PSU tier list 2.0

[logainofhades]

The point deductions, for what type of modularity it has, is asinine. Deducting points for price is too.

 

[turkey3_scratch]

Naaa, I got you for that. Lol.

I've told him multiple things, but mostly miscalculations on his voltage percentages, to which he gets sort of mad at me and thinks I'm a you-know-what bag, but too many people seem to shy away from getting an account at that website and participating in these discussions there. I mean, it's literally so easy to contact Oklahomawolf/Tazz/Jonny, they're on every day a ton.

 

[MasterMace]

This guy here claims to have had two (2) GTX 780's destroyed by a Corsair CX 750M PSU (Tier 4).

http://www.mmo-champion.com/threads/2082599-GPU-burned-out-faulty-socket-to-blame?p=42613732#post42613732

Jesus Christ, I knew Corsair CX 's were bad. What do you PSU experts say?

First, i wouldn't suggest a CX above 430W. (Can't view the article from location) Second I wouldn't suggest running 2 way SLI/XFire/EMA for that high of a Graphics on a PSU under 850W.

 

[logainofhades]

CX750M is not a good unit to begin with, and underpowered for those power hungry cards. I would say at least an Evga B2 850, to handle them.

 

[Karadjgne]

I'm surprised the psu outlived the gpus.

 

[turkey3_scratch]

Well the new CX750M is suppose to last 5 years loaded to 750W 24/7 with 40C ambient temps.

The old one was 3 year warranty I think? Then that's supposed to last 3 years running 24/7 loaded to 750W with 30C ambient temps.

 

[Someone Somewhere]

I would have to read their warranty, but I would be surprised if it specifies 24/7.

 

[turkey3_scratch]

I'm just saying what the definition of a 750W power supply is. McSteel on Jonnyguru told me what a 750W power supply really is just means that the manufacturer says its good for the warranty period running at the rated temp 24/7. Because for many a long while I was seeking out what rated wattage really was, so he (who I consider #1 PSU authority) has finally told me. Of course the whole system is still corrupt since we are taking the manufacturer's word for it, so there's not really a standard. They can still slap whatever number they want in the label and people think it actually means something.

This has caused me to devise the Turking Test 1.0 which proves my hypothesis that the wattage rating of a power supply, while it technically means what I said above, has no real meaning or solidity. You take two power supplies. You don't know anything about them because brand names and labels are scratched off. You are given all the testing equipment in the world. Then you have to test these power supplies and determine their rated wattage yourself. Then the ultimate test is: will the rated wattage you determine for them be = to the actual rated wattage of them before their labels were scratched off?

The answer is probably no. You're not going to end up with the same rated wattage since you'll be left confused as heck determining how to rate it. Even if you get lucky, your method you used will not maintain its consistency for a variety of other power supplies. There will always be power supplies I can pick and choose to prove the methodology incorrect. Because there is no proper methodology for determining rated wattage. It would be wonderful if there was an actual thing reviewers tested called wattage but they don't because there is no such thing.

Oh wait did I go off on a side train again? Oh well it was fun. That is my finalized Turking Test. Version 1.0.

 

[jimmysmitty]

It's flawed in that every single unit has not been tested and reviewed, which is not the fault of the list. Huntkey is a perfect example. There is 1 decent unit, the Huntkey Jumper, which happened to be reviewed, and is good, the rest of Huntkey I wouldn't trust to be a reliable paperweight. If not for that 1 review, the Jumper too would have been lumped with the rest. Lousy.

Problem is that not every PSU is given to be nor could anyone afford to review every unit. However take history and quality from OEMs and you can discern a lot from it. We know that historically Seasonic PSUs are well built and perform well so a new Seasonic will probably perform well and be built well.

There is no all in one solution for what is a good or bad PSU but there are some ideas. None are perfect but I would rather steer someone to a PSU I know that is a great product rather than try t assume one without any reviews or information may be good.

 

[Karadjgne]

Exactly. That was a response to the prior post. Quotes on a mobile is an exercise in futility, so I expect things to be taken out of context every now and then

 

[turkey3_scratch]

Problem is that not every PSU is given to be nor could anyone afford to review every unit.

Except the PSU companies themselves. They review every unit on the market basically. But we can't see that data 😀

 

[Karadjgne]

I seriously doubt that any given warranty is based on 100% load at 24/7. You'd have to live in the north pole with the doors and windows open in winter to keep a 750w unit at 30°C or below at 100% load for longer than a few minutes. 24/7 at idle or nominal usage of 50% load would be more likely. Corsair RM750 units failed 100% load testing due to shutdowns caused by reaching thermal limits, what makes you think a CX will last 3 solid years of such abuse?

 

[turkey3_scratch]

I seriously doubt that any given warranty is based on 100% load at 24/7. You'd have to live in the north pole with the doors and windows open in winter to keep a 750w unit at 30°C or below at 100% load for longer than a few minutes. 24/7 at idle or nominal usage of 50% load would be more likely. Corsair RM750 units failed 100% load testing due to shutdowns caused by reaching thermal limits, what makes you think a CX will last 3 solid years of such abuse?

The temp rating is for ambient temps not the temps inside of the PSU, especially considering that the temps inside a PSU are not uniform and some areas could be 70C-90C even.

The CX750M did not "fail" load testing in my opinion. Oklahomawolf clearly shoved intake temperatures that were too hot into it, resulting in it to shut down. Which is a good thing. It's a shame people turn this into a negative thing. You have EVGA power supplies like the P2 which lack any form of OTP. If you shove hot enough temps in there, by golly it could very well blow up; same with the G2 units since they (the 650W and 550W I know for sure) lack OTP. Now those high-end units might not be subject to such temps, but take into account something like the EVGA B units without OTP. I can see bad scenarios where heat is pumped into them and they blow up because of no OTP. The bridge rectifier would have blown on that CX had it not shut down. I still have speculation that it would blow under the right temps and overload but that's a discussion for another time.

I never said the CX will last 3 years under such conditions. It's supposed to, though. I'm using it as a hypothetical point on how there needs to be more standards and how everything is corrupt. Temperature ratings themselves are ridiculous, no standards by which they are made. Load tables - ridiculous - no standards by which they are made. Rated wattage - ridiculous - no standards by which it is made.

Intel clearly did not do a good job of outlining the specifics of this stuff in their spec. The number one thing people look at is the wattage of a power supply and it really doesn't mean anything since it just has no standard to it. And Intel is firm in that they will never revise the ATX spec again. They say that if someone wants to make a new spec they can do it themselves and will have to get all the manufacturers to tag along. So seeing the elimination of the 3.3V rail and such stuff will not happen. But this thread I made on Jonnyguru exploded into a bunch of people talking about what their new spec would be. Most people want the PSU to be 12V only. Or something like 21V. There's a lot of technicality to it all. http://www.jonnyguru.com/forums/showthread.php?t=13650

Maybe AMD will make a new spec? Doubtful. That won't ever happen. It's a shame that change is so difficult. The minor rails are nearly pointless these days. It's why it's hard to recommend group regulated power supplies, almost everything is 12V.

 

[Karadjgne]

Well the new CX750M is suppose to last 5 years loaded to 750W 24/7 with 40C ambient temps.

The old one was 3 year warranty I think? Then that's supposed to last 3 years running 24/7 loaded to 750W with 30C ambient temps.

End quote.


http://m.hardocp.com/article/2013/11/13/corsair_rm750_750w_power_supply_review/4#.V_R42HQo7qA
OTP tripping at less than 50% load at 45°C (can we say ambient temps in India during the summer for some ppl?). Failed in my book. And that's assuming the psu is fed from the outside ambient temps, not like some ppl need a psu that's fed from case air or even if it's a top load, which would be temp suicide. Seems the CX would be the superior psu here, as it doesn't include a really dumb smart fan.

 

[turkey3_scratch]

Yeah I don't like the PSU myself.

 

[Karadjgne]

It's tier 4 for that reason alone, above any and all decent performance when under normal conditions. Even when stacked against the El cheapo XFX 550w, which can handle full loads at 50°C ambient, and then some, it really starts to look bad. And it's not the only psu by any means to go down the toilet when things get warm. That's one test I totally agree with, the hot box. If a psu can't pass brutal, it's a failure. Corsair has the unenviable position on the top of my stink list simply because they, like Seasonic, have pretty much world wide distribution, which includes India and other extreme climate countries. If it won't work 100% at 50°C, you better hope the pc is in an air-conditioned room. A 30°C warranty? That's a joke.

 

[turkey3_scratch]

It's tier 4 for that reason alone, above any and all decent performance when under normal conditions. Even when stacked against the El cheapo XFX 550w, which can handle full loads at 50°C ambient, and then some, it really starts to look bad. And it's not the only psu by any means to go down the toilet when things get warm. That's one test I totally agree with, the hot box. If a psu can't pass brutal, it's a failure. Corsair has the unenviable position on the top of my stink list simply because they, like Seasonic, have pretty much world wide distribution, which includes India and other extreme climate countries. If it won't work 100% at 50°C, you better hope the pc is in an air-conditioned room. A 30°C warranty? That's a joke.

No the warranty works if your temps are over 30C. I don't think you are thinking I was talking about what I think I was talking about. Oh well.

I think what I was trying to say is warranty is based off that rated temp. Gah, nevermind, my point is the manufacturers can do whatever they want and don't have to follow standards when it comes to this stuff. That really me.

 

[Someone Somewhere]

The answer is probably no. You're not going to end up with the same rated wattage since you'll be left confused as heck determining how to rate it. Even if you get lucky, your method you used will not maintain its consistency for a variety of other power supplies. There will always be power supplies I can pick and choose to prove the methodology incorrect. Because there is no proper methodology for determining rated wattage. It would be wonderful if there was an actual thing reviewers tested called wattage but they don't because there is no such thing.

This applies to literally anything you measure, except possibly basic counting.

How long is this piece of string? Depends what force you put on it (and if you test different pieces of string, how you choose that force for different sized string), temperature, humidity, whether you specifically unkink and straighten it, whether you measure from the longest part or where it reaches full thickness or what, and loads of other factors.

Generally, there is a standard for a class of product that specifies exactly how to measure this - AS/NZS, EN, BS, probably not UL as they're more safety.

I don't know if there's one specifically for computer power supplies (probably too niche), but there will be for mains-powered DC-output supplies in general.

The ATX spec probably provides information on how to rate it.

 

[ko888]

Why all this misinformation about the published temperature rating?

All it is, is a temperature de-rating specification.

http://www.power-eetimes.com/content/power-supply-de-rating-practice

 

[turkey3_scratch]

Why all this misinformation about the published temperature rating?

All it is, is a temperature de-rating specification.

http://www.power-eetimes.com/content/power-supply-de-rating-practice

The graphs on that page imply a power supply is going to run out of power someone or hit a peak. That won't happen and it's one reason why we have OPP and OCP. Now, if they define output power as the point at which some component blows up, that's different; that just means something blew up, in which case you'd hope a power supply would have OTP kick in before such a thing were to happen.

I think my same test applies to temperature rating. You are given two random power supplies with their labels scratched off. You have all the equipment at your side to test them. In the end, you have to determine the rated wattage and rated temperature.

1) Will your determined rated temperature value match that of the actual power supply before the label was scratched off?
2) What methodology was used to determine this?
3) Is this methodology consistent for every power supply, or are there counterexamples with which this methodology does not work?

If any one of those is false, then rated temperature does not have a true definition. Unless those three things can be proven, rated temperature (and rated wattage) are in my book still dumb numbers. So far, though, nobody has been able to answer my test and disprove it, so my Turking Test 1.0 stands strong. Applies to both rated wattage and rated temperature. Come on somebody, give me a shot! Tell me your methodology, what you would do.

For all of you who think I'm nuts, here is my philosophical argument for why rated wattage and rated temperature have no meaning. It is a completely valid argument. The form of the argument is in variables below.

1) If there is a methodology to test an arbitrary power supply with a scratched off label with profesisonal equipment and come up with the original values for rated temperature and rated wattage on the label, then rated wattage and temperature does have a meaning pertaining to that specific power supply.

2) There is a methodology to test an arbitrary power supply with a scratched off label with profesisonal equipment and come up with the original values for rated temperature and rated wattage on the label.

3) So, rated wattage and temperature does have a meaning pertaining to that specific power supply.

4) If rated wattage and temperature do have a meaning pertaining to that specific power supply and the same methodology applies to every power supply, then rated temperature and rated wattage have a consistent meaning for every power supply.

5) The same methodology does not apply to every power supply.

6) So, rated temperature and rated wattage have no consistent meaning.

7) If rated temperature and rated wattage have no consistent meaning, then they should not be used as parameters of comparison for power supplies.
_____________________________
C) Therefore, rated temperature and rated wattage should not be used as parameters of comparison for power supplies.

1) If A, then B.
2) A
3) So B
4) If B and C, then D
5) Not C
6) So not D
7) If not D, then F.

C) F

 

[Karadjgne]

Depends on the definition of rated wattage and how you apply it. To me, rated wattage isn't output. It's input. When all things are connected to the pc that need to be connected to make a pc, the rated wattage is what can be seen at input when the psu is at full load equitable load on all its rails. Not necessarily the full load of each rail, but full cross load. Minus, of course, the efficiency.

So hook everything up to a test bench, crank it up, see how much it pulls from the wall, multiply by efficiency and slap on a wattage sticker.

So a bench load pulling 630w from the wall with a supposed bronze rating would be a 520w psu. Sticker or not.

 

[CountMike]

Sticker or not, it always shows rating for full power usage and not to be used like that full time. Adding at least 25% to estimated maximum usage should ensure best stability on particular PSU. Lower tier PSUs may need even more since manufacturers tend to over estimate (to say at least) capabilities of their products.

 

[Someone Somewhere]

So far, though, nobody has been able to answer my test and disprove it, so my Turking Test 1.0 stands strong.

Absence of evidence is not evidence of absence.

1) If there is a methodology to test an arbitrary power supply with a scratched off label with profesisonal equipment and come up with the original values for rated temperature and rated wattage on the label, then rated wattage and temperature does have a meaning pertaining to that specific power supply.

They are inter-related; not directly measurable independently.

They are also statistical (test hundreds of PSUs, sample size of one is automatically useless), heavily rounded, and rely on, amongst other things, expected duty cycle and design life (do you want to wait five years?). You cannot easily go back from final product to design criteria; it's a one-way function.

7) If rated temperature and rated wattage have no consistent meaning, then they should not be used as parameters of comparison for power supplies.

Merely because something is not easy to measure consistently does not mean that it does not exist. Plus, you still fail to provide any better alternative- or do I remember you wanting them to get rated in amps at 12V, which has exactly the same problems?

 

[turkey3_scratch]

Sticker or not, it always shows rating for full power usage and not to be used like that full time. Adding at least 25% to estimated maximum usage should ensure best stability on particular PSU. Lower tier PSUs may need even more since manufacturers tend to over estimate (to say at least) capabilities of their products.

The RM650x can handle 800W continuously. So 650W in the case of it does not seem to be full power. A PSU like it should be fine 24/7 under 650W of load for 10 years.

@Someone Somewhere: no, I never wanted them to get rated in amps at 12V. In fact I don't agree with how amp ratings are done either.

The big void in all these ratings is we don't know the target performance the manufacturer is going for because they don't specify that on the label. Is this 550W PSU going for 2% voltage regulation? You may have a 550W Platinum PSU with 2% which is actually a 650W Gold PSU with 3% regulation and 750W Bronze PSU with 5% regulation. But we really don't know what the performance target is. We're not told that. Of course that's a very hypothetical scenario with the Platinum > Gold > Bronze.

 

[timeconsumer]

Coolmax's target performance: Will 51% of the models run at the numbers on the box without exploding?