Discussion in 'Reviews' started by crmaris, Nov 4, 2012.
To read this review go to: http://www.techpowerup.com/reviews/Corsair/AX760i/
"The following results were obtained at 40°C - 45°C ambient".... well -there was some heat goin' on
Damn! Yea 104-113 deg? did you turn on the volcano in your house?
I wonder how this will compare to a retail AX760i.
Nice review, only thing I wish that you would had is cable length.
I give the cable length (and not only), in the second page.
Thank You, don't know how I missed that!
Very nice PSU though damnnnnnn $229 for a 760w unit
Thats because it's digital.
Seriously, is the only thing that makes it digital is the use of software interaction? The Corsair Link isn't that appealing at upwards of $100 and from this review, not accurate. Is it just a marketing fad we'll have to endure?
FYI, Digital is defined as discrete with respect to time and amplitude. Analog is continuous time and amplitude.
The DSP's main advantages are the ultra high efficiency, the tight voltage regulation, the low ripple and the high PF. The monitor/control software is just a bonus since the digital control easily allows for it (you can take fairly easily data out of it and also control some of its functions since in essence it is a programmable processor and not a dump controller that has no I/O ports with the outside world).
So you don't pay actually for the software but you pay for the performance that the DSP offers to the PSU.
Nice looking PSU. I love corsair for their all-black cables, which look really cool in the case.
This is why I love your reviews, thanks for the sweet reply. Also your BJT avvy is pretty kick ass too.
Over Current Protection point are missing in the review
PSU reviews on the German www.technic3d.com include measured values for the 3.3V, 5V and individual 12V rail over current protection points. For example OCP kicks in at 132A on the 12V line for a Corsair HX1050 PSU, see http://www.technic3d.com/review/netzteile/1272-corsair-hx-1050-watt-netzteil-im-test/5.htm
Measured OCP values help to determine whether a PSU is capable of running specific components. In our case it helps to calculate how many SATA drives can be attached. For most PSU's the OCP on 5V rail kicks in when booting/waking many drives at once.
Why do PSU reviews at techpowerup.com not measure over current protection points in general?
What is the 5V (and when possible 12V) rail its measured over current protection points for the Corsair AX760i?
Quite a few review sites lack that information including this one and johnny guru.
I don't measure and OCP because the day has only 24h and also I don't find it so important. If in a PSU I see OCP triggering probs then I measure it.
Do they measure inrush current, hold-up time, run transient tests, crossload tests, provide the full data on tables so everyone can check on their readings/get additional info, make an internal analysis, use a hot-box to acquire results etc. etc.? There are far more important tests than OCP. Also why measure OCP on the minor rails. No modern system will draw more than 10A out of these rails unless you short them.
Also a SINGLE RAIL PSU doesn't actually have OCP but OPP!! It is meaningless for a single rail and high capacity PSU to have OCP. So in the HX1050 review they actually measured the OPP point and not OCP.
Now that's a psu! The voltage stability is almost unbelievable.
Big thanks to crmaris for another great review! Very much appreciated! I am in a market for a psu right now, going through many reviews, and I think your reviews are the most detailed and helpful of all I've read so far. If only all the interesting psu's had such quality reviews available online...
There is, perhaps, one thing that i feel is missing there. I suspect that the reason may have the same roots with the absense of OCP trigger points, but I do feel like asking about it, because, as a person with little knowledge in the field, I'd really like to hear crmaris opinion on this. I think that the first among the important characteristics of a psu is the degree of safety for the load, that it provides. And the primary cause of permanent damage to the load (our pc hardware) is caused by applying non-spec voltages to it. The main mechanism to protect from this is the OVP/UVP protection of a PSU. And if I understand it correctly, the standard requirements for these protections are quite liberal, so there may be up to +/-30% overvoltage or undervoltage situation, and the protection will still not activate.. in a standard-compliant psu. As users, we are obviously interested in these protections being functional, and set as tight as possible. The information on the setting of these protections is, however, not available. I didn't see manufacturers mention it, nor any single review reporting them. I wonder why is this? Maybe I am misunderstanding the importance of this.. or is it somehow prohibitively hard to test these protecions? Or are all the manufacturers really the same in this aspect? It would be great if you could clarify this for me, please..
all system components (mainboard, VGAs, etc) have their DC-DC converters that transform +12V to the voltages that these components need, so there is quite a headroom even for some, short, overvoltages.
Also ATX spec states exactly the limits for the OVP which is 15V for the 12V rail, 6.3V for 5V, 4.2V for 3.3V and 6.3V for 5VSB. However it also gives the minimal and maximum allowed values of OVP for each rail and for example at +12V min is 13.4V and max is 15.6V. So yeap the range is quite large.
Every PSU that is ATX compliant should follow the ATX guidlines including of course the output protection limits. However as far as I remember I never encountered a PSU that reached dangerous voltage levels (but many times I have seen PSUs with dangerously high ripple levels). Usually they blow before they overvolt components.
According to my experience the most sneaky thing in a PSU is ripple since it highly stress components like caps and if it goes over the roof it can easily kill or greatly shorten the lifespan of sensitive parts. Also you may test a PSU cold and find normal ripple levels but once you test it with over 35-40C ambient ripple can even exceed 200 mV.
So for me ripple goes first, then response to dynamic tests, then turn on transient reponse, v. reg. and the rest. Also an experienced reviewer during testing can spot the possible defects of a PSU and mention them afterwards (e.g. a low OTP point that can cause unexpected shut downs or a low OCP triggering point with the same effects). Finally highly stressing a PSU at >45C gives you a crystal clear view of its possible defects/problems, since it will simply not survive the tests or it will fail on them. This is why testing at high ambient is crucial, especially for a PSU.
Thank you for the fast and detailed reply!
I was rather worried about some persistent overvoltages, that might (perhaps?) develop in the long lifetime of a PSU (which I feel to be 5-7 years for a good PSU.. I'd probably replace or maybe recap(?) it after that period to stay on the safe side..). Like some malfunction in the feedback loop or somethiing, that could maybe lead it to outputting persistent 14V on the +12V rail, for example OVP at 13.4V would save the day
So I am actually happy to read, that you
I think that is kind of sweet
Speaking of ripple - yeah.. that is the tricky way of applying non-spec voltages to the components... I did recognize that as a crucial factor, but since there is no explicit protection against "overripple", i decided that the best (and only?) way to protect against it is to pick a PSU with 1) very good initial ripple suppression (as per your reviews, for example ) 2) Japanese 105°C caps (because I am under the impression that bad caps are what can cause "overripple" in a PSU which initially had good ripple suppression). Plus, providing the best thermal environment for the PSU.. within reason, that is I'm considering drilling some holes in the bottom of my PC case to let it actually get that <25°C outside air for cooling... If extra 10°C slice capacitor life by half, it actually can be very important to give it the coolest available air to breathe... Perhaps similar reasoning could be used to justify more efficient PSUs..? Although their fan profiles are generally much more relaxed, so...
Yeah, it's great that there are such wonderful reviewers, like yourself, who can do and do that proper testing. It is a really time consuming task though, I imagine, so one can't get all the interesting PSUs reviewed..
That phrase of your's also reminded me, how you always keep saying how you believe OTP to be a crucial protection in a PSU. I also feel that leaving out OTP is not a good practice overall.. which is why I favor Seasonic-branded units over the Corsair-branded ones.. Corsair seems to have different beliefs about OTP. Although these new AXi's do have it..coupled with cosmic price, especially for my region... I wonder if you could give a brief comment about why you believe OTP to be so important?
In my regular work I encountered many PSUs (not commercial ones but industrial grade) blown or damaged due to overheat, so I extremely value OTP. From my experience the number #1 enemy for a PSU is heat, so naturally I consider OTP as crucial.
Where is the OPP in the AX760i/AX1200i review?
It seems that you find OPP (over power protection) more important over OCP.
After a second read of both the AX760i the AX1200i review, I couldn't find a measured OPP value either.
When available, where can I find the 5V and 12V OPP in the review(s)?
nope I don't find OPP more important and I wonder from where you figured this out. Till recently I didn't even conduct overpower tests since I believe that if you need a larger PSU simply buy a new one and do not abuse your current.
Because of some demands I included an 110% overload test in my methodology but I am not going to enter the Watts race since even if I manage to overload a sample of mine to 150%, for example, this doesn't mean that another user with the same PSU will manage to do so. And if its PSU blows or fails rather quickly then most likely he will say "but I have read that this PSU can easily do 150% of its max rated power". Needless to say that the warranty doesn't cover such cases. Also I can understand in a degree the reason behind overloading small PSUs and see if they can handle the extra load. The thing I can't understand is why overload a >1000W beast! What system will be able to stress enough a 1200W PSU for example? In a review I am working right now I test a 1350W PSU and even with the mere 110% overpower percentage I use the output power increases to 1485W! But since I conduct this test to all PSUs I am forced to conduct to this PSU too, although it is practically pointless.
OPP is for all rails combined and not for each rail individually. It has to do, in other words, with the combined power that all rails deliver.
I never understood why they push big PSU's much either. It's kinda fun to see but as you say has little to zero practical value. Why buy a PSU with the intentions of pushing it above spec anyway?
In my understanding, it makes sense to overpower a PSU during testing in order to confirm, that it will either operate safely, or safely shut down due to OPP (as it should), rather than burn or explode or whatever
Besides, isn't overpowering possible without the intension to do so? Some malfunction like a short on a PCB (which may have high enough resistance not to trip the SCP) could potentially create abnormal load.. that's what one can read in jonnyguru's faq, at least. If I got that correctly.
Also, I understand that OPP essentially limits the current that the PSU can draw from the wall; this could be helpful in case of abnormally low voltage in the mains, which might make psu draw more current than it is designed to handle at otherwise normal load...
I am not an expert, and may be saying non-sense above; I will happily stand corrected in that case.
if a short circuit happens in the primary then immediately the fuse of the PSU will blown before any protection kicks in.
Now if a short circuit happens in the secondary side then the SCP (Short Circuit Protection) of the supervisor IC will handle the job and nowadays all supervisor ICs include this protection.
As for OPP I still see no point in testing this since some PSUs have a high OPP while some others have a low one and also the ambient in which you conduct the tests plays a significant role in this (if a PSU has a high OPP trigger point and you stress it under 50C then most likely it will fail). And even two PSUs of the same model may have a different trigger OPP point so I still find it useless to test it.
I already test all PSUs in the worst possible conditions and for long time, so if a PSU survives my tests then it is rock stable. But if I start to overpower it with 45-50C ambient then most likely I will degrade it in such a degree (if it doesn't fail) that it won't survive the test and the PSU can't be blamed for this or its manufacturer since operation at 50C is not required by ATX by only recommended. Most of the times the manufacturer sets an OPP trigger point for normal conditions or at least a little above the normal ones, but not for the worst case scenario which I use.
Also OPP takes readings from the APFC converter where the voltage is steady regardless the mains voltage.
Always love your PSU reviews Cmaris. I guess it will be a bit before prices come down. However for the lower heat and energy cost, it is almost worth it.
Separate names with a comma.