Corsair sf750 review in Spanish (complete analysis)

Corsair has been offering its range of SF fonts to the market for almost three years, which stand out for being much smaller than normal when making use of an SFX format, intended for smaller boxes. Just a few months ago, the brand expanded this range with 80 Plus Platinum efficiency models , highlighting a new model that we are happy to analyze today: the SF750 Platinum, a strong bet with no less than 750W of power in a chassis of less than 1 liter volume.

Regarding its predecessors, in addition to the improvement in efficiency, we have improvements in sound and internal quality, the use of cabling with “sleeving”, and the inclusion at last of an SFX to ATX adapter. 100% modular cable management, semi-passive ventilation mode and 7-year warranty are maintained.

While there isn't too much variety on the SFX font market, the competition is still fierce. Will Corsair manage to stand out from its competitors with its new bet for the high-end? Let's see it!

We thank Corsair for the trust placed in sending this source for analysis.

Technical Specifications Corsair SF750 Platinum

External analysis

The box summarizes the most relevant characteristics of the source. On the back, this information is expanded giving also the efficiency and sound data. It highlights the use of a semi-passive mode that should keep the fan off up to 300W of consumption.

From the first moment, we are surprised by the abysmal difference in box dimensions, compared to another ATX source from the brand. And, as we say, here is the key to this font: its extremely small format with only 125x100x63.5mm dimensions (from the font).

The open box shows us excellent protection that will ensure that the source reaches your hands in perfect condition.

In addition to the PSU and cables, we found a bag with all the wiring, velcro straps, nylon cable ties, a user manual, a warranty information document, a Corsair sticker and a set of screws.

We take the fountain out of its bag and look at our little protagonist.

The difference in size when we buy this SF750 with an ATX format AX850 is abysmal. And it is that this SF750 reaches a very high power density, of no less than 945W per liter. This is a 'dummy' calculation to illustrate its ultra-compact size.

To see even more clearly the small size it has, we can compare it with everyday objects such as a normal-size smartphone.

In general, the external appearance of the SF750 is refined but at the same time striking. The truth is that we liked it a lot, and we will like it more if the beauty continues inside.

We highlight the inclusion on the front of a sticker (which we will obviously remove later) that indicates in English that the source fan will not work at low to moderate loads, so that no user will be frightened when viewing it off. The Spanish translation, on the other hand, has nothing to do with what we have just indicated, as it is very poorly done.

As expected in a high-end-oriented product, this source is completely modular, using Corsair type 4 cables which are universal among most of their ranges, and can be purchased separately (you can check compatibility here)

Cabling management

The included cables have black sleeving, so they go one step beyond traditional or flat meshes. This sleeving, as we saw in the RM850x White, is of very good quality and is very good at handling.

The wiring provided is as follows: 2 EPS, 4 PCIe, 8 SATA and 3 Molex connectors. In the case of PCIe connectors, these are arranged in two cables and not four, something that we did not like too much but at least it does not happen in the SF450 Platinum and SF600 Platinum. Anyway, buying this 750W model in front of its younger sisters would make sense in the case of including a graphics card that consumes a lot, so at least 2 PCIe would be used.

Cable length: shorter than normal

	ATX	CPU	PCIe	SATA	Molex
Corsair SF750	300mm	400mm	400mm + 100mm	445mm	330mm
Most used lengths in SFX fonts	~ 300mm	~ 350-400mm	~ 350-550mm + 100-150mm	600-900mm	600-700mm
Corsair AX850 Length	610mm	650mm	775mm	800mm	750mm

As you can see, the difference in cable length between most SFX sources and the ATX that we have taken as a reference (Corsair AX850) is very large. But what matters most to us is the difference between the SF750 and its competitors: we can see that the PCIe, SATA and Molex cable strips are shorter than we would like.

We understand that this has the motivation to optimize the length for SFX boxes, being fair and necessary for most of them, but it may be inconvenient in some cases. So, our recommendation is to investigate if this length of wiring is sufficient for the SFF box you are considering, and if not, consider buying extensions, since there are quality kits and brands recognized for just 30 euros.

Does this font work for an ATX box?

Certainly, this source includes an SFX to ATX adapter, so we can mount it without problems in any box that requires an ATX source.

However, the length of the cabling would make it impossible to mount on most ATX mid-towers. This occurs with almost any SFX font and would involve having to use extensions.

Internal analisis

The manufacturer of this SF750 Platinum is the same as the rest of Corsair's SFX ranges (SF450 / SF600 Gold or Platinum): Great Wall. It is a manufacturer that we already had experience with the Riotoro Onyx that surprised us so much in terms of quality. We hope to repeat sensations in the SF750.

Within the primary filtering we highlight the inclusion of an MOV, NTC and relay, protection components that should not be missing in any high-end source.

A detail that is obvious when analyzing the interior is the abundance of a " strange " dark gray substance that some identify as something negative. Nothing is further from reality. It is a special glue that physically fixes components characterized by vibrating and emitting electrical noise such as coils, to avoid this effect. The biggest concern that comes to mind when seeing its use is if this worsens the cooling, but no: it is a tail with great thermal conduction.

The capacitors used are entirely Japanese, the primary being a 420V 470uF Nippon Chemi-Con (in Cybenetics' hold-up time test we will see if this capacity is sufficient), which resists temperatures up to 105ºC.

The MOSFETs used on the primary side are also of the highest quality, as we have German Infineon and Japanese Fuji Electric transistors.

On the secondary side, we are impressed by the use of almost 100% solid capacitors, of much greater durability compared to the electrolytic ones that tend to be the majority in other PSUs. They are also Japanese and most are from Nippon Chemi-Con or Nichicon. In the image we can also see the DC-DC converters that allow an excellent regulation of voltages.

Being a 750W source, there is a lot to filter in very little space, hence these solid capacitors have been arranged throughout the PSU.

Let's now see what is hidden under the chassis. The first thing we find is an aluminum plate that has the purpose of turning the chassis itself into a heatsink for the PSU, as it happens in other small sources such as laptop chargers. Under this plate is a thermal pad.

Specifically, the main purpose of the board is to cool the 12V rail MOSFETs, a very relevant component that gets quite hot. Here, we continue to see how quality components are used, in this case signed by the American Alpha & Omega.

The weld quality is very good, as we expected from GreatWall, a very good job. In it we highlight the chip in charge of the protections, an Infinno IN1S4291 of which we lack information.

As always, we finish with the fan, a Corsair NR092L 92mm in diameter. This model makes use of "rifle" bearings, so it is of good quality, although it does not reach the FDB of other models of the brand itself. Even so, later we will check that it is governed by a well-implemented semi-passive mode so it will not always be working.

Cybenetics performance tests

As we already indicated in our specification table, this power supply has a certification of efficiency and loudness issued by Cybenetics. This company stands out for carrying out more advanced and complete tests than 80 Plus (as they test more efficiency points and 80 Plus does not check loudness), but also because detailed tests with all the tests carried out are published on its website.

Since Cybenetics allows their data to be used with the corresponding attribution, we will show them in this review and explain them. Our goal is for everyone to understand the meaning of all these tests, as the data alone can be incomprehensible to many users. In addition, Cybenetics has equipment that exceeds € 30, 000-50, 000 in cost, enabling them to do some of the most reliable testing in the world.

Cybenetics Testing Glossary

As the tests carried out by Cybenetics have some complexity, we explain in these tabs what is measured and what is its importance.

This is information that we will include in all our reviews with data from Cybenetics so, if you already know how the test structure works, you can continue reading. If not, we recommend you take a look at all the tabs to find out what each test is about.;)

• Glossary of terms Voltage regulation Ripple Efficiency Loudness Hold-up time

Let's go with a small glossary of some terms that could be somewhat confusing:

• Rail: PC sources that follow the ATX standard (like this one) do not have a single outlet, but several, which are distributed in " rails ". Each of those rails outputs a specific voltage, and can supply a specific maximum current. We show you the rails of this Thor in the image below. The most important is 12V.

  

  Crossload: When testing a power supply, the most common is that the loads made on each rail are proportional to their "weight" in the power distribution table of the source. However, it is known that the actual loads of the equipment are not like this, but are usually very unbalanced. Therefore, there are two tests called "crossload" in which a single group of rails is loaded.

  On the one hand, we have the CL1 that leaves the 12V rail unloaded and gives 100% at 5V and 3.3V. On the other, the CL2 that 100% loads the 12V rail leaving the rest unloaded. This type of testing, of limit situations, truly shows if the source has a good regulation of voltages or not.

The voltage regulation test consists of measuring the voltage of each source rail (12V, 5V, 3.3V, 5VSB) in the different load scenarios, in this case from 10 to 110% load.

The importance of this test lies in how stable all the voltages are maintained during the tests. Ideally, we like to see a maximum deviation of 2 or 3% for the 12V rail, and 5% for the rest of the rails.

What does not matter so much is 'what voltage is it based on', although it is a fairly widespread myth, it should not matter to us that the 11.8V or the 12.3V are around for example. What we do demand is that they be kept within the limits of the ATX standard that governs the correct operation rules of a PSU. The dashed red lines indicate where those limits are.

Vulgarly, it can be defined as the "residuals" of alternating current that remain after the transformation and rectification of the household AC into low-voltage DC.

These are variations of some millivolts (mV) that, if they are very high (being able to say that there is a "dirty" energy output) can affect the behavior of the equipment components and in some cases damage fundamental components.

A very guiding description of what a source's ripple would look like on an oscilloscope. In the graphs below what we show is the variation between peaks like the ones seen here, depending on the source load.

The ATX standard defines limits of up to 120mV on the 12V rail, and up to 50mV on the other rails we show. We (and the community of PSU specialists in general) consider that the 12V limit is quite high, so we give a "recommended limit" of just half, 60mV. In any case you will see how the majority of sources that we test give excellent values.

In the transformation and rectification processes from household alternating current to the low voltage direct current required by the components, there are various energy losses. The efficiency concept allows quantifying these losses by comparing the power consumed (INPUT) with that delivered to the components (OUTPUT). Dividing the second by the first, we obtain a percentage.

This is precisely what 80 Plus proves. Despite the conception that many people have, 80 Plus only measures the efficiency of the source and does not do any quality testing, protections, etc. Cybenetics tests efficiency and sound, although it altruistically includes the results of many other tests such as the ones we showed you in the review.

Another very serious misconception about efficiency is believing that this determines what percentage of your "promised" power the source can deliver. The truth is that the "real" power sources announce what they can give at the START. In other words, if a 650W source has 80% efficiency at this load level, it means that if the components demand 650W, it will consume 650 / 0.8 = 812.5W from the wall.

Last relevant aspect: the efficiency varies depending on whether we are connecting the source to a 230V electrical network (Europe and most of the world), or to 115V (mainly the US). In the latter case it is less. We publish Cybenetics data for 230V (if they have it), and since the overwhelming majority of sources are certified for 115V, it is normal for 230V to fail to meet the 80 Plus requirements advertised by each source.

For this test, Cybenetics tests the PSUs in an extremely sophisticated anechoic chamber with equipment worth tens of thousands of euros.

It is a room isolated from outside noise almost entirely, suffice it to say that it has a 300kg reinforced door to illustrate the great isolation it has.

Within it, an extremely accurate sound level meter capable of measuring below 6dbA (most have at least 30-40dBa, much more) determines the loudness of the power supply in different load scenarios. The speed the fan reaches in rpm is also measured.

This test basically measures how long the source is able to hold on once it is disconnected from the current while at full load. It will be a few crucial milliseconds to enable a safer shutdown.

The ATX standard defines 16 / 17ms (according to test) as a minimum, although in practice this will be more (we will not always be charging the PSU at 100% so it will be greater), and there are usually no problems with lower values.

We recommend you take a look at the test report published by Cybenetics:

Link to the full Cybenetics report for the SF750 Cybenetics official website

Voltage regulation

The voltage regulation results are magnificent, since deviation is barely observed with a paltry 0.16% in 12V and less than 0.5% in the rest of the rails. In short, a source with "pinned" voltages.

Curly

We also see very good data on ripple, as expected. The values ​​obtained in 12V are not at the forefront of what is seen in other ATX sources, but it remains within absolutely perfect margins that will never give a problem, even in the case of an overclock.

Let's also keep in mind that this SF750 does not include annoying capacitors in the cables, which are used to improve the Ripple but at the same time make the mounting experience worse.

Efficiency

Efficiency stays extremely close to the values ​​required by 80 Plus Platinum at 230V, far exceeding them except at 50% load where Cybenetics measured 93.57% of the required 94%. It is an excellent result considering that this source, like the vast majority, is certified at 115V, where the requirements are lower. ( At 230V, the power supplies are more efficient, but the 80 Plus requirements are quite stricter)

Fan speed and loudness:

Semi-passive mode keeps the fan off up to 40% load. So, it starts at about 1300rpm, revolutions that are not as high as it seems, being a 92mm fan. We will confirm this later with our own experience in terms of loudness, but the data is already quite decent: at these revolutions, the source produces 13.2 dBa of noise, which is not bad at all.

When we start to hover around 100% load, and as expected, the fan already exceeds 3000rpm being much louder.

Hold-up time:

Hold-up time Corsair SF750 (tested at 230V)	11.60 ms
Data extracted from Cybenetics

As we expected after seeing the small primary capacitor, the hold-up time is far from the ideal of 16 or 17ms and we have only 11.60 in the SF750 Platinum. We assume that it was a question of space, since it was difficult to include another or a larger primary capacitor. In any case, they are not worrying data since the hold-up time will always be higher when we are not fully loaded.

We reiterate our thanks to Cybenetics for allowing the use of this test data and invite you to learn more about them here.

Our experience in terms of semi-passive mode and loudness

Corsair continues to demonstrate that it holds the key to developing the best semi-passive modes we've seen in power supplies.

Also, when the fan has to run, it is barely audible despite being only 92mm in diameter.

Most semi-passive ventilation modes we've tried have a problem: the ease with which the fan goes on and off “loops” every few seconds. This causes, except in some types of fan, that the durability of the fan is reduced in such a way that using the semi-passive mode does not make sense. This is caused by the use of ventilation profiles that turn the fan on or off depending on a certain temperature: when the source heats up after applying load, the fan turns on. When the fan reduces the temperature below the threshold, it turns it off. Then it reheats, and it lights up… like this continuously.

What we have just explained is something that does not happen with the semi-passive modes used by Corsair in most of its sources, including the SF750. Instead of using a simple ventilation curve, the brand implements a much more expensive digital microcontroller that allows the ventilation curve to be programmed efficiently. Thus, when the fan has to be turned on because the source has heated up, even if the temperature drops, the fan will not turn off until ensuring that the source is truly cooled, preventing one of these loops from entering.

In short, it is about going from a simple curve that only takes into account the temperature, to an algorithm that regulates the semi-passive mode depending on the temperature, the load and the time of use, ensuring that the maximum objectives are met. Quiet and maximum fan durability, which most semi-passive modes breach.

Leaving the theory and moving to practice, we've been playing for hours, maintaining the source in different loading scenarios, with and without Overclocking… and the semi-passive mode has always been more than successful. What's more, when the fan turns on, the loudness is very low (as the Cybenetics tests already showed), which reminds us of other very silent ATX sources.

Final words and conclusion about Corsair SF750

Corsair has put all the meat on the grill and has decided to launch the best possible evolution of its SF 80 Plus Gold fountains. The brand and manufacturer of this font, Great Wall, have done a simply excellent job of launching a product that could compete from you to you with the high-end range of ATX fonts, thanks to the level of the internal components, the loudness, electrical efficiency and performance.

It is also noteworthy to comment on the great milestone of achieving a 750W source in a format as restrictive as SFX. They haven't even settled for the slightly longer SFX-L format to do it. All maintaining such a quiet operation that reminds us of an ATX source, despite the limitations of this limited space and the use of a fan of only 92mm.

The only point that we can qualify as “negative” is the relatively short length of the wiring, we understand that it is suitable for most boxes with space for SFX source, but in some cases they may be somewhat short.

We recommend reading our updated guide to the best power sources

The price in stores of this SF750 is around 145 euros, although we have seen it higher in some stores (hopefully it will drop). For those who are not going to need such an excessive power, which is the vast majority, the SF600 Platinum offers good savings at 120 euros, and the SF450 Platinum at 110. All three share quality and performance.

After carefully analyzing what this font offers, and after an exhaustive comparison with the different models on the market, we have concluded that at the time of writing this review, the Corsair SF750 is the best SFX font on the market: No SFX or SFX-L model comes close to matching it in quality, performance, or power density at this price. There are quite competent alternatives on the market, but we have not found any equivalent to the SF750. We don't know if any better model will land in the future.

Advantage

• Excellent internal quality with the best components and a wonderful construction. 80 Plus Platinum efficiency with 93.5% maximums at 230V. 100% modular wiring with sleeving, to date the only range of SFX sources that includes it is probably the source with Highest power / volume ratio on the market, achieving 750W in SFX format, compared to 700W SFX-L from other brands. Hyper-silent operation at all loads with a good quality fan, practically on par with the best ATX sources. 7 year warranty. Reasonable price for an SFX model with these benefits.

Disadvantages

• The wiring length is ideal for most SFF configurations, but some boxes would require extensions, with no option to disable semi-passive mode, although fortunately it works very well.

The Professional Review team awards you the Platinum Medal and Recommended Product.

Corsair SF750

INTERNAL QUALITY - 96%

SOUNDNESS - 98%

WIRING MANAGEMENT - 90%

PROTECTION SYSTEMS - 95%

PRICE - 96%

95%