Am I correct in thinking that this PSU, and other recent models capable of producing their entire output rating at 12V (give or take rounding on the amperage) are pure 12V designs internally that then use DC-DC converters to make the 3.3/5V rails instead of the traditional design that had separate circuitry for the 3.3/5v and 12v rail. With the older design having problems in cross load tests (when one side was maxed and the other only had a minimum load), the change in design seems a reasonable way to handle the shift of almost everything on the mobo to 12V operation; but I haven't seen it confirmed anywhere.
Yes, many modern power supplies are built on a platform that only does AC to 12V DC, then use DC to DC conversion to get the 5V and 3.3V rails. I can think of all kinds of reasons that might provide cleaner power or be more efficient, but I'll leave that discussion to the experts.
Why are the PSU reviews disproportionately focusing on 1kW and higher units? Even a high end SLI/Crossfire rig will struggle to load these PSUs. Especially with Intel and NVIDIAs latest power efficient architecture. How about some more reviews on units likely to be utilised by us mere mortals not doing quad Titan SLI?
I agree. It's cool to read about ridiculously high end units. But few people are using dual graphics cards, and even fewer are going with more. I'd love to see reviews on lower priced units.
For example, I recently built a computer with <400W power requirements. So, I got a $40 EVGA 500W supply - the cheapest 80+ supply I could find. There were approximately zero reviews of that unit on review sites (but plenty of user reviews on Newegg, for what that's worth).
For anyone who's wondering, that supply (100-W1-0500-KR R) is nice and quiet, even when feeding an overclocked FX-8350.
I agree entirely; a review of PSUs starting from the absolutely cheapest, determining which ones work adequately, would be really useful. An i7/4790K with 16G memory and an SSD is a 150W system; to learn, with the kind of confidence that one of Anandtech's comprehensive reviews provide, that a $20 PSU from newegg is adequate would be quite helpful.
Hear hear! The number of people who look at this site that really really need such a PSU is probably less than 5%. Please lets have more reviews in the 450-750W range.
Agreed. I don't use graphics cards any longer for computing, Intel HD Graphics will do. Thus, a PSU's total wattage isn't a main focus while the quality of it is. 650 watts is our limit in the long run for a non-discrete graphics machine and 750 to 850 with one.
300-550W 80 Plus Gold/Plat fully modular active cooling (preferrably a hybrid fan) available as SFX
I would imagine it is easy for a company like Seasonic to put out such a device. If only their lower wattage Plat models had some sort of active cooling, I would be less reluctant to buy one.
Personally, I have always gone the opposite route. The power supply is the most underrated and unappreciated component in PC building. The first reason is upgrade path. you never know if you will add other components later. As an avid AMD user, they have stuck with one socket for awhile, so having a beefier PS unit that can handle a dramatic change in TDP (with overclocking) pays dividends. The second and most important reason is the strain on the PSU itself. My rule of thumb is to try and stay under half of the rated watts of the unit. Though truth be told, the Amperage is the more important feature to account for. If i calculate my power needs to be no more than 400 watts @ 30 to 40 amps, I will always go for a min of 800 watts @ 60-80 amps. the reason is again strain. Just my personal preference. I use the car comparison the most for people building for the first time. If my car can go 150 MPH, i can't expect to maximize longevity by going that fast continuously. but at 75 MPH all parts of the car can handle that for much longer.
There's more to power supplies then raw numbers. Actually I want to see how smooth and accurate the power is myself. Having more power then you need isn't bad, just cost money. Power supplies also lose power over time as they age but it takes quite a bit to make any real notice. Still, I like to keep them running well under their power band for both heat and noise. There are plenty of 500-1kW units out there that fit the budget lineup. I'm running an SLI setup now with a 750W unit and still have plenty of headroom.
But what I am interested in is the power clarity and accuracy, especially in these gold and higher rated units. And unfortunately, the extremely low power supplies tend to be the cheapest, noisiest (signals), and more inaccurate units. Since power is the core of any system, stable and accurate power is paramount to a good and stable system. It's one area that I refuse to skimp on in any build. But still don't go insane at the same time. But I'd easily spend $100 for a solid, accurate, tested unit instead of a $40 Joe's Power Supply version...
What I'd like to see is how that $40 supply compares to a $100 supply - and that would include power delivery characteristics. If Anandtech would test some of those cheaper units, maybe we'll find that spending $100 on a 500W power supply is really unnecessary.
Also, note that a mildly overclocked i7-4790K with a GTX 980 fits comfortably within a 500W power envelope.
Between my wife and 3 kids, and my work computers (I work at home) I have 10 desktop machines at home, all fast and all used for gaming or running virtual machines or both. Most system are overclocked, and have hungry video cards. My system has a Corsair 650W system I bought 8 years ago, but most of the other systems have more modest PSUs, most of which I bought on sale for ~$40 (typically Antec Earthwatts or Rosewill 500W units). I've never had any stability problems with any of them. My server has a 500W PowerMan PSU with 6 hard drives, 2 SSDs, and a devil's Canyon i5 overclocked to 4.5 Ghz (can go higher) - this PSU came free with an InWin case I bought 5 years ago. If you have a single computer, then it makes sense to pay $90-120 on a good power supply, but when you have 10 computers to pay for, you'll find $40 (on sale) PSUs can work just fine. The motherboard already has plenty of power circuitry on it to deal with ripple, etc... I have the $45 Rosewill 450W PSU powering a hex-core OC AMD cpu + OC 560Ti video card on one system, and an overclocked G3258 + HD6850 on another system.
The Seasonic SSR-360GP is by all accounts a quality PSU, and costs $60, so you don't have to spend $100 to get quality unless you need higher wattage. People who want modular cables could go with the SSR-450RM ($80), which is probably equal in quality, although I haven't purchased it and therefore didn't read the reviews carefully. Neither of these products were reviewed by Anandtech. In fact, I've never consulted an Anandtech review when buying a PSU.
Well, the primary reason is that the manufacturers generally like to showcase the best they have. One company representative argued, and I partially agree, that "Someone who goes and buys a $30 PSU will not bother to check the reviews to begin with. If they are to spend >$100, they both know what they want and will check it out thoroughly before placing an order.". The other half, I will add it myself - manufacturers like to show the best they have because it reflects on their brand as a whole. If a 1.5kW PSU aces the tests, some people would (erroneously) think that their $30 unit is just as good.
I do agree that the common home PC does not require more than a 500W PSU, tops. The problem is that there are actually very very very few such PSUs coming from any reputable manufacturer and even those are very basic units, based on designs and platforms that may be a decade old. There is nothing really interesting about them and the users will not actually care (although they should) if, for example, the maximum ripple is 80mV instead of 40mV.
I do try to source "small" PSUs when possible - hence the reviews of the Antec Edge 550W, the Nightjar NJ520, the Dark Power Pro 10 650W. It's not even half of the units I reviewed, I know, but it is not always possible to source small units, neither I can write a dozen PSU reviews per month.
I disagree that people going for budget PSUs won't bother to read reviews. I for one tried, until I realized such reviews were nonexistent. If someone's going so far as to pick individual parts for a computer, there's a good chance they'll care about a component as important as the PSU.
Even if you don't find them interesting enough to write a review on, making some test data available would be very useful.
Also, like another poster said, there's more of a risk factor with low priced PSUs. That makes reviews (or even just raw data) even more critical.
It'd be crazy to ask for 12 PSU reviews a month, but it'd be awesome if a PSU with a popular internal design gets tested, and models using the same design are just listed.
Yup. And while I know loaned/donated hardware is the default for most reviews; in a roundup review, spending $30-60 for a single no-name model to show what spending a little more for a brand name model gives should be an acceptable use of editorial dollars. Two cheap models might be worth it as well; one obvious garbage box to be sacrificed on the alter of magic smoke, and a second that while expected to survive is a few steps down on the performance scale (eg bare 80plus). I wonder if you might be able to fiangle the latter out of one of the OEMs: "I want a baseline model to demonstrate why your 80+ gold model is worth the extra $20."
If I may add one more thing, I have never once looked at a 1000W power supply and said "this unit performs well, so this brand's 400-600W units must be good too".
When I went shopping for a PSU, I went for a $40 500W model with a plain 80+ rating because there were practically no reviews of sub-$100 400-600W units (and I didn't want a $20 no-name model that might have 45% efficiency). But if reviews showed that spending $100 gave a big improvement over lower priced units, I might have decided differently.
The other half is that high efficiency PSUs only pay for themselves under largish 24/7 loads at typical US power prices. I played around with the numbers recently for my new build:For a 24/7 load (distributed computing), $0.12/kwh for electricity, modular designs only, and wanting ~200W headroom to keep the PSU fan from adding to the system noise; and over 6 years a 91% platinum PSU only broke even over an 88% gold unit if I ran SLI/xFire for about half the time. In a pure single GPU config the gold model barely broke even vs a cheaper silver model. I ended up going with the platinum model because I expect the 5k monitor I'm planning on buying in the next 12-24 months almost certainly will require SLI/xFire for a few years; combined with a housing situation where my landlord is responsible for the heating bill while air conditioning is on me tipping the scale.
If your PC spends most of its day off/idle a high efficiency PSU isn't going to give much of a return except possibly at very high electricity rates. If an OEM ever makes one, a small 80+ titanium unit might be worthwhile for mainstream users just due to the large boost it requires at a 10% load.
Good point. Power requirements for PCs with massive storage have also come down quite a bit now that 3 2TB drives can be replaced with a single 6TB drive.
I have 8 hard drives (1 SDD for the OS) in my file server. If they increase the size I'll just get 8 6TB drives. Right now I'm slowly migrating to 4TB models (up from 1.5TB and 2TB models). Hard drives don't consume a lot of energy regardless.
My home server has 5 hard drives (4x 3TB array plus the OS drive) and i'm using a 450W PSU. The system idles at 60W measured at the wall. It hits about 75W when the array is maxing out the gigabit network connection. The only time you really need to worry about power requirements for "massive storage" is during boot, but if you stagger the spin-up then it's not a problem. Not that I've had to bother with my current system.
"The high ambient temperatures have a significant impact on the electrical performance of the Photon 1050W, reducing its energy conversion efficiency by an average of 2%. The drop is higher as the load increases, reaching a massive 3.7% drop with a load of 1050W."
C'mon now, an increase from 2% to 3.7% is NOT a "massive" difference.
You wouldn't be saying that if it was the APR % difference in your mortgage for example... If you were at 2% and they raised you to 3.7%, you'd @#$% the bed. Everything is perspective.
C'mon now, don't conflate two items that are completely different. Changes in a mortgage APR don't equally equate into slight changes in a power supply's energy conversion efficiency.
If two power supplies were being compared and one was 3.7% while the other was 5.0%, would you still describe the 3.7% variance as "massive"? How would you describe the 5.0% variance?
E.Fyll did a good job with his review; I just thought the word "massive" was a bit hyperbolic in the context used.
Actually, it is an 85% increase over the average efficiency drop. That, alongside the fact that everything above 1-1.5% for that particular test is very high, is pretty much massive.
Hey, I knew someone would give me the actual % difference. :)
Not to be overly pedantic, I was just thinking a different word would be better - like "significant". What if you had another test that showed a 10% variance? What word would be appropriate then - "ginormous"? Words that describe the scale of difference begin to lose impact when they start out with a descriptive word that is a little too large in scale itself.
Anyway, just having a little fun here. E.Fyll, you did a nice job with the review!
I also agree with the opinions given here that we need more 500-600W reviews. That is what I think most buy when they are building their own PCs with a single GPU, and that's where we need the expert guidance/reviews to make the most informed purchase.
Each word is appropriate for a specific purpose. In this case, an energy conversion efficiency drop from 2% to 3.7% just from the increase of the ambient temperature is massive. If there was a 10% drop in that test, something would be horribly wrong.
As you mentioned, the scale itself is important. However, the scale is not 0-100%. Even the entirety of the ECE scale itself is 60-100% (you cannot really have 0.1% efficiency with such a PSU). In this specific test, the maximum of the scale that is being described is perhaps a 0% to 5.5-6% drop (you cannot have a 90% drop either), meaning that going from 2% to 3.7% is a move a whole 1/3 down the scale. The large drop at maximum power output is also a sign of overloading - units rated at 50°C do not (should not) behave like that, the efficiency decline due to the higher temperature is even across the entire load range.
You just gave me an idea for a new graph. Thank you.
You see, everything is relative. An efficiency loss of 1.7% does not initially sound much, I agree. However, when you consider that the losses nearly double and that a good PSU should not even reach 1% during that test, let alone 3.7%, the huge difference becomes obvious.
Actually, a 3.7% decrease in efficiency REALLY IS MASSIVE.
For example, a 3.7% DECREASE IN EFFICIENCY from 90.0% to 86.3% means that the waste heat produced by the PSU increases from 10.0% to 13.7%, or a 37% INCREASE IN HEAT OUTPUT.
I would call 37% a MASSIVE INCREASE in anything!
The amplification factor, call it A, is obviously given by:
A = 100% / (100% - Efficiency)
In my example, A = 10. In an 80+ Platinum rated PSU operating at 93.3% Efficiency, A = 15. Even in a plain vanilla 80+ PSU operating at only 80% Efficiency, A = 5.
So the % increase in HEAT is roughly 10X the % decreased in EFFICIENCY. It would behoove you to be mindful of this.
But the drive for HIGH EFFICIENCY is mainly motivated by the desire for LOWER HEAT PRODUCTION, hence LOWER COOLING REQUIREMENTS, hence LOWER NOISE.
The savings on your electricity costs are merely a fringe benefit.
One other thing... Seeing as how a name-brand high-efficiency PSU with a 7-YEAR WARRANTY will likely last 10-20 years before failure, and moreover that the QUALITY OF THE POWER DELIVERY (voltage accuracy and ripple) strongly influence the MTBF of the other components in your computer build, WHY CHEAP OUT ON YOUR PSU ???
My father once bragged to me about how much money he saved by only buying a cheap $20 watch EVERY YEAR. I responded by showing him my $200 Citizen solar-powered watch, with the titanium bracelet and sapphire crystal, which was 10X as accurate, looked way nicer, and over 10 years would cost the same, and over 20 years half as much as he would spend on cheap, cheap-looking, and minimally accurate watches.
Better to buy a GOOD one once, than to buy a CHEAP one over and over again.
That's my philosophy for any (so-called) durable goods.
Considering its price and features, I feel that this PSU deserved at least a Bronze award.
E. Fylladitakis, please can you add a cable/connector table to your future reviews, similar to e.g. JonnyGURU? It's a bit of a pain to have to find the manufacturer's page, and available connectors play a big part in the selection criteria for a PSU. For example this model has 8x 6+2-pin PCIe and 15 SATA connectors, which is quite generous for a 1050W unit.
I simply do not do awards. Not at all, ever. Every specific product has its own advantages, disadvantages and market potential. For user X, a "bronze" product might be much better than a "gold" product, even if they are similar, just because the former suits the needs of the user better. Everything is relative.
I'll consider the tables. It is not difficult, I just thought that it is something too redundant.
While we're suggesting things to review (like midrange and lower end units), there's something I'm not seeing in reviews that really should be in there: UPS compatibility testing.
Active power factor correction (PFC) doesn't seem to be a feature on this unit, but it's becoming increasingly common, and many of those units don't seem to like the modified sinewave output of cheaper UPS units. Seeing as this is the sort of incompatibility that could be very bad to learn about the hard way, that would be a good quick test to add into PSU reviews.
All modern switching PSU include (or should) APFC. It is actually illegal to sell a PSU without APFC in Europe nowadays, for example.
Unfortunately, that is an issue of having a bad UPS, not a bad PSU. A cheap UPS will certainly output a modified sine wave. If the output is that bad that would cause overvoltage within the APFC circuit and blow the capacitors, it is not because the PSU is bad but because the output of the UPS is awful. A waveform that would increase the voltage at the APFC capacitors from the expected maximum of 340V to >450V does not even remotely pass as a "modified sine wave".
The only PSU that I have ever seen being too "sensitive" about modified sine wave signals was an Enermax design nearly a decade ago. Even then, Enermax enhanced the tolerance of the unit on its second revision. Still, it was not the PSU's fault that some UPS had terrible outputs. That's where the major difference between a $50 600VA Chinese off-brand UPS and a Schneider $200 600VA UPS usually lies...
Generally speaking, the voltage rating of the input capacitors usually hints the tolerance of the PSU when handling such signals. The higher, the better.
I heartily agree it should be tested whether a PSU will work with an old non-sine UPS. Why should we have to dump a working UPS and buy a $150+ new sinewave one when we get a new PSU? I'm all in favor of PFC, but I would have to return any one that wouldn't work with my existing inventory of older UPS.
It would be nice to see a review of the Corsair RM 850. Techpowerup's review suggests that it is the quietest actively-cooled PSU available. While they gave the RM 750 an even better rating for the average noise level throughout its operating range, the 850 delivers more watts before noise ramps up. So, it appears to be the best option for those who want to avoid noise while still having the ability to go beyond what passive PSUs offer.
However, the concern is the quality of the capacitors primarily. It would be good to see another review site verify or contradict Techpowerup's results.
Yes. I've heard the Corsair RM model's are debatable because they are manufactured by Great Wall or some other maker. Folks tend to suggest on the Corsairs made by Seasonic are worthy such as the AX models. I haven't had any issues with their TX and TX-v2 PSU's in the past years but now the RM has replaced those. There's lot of features with the RM and reviews give them high points for total output accuracy while some say they a few percentage of them don't last or have problems.
It would be reasonable to ask many of the hardware review people to do long term 60% power usage with daily on-off cycle testing combined for a final determination of longevity and/or problems. Most will say all PSU's loose total power output after a few years which stands to say why folks should purchase a PSU with about double the power they really need to allow for that and to allow the unit to operate and full efficiency.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
42 Comments
Back to Article
DanNeely - Tuesday, March 24, 2015 - link
Am I correct in thinking that this PSU, and other recent models capable of producing their entire output rating at 12V (give or take rounding on the amperage) are pure 12V designs internally that then use DC-DC converters to make the 3.3/5V rails instead of the traditional design that had separate circuitry for the 3.3/5v and 12v rail. With the older design having problems in cross load tests (when one side was maxed and the other only had a minimum load), the change in design seems a reasonable way to handle the shift of almost everything on the mobo to 12V operation; but I haven't seen it confirmed anywhere.MrCommunistGen - Tuesday, March 24, 2015 - link
Yes, many modern power supplies are built on a platform that only does AC to 12V DC, then use DC to DC conversion to get the 5V and 3.3V rails. I can think of all kinds of reasons that might provide cleaner power or be more efficient, but I'll leave that discussion to the experts.DanNeely - Tuesday, March 24, 2015 - link
Is the -12V (used for RS232) done that way too, or does it still get a special side circuit like the 5V standby.Gigaplex - Tuesday, March 24, 2015 - link
Why are the PSU reviews disproportionately focusing on 1kW and higher units? Even a high end SLI/Crossfire rig will struggle to load these PSUs. Especially with Intel and NVIDIAs latest power efficient architecture. How about some more reviews on units likely to be utilised by us mere mortals not doing quad Titan SLI?chlamchowder - Tuesday, March 24, 2015 - link
I agree. It's cool to read about ridiculously high end units. But few people are using dual graphics cards, and even fewer are going with more. I'd love to see reviews on lower priced units.For example, I recently built a computer with <400W power requirements. So, I got a $40 EVGA 500W supply - the cheapest 80+ supply I could find. There were approximately zero reviews of that unit on review sites (but plenty of user reviews on Newegg, for what that's worth).
For anyone who's wondering, that supply (100-W1-0500-KR R) is nice and quiet, even when feeding an overclocked FX-8350.
TomWomack - Tuesday, March 24, 2015 - link
I agree entirely; a review of PSUs starting from the absolutely cheapest, determining which ones work adequately, would be really useful. An i7/4790K with 16G memory and an SSD is a 150W system; to learn, with the kind of confidence that one of Anandtech's comprehensive reviews provide, that a $20 PSU from newegg is adequate would be quite helpful.jabber - Tuesday, March 24, 2015 - link
Hear hear! The number of people who look at this site that really really need such a PSU is probably less than 5%. Please lets have more reviews in the 450-750W range.jabber - Tuesday, March 24, 2015 - link
Oh and some nice reviews of specialist 200-400W would be good too.FunBunny2 - Saturday, March 28, 2015 - link
Yeah, but... AT's New Overseers clearly are going down the Apple Road: cater to the 20% rich and dumb enough to drop large coin on useless bling.gsuburban - Sunday, June 28, 2015 - link
Agreed. I don't use graphics cards any longer for computing, Intel HD Graphics will do. Thus, a PSU's total wattage isn't a main focus while the quality of it is. 650 watts is our limit in the long run for a non-discrete graphics machine and 750 to 850 with one.etamin - Tuesday, March 24, 2015 - link
I'm still waiting for a high quality PSU that is:300-550W
80 Plus Gold/Plat
fully modular
active cooling (preferrably a hybrid fan)
available as SFX
I would imagine it is easy for a company like Seasonic to put out such a device. If only their lower wattage Plat models had some sort of active cooling, I would be less reluctant to buy one.
phantomferrari - Wednesday, March 25, 2015 - link
You mean like these :)http://silverstonetek.com/product_power.php?tno=7&...
Jahzah_1 - Wednesday, March 25, 2015 - link
Personally, I have always gone the opposite route. The power supply is the most underrated and unappreciated component in PC building. The first reason is upgrade path. you never know if you will add other components later. As an avid AMD user, they have stuck with one socket for awhile, so having a beefier PS unit that can handle a dramatic change in TDP (with overclocking) pays dividends.The second and most important reason is the strain on the PSU itself. My rule of thumb is to try and stay under half of the rated watts of the unit. Though truth be told, the Amperage is the more important feature to account for. If i calculate my power needs to be no more than 400 watts @ 30 to 40 amps, I will always go for a min of 800 watts @ 60-80 amps. the reason is again strain. Just my personal preference. I use the car comparison the most for people building for the first time. If my car can go 150 MPH, i can't expect to maximize longevity by going that fast continuously. but at 75 MPH all parts of the car can handle that for much longer.
SirGCal - Tuesday, March 24, 2015 - link
There's more to power supplies then raw numbers. Actually I want to see how smooth and accurate the power is myself. Having more power then you need isn't bad, just cost money. Power supplies also lose power over time as they age but it takes quite a bit to make any real notice. Still, I like to keep them running well under their power band for both heat and noise. There are plenty of 500-1kW units out there that fit the budget lineup. I'm running an SLI setup now with a 750W unit and still have plenty of headroom.But what I am interested in is the power clarity and accuracy, especially in these gold and higher rated units. And unfortunately, the extremely low power supplies tend to be the cheapest, noisiest (signals), and more inaccurate units. Since power is the core of any system, stable and accurate power is paramount to a good and stable system. It's one area that I refuse to skimp on in any build. But still don't go insane at the same time. But I'd easily spend $100 for a solid, accurate, tested unit instead of a $40 Joe's Power Supply version...
chlamchowder - Tuesday, March 24, 2015 - link
What I'd like to see is how that $40 supply compares to a $100 supply - and that would include power delivery characteristics. If Anandtech would test some of those cheaper units, maybe we'll find that spending $100 on a 500W power supply is really unnecessary.Also, note that a mildly overclocked i7-4790K with a GTX 980 fits comfortably within a 500W power envelope.
kmmatney - Tuesday, March 24, 2015 - link
Between my wife and 3 kids, and my work computers (I work at home) I have 10 desktop machines at home, all fast and all used for gaming or running virtual machines or both. Most system are overclocked, and have hungry video cards. My system has a Corsair 650W system I bought 8 years ago, but most of the other systems have more modest PSUs, most of which I bought on sale for ~$40 (typically Antec Earthwatts or Rosewill 500W units). I've never had any stability problems with any of them. My server has a 500W PowerMan PSU with 6 hard drives, 2 SSDs, and a devil's Canyon i5 overclocked to 4.5 Ghz (can go higher) - this PSU came free with an InWin case I bought 5 years ago. If you have a single computer, then it makes sense to pay $90-120 on a good power supply, but when you have 10 computers to pay for, you'll find $40 (on sale) PSUs can work just fine. The motherboard already has plenty of power circuitry on it to deal with ripple, etc... I have the $45 Rosewill 450W PSU powering a hex-core OC AMD cpu + OC 560Ti video card on one system, and an overclocked G3258 + HD6850 on another system.KAlmquist - Tuesday, March 24, 2015 - link
The Seasonic SSR-360GP is by all accounts a quality PSU, and costs $60, so you don't have to spend $100 to get quality unless you need higher wattage. People who want modular cables could go with the SSR-450RM ($80), which is probably equal in quality, although I haven't purchased it and therefore didn't read the reviews carefully. Neither of these products were reviewed by Anandtech. In fact, I've never consulted an Anandtech review when buying a PSU.E.Fyll - Tuesday, March 24, 2015 - link
Well, the primary reason is that the manufacturers generally like to showcase the best they have. One company representative argued, and I partially agree, that "Someone who goes and buys a $30 PSU will not bother to check the reviews to begin with. If they are to spend >$100, they both know what they want and will check it out thoroughly before placing an order.". The other half, I will add it myself - manufacturers like to show the best they have because it reflects on their brand as a whole. If a 1.5kW PSU aces the tests, some people would (erroneously) think that their $30 unit is just as good.I do agree that the common home PC does not require more than a 500W PSU, tops. The problem is that there are actually very very very few such PSUs coming from any reputable manufacturer and even those are very basic units, based on designs and platforms that may be a decade old. There is nothing really interesting about them and the users will not actually care (although they should) if, for example, the maximum ripple is 80mV instead of 40mV.
I do try to source "small" PSUs when possible - hence the reviews of the Antec Edge 550W, the Nightjar NJ520, the Dark Power Pro 10 650W. It's not even half of the units I reviewed, I know, but it is not always possible to source small units, neither I can write a dozen PSU reviews per month.
chlamchowder - Tuesday, March 24, 2015 - link
I disagree that people going for budget PSUs won't bother to read reviews. I for one tried, until I realized such reviews were nonexistent. If someone's going so far as to pick individual parts for a computer, there's a good chance they'll care about a component as important as the PSU.Even if you don't find them interesting enough to write a review on, making some test data available would be very useful.
chlamchowder - Tuesday, March 24, 2015 - link
Also, like another poster said, there's more of a risk factor with low priced PSUs. That makes reviews (or even just raw data) even more critical.It'd be crazy to ask for 12 PSU reviews a month, but it'd be awesome if a PSU with a popular internal design gets tested, and models using the same design are just listed.
Finally, if a PSU does fail (http://www.bit-tech.net/hardware/psus/2014/06/20/5... that'd be very, very good to know.
DanNeely - Tuesday, March 24, 2015 - link
Yup. And while I know loaned/donated hardware is the default for most reviews; in a roundup review, spending $30-60 for a single no-name model to show what spending a little more for a brand name model gives should be an acceptable use of editorial dollars. Two cheap models might be worth it as well; one obvious garbage box to be sacrificed on the alter of magic smoke, and a second that while expected to survive is a few steps down on the performance scale (eg bare 80plus). I wonder if you might be able to fiangle the latter out of one of the OEMs: "I want a baseline model to demonstrate why your 80+ gold model is worth the extra $20."chlamchowder - Wednesday, March 25, 2015 - link
If I may add one more thing, I have never once looked at a 1000W power supply and said "this unit performs well, so this brand's 400-600W units must be good too".When I went shopping for a PSU, I went for a $40 500W model with a plain 80+ rating because there were practically no reviews of sub-$100 400-600W units (and I didn't want a $20 no-name model that might have 45% efficiency). But if reviews showed that spending $100 gave a big improvement over lower priced units, I might have decided differently.
DanNeely - Tuesday, March 24, 2015 - link
The other half is that high efficiency PSUs only pay for themselves under largish 24/7 loads at typical US power prices. I played around with the numbers recently for my new build:For a 24/7 load (distributed computing), $0.12/kwh for electricity, modular designs only, and wanting ~200W headroom to keep the PSU fan from adding to the system noise; and over 6 years a 91% platinum PSU only broke even over an 88% gold unit if I ran SLI/xFire for about half the time. In a pure single GPU config the gold model barely broke even vs a cheaper silver model. I ended up going with the platinum model because I expect the 5k monitor I'm planning on buying in the next 12-24 months almost certainly will require SLI/xFire for a few years; combined with a housing situation where my landlord is responsible for the heating bill while air conditioning is on me tipping the scale.If your PC spends most of its day off/idle a high efficiency PSU isn't going to give much of a return except possibly at very high electricity rates. If an OEM ever makes one, a small 80+ titanium unit might be worthwhile for mainstream users just due to the large boost it requires at a 10% load.
JeffFlanagan - Tuesday, March 24, 2015 - link
Good point. Power requirements for PCs with massive storage have also come down quite a bit now that 3 2TB drives can be replaced with a single 6TB drive.bigboxes - Tuesday, March 24, 2015 - link
I have 8 hard drives (1 SDD for the OS) in my file server. If they increase the size I'll just get 8 6TB drives. Right now I'm slowly migrating to 4TB models (up from 1.5TB and 2TB models). Hard drives don't consume a lot of energy regardless.Gigaplex - Tuesday, March 24, 2015 - link
My home server has 5 hard drives (4x 3TB array plus the OS drive) and i'm using a 450W PSU. The system idles at 60W measured at the wall. It hits about 75W when the array is maxing out the gigabit network connection. The only time you really need to worry about power requirements for "massive storage" is during boot, but if you stagger the spin-up then it's not a problem. Not that I've had to bother with my current system.romrunning - Tuesday, March 24, 2015 - link
"The high ambient temperatures have a significant impact on the electrical performance of the Photon 1050W, reducing its energy conversion efficiency by an average of 2%. The drop is higher as the load increases, reaching a massive 3.7% drop with a load of 1050W."C'mon now, an increase from 2% to 3.7% is NOT a "massive" difference.
SirGCal - Tuesday, March 24, 2015 - link
You wouldn't be saying that if it was the APR % difference in your mortgage for example... If you were at 2% and they raised you to 3.7%, you'd @#$% the bed. Everything is perspective.romrunning - Tuesday, March 24, 2015 - link
C'mon now, don't conflate two items that are completely different. Changes in a mortgage APR don't equally equate into slight changes in a power supply's energy conversion efficiency.If two power supplies were being compared and one was 3.7% while the other was 5.0%, would you still describe the 3.7% variance as "massive"? How would you describe the 5.0% variance?
E.Fyll did a good job with his review; I just thought the word "massive" was a bit hyperbolic in the context used.
E.Fyll - Tuesday, March 24, 2015 - link
Actually, it is an 85% increase over the average efficiency drop. That, alongside the fact that everything above 1-1.5% for that particular test is very high, is pretty much massive.romrunning - Tuesday, March 24, 2015 - link
Hey, I knew someone would give me the actual % difference. :)Not to be overly pedantic, I was just thinking a different word would be better - like "significant". What if you had another test that showed a 10% variance? What word would be appropriate then - "ginormous"? Words that describe the scale of difference begin to lose impact when they start out with a descriptive word that is a little too large in scale itself.
romrunning - Tuesday, March 24, 2015 - link
Anyway, just having a little fun here. E.Fyll, you did a nice job with the review!I also agree with the opinions given here that we need more 500-600W reviews. That is what I think most buy when they are building their own PCs with a single GPU, and that's where we need the expert guidance/reviews to make the most informed purchase.
E.Fyll - Wednesday, March 25, 2015 - link
Each word is appropriate for a specific purpose. In this case, an energy conversion efficiency drop from 2% to 3.7% just from the increase of the ambient temperature is massive. If there was a 10% drop in that test, something would be horribly wrong.As you mentioned, the scale itself is important. However, the scale is not 0-100%. Even the entirety of the ECE scale itself is 60-100% (you cannot really have 0.1% efficiency with such a PSU). In this specific test, the maximum of the scale that is being described is perhaps a 0% to 5.5-6% drop (you cannot have a 90% drop either), meaning that going from 2% to 3.7% is a move a whole 1/3 down the scale. The large drop at maximum power output is also a sign of overloading - units rated at 50°C do not (should not) behave like that, the efficiency decline due to the higher temperature is even across the entire load range.
You just gave me an idea for a new graph. Thank you.
You see, everything is relative. An efficiency loss of 1.7% does not initially sound much, I agree. However, when you consider that the losses nearly double and that a good PSU should not even reach 1% during that test, let alone 3.7%, the huge difference becomes obvious.
Dr.Neale - Thursday, March 26, 2015 - link
Actually, a 3.7% decrease in efficiency REALLY IS MASSIVE.For example, a 3.7% DECREASE IN EFFICIENCY from 90.0% to 86.3% means that the waste heat produced by the PSU increases from 10.0% to 13.7%, or a 37% INCREASE IN HEAT OUTPUT.
I would call 37% a MASSIVE INCREASE in anything!
The amplification factor, call it A, is obviously given by:
A = 100% / (100% - Efficiency)
In my example, A = 10. In an 80+ Platinum rated PSU operating at 93.3% Efficiency, A = 15. Even in a plain vanilla 80+ PSU operating at only 80% Efficiency, A = 5.
So the % increase in HEAT is roughly 10X the % decreased in EFFICIENCY. It would behoove you to be mindful of this.
So
Dr.Neale - Thursday, March 26, 2015 - link
Sorry for the typos at the end.But the drive for HIGH EFFICIENCY is mainly motivated by the desire for LOWER HEAT PRODUCTION, hence LOWER COOLING REQUIREMENTS, hence LOWER NOISE.
The savings on your electricity costs are merely a fringe benefit.
One other thing... Seeing as how a name-brand high-efficiency PSU with a 7-YEAR WARRANTY will likely last 10-20 years before failure, and moreover that the QUALITY OF THE POWER DELIVERY (voltage accuracy and ripple) strongly influence the MTBF of the other components in your computer build, WHY CHEAP OUT ON YOUR PSU ???
My father once bragged to me about how much money he saved by only buying a cheap $20 watch EVERY YEAR. I responded by showing him my $200 Citizen solar-powered watch, with the titanium bracelet and sapphire crystal, which was 10X as accurate, looked way nicer, and over 10 years would cost the same, and over 20 years half as much as he would spend on cheap, cheap-looking, and minimally accurate watches.
Better to buy a GOOD one once, than to buy a CHEAP one over and over again.
That's my philosophy for any (so-called) durable goods.
The_Assimilator - Tuesday, March 24, 2015 - link
Considering its price and features, I feel that this PSU deserved at least a Bronze award.E. Fylladitakis, please can you add a cable/connector table to your future reviews, similar to e.g. JonnyGURU? It's a bit of a pain to have to find the manufacturer's page, and available connectors play a big part in the selection criteria for a PSU. For example this model has 8x 6+2-pin PCIe and 15 SATA connectors, which is quite generous for a 1050W unit.
E.Fyll - Wednesday, March 25, 2015 - link
I simply do not do awards. Not at all, ever. Every specific product has its own advantages, disadvantages and market potential. For user X, a "bronze" product might be much better than a "gold" product, even if they are similar, just because the former suits the needs of the user better. Everything is relative.I'll consider the tables. It is not difficult, I just thought that it is something too redundant.
seerak - Tuesday, March 24, 2015 - link
While we're suggesting things to review (like midrange and lower end units), there's something I'm not seeing in reviews that really should be in there: UPS compatibility testing.Active power factor correction (PFC) doesn't seem to be a feature on this unit, but it's becoming increasingly common, and many of those units don't seem to like the modified sinewave output of cheaper UPS units. Seeing as this is the sort of incompatibility that could be very bad to learn about the hard way, that would be a good quick test to add into PSU reviews.
E.Fyll - Wednesday, March 25, 2015 - link
All modern switching PSU include (or should) APFC. It is actually illegal to sell a PSU without APFC in Europe nowadays, for example.Unfortunately, that is an issue of having a bad UPS, not a bad PSU. A cheap UPS will certainly output a modified sine wave. If the output is that bad that would cause overvoltage within the APFC circuit and blow the capacitors, it is not because the PSU is bad but because the output of the UPS is awful. A waveform that would increase the voltage at the APFC capacitors from the expected maximum of 340V to >450V does not even remotely pass as a "modified sine wave".
The only PSU that I have ever seen being too "sensitive" about modified sine wave signals was an Enermax design nearly a decade ago. Even then, Enermax enhanced the tolerance of the unit on its second revision. Still, it was not the PSU's fault that some UPS had terrible outputs. That's where the major difference between a $50 600VA Chinese off-brand UPS and a Schneider $200 600VA UPS usually lies...
Generally speaking, the voltage rating of the input capacitors usually hints the tolerance of the PSU when handling such signals. The higher, the better.
kaborka - Wednesday, March 25, 2015 - link
I heartily agree it should be tested whether a PSU will work with an old non-sine UPS. Why should we have to dump a working UPS and buy a $150+ new sinewave one when we get a new PSU? I'm all in favor of PFC, but I would have to return any one that wouldn't work with my existing inventory of older UPS.Oxford Guy - Wednesday, March 25, 2015 - link
It would be nice to see a review of the Corsair RM 850. Techpowerup's review suggests that it is the quietest actively-cooled PSU available. While they gave the RM 750 an even better rating for the average noise level throughout its operating range, the 850 delivers more watts before noise ramps up. So, it appears to be the best option for those who want to avoid noise while still having the ability to go beyond what passive PSUs offer.However, the concern is the quality of the capacitors primarily. It would be good to see another review site verify or contradict Techpowerup's results.
gsuburban - Sunday, June 28, 2015 - link
Yes. I've heard the Corsair RM model's are debatable because they are manufactured by Great Wall or some other maker. Folks tend to suggest on the Corsairs made by Seasonic are worthy such as the AX models. I haven't had any issues with their TX and TX-v2 PSU's in the past years but now the RM has replaced those. There's lot of features with the RM and reviews give them high points for total output accuracy while some say they a few percentage of them don't last or have problems.It would be reasonable to ask many of the hardware review people to do long term 60% power usage with daily on-off cycle testing combined for a final determination of longevity and/or problems. Most will say all PSU's loose total power output after a few years which stands to say why folks should purchase a PSU with about double the power they really need to allow for that and to allow the unit to operate and full efficiency.