additional fan. Moreover, Wintech uses the same platform in PSUs that are cooled by 80mm fans only.

I couldn’t discover the manufacturer of the fan. It only bears an “Ultra” sticker. A bit more than one third of the fan is covered with a piece of translucent plastic. It is where the fan has contact with the external perforated panel of the PSU.

Having a lower wattage in comparison with the above-described models, this PSU has a higher allowable load on the +12V rail. It is 32A. The combined load on the +5V and +3.3V is just a little over the requirements of the ATX12V 2.0 standard, in which this load was greatly reduced relative to the previous versions.

The PSU offers the following cables and connectors:

Instead of separate wires, we have flat silver cables here. The manufacturer claims such cables help to maintain proper airflows inside the system case, but I doubt that airflows can be affected much by a simple change of the shape of cables. I think the main advantage of FlexForce cables, as they are officially dubbed, is of an aesthetic nature. They do have drawbacks, by the way. First, you may find the lack of color-coding of the wires inconvenient. And second, a flat cable can be bent in one direction only, so you may have some troubles trying to lay them out neatly in the case.

The cross-load characteristic of this PSU is average. On one hand, the diagram covers almost the entire range of loads allowable for this model, but on the other hand, all the three voltages vary greatly and deflect rather far from their nominal values under loads typical of modern computers.

At a load of 475W the voltage ripple was 40, 85, 26 millivolts on the +5, +12, +3.3V rail, respectively. A low-frequency, 100Hz constituent accounts for the most of the pulsation.

The fan speed management system works rather oddly in this power supply. The speed is altogether constant at 1000rpm under loads of below 150W. It then grows up quickly under higher loads, reaching 2100rpm. As a result, the PSU is quiet under small loads, but audible under a load of 250-300W. Moreover, the plastic sheet that partially covers the fan rattles as in some other power supplies from Wintech. To solve this problem, tighten the screws that hold the sheet and add a few drops of glue – or just remove that piece of plastic altogether!

The PSU is 81% efficient at best, which is a good, but not record-breaking result. The power factor varies from 0.65 to 0.67 (as I said above, this unit lacks any type of power factor correction).

The X-Finity ULT-FX500 has done as well in my tests as the other Wintech units I’ve met. It is an average-quality product without any serious advantages, but with acceptable parameters. The drawbacks of the ULT-XF500 in particular are a strong 100Hz output ripple, lack of any type of power factor correction, and a rapid increase of the fan speed at higher loads, which makes the PSU rather noisy. If these drawbacks do not disturb you, you may be pleased with the ULT-FX500, which is a good enough modern power supply.

Ultra Products X-Finity ULT-XF600 (600W)

The ULT-XF600 is manufactured by Wintech and is based on the same components as the previous model. The difference between the two is mainly in the cooling system design. For some unclear reason the senior model is cooled with two coaxially mounted 80mm fans rather than with a single 120mm fan.

The steel case of the PSU looks matte in the snapshot but it is steel indeed.

Like the ULT-XF500, this power supply supports input voltages of 110V and 220V. The switch is placed in its traditional location here, next to the mains connector.

It’s easy to see the ULT-XF500 and ULT-XF600 are based on identical PCBs but differ in the ratings of the components (the high-voltage capacitors on the input of the senior model are 1500µF rather than 1000µF now, and the heatsinks are larger). One of the fans is smugly nestled in the gap between the PCB and the panel that was empty in the ULT-FX500; the engineers had to make the case a little longer to fit the second fan.

The load capacity of the PSU has increased according to its wattage. The allowable current on the +12V rail is 3A higher in comparison with the junior model, but the max load on the +5V and +3.3V rails hasn’t changed much.

The PSU is equipped with the following cables and connectors:

Like those of the previous model, the cables are flat and silvery. Curiously enough, they have put multicolored marks on the wires where the cables are soldered to the power supply’s PCB.

Alas, the cross-load characteristic of this power supply is worse than its predecessor’s: the +5V and +3.3V voltages are rather stable, but the +12V is obviously set too low. As a result, this voltage is going to sag to 11.5-11.6V in a modern computer where it has to bear the main load while the low-voltage rails are under a load of 30-50W only.

At a load of 560W the voltage ripple is 35 millivolts on the +5V rail, 70 millivolts on the 35V rail, and 40 millivolts on the +3.3V rail. The higher-capacity capacitors on the input do not help the PSU cope with a low-voltage pulsation which still persists.

The PSU is cooled with two 80x80x25mm fans of an obscure origin (they are labeled “Ultra”, but I doubt Ultra Products itself manufactures fans and transports them to Taiwan for Wintech). The speed of the fans depends almost linearly on the PSU temperature. The unit is quiet, almost silent, at low loads, but at loads of 200-250W the fans are perfectly audible.

The efficiency and power factor graphs for this unit are almost the same as the ones for the ULT-XF500. This is expectable considering their similar circuit design. The efficiency is 81% at the maximum.

So, the X-Finity ULT-XF600 power supply has the same pros and cons as the junior, ULT-XF600 model. This PSU doesn’t have any exceptional traits (I don’t count in the silvery cables among such since owners of non-transparent system cases won’t notice them at all while other users are offered much more exciting options like shining cables, for instance). Its +12V voltage is set too low; it has a rather strong low-frequency pulsation on the output and is rather loud under high loads. If you are not worried about these drawbacks, the PSU will suit you fine. It is quite capable of powering up a modern computer.

Zippy Emacs HP2-6500PE (500W)

I guess many of our readers have heard the name of Zippy Technology, yet the company is mostly known as the manufacturer of server-oriented Emacs power supplies. Some time ago they decided to enlarge their product range to cover home users, too. To address the gaming community they released a series of power supplies under the name of “Gaming Power”.

The 500W HP2-6500PE model is a midrange Gaming Power product. The series includes models a hundred watts more and less powerful than this.

The PSU has a black steel case of a non-typical design with an L-shaped rather than U-shaped cover. When removed, the cover exposes the PSU’s contents from the top and back.

The HP2-6500PE betrays its server origin right away: it uses an 80mm fan under a pressed-out (rather than wire) grid; the mains connector has lugs for a spring that would hold the cable down and there is a modest green power indicator above it. Zippy seems to have just re-released an ordinary server power supply, having painted it black and put it into a beautiful box. I’m afraid that’s not exactly what home users need whose requirements differ from server integrators’.

The PSU is based on a Fairchild ML4800CP chip, which is a PWM controller and an active PFC controller in a single case. The components are packed densely; the heatsinks are large and occupy almost all of the free space (not surprising, considering the 80mm fan). The odd plastic square on the heatsinks is used for the top cover which is fastened to it with two screws.

The power supply is universal in its specs. On one hand, it can yield up to 432W through the +12V rail. On the other hand, the +5V rail has a very high load capacity by today’s standards, but modern computers do not actually need that.

The PSU is equipped with the following cables and connectors:

The only problem you may have with the connectors available here is that there is a 24-pin EPS12V-compliant mainboard connector while most other PSUs offer splittable connectors with a detachable 4-pin part.

It’s all splendid otherwise: two connectors for graphics cards (and you wouldn’t probably want to buy a PSU of that high wattage for a computer with a single graphics card) and four power connectors for SATA drives – just what a high-wattage modern power supply should be equipped with.

The cross-load characteristic of the PSU looks well, although not quite ideally. The main problem is that the +12V voltage bottoms out under +12V-oriented loads. And it is typical of all modern computers, especially of those with a SLI or CrossFire graphics subsystem, to consume a lot from the +12V power rail and little from the other ones. On the other hand, the +12V voltage goes out of the acceptable limits at greatly misbalanced loads only, so this is not such a terrible defect.

I tested the PSU with an APC SmartUPS SC 620 uninterruptible power supply. The UPS would indicate overload at a load of about 365W (AC power source) and 320W (when switching to the batteries). There were no problems under loads of below 320W: both the power supply and the UPS correctly worked through disconnection from the 220V electric network.

At full load (500W) the voltage ripple was 37 millivolts on the +5V rail (mostly low-frequency pulsation as the lower of the two oscillograms shows), 71 millivolts on the +12V rail, and 32 millivolts on the +3.3V rail. The pulsation becomes weaker at lower loads.

The unit is cooled with an NMB-Matsushita 3110GL-B4W-B54 fan whose speed is adjusted depending on the temperature. The PSU is among the quieter models with 80mm fans, but it cannot compete with 120mm fans. The fan is audible at work.

The efficiency of the PSU is good at 84%. The power factor is less good, being 0.97 at best, which is lower than with many other PSUs with active PFC. The difference is small, though.

So, it is clear the Zippy HP2-6500PE is originally a server power supply which has been released anew in a colorful box for home users. The server origin of this model is betrayed by its appearance as well as parameters, particularly by the high load capacity of the +5V rail. This is a high-quality and reliable power supply capable of coping with any modern computer, but you won’t like it if silence is your priority. The speed of its fan is too high even under low loads. Owners of powerful, probably overclocked computers with a top-end processor and a SLI or CrossFire subsystem consisting of two top-end graphics cards should be aware of the sagging +12V voltage. This voltage is likely to go down to 11.5-11.6V with your configuration, potentially spoiling your overclocking attempts.

Zippy Emacs PSM-6600PE (600W)

Although this is the topmost model of the Gaming Power series, it looks even more like a server power supply than the HP2-6500PE: an unpainted steel case, a punched-out fan grid, lack of any auxiliary elements – there is not even a power on/off switch here. That’s not how we’ve come to think of an expensive power supply for a home PC. Well, you cannot admire the beauty of a highlighted PSU fan anyway unless you have a system case with a transparent panel, so why spend extra money for it? Let’s better talk about things that are really important. I mean the parameters of this power supply.

This model differs internally from the 500W model, although they are based on similar platforms. The PSM-6600PE uses an ML4800CP chip, too, as the main controller.

A steel bracket with a gray insulation on top is fastened to the heatsink with the diode packs. It drew my attention, but I could find no purpose for it other than providing more mechanical robustness to the PSU. It cannot take any great part in cooling the PSU since its thermal conductivity and the thermal conductivity of the PSU’s panel it borders upon is too low.

The specification says the allowable load on the +12V rail has grown more in comparison with the previous model than the total wattage – by 120 watts, to be exact. The load capacity of the +5V rail has diminished to 30A, but you shouldn’t worry about that. As I have repeatedly said in my reviews, modern computers do not consume much power from the low-voltage rails anyway.

The PSU offers the following cables and connectors:

So, we see the same picture as in case of the HP2-6500PE PSU: there are all the necessary cables present, but if you have an old mainboard with a 20-pin power connector, you’ll have to use an adapter (although i really doubt that the owners of older mainboards will ever go for a powerful PSU like that anyway).

There is no plastic ring around the wires where they go out of the PSU case. Some users have reported they have seen wires losing their insulation due to such design, but I do not think this is a really big problem. It’s really hard to cut the thick and durable insulation with a smooth rolled-in metal edge of the PSU case.

The cables are all hidden in plaited tubes.

The cross-load diagram for this power supply is a treat to my eye. Featuring independent voltage regulation, the PSU never allows any output voltage go out of the acceptable range. Moreover, the output voltages do not deflect by more than 2% from the nominal value, the allowable deflection being 5%! That’s an excellent performance, among the best results I’ve ever seen in my tests.

At full load (600W) the output voltage ripple was 22, 50 and 15 millivolts on the +5V, +12V and +3.3V rail, respectively. The pulsation spectrum differs: only high-frequency pulsation on the +5V and +3.3V, and mostly low-frequency, 100Hz pulsation on the +12V (the lower oscillogram).

When working with an APC SmartUPS SC 620, the UPS would indicate overload at a load of about 365W (AC power source) and 250W (when switching to the batteries). The result is 70W worse than that of the previous model. So when there was a load of 250W on the PSU, the UPS’s overload protection woke up when switching to the batteries and the UPS shut down the computer.

The PSU is rather quiet at power loads of 250W and less (but I can’t call it silent because the hiss of the air is distinctly heard; most PSUs with 120mm fans are much quieter). When the load goes up, the PSU is not just noisy, but actually loud.

To make it clear, I’ve got two regular system cases on my desk and four powerful fans (two 92mm Thermaltake and two 80mm Delta Electronics) to cool the testbed. This arrangement isn’t very quiet even in the big room where there are a few more people and half a dozen more computers besides myself and mine. And the PSM-6600PE managed to make itself heard and prominent above everything else with its irritating loud high-frequency hiss of its fan! By the way, a Sanyo Denki SanCooler 80 fan (part number 9A0812S402) is installed in it.

The PSU is as efficient as the HP2-6500PE (about 84% at the maximum), and its power factor is higher at 0.99.

So, the top-end model in the Gaming Power series leaves an ambiguous impression. It really has excellent electrical parameters like super-stable output voltages, good efficiency, low voltage ripple at high output power, but it is so loud at high loads that I doubt you’d want to use it not only at home but even in office. The problem is not about the noise as such, but about its spectrum, which is shifted upwards and is thus distinguishable against the common chorus of PC fans. It is really annoying for the human ear. You can also encounter problems using this power supply with an UPS. I’d recommend a high-power and high-quality UPS for this power supply for the switching-over to the batteries was done correctly.

On the other hand, the noise factor becomes less important if we talk about a powerful gaming machine with two graphics cards. Few people who own such computers do anything to improve their cooling while the stock coolers of modern top-end graphics cards are generally rather noisy. In this case, the excellent electrical parameters of the PSM-6600PE may outweigh its noisiness.


Of the five PSU manufacturers whose products have been presented to you in this article, only two have taken part in our earlier reviews. These are A.C.Ryan and CoolerMaster.

Alas, A.C.Ryan has nothing to be proud of. I already noticed some drawbacks in their models of the Ryanpower2 series before, but the new and allegedly higher-wattage model even adds new drawbacks to the older ones. The new PSU can’t even yield the declared maximum output power at all! The included cables are scanty; the available connectors do not permit to connect a modern computer without adapters. And I wouldn’t be that much disappointed if it were not for the boastful words the PSU description is interspersed with.

CoolerMaster’s 550W model is, on the contrary, a lucky continuation of the older 450W model. The RS-550-ACLY can easily power up almost any modern computer and comes with as many cables and connectors as you need to attach any peripherals without adapters. This PSU has a single drawback. It is noisier than the previous model since its fan rotates at a 400rpm higher speed under all loads.

Sunbeamtech did well among the debutants. Although this manufacturer is not widely known, its model proved to have good technical parameters as well as rich and curious accessories and functionality. Telling you the truth, the PSU from A.C.Ryan should have been like that if its developers justified the writings of its marketing department. The PSU offers cables just for everything while the temperature & fan management system deserves my special praises. I’ve met only two PSUs with temperature control before: Magnum and SinTek WIN550XSPX-X. Both display the temperature value on an integrated screen placed in a very inconvenient location. As opposed to that, the SUNNU550 can be attached not only to its native panel for a 3.5” bay, but to control panels from other manufacturers, too.

The products from Zippy are good, too. At least they do have excellent electrical parameters, but may be found intolerably noisy by many home users (especially the senior 600W model with its annoyingly-sounding fan). I think Zippy would do well not to adapt server PSUs for pretty-looking boxes, but to release a real home product with a quiet 120mm fan, as the major manufacturers do.

The last goes Ultra Products whose power supplies are actually manufactured by the well-known Wintech. These are average products with some pretensions (silvery cables!), but without anything really exceptional about them. Their quality of manufacture and their electrical characteristics are just average.


Click here for the cross-load characteristics of the tested PSUs.

Click here to download the viewer of the cross-load data.