Definitely true. The 80Plus loading methodology differs from the standard ATX loading, and that can easily account for the 1-2% difference between what Christoph measured and what CM reports.
They don't list the temperature there but IIRC it is 25c (will have to check when I get home). Also, 80 Plus Silver is 85-88-85% not 82-85-82% as it seems to be indicated.
I was just about to mention this as there's some misinformation on 80plus scattered about in the article. Hopefully you'll clean that up soon. To me it's a pretty big achievement to get 85-88-85 at 25C on 115V on a huge 900w psu like this one. One thing I've noticed is that some of the new low power psu models are showing extremely high low wattage efficiency such as 90-92%. Some of these include the two dell 80plus silver models and some of the new actel ones. 80plus Bronze is much easier to achieve as most 80plus psu's are already near 85% in the middle range.
"80Plus Bronze requires at least 82% efficiency at 20% load, 85% efficiency at 50% load, and 82% efficiency at 100% load. 80Plus Silver bumps the requirements up to 85%/88%/85% for the same 20%/50%/100% loads. In short, the Cooler Master 900W UCP had better be able to reach 88% efficiency."
I'm not sure how that's at all confusing. As to the UCP 900W actually achieving Silver in our tests, it does appear to fall just short at the 20% load mark, but only by ~1% (84% efficiency when it's supposed to be 85%). I'm not going to worry overly much on that point, as there's a certain margin of error in testing.
"Before you get too carried away with the high efficiency, however, we need to rain on your parade a bit. The 80Plus Silver badge means that most users will get at least 82% efficiency, but even 20% load represents a power output of 180W, which for an idle system represents a significant amount of high-end hardware."
Since that is NOT what 80Plus Silver means it could be a bit misleading.
There are four 12v rails, yet you only show regulation on a single rail, not even labeled at that... where are the voltage regulation line graphs for the other 3 rails, if there are in fact 4 rails?
+12V rails are rarely independent. Usually "multiple" +12V rails is just a +12V rail split up into four, six, etc. with an over current protection circuit in place for each. If there's any "regulation" difference between one +12V rail and another, it's usually caused by resistance between the +12V source and the end of the connector and NOT actual poor voltage regulation. So the best course of action would actually be to average out the results or combine +12V rails into one.
ehm we combined them into one graph, that's why they are so thick ;) The graph shows in which area all of the rails have been regulated. We had shown differently before but with six rails for example you cant see anything anymore...
The 12V rail problem is that 12V1 (rated at 25A) supplies the power for the 24-pin connector, the 4-pin ATX12V connector, and all the SATA and Molex connectors. Meanwhile, 12V2 *only* powers the EPS12V connector (which quite a few people won't even use!), and 12V3 and 12V4 are dedicated to the PEG connectors.
Basically, there's a lot of stuff coming off of the main 12V rail, and thus it's going to be virtually impossible to come anywhere near the rated output unless you happen to have an EPS12V connector on your motherboard. More important is that with the right combination of hardware (i.e. quad-core overclocked CPU, a high-end GPU, and several HDDs) you could easily overload 12V1.
The EPS12V is the 8-pin CPU connector, correct? The same one that seems to be far more common these days on the class of motherboards likely to be used with a 900W PSU than the 4-pin connector?
I don't know that I would call EPS12V "common". It's used on some high-end mobos, but not on others. It was initially more of a workstation/server connector. Some PSUs have a 4/8-pin cable that works with either ATX12V or EPS12V, but it seems Cooler Master decided to go with a dedicated ATX12V and a dedicated EPS12V. It would have made a lot more sense IMO if they had all of the peripherals on the same rail as the EPS12V (and ATX12V for that matter).
Anandtech are not interested in presenting a complete professional review of PSU's, instead they have stated we should "trust them" on ripple and noise questions.
I think that Anandtech PSU reviews should look at JonnyGuru.com and HardOcp.com and take notes on how to do things right.
"We see very little voltage drop on the rails, which is very good. All of the rails are within 2% to 2.5% of regulation, starting slightly higher than the target voltage and dropping slightly below target at higher loads. Ripple on the 12V rails is also exceptionally good, measuring at most 8mV. Lower voltage rails also perform decently, though not quite as good, measuring up to 21mV. The ability of this power supply to deliver the required wattage with little voltage drop even at higher loads leaves us with a good impression."
Posting a graph that shows ripple isn't really any more informative than what we did right there: up to 8mV ripple on the 12V rails and 21mV on the 3.3V and 5V lines. You might want to read the text a bit more closely before firing off complaints. We state in the conclusion that the PSU has very good ripple results, and I guess that's probably as far as you got? The previous several pages of test results are there for a reason.
A graph educates people, and shows that you've performed the necessary readings to construct it. Ripple/Noise is arguably one of the most important graphs to have on a review.
Sure, you had 'something' in this review, which is an improvement over the last few months.
Do you have any evidence that 78mV of ripple has caused any problems by itself, isolated as the variable? The powered parts themselves cause more than 78mV of ripple local to them, where it might matter if it ever did - this if you had a hypothetically perfect 0mV ripple PSU.
I too have only had good experiences with their products and rebates. I have heard that others have experienced problems with rebates, but I don't usually hear many complaints regarding their products.
I guess that some people must get a different experience than others, the question is which one of us had the normal experience and which one of us got lucky. Until I have a bad experience, I am sticking with my Coolermaster products.
You forgot to mention the business model coolermaster has relied on. Sell your product that has a worth of about $25 for retail price of $75 and then offer a $65 rebate that will likely not get paid!
Then, when the customer has problems with the product, just make the process to repair or replace so difficult, that only the most persistant buyers will percivere the process to get something done!
It has everything to do with any Coolermaster product review! If the company is poor on customer support, then it affects the value equation for prudent buyers. Sorry you don't understand!
In the past 12 months I've bought 2 Coolermaster PSU and 1 case, all with rebates making close enough to free. They were all an excellent compromise for the price and all rebate checks were received.
Yes you may see lots of people that didn't get a rebate check but remember that when something normally costing about $40-50, not $25, ends up around $0 to $10 after a rebate, you have a very large # of people who take advantage of it so even 100 reports of not getting a rebate check may be a small % of total buyers.
I happen to be typing on a system that has a 600W Coolermaster Extremepower PSU in it, while it isn't very old yet at about 8 months, it has done fine thus far with overclocked CPU, video, 4 hard drives. Granted it's only peaking at roughly 300W consumption but a little common sense is good in this area, one doesn't buy a $50 PSU then expect to get 600W out of it long term.
I really appreciate the test articles and the time spent on the testing setup. Thanks for the heads up about the 80plus silver certification/marketing ploy. I now see that this is not the PSU for my application.
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
33 Comments
Back to Article
andlcs - Wednesday, July 30, 2008 - link
Solid polymer capacitors are ELITE brand.http://www.chinsan.com/product/index.asp?id=22">http://www.chinsan.com/product/index.asp?id=22
sheh - Tuesday, July 29, 2008 - link
20% load efficiency at is lower than 85% regardless of input voltage.strikeback03 - Tuesday, July 29, 2008 - link
IIRC they test by the ATX methodology, not the 80Plus loading, which might account for the difference.MrOblivious - Tuesday, July 29, 2008 - link
Or it could be the unit to unit variation, the different temperature, or the different load pattern, etc. ;)JarredWalton - Tuesday, July 29, 2008 - link
Definitely true. The 80Plus loading methodology differs from the standard ATX loading, and that can easily account for the 1-2% difference between what Christoph measured and what CM reports.MrOblivious - Tuesday, July 29, 2008 - link
80Plus does list what the requirements for certification on each test report (115v 60hz) and the load steps used to accomplish them. You can see the UCP report here: http://www.80plus.org/manu/psu/psu_reports/SP215_C...">http://www.80plus.org/manu/psu/psu_repo..._MASTER_...They don't list the temperature there but IIRC it is 25c (will have to check when I get home). Also, 80 Plus Silver is 85-88-85% not 82-85-82% as it seems to be indicated.
tomoyo - Tuesday, July 29, 2008 - link
I was just about to mention this as there's some misinformation on 80plus scattered about in the article. Hopefully you'll clean that up soon. To me it's a pretty big achievement to get 85-88-85 at 25C on 115V on a huge 900w psu like this one. One thing I've noticed is that some of the new low power psu models are showing extremely high low wattage efficiency such as 90-92%. Some of these include the two dell 80plus silver models and some of the new actel ones. 80plus Bronze is much easier to achieve as most 80plus psu's are already near 85% in the middle range.JarredWalton - Tuesday, July 29, 2008 - link
"80Plus Bronze requires at least 82% efficiency at 20% load, 85% efficiency at 50% load, and 82% efficiency at 100% load. 80Plus Silver bumps the requirements up to 85%/88%/85% for the same 20%/50%/100% loads. In short, the Cooler Master 900W UCP had better be able to reach 88% efficiency."I'm not sure how that's at all confusing. As to the UCP 900W actually achieving Silver in our tests, it does appear to fall just short at the 20% load mark, but only by ~1% (84% efficiency when it's supposed to be 85%). I'm not going to worry overly much on that point, as there's a certain margin of error in testing.
MrOblivious - Tuesday, July 29, 2008 - link
"Before you get too carried away with the high efficiency, however, we need to rain on your parade a bit. The 80Plus Silver badge means that most users will get at least 82% efficiency, but even 20% load represents a power output of 180W, which for an idle system represents a significant amount of high-end hardware."Since that is NOT what 80Plus Silver means it could be a bit misleading.
JarredWalton - Tuesday, July 29, 2008 - link
Ah... I see my typo. I put that paragraph in, and apparently hit a 2 instead of 5. Should be 85% there, you're right....MrOblivious - Tuesday, July 29, 2008 - link
Sorry meant to say seems to be indicated in the article in my last line.Adamantine - Tuesday, July 29, 2008 - link
There are four 12v rails, yet you only show regulation on a single rail, not even labeled at that... where are the voltage regulation line graphs for the other 3 rails, if there are in fact 4 rails?jonnyGURU - Tuesday, July 29, 2008 - link
+12V rails are rarely independent. Usually "multiple" +12V rails is just a +12V rail split up into four, six, etc. with an over current protection circuit in place for each. If there's any "regulation" difference between one +12V rail and another, it's usually caused by resistance between the +12V source and the end of the connector and NOT actual poor voltage regulation. So the best course of action would actually be to average out the results or combine +12V rails into one.More on "multiple" +12V rails: http://forums.anandtech.com/messageview.aspx?catid...">http://forums.anandtech.com/messageview.aspx?catid...
Christoph Katzer - Tuesday, July 29, 2008 - link
ehm we combined them into one graph, that's why they are so thick ;) The graph shows in which area all of the rails have been regulated. We had shown differently before but with six rails for example you cant see anything anymore...SilthDraeth - Tuesday, July 29, 2008 - link
I read the review and I saw you nitpick about a few things, but I didn't read about any real problem.JarredWalton - Tuesday, July 29, 2008 - link
The 12V rail problem is that 12V1 (rated at 25A) supplies the power for the 24-pin connector, the 4-pin ATX12V connector, and all the SATA and Molex connectors. Meanwhile, 12V2 *only* powers the EPS12V connector (which quite a few people won't even use!), and 12V3 and 12V4 are dedicated to the PEG connectors.Basically, there's a lot of stuff coming off of the main 12V rail, and thus it's going to be virtually impossible to come anywhere near the rated output unless you happen to have an EPS12V connector on your motherboard. More important is that with the right combination of hardware (i.e. quad-core overclocked CPU, a high-end GPU, and several HDDs) you could easily overload 12V1.
strikeback03 - Wednesday, July 30, 2008 - link
The EPS12V is the 8-pin CPU connector, correct? The same one that seems to be far more common these days on the class of motherboards likely to be used with a 900W PSU than the 4-pin connector?JarredWalton - Wednesday, July 30, 2008 - link
I don't know that I would call EPS12V "common". It's used on some high-end mobos, but not on others. It was initially more of a workstation/server connector. Some PSUs have a 4/8-pin cable that works with either ATX12V or EPS12V, but it seems Cooler Master decided to go with a dedicated ATX12V and a dedicated EPS12V. It would have made a lot more sense IMO if they had all of the peripherals on the same rail as the EPS12V (and ATX12V for that matter).Bozo Galora - Tuesday, July 29, 2008 - link
Weren't you going to add ripple and noise tests?Or do I have the wrong recollection?
These guys say it had 78mv on 12V line
http://www.techpowerup.com/reviews/CoolerMaster/UC...">http://www.techpowerup.com/reviews/CoolerMaster/UC...
Amart - Tuesday, July 29, 2008 - link
Anandtech are not interested in presenting a complete professional review of PSU's, instead they have stated we should "trust them" on ripple and noise questions.I think that Anandtech PSU reviews should look at JonnyGuru.com and HardOcp.com and take notes on how to do things right.
JarredWalton - Tuesday, July 29, 2008 - link
"We see very little voltage drop on the rails, which is very good. All of the rails are within 2% to 2.5% of regulation, starting slightly higher than the target voltage and dropping slightly below target at higher loads. Ripple on the 12V rails is also exceptionally good, measuring at most 8mV. Lower voltage rails also perform decently, though not quite as good, measuring up to 21mV. The ability of this power supply to deliver the required wattage with little voltage drop even at higher loads leaves us with a good impression."Posting a graph that shows ripple isn't really any more informative than what we did right there: up to 8mV ripple on the 12V rails and 21mV on the 3.3V and 5V lines. You might want to read the text a bit more closely before firing off complaints. We state in the conclusion that the PSU has very good ripple results, and I guess that's probably as far as you got? The previous several pages of test results are there for a reason.
Amart - Wednesday, July 30, 2008 - link
A graph educates people, and shows that you've performed the necessary readings to construct it. Ripple/Noise is arguably one of the most important graphs to have on a review.Sure, you had 'something' in this review, which is an improvement over the last few months.
Christoph Katzer - Tuesday, July 29, 2008 - link
With 78mV no company in the high-end field would have released the product...mindless1 - Tuesday, July 29, 2008 - link
Do you have any evidence that 78mV of ripple has caused any problems by itself, isolated as the variable? The powered parts themselves cause more than 78mV of ripple local to them, where it might matter if it ever did - this if you had a hypothetically perfect 0mV ripple PSU.Zak - Tuesday, July 29, 2008 - link
I found CM products to have competitive prices - but I don't shop for high end stuff - and I've always received their rebates.nycromes - Tuesday, July 29, 2008 - link
I too have only had good experiences with their products and rebates. I have heard that others have experienced problems with rebates, but I don't usually hear many complaints regarding their products.I guess that some people must get a different experience than others, the question is which one of us had the normal experience and which one of us got lucky. Until I have a bad experience, I am sticking with my Coolermaster products.
Glenn - Tuesday, July 29, 2008 - link
You forgot to mention the business model coolermaster has relied on. Sell your product that has a worth of about $25 for retail price of $75 and then offer a $65 rebate that will likely not get paid!Then, when the customer has problems with the product, just make the process to repair or replace so difficult, that only the most persistant buyers will percivere the process to get something done!
Hey, it must be working for them!
JEDIYoda - Tuesday, July 29, 2008 - link
so what does anything you say have to do with this review?For that matter can you back up what your claiming with links??
Glenn - Tuesday, July 29, 2008 - link
It has everything to do with any Coolermaster product review! If the company is poor on customer support, then it affects the value equation for prudent buyers. Sorry you don't understand!Do some reading-
JEDIYoda - Wednesday, July 30, 2008 - link
I do understand very well! It is you that has a axe to grind against the Coolermaster products.The company support and how they handle problems has nothing at all to do with this review of the product!!
Yoyr bitch is against the compant itself...the 2 are very seperate!!
As usual a very well done review!!
mindless1 - Tuesday, July 29, 2008 - link
In the past 12 months I've bought 2 Coolermaster PSU and 1 case, all with rebates making close enough to free. They were all an excellent compromise for the price and all rebate checks were received.Yes you may see lots of people that didn't get a rebate check but remember that when something normally costing about $40-50, not $25, ends up around $0 to $10 after a rebate, you have a very large # of people who take advantage of it so even 100 reports of not getting a rebate check may be a small % of total buyers.
I happen to be typing on a system that has a 600W Coolermaster Extremepower PSU in it, while it isn't very old yet at about 8 months, it has done fine thus far with overclocked CPU, video, 4 hard drives. Granted it's only peaking at roughly 300W consumption but a little common sense is good in this area, one doesn't buy a $50 PSU then expect to get 600W out of it long term.
Glenn - Tuesday, July 29, 2008 - link
http://forum.coolermaster.com/viewforum.php?f=29">http://forum.coolermaster.com/viewforum.php?f=29cparka23 - Tuesday, July 29, 2008 - link
I really appreciate the test articles and the time spent on the testing setup. Thanks for the heads up about the 80plus silver certification/marketing ploy. I now see that this is not the PSU for my application.