What's the obsession with PSUs?

[iandh]

The new HX 750 and 850 seem promising

I've been eyeballing them....

 

[n-ster]

Yea one of them may be the one to power my future i7 or 720BE rig

 

[angelkiller]

Well, there are two good reasons, although I've heard many bad ones:

1) Most PSU's hit their peak efficiency at anywhere from 30-70% of their max load rating, so that would be the 400w on 750w PSU you speak of. Assuming prices are all similar, it is good to tune your PSU's peak efficiency point to your average load draw. Not all PSU's are like this, but most are. Any good PSU review will show load vs. efficiency charts, and sometimes there can be quite a large difference, such as 87.5% @ 350w and 81% @ 750w

2) As a person who manufactures electronics for a living, and has lots of experience in transformers and amplifiers, I can wholeheartedly say it is much better for the PSU, and much better for your computer to not be running your PSU to the bleeding edge. Only the very best PSU's are able to run at 100% load and not have their lifetime adversely affected. Your power will be more stable, which can make for a more stable system, especially at OC, and your PSU will lead a happy and stress free life.

In general, based on experiences in my career, I think it is wise to run a PSU rated to at least 20% more than your actual load. Going beyond 50% can actually be bad though because you start to drop below the efficiency peak, but this really depends on PSU. For instance, at work, if I have a component that draws 6A @ 24v, then I will typically purchase a 10A switching power supply for the sake of reliability.

Hi, let me first say that I agree with what you're saying and you obviously know what you're talking about. But I have a couple of questions/clarifications that I'd like to ask.

You said only the best PSUs can run at 100%. Now, wouldn't it make sense that all PSUs should be able to run at 100%? Isn't that what the rating is for? So the rating tells the most power the PSU can deliver without adverse affects. (But of course not every PSU lives up to its specs)

I guess my issue here is what people call 'stress' on a PSU. (Everything I mention here is theoretical.) What is 'stress' on a PSU? Lets say a certain PSU is rated for 500W. Doesn't that mean it can give 500W continious, with no issues? If so, how would 500W be 'stressful'? I could understand that drawing 600W would be stressful, because the components are giving more than 100%. But if the components are running at less than 100%, let's say 80%, how is that stressful? Technically, it is stress because the parts are doing work, but the parts arn't working to thier capacity, which means they shouldn't be having problems.

I guess my issue is with ratings in general. If it's rated for a certain amount, but can't safely give that amount, doesn't that defeat the purpose of the rating? So what's wrong with running a PSU at 90% if it's supposed to be able to give 100%? (I know it less efficient and such) So if I can't run my 500W PSU at 100%, then what's the point of it being a 500W PSU?

I'm really sorry if this sounds like I'm contradicting you. I'm not!! This was supposed to be a follow-up question. Thanks for any help.

 

[silkstone]

Hi, let me first say that I agree with what you're saying and you obviously know what you're talking about. But I have a couple of questions/clarifications that I'd like to ask.

You said only the best PSUs can run at 100%. Now, wouldn't it make sense that all PSUs should be able to run at 100%? Isn't that what the rating is for? So the rating tells the most power the PSU can deliver without adverse affects. (But of course not every PSU lives up to its specs)

I guess my issue here is what people call 'stress' on a PSU. (Everything I mention here is theoretical.) What is 'stress' on a PSU? Lets say a certain PSU is rated for 500W. Doesn't that mean it can give 500W continious, with no issues? If so, how would 500W be 'stressful'? I could understand that drawing 600W would be stressful, because the components are giving more than 100%. But if the components are running at less than 100%, let's say 80%, how is that stressful? Technically, it is stress because the parts are doing work, but the parts arn't working to thier capacity, which means they shouldn't be having problems.

I guess my issue is with ratings in general. If it's rated for a certain amount, but can't safely give that amount, doesn't that defeat the purpose of the rating? So what's wrong with running a PSU at 90% if it's supposed to be able to give 100%? (I know it less efficient and such) So if I can't run my 500W PSU at 100%, then what's the point of it being a 500W PSU?

I'm really sorry if this sounds like I'm contradicting you. I'm not!! This was supposed to be a follow-up question. Thanks for any help.

I know the question wasn;t directed at me, but here's a couple of answers.
The testing they do for psu's is usually at 25C. Now that is an underestimate for the noraml operating temperature and most will generally run somewhere between 30C-40C. For every degree in temperature increase the PSU actually looses some of its rated power output, in the range of 1-3w per degree C. That's one of the reasons why good quality PSU will come with a very hefty heatsink inside.
Edit - Also the specs on PSU's are kind of vague... yes most can deliver what they are rated at, or close to, plus or minus 10%. But it's all about how that power is distributed over the various rails that is important. Also the minimum amount of power a PSU can deliver over a single rail is important, and so if you draw less power than is rated you can have big stability issues.

The second point about electrical stress. The components get stressed all the time, it's not exactlythe same as mechanical stress but for all intents and purposes you can imagine it working the same way. The majority of stress comes from either switching the PSU on, where the components are actually going from 0w to 200-300w. This stress is also from increased power demands while the computer is operating, e.g. opening a video encoder, running a game, and also to a lesser extent opening a program. This is where the demand on the power supply many jump from 200w upto 400w.

Stress on a PSU isn;t actually caused by running the PSU at high demand for a duration (although this does indirectly play a part in "stress")
The cause of this stress is actually due to moving parts believe it or not (hence the simile with mechanical stress being somewhat valid.) Even tho PSU contain no moving parts barring the fan, the actual components will expand and contract depending on the temperature, now this can be very bad for certain electronic components, mainly capaictors, and is the leading cause of fatigue.

Hope this helps with your Q.