FSP Group Power Supply Units Roundup

FSP Group, one of the leading manufacturers of power supplies for computer as well as for other household, industrial and even medical equipment, is better known to people who are assembling computers. Today we would like to introduce to you a number of workhorses from this manufacturer boasting high mounting quality and reliable performance.

by Oleg Artamonov
05/24/2005 | 01:22 PM

FSP Group, one of the leading manufacturers of power supplies for computer as well as for other household, industrial and even medical equipment, is better known to people who are assembling computers. This company is just mostly turning out OEM products, shipping them without colorful boxes, long user guides and enticing ads that off-the-shelf products from better-promoted brands come with.

 

Anyway, the excellent reputation of power supplies from FSP Group (also known as Sparkle Power or SPI Inc.) compels me to write a review about them. Because if you are looking for a good and solid power supply without those freakish Christmas-tree-like lights, then such products may be just what you need.

The units from FSP covered here can be divided into three categories by different versions of the ATX standard (as you remember, these versions differ mainly in the distribution of the load across the PSU’s power rails and, starting from version 2.0, in the replacement of the older 20-pin mainboard power connector with a 24-pin one).


First, we’ve got two inexpensive models of the ATX12V 1.2 standard that can be referred to as the “GTF series” (by the suffix in the model names). Despite the declared wattages of 300 and 350W, their allowable load currents correspond to the standard’s requirements to 250 and 300W units, respectively. Besides the main label, there was also a sticker with the text “+12V/18A MAX” on the 300W model, so I thought it proper to put two load currents into the table above. I will dwell upon this point in more detail below.

The next three units are a step above the previous two – they comply with the ATX12V 1.3 standard that features a higher load capacity of the +12V rail. By the way, don’t be surprised to see the Zalman ZM400B-APS here. This power supply is actually manufactured by FSP Group and is fully analogous to the FSP400-60PFN model. The three models all exceed a little the requirements of the standard (I put the requirements to the 300W unit into the table for the sake of reference since this version of the standard doesn’t describe units of a higher wattage).

The last three PSU models belong to the latest version of the standard, ATX12V 2.0. So, they have two +12V outputs, and there are two numbers in the corresponding column of the table, the maximum load capacity of a unit on the +12V rail equaling their sum. To be exact, the top model, FSP460-60PFN, formally belongs to the server standard EPS12V and is targeted at entry-level servers. But from the home user’s point of view it doesn’t differ from an ATX12V 2.0 power supply: it has the same two +12V outputs and a 24-pin connector for the mainboard, too. These three units all fully comply with the requirements of the standard.

FSP ATX-300GTF (300W) and ATX-350GTF (350W)

These two units are presently the two juniors in FSP Group’s product line-up. They are almost identical both internally and externally, so the snapshot above shows the 350W model only. The 300W one outwardly differs with the label alone.


FSP ATX-300GTF


FSP ATX-350GTF

The PSUs are designed in the traditional way with one main regulator (it’s based on the FSP3528 chip that seems to be a standard PWM controller remarked for FSP) and an auxiliary regulator on the +3.3V rail. The 300W and 350W models only differ in the ratings of some elements. For example, the former has 680µF capacitors on the input of the high-voltage rectifier, while the latter has 820µF capacitors there. The fan speed control circuits also differ. In both models, however, this circuit resides on an independent card fastened to the heatsink with the diode packs so it’s not impossible to equip the ATX-350GTF with the speed-control circuit like the one shown on the snapshot of the ATX-300GTF and vice versa. The type of the installed card probably depends on the PSU production date and/or availability of particular parts at the factory storehouse.

The PSUs are equipped with a three-section line filter (although a simple two-section filter is commonly thought sufficient for a computer power supply) and a passive PFC throttle. Each PSU has such cables:

The wires all have a 12AWG section (about 0.8 sq. mm) and are bound together with nylon ties.

As I have written in the introduction, there was a sticker with the text “+12V/18A MAX” on the ATX-300GTF model. So, I decided to check up if the PSU could really yield such a current. The higher-wattage ATX-350GTF unit didn’t have such a sticker, so the diagrams below are somewhat paradoxical: the higher-wattage PSU has a smaller load on the +12V rail. But I want to remind you once again that I don’t pursue the goal of finding the limit the power supply is going to crash at. I don’t go beyond the max ratings indicated on the PSU’s label, and the limit for the 12V rail of the ATX-350GTF is declared to be 15 amperes.

As you can see, both PSUs successfully – for their class, of course – maintain stability of the +12V as well as +5V rail. The ATX-300GTF quite easily held the 18amp load (yes, the voltages go out of the acceptable limits when there’s a tenfold or bigger misbalance between loads on different rails, but it is not a serious drawback considering the category this PSU belongs to). The PSU also had no problems maintaining the 18amp load on the +12V rail for a long time. I dismantled the PSU and found a SBL2060CT diode pack on the +12V rail. It can hold such a current all right.

The oscillograms of the pulsations on the outputs of both PSUs are absolutely identical, so I only publish the performance graph of the ATX-350GTF. At the maximum load power the ripple is no more than 20 millivolts on the +5V rail and 35 millivolts on the +12V rail. This is far on the safe side of the limits.

The two models display a different correlation between the fan speed and the temperature of the PSU due to the use of different control circuits. The fan of the ATX-300GTF works perceptibly quieter, especially at low loads. At loads below 200W this fan is practically silent whereas the fan of the ATX-350GTF accelerates to 2000rpm at a 100W load already. Both PSUs are not quiet at high loads. But as I mentioned above, it is a separate card that controls the fan speed. It can be easily replaced, so different batches of these PSUs may display varying speed/load correlations.

The efficiency and power factors of the PSUs are absolutely identical, so I publish the ATX-350GTF graphs only. These characteristics of the PSUs just fit into the requirements of the standard, but not more: the efficiency factor at full load equals the allowable minimum of 68%. The passive PFC device helps to fit into the EU requirements (EN 61000-3-2) as concerns the harmonics in the consumed current. It but slightly increases the power factor proper, so it’s of small practical use.

Thus, the ATX-300GTF and the ATX-350GTF are high-quality units for entry-level systems. They are almost blameless in their own class, having stable voltages and small pulsations. The units are not noiseless. At high loads their fans accelerate to rather high speeds (this is in part due to the low efficiency factor, of course), but their noise is going to be quite acceptable in a relatively low-power computer (by today’s standards, I mean).

FSP FSP300-60PN(PF) (300W) and FSP350-60PN(PF) (350W)

Like the two above-described models of the GTF series, this pair of PSUs are also identical externally and almost identical internally. Unlike the GTF series, these PSUs are equipped with a 12cm fan installed in the bottom panel (here and elsewhere I refer to the PSU’s panels as they would be after installation of the PSU into a standard “tower” system case).


FSP350-60PN(PF)

Internally the PSUs are very similar: they are based on identical PCBs but the higher-wattage model has bigger input capacitances (820µF against 680µF) and a larger heatsink with the switching transistors. The central plate of the heatsinks is about 4 millimeters thick.

Like with the GTF series, these PSUs have a three-section line filter and a passive PFC throttle at the input. The regulator based on the KA3511 chip resides on an independent small card positioned perpendicularly to the main one.

The PSUs have the following cables:

The wires have an 18AWG section and are bound together with nylon ties.

Both units correspond to the ATX12V 1.3 standard. It means that, unlike the previous models, they must provide a load current of 18 amperes or more on the +12V rail. The maximum load currents of the PSUs are the same, the top model differing from the junior one only in the allowable total load power.

The stability of the output voltages is good. Of course, these PSUs cannot rival models with auxiliary voltage regulators, but they do perform well in their own class.

The output voltage ripple is the same with both models (at the same load, of course). So, the oscillogram above shows the pulsation of the top model, taken under this model’s maximum allowable load of 350W. The voltage swing is somewhat bigger than we have seen with the models of the GTF series (particularly, you can now see minor spikes at the moments the inverter’s transistors are switched), but this parameter anyway complies with the requirements of the standard.

Both PSU models display similar-shaped dependencies of the fan speed on the load (the regulators are integrated directly into the PSU’s circuit, so they are identical in both PSUs), but the curve of the junior model is somewhat shifted to the left. The last fact may be due to a random discrepancy in the ratings of the elements employed. The PSUs work silently only under small loads. When the load is high, the fans are quick to speed up to their full capacity, i.e. a little above 1200rpm. The air stream from the 12cm blades creates some perceptible noise at such a speed. Curiously enough, the manufacturer of the fans – the fans are Yate Loon D12BM-12 – rates them for 1700rpm, but I don’t have a reason to mistrust the showings of my tachometer.

The power and efficiency factors of the two units are identical, too. Like with the above-described GTF series units, the power factor is higher than that of units without PFC, yet it doesn’t exceed 0.8. The efficiency isn’t high, either. It is 71% at the max load (the ATX12V 1.3 standard is stricter than version 1.2 here and demands an efficiency of no less than 70% at full load).

So, these power supplies are going to serve well in a low-end or midrange computer thanks to their increased load capacity on the +12V rail. Yet, if you need a high current on this very rail, it would be wise for you to consider PSUs of the new version of the standard, ATX12V 2.0, which are discussed later in this review. ATX12V 1.3 units occupy a rather narrow market niche, by the way. On the one hand, many low-end computers are quite satisfied with ATX12V 1.2 PSUs (see the above-described ATX-350GTF, for example). But on the other hand, ATX12V 2.0 models are the best choice for a modern computer. So, you may want to consider 1.3 units as an upgrade to your current system (if your older PSU has crashed or something) that consumes much power from the +5V rail since ATX12V 2.0 power supplies have a low allowable load on this rail and are not quite suited for +5V-oriented computers.

The drawbacks of the reviewed two PSUs from FSP Group are, like with the GTF series, the relatively low efficiency and the loud fan that accelerates to high speeds under high loads.

Zalman ZM400B-APS (FSP400-60PFN, 400W)

We already reviewed this model in our article on Zalman’s power supplies (see our review called 3 Zalman Power Supply Units Roundup). But we’ve changed our test methods since then and now measure such things as cross-load characteristics, efficiency and power factors. So, it makes sense to re-review this PSU, especially considering that it is in fact the FSP400-60PFN model from FSP Group and thus ideally fits into the scope of this article. It complements the model range of power supplies FSP Group is selling under their own brand.

Like the aforementioned models of the PN(PF) series, the ZM400B-APS is an ATX12V 1.3 device, although designed in quite a different way.

What strikes you immediately in this unit is the massive heatsinks that take up almost all free space inside. T- or L-shaped heatsinks usually have a thin horizontal bar and without much ribbing (see the PSUs described above, for example), but here its thickness is the same as that of the heatsink’s main bar.

The second distinguishing feature of the PSU is its active power factor correction device. This vertically positioned card can be clearly seen in the snapshot above. Among other amenities, the active PFC permits to get along without a switch for choosing the mains voltage. This PSU can take from 90 to 240 volts on its input.

The following cables are present:

The wires have an 18AWG section, except the wires to the ATX12V connector which have a thinner, 20AWG section. This is quite normal for units of the ATX12V 1.3 standard.

The PSU maintains stability on both +12V and +5V rails superbly. It’s head above the above-described models in this parameter and closely approaches PSUs with dedicated voltage regulation.

The ZM400B controls the fan rotation speed smoothly, without those discrete steps as we have seen with the PN(PF) series. The max rotation speed is only 2050rpm. As a result, the fan is quiet even at full load, despite its small size (80mm, but it’s a quality thing – NMB 3110GL-B4W-B30).

On the other hand, the use of a small fan worsens not only the cooling of the PSU proper, but also the ventilation of the whole system case, so you may want to put additional system fans into your computer. But two low-speed fans are anyway much quieter than a high-speed one.

The efficiency factor of this model is higher than that of the previous models, but it couldn’t hit the 80% mark anyway. The power factor, despite the active PFC device, isn’t impressive. Those 0.93-0.94 look good against PSUs with passive PFC, but not too well against many other models with active PFC.

In my earlier review I called this PSU an excellent choice for top-end computers, but things have changed since then. PSUs of the ATX12V 2.0 standard are widely available now and suit much better for the latest generation of computer systems. Thus, the ZM400B-APS, a.k.a. FSP400-60PFN, is still a high-quality and quiet PSU with excellent characteristics, but I wouldn’t now recommend it for today’s top-end computers. Such computers may put a higher load on the +12V rail than this PSU can sustain. The ZM400B-APS is also going to make a good PSU for powerful last-generation systems based around the Socket A platform with a mainboard that powers the CPU via the +5V rail. A high load on this rail leads to a distortion of the output voltages with many PSUs, while the ZM400B-APS is going to handle this situation without problems.

FSP FSP300-60THN-P (300W) and FSP400-60THN (400W)

These two units (the snapshot above shows the FSP300-60THN-P, but the 400W model looks exactly the same) bear a strong resemblance to the earlier-described PSUs of the PN(PF) series, but these two belong to the latest version of the standard, namely ATX12V 2.0.

The units are almost identical inside, like the PN(PF) pair again. But on a closer examination you can note certain differences in the placement of the components. The two models differ mainly in the ratings of the elements employed and have absolutely the same PCBs.

The junior model of the two is equipped with a passive PFC device, while the senior model has none (but it may come with a passive PFC, too). As for other and less conspicuous differences, the 400W model has higher-capacity capacitors on the input of the inverter (1000µF against 820µF), a larger line-filter throttle and an additional diode pack on the bottom heatsink that ensures the necessary load current on the +12V rail. Here, the 300W model has one diode pack on this rail, while the 400W model has two packs in parallel.

The FSP300-60THN-P has surprisingly short cables:

As you see, this PSU will only suit for small- or average-size system cases. If you’ve got a large case, the cables may prove to be too short. Fortunately, the cables of the FSP400-60THN are longer: the ATX and ATX12V cables are about 50cm long, the peripheral power cables are about 40cm to the first connector and 20cm more to the second one. Unfortunately, the 400W model also has just a single SATA power cable which is obviously insufficient for modern systems.

The regulator is located on a small card installed perpendicularly to the main card. It is based on a chip marked as FSP3529. The fan-speed regulator is also implemented on a separate card, and these cards of the two PSU models are identical.

The PSUs fully comply with the ATX12V 2.0 standard in their declared load characteristics. This means a very high load capacity of the +12V rail, but also low load capacities of the +5V and +3.3V rails (don’t even look at the current, but at the low total wattage of these two rails – it’s smaller than that of 250W units of the older version of the standard) since modern components are supposed – quite correctly, by the way – to tend to the +12V rail. As a side remark, the Power Supply Design Guide, the fundamental document on every thing concerning the parameters of power supplies, says the max combined load on the +5 and +3.3V rails should be 130 watts, but the declared current of 28 amperes on the +5V rail already gives a higher wattage, so the parameters of the FSP400-60THN – where this load is 150 watts – seem more logical. Anyway, law is law, and the current requirements of the standard are just as they are listed in the table at the beginning of this article.

The load current on the +12V rail is reduced by 1amp for the 400W model in the diagrams above since its protection would get to work as soon as 29 amperes was achieved. So I had to reduce the load a little to record the cross-load characteristic without problems. As you can see, the stability of the output voltages is good, especially with the senior model. This PSU is even among the few units whose cross-load characteristic fully covers the area recommended by the standard (the CLC of many units, including the FSP300-60THN-P, doesn’t meet the recommendation at high loads on the +5V rail; yet, this is such a common thing that I don’t regard it as a serious drawback but rather think that the standard is somewhat too strict here).

The oscillograms of the output voltages look neat, too. The two models produced almost identical oscillograms, so I publish only one picture. The ripple at the max load isn’t even as high as half of the acceptable peak.

The PSUs also use identical fans (Yate Loon D12BM-12) and identical fan-control circuits, so the diagram above shows you the senior model. The graph of the junior model coincides with that quite accurately (at loads below 300 watts, of course). We can see some progress over the PN(PF) series: the fan speed is now controlled flexibly and smoothly, so the speed reaches the maximum only when it is really needed. Overall, the THN series units are quieter at work, even though they use the same fans.

The new PSUs are better than the previous series from the efficiency standpoint, too. Their efficiency is now 80-82%. That’s not a record, but a good result anyway, since the standard recommends 80% and higher efficiency and demands no less than 70%. The two graphs also show the power factor difference between a PSU with passive PFC (its efficiency, but also the power factor are somewhat better than those of the previous models) and without PFC. As you see, there’s some gain from passive correction, although I have repeatedly complained about the small value of this gain.

Thus, the THN series is quite appealing. It includes powerful PSUs capable of meeting the needs of the absolute majority of modern computers. Externally designed like the previous PN(PF) series, and with the same fans, these power supplies turn to be quieter at work due to the improved control over the fan speed.

As for disadvantages, the 300W model has too short cables and only one SATA power cable and the total number of connectors is rather small for a PSU of such wattage. It would be better to have at least two SATA power connectors, six Molex connectors and, preferably, an additional 6-pin connector for the graphics card.

FSP FSP460-60PFN (460W)

This is one of the top-end models in the PSU model range from FSP Group. It formally conforms to the EPS12V standard and is intended for entry-level servers. However, there’s no fundamental difference between EPS12V and ATX12V 2.0 units from the end-user’s point of view, so nothing prevents you from using this PSU in an ordinary desktop computer.

Externally this PSU resembles the above-described FSP400-60PFN (Zalman ZM400B-ASP, that is). Its internal design is original, however, and has no analogs among other PSUs reviewed here. The FSP460-60PFN is made on two horizontal full-size cards (I mean these cards occupy the entire case) on which small vertical cards are additionally fastened.

The bottom card carries input filters, an active PFC, high-voltage capacitors and the inverter’s switches. The top card accommodates a power transformer, output diode packs, auxiliary regulators’ throttles, and output capacitors. The PSU is designed with additional regulation of the output voltages by means of magnetic amplifiers. This design should ensure an ideal cross-load characteristic.

Cooling becomes an issue with such a high component density. The FSP460-60PFN uses short but very thick T-shaped heatsinks with an additional plate fastened above them (the snapshot above shows you the PSU with that plate in place). The plate is additionally fastened to the case of the PSU. The case is made of steel rather than of aluminum, so it’s not effective as a heat-spreader. Anyway, you shouldn’t worry about the bottom panel of the PSU becoming very hot at work – it is a consequence of the heatsink being pressed right to it.

The number of cables and the length of those cables do impress:

The mainboard power cable is hidden into a meshed pipe, while the rest of the cables are bound together with nylon ties. A massive ferrite ring is placed on each of the cables where they leave the PSU. This ring acts as a simple filter. Of course, there are also normal filters at the PSU’s output – throttles as well as capacitors. The total capacitance of the latter is astonishing: six 3300µF capacitors, one 4700µ capacitor and two 2200µF capacitors.

As prescribed by the EPS12V and ATX12V 2.0 standards, the PSU has two +12V rails with independent over-current protection. Such protection is also installed on the other output lines (many PSUs have protection against an overshoot above the allowable total wattage, but not against over-current on some particular rails). You can see the six paired shunts in the photograph of the open PSU.

The cross-load characteristic of this PSU is perfect. Well, that’s exactly what we must expect from a circuit with dedicated voltage regulation. The load capacity of this PSU on the +5 and +12V rails is quite high by the measures of the EPS12V and ATX12V 2.0 standards.

The high-frequency ripple is small: less than 15 millivolts on the +5V rail and less than 50 millivolts on the +12V rail, the allowable maximums being 50 and 120 millivolts, respectively.

This PSU uses a high-quality 80mm fan from Nidec. Its speed is flexibly adjusted depending on the load. Cooling such a PSU requires a strong air stream so the fan speed is higher than 2000rpm even at small loads. The noise mostly consists of the hissing of the air – the buzz of the fan’s blades is hardly audible.

The efficiency factor of the FSP460-60PFN is average, amounting to almost 79% at full load. This is no record, of course, but a good number nonetheless against the other PSUs from FSP Group that don’t boast high efficiency. The power factor is as high as 94% thanks to the active correction. This is not a record, too, but much better than units with passive PFC have.

So, the FSP460-60PFN is an excellent power supply for entry-level servers as well as powerful workstations. It gives you high wattage and excellent load characteristics but at the expense of a rather loud fan. Another big plus of this model is the amazing length of its cables and numerous connectors. Owners of large system cases are going to appreciate this.

Conclusion

I have described and tested different power supply series now being manufactured by FSP Group. The tested units all boast good performance and high build quality. Added their modest pricing, they can be recommended for use in computers of various power needs. Yet it is possible to single out several groups of models that are preferable for particular applications.

First, the ATX-300GTF and ATX-350GTF units are entry-level models that comply with the ATX12V 1.2 standard that is already becoming obsolete. Being a combination of low price and good quality, they will suit perfectly as a power supply for a new, but relatively low-power computer or to replace a disabled PSU in a last-generation computer.

Second, the FSP300-60PN(PF) and FSP350-60PN(PF) models belong to the newer, 1.3 version of the ATX12V standard. They don’t present much interest to the user since they occupy a small market niche. They don’t offer any significant advantages to low-power computers if compared to the GTF series, while the owners of new midrange and top-end computers should consider power supplies of the ATX12V 2.0 standard. I should also note that the 12cm fans of these PSUs are noisier than they might be due to the poor stepping control over their speed.

Third, the FSP400-60PFN model that is also selling as Zalman ZM400B-APS belongs to the ATX12V 1.3 standard, too. This model features an excellent quality of manufacture and a very quiet fan. It doesn’t suit well for powerful computers of the latest generation as it doesn’t comply with the latest version of the ATX standard.

Fourth, the two models of the ATX12V 2.0 standard – FSP300-60THN-P and FSP400-60THN – pleased me a lot, especially the latter model. Alas, the former is much worse in the cross-load characteristics as well as in the ease of use due to the short cables. Well, you can find faults with the cables of the top model, too. The number of available power connectors may be insufficient for some modern systems. If that’s not a problem to you, the models of the THN series are going to be a good choice for a high-power modern computer. They are also free from the defect of the PN(PF) series: despite the use of the same fans, they often work quieter thanks to a more efficient control over their rotation speeds.

Then, the FSP460-60PFN model is an excellent power supply for entry-level servers and workstations. Its characteristics are superb, but its fan is rather noisy at work. For that reason I wouldn’t recommend you to use this PSU in a home computer. The FSP400-60THN would be a better choice, since it can handle nearly any modern computer system. If the noise doesn’t scare you, this server-oriented PSU will work nicely in a home computer, too.

Generally speaking, the power supplies from FSP Group reviewed in this article are typical workhorses. They don’t try to astonish you with an extravagant exterior, shiny cases and blue highlighting, but they do have very good characteristics and ensure stable operation of your computer. So, if you don’t have any extra requirements concerning the exterior of the power supply and if you are not into matching the colors of the screws, connectors and shining-in-ultraviolet cables, you should definitely consider the above-described models when shopping for a new power supply.

Appendix

Some readers requested me to modify the cross-load characteristic diagrams in some way or another like to use the same scale in all of them, to change the delays of the animation or even to publish three different pictures for the three voltages, to mark the load value in the extreme points of the diagram, to denote somehow the requirements of the standard or some other etalon and so on and so forth… Unfortunately, it is simply impossible to satisfy all these requests as some of them are contradictive and others would greatly worsen the look of the diagrams for some PSUs. So I now offer you a compressed file with source data for the cross-load diagrams and a simple program for choosing the diagram-building parameters manually. Our readers who want to have the most detailed information on a particular PSU may find it useful.

The cross-load characteristics of the tested power supplies (click here)

The program to view the source data (click here)