End Power Supply Myths!

[Phlargo]

Everyone has an opinion about power supplies, yet I have known very few signs that clearly show that a power supply is malfunctioning or not putting out enough juice. The only confirmed power supply deaths or inadequacies can be described in a few short examples:

1. Something in the system stops working - a replacement power supply fixes the problem when nothing else does. (after you've isolated all other components)
2. When buying a new piece of hardware, it indicates in needs X amps to run on the 12V rail. My power supply has X-1 amps on the 12 volt rail. It doesn't work, power supply isn't enough.
3. My friend got a message on his computer saying that in order to run SLI on his video cards, he'd need more power (this is not confirmed - I wasn't there and have never heard of such a thing).

As far as I know and have been concerned, any other discussion about power supplies has been based almost entirely on myth and emotional responses as opposed to an actual factual basis that dictated a power supply replacement/upgrade was necessary. I read 5 posts a day where people suggest a new power supply without any real reason to do so. In fact, based on such a post, some months ago, I purchased a high performance power supply which I subsequently determined to be completely unnecessary. Is it our job to support the premium power supply industry without criticism?

Review sites dictate useful factual data about power supplies - output, efficiency, flexibility, temperature. Anandtech has also been really good about doing power analysis recently. I think it is time we as a community begin separating power supply fact from power supply fiction.

I would like to start a discussion regarding the rational basis by which those decisions involving power supply replacement/upgrade should be made. Here are a few questions which I think would be appropriate for starting out:

1. What specific symptoms have you experienced which indicated that you were in need of a power supply replacement or upgrade?
2. What methods and information did you use to determine the appropriate replacement power supply?
3. What power supply information resources do you consume?

I am open to further questions and comments on the issue. I think it's time we debunked some of the rumors, now that we have this dedicated forum!


Useful Links provided so far:

Power Supply Myths according to PC Power and Cooling: This article is presented from the position of PC Power & Cooling and should be interpreted as marketing material. While it may contain useful information, some of the facts appear to be jaded to show PC Power & Cooling design and power supplies in an exceeding favorable light. YMMV.
http://www.pcpower.com/technology/myths/

Power Supply Need/Usage Calculators:
http://www.extreme.outervision.com/PSUEngine
http://legionhardware.com/document.php?id=635

Power Supply Review Sites:
http://www.jonnyguru.com

Consumer Experiments with Power Supply Usage:
http://www.overclockers.com/articles1452/

 

[MotF Bane]

http://www.pcpower.com/technology/myths/

 

[Phlargo]

That is a great site to get started with. They have a lot of information and bring a lot to the table. I'd like to bring that discussion here. It's too passive if it's on someone else's site - we have an incredible wealth of technical knowledge and experience on the anandtech forums - there is no reason we shouldn't try and distill a far more eloquent and comprehensive power supply discussion.

I personally enjoy the site even if it's from a commercially-driven perspective. They've got a great position - why do we need 5 small 12V rails when we can have one really big one

 

[alaricljs]

PSU Calculator

I used it for my server and current PC... good stuff considering people were going nuts over my using some cheapo 300watt PSU for a system with 8 HDDs. The calc was spot on according to my Kill-O-Watt and some basic power loss math.

 

[Aluvus]

Originally posted by: Phlargo
2. When buying a new piece of hardware, it indicates in needs X volts to run on the 12V rail. My power supply has X-1 volts on the 12 volt rail. It doesn't work, power supply isn't enough.

I assume you mean amps, not volts. A difference from nominal of 1 V on any rail would be rather unsettling.

And while we're here, let's bust another myth and note that those "needs" are just best-guess estimates. Of mixed quality.

 

[yuppiejr]

Shameless plug for the best PSU review site on the net, IMHO: www.jonnyguru.com (no affiliation, I just find it a particularly good source of information on the subject).

I'd take most of the PCP&C "myths" page with a grain of salt, their info is skewed to support the superiority of their PSU designs though the last point on 12v rail amperage is right-on. Mind you, most current gen "multi-rail" PSU's don't seem to have actual limiters on their rails and are actually single rail PSU's incognito. The bit about 80 vs 120 mm fans and "loss of wattage as heat goes up" seems to be pretty soundly contradicted by reputable PSU review sites out there that aren't trying to sell you something... (see link above...) The SLI bit is just plain nonsense and that warranty scare tactic is just marketing garbage.

Point 1 is a horrible exaggeration of value - unless you've got a SERIOUS machine (heavily overclocked C2D/AM2 w/ water cooling, SLI video cards, lots of HDD's, etc..) that's on 24 hours a day running 100% load for 3 years you are not going to see the $218 savings they talk about.

 

[Phlargo]

Originally posted by: Aluvus

Originally posted by: Phlargo
2. When buying a new piece of hardware, it indicates in needs X volts to run on the 12V rail. My power supply has X-1 volts on the 12 volt rail. It doesn't work, power supply isn't enough.

I assume you mean amps, not volts. A difference from nominal of 1 V on any rail would be rather unsettling.

And while we're here, let's bust another myth and note that those "needs" are just best-guess estimates. Of mixed quality.

Thanks.. I'll correct it in the original.

 

[Phlargo]

Originally posted by: yuppiejr
Shameless plug for the best PSU review site on the net, IMHO: www.jonnyguru.comwww.jonnyguru.com (no affiliation, I just find it a particularly good source of information on the subject).

I'd take most of the PCP&C "myths" page with a grain of salt, their info is skewed to support the superiority of their PSU designs though the last point on 12v rail amperage is right-on. Mind you, most current gen "multi-rail" PSU's don't seem to have actual limiters on their rails and are actually single rail PSU's incognito. The bit about 80 vs 120 mm fans and "loss of wattage as heat goes up" seems to be pretty soundly contradicted by reputable PSU review sites out there that aren't trying to sell you something... (see link above...) The SLI bit is just plain nonsense and that warranty scare tactic is just marketing garbage.

Point 1 is a horrible exaggeration of value - unless you've got a SERIOUS machine (heavily overclocked C2D/AM2 w/ water cooling, SLI video cards, lots of HDD's, etc..) that's on 24 hours a day running 100% load for 3 years you are not going to see the $218 savings they talk about.

I think you make an important point, Yuppiejr. PCP&C is a commercial website selling a product. There is a clear need for distinguishing the quality of their information from that of a more neutral 3rd party source.

I also agree with your substantive point - some of the myths that PCP&P espouse are dependent on very specific conditions and may not be true generally.

 

[CrystalBay]

See that's the thing, both OCZ and PCP&C are excellent at marketing. On the other hand they also have very good products.

 

[jonnyGURU]

Yeah... thanks for the link BTW....

I have issues with the whole PCP&C myths page. In fact, when I first saw this thread, I was thinking it was a "end the power supply myths put forth by PCP&C."

PCP&C's "myths" are myths as they apply to PCP&C's bottom line. Of course everything they say and do is right. They want you to buy they're product.

And the Extreme Outervision calculator, although I thoroughly respect all of the hard work behind it, gives out very inflated numbers for "power requirements." The fact that an SLI, quad core PC needs an 800W power supply when overclocking is in itself a myth.

 

[jpeyton]

Originally posted by: MotF Bane
http://www.pcpower.com/technology/myths/

Some of those 'myths' are myths in themselves.

3. DON'T LOSE POWER WITH MODULAR PLUGS

This has been proved false; power supply performance with/without modular plugs is virtually identical.

5. AN SLI CERTIFIED POWER SUPPLY WILL ALWAYS POWER HIGH-END GRAPHICS CARDS?

They quote an poorly written review from ExtremeTech as evidence that their 850W unit can power what competitor's 600W units can't. An FX-60/SLI setup doesn't use anywhere near 600W at load, and any quality competitor's model should power it fine (620HX for example, will power that without breaking a sweat).

6. SHOULD AN ATX POWER SUPPLY BE COOLED WITH A 120MM FAN?

PCP&C 'silencer' models with 80mm have always done poorly in SPCR tests compared to Seasonic/Corsair PSUs with 120mm fans. There are a lot of well-designed, 120mm fan units on the market today.

 

[CrystalBay]

Yeah, jonnyGURU's is not a marketing myth site ...

 

[Phlargo]

Well I appreciate all your feedback and I understand the controversy around the PCP&P link - I think for now I'll hang on to the PCP&P link - I say it is according to them and they do have some decent points. In the end, this is a thread about being a critical consumer (and being critical in general), so I think that the more information we are able to bring together here, the more useful it is for a potential buyer.

Glad to have you here, jonnyGURU along with your fan club You run a great site.

I think we are wise to dispel the myths about the myths! Keep up the good work

 

[BladeVenom]

Here's some interesting perspectives:
"How Much Power Do You Really Need For Your PC? Yet Another Experiment"
How Much Power Is Enough?

 

[yuppiejr]

Originally posted by: Phlargo
Well I appreciate all your feedback and I understand the controversy around the PCP&P link - I think for now I'll hang on to the PCP&P link - I say it is according to them and they do have some decent points. In the end, this is a thread about being a critical consumer (and being critical in general), so I think that the more information we are able to bring together here, the more useful it is for a potential buyer.

Glad to have you here, jonnyGURU along with your fan club You run a great site.

I think we are wise to dispel the myths about the myths! Keep up the good work

While I appreciate your "critical consumer" perspective, my point of contention with the PCP&C Myths page as a source of useful information on PSU's stands. If the link were identified as a good example of "marketing FUD disguised as useful information to sway your purchase decision to our product," I'd have no issue with the link. Out of 8 "facts", 7 were serious marketing FUD or in some cases factually wrong.

Don't get me wrong, I think PCP&C produces a superb line of power supplies and I don't think you'll ever go wrong with one of their products in your machine if you've got the money to spend. My father swears by them and won't touch a PSU by any other brand. They are a premium priced product and the marketing machine has always done what they can to "educate" the consumer as to the superiority of their design even when it requires some significant stretches of fact to get there. I would make the comparison to Bose in the A/V industry, though I'd say the difference is the PCP&C is a great product with a great marketing machine behind it while Bose is an average product that's exceptionally well marketed. Perception = reality and all that jazz...

 

[Phlargo]

Originally posted by: BladeVenom
Here's some interesting perspectives:
"How Much Power Do You Really Need For Your PC? Yet Another Experiment"
How Much Power Is Enough?

Yeah.. these are great resources - I especially found the Legion Hardware article useful. Thanks - I'll add them to primary post

 

[Phlargo]

Originally posted by: yuppiejr

Originally posted by: Phlargo
Well I appreciate all your feedback and I understand the controversy around the PCP&P link - I think for now I'll hang on to the PCP&P link - I say it is according to them and they do have some decent points. In the end, this is a thread about being a critical consumer (and being critical in general), so I think that the more information we are able to bring together here, the more useful it is for a potential buyer.

Glad to have you here, jonnyGURU along with your fan club You run a great site.

I think we are wise to dispel the myths about the myths! Keep up the good work

While I appreciate your "critical consumer" perspective, my point of contention with the PCP&C Myths page as a source of useful information on PSU's stands. If the link were identified as a good example of "marketing FUD disguised as useful information to sway your purchase decision to our product," I'd have no issue with the link. Out of 8 "facts", 7 were serious marketing FUD or in some cases factually wrong.

Don't get me wrong, I think PCP&C produces a superb line of power supplies and I don't think you'll ever go wrong with one of their products in your machine if you've got the money to spend. My father swears by them and won't touch a PSU by any other brand. They are a premium priced product and the marketing machine has always done what they can to "educate" the consumer as to the superiority of their design even when it requires some significant stretches of fact to get there. I would make the comparison to Bose in the A/V industry, though I'd say the difference is the PCP&C is a great product with a great marketing machine behind it while Bose is an average product that's exceptionally well marketed. Perception = reality and all that jazz...

I see what you're getting at. I'll note that the PCP&C article is somewhat disputed as to its position and to take its advice with a grain of salt.

 

[aatf510]

Originally posted by: alaricljs
PSU Calculator

I used it for my server and current PC... good stuff considering people were going nuts over my using some cheapo 300watt PSU for a system with 8 HDDs. The calc was spot on according to my Kill-O-Watt and some basic power loss math.

This PSU Calculator being accurate? You are kidding right?
This is one of the most misleading wattage claim out there...

 

[imported_Willie]

Originally posted by: Phlargo

Originally posted by: yuppiejr

Originally posted by: Phlargo
Well I appreciate all your feedback and I understand the controversy around the PCP&P link - I think for now I'll hang on to the PCP&P link - I say it is according to them and they do have some decent points. In the end, this is a thread about being a critical consumer (and being critical in general), so I think that the more information we are able to bring together here, the more useful it is for a potential buyer.

Glad to have you here, jonnyGURU along with your fan club You run a great site.

I think we are wise to dispel the myths about the myths! Keep up the good work

While I appreciate your "critical consumer" perspective, my point of contention with the PCP&C Myths page as a source of useful information on PSU's stands. If the link were identified as a good example of "marketing FUD disguised as useful information to sway your purchase decision to our product," I'd have no issue with the link. Out of 8 "facts", 7 were serious marketing FUD or in some cases factually wrong.

Don't get me wrong, I think PCP&C produces a superb line of power supplies and I don't think you'll ever go wrong with one of their products in your machine if you've got the money to spend. My father swears by them and won't touch a PSU by any other brand. They are a premium priced product and the marketing machine has always done what they can to "educate" the consumer as to the superiority of their design even when it requires some significant stretches of fact to get there. I would make the comparison to Bose in the A/V industry, though I'd say the difference is the PCP&C is a great product with a great marketing machine behind it while Bose is an average product that's exceptionally well marketed. Perception = reality and all that jazz...

I see what you're getting at. I'll note that the PCP&C article is somewhat disputed as to its position and to take its advice with a grain of salt.

It's not somewhat disputed......much of what is on their myth page has been shown to be completely false through testing and real world use. Corsair powered a C2D system that had two 8800GTS video cards running in SLI, along with powering a water cooling system, with one of their 620HX power supplies.....something that PCP&C says cannot be done. Jonny did conclusive testing that showed their "Modular is bad" myth is just complete BS.

The list goes on....but to think anything on that "Myths Exposed" page is anything other than fear tactics from PCP&C being used to sway one to buy their power supplies is naive, to say the least.

 

[jonnyGURU]

Originally posted by: Phlargo

Originally posted by: BladeVenom
Here's some interesting perspectives:
"How Much Power Do You Really Need For Your PC? Yet Another Experiment"
How Much Power Is Enough?

Yeah.. these are great resources - I especially found the Legion Hardware article useful. Thanks - I'll add them to primary post

Unfortunately, those editorials don't take into account the condition that PSU would be in after a year's use.

 

[Acanthus]

The biggest mistake i see is people grossly overshooting their power requirements on gaming machines.

Oftentimes you see expensive OCZ or Corsair 650w (or more) PSUs on machines with one graphics card and one or two HDs, when a 500w thermaltake for 1/2 the price will run the same machine like a champ.

 

[jonnyGURU]

Originally posted by: Acanthus
The biggest mistake i see is people grossly overshooting their power requirements on gaming machines.

Oftentimes you see expensive OCZ or Corsair 650w (or more) PSUs on machines with one graphics card and one or two HDs, when a 500w thermaltake for 1/2 the price will run the same machine like a champ.

True. There IS a happy medium. Some people can't seem to find it. There's people that tell you you only need a good 300W, while others run out and buy 1200W power supplies.

There is something to be said about using a big PSU on a little comptuer. The question came up in my forums as to why one PSU would be rated at a "higher tier" than another simply because it produced more power (i.e. Why would a Thermaltake 1200W really be "better" than a Seasonic 380W even for someone that only needed a 380W.) The following analogy seemed to set that straight and is a good argument against those who like to beat people up and say people buy big PSU's because of e-Penis status:

...regulation is relative. A PSU that's "only" 700W that has 2% regulation, but a PSU that can put out almost twice that that still has 2% regulation is going to be almost twice as good as far as regulation is concerned.

I'm going to use two fictional PSU's to make this point.

The Spectre 600W vs. the Wolf 1200W.

The Spectre 600W dropped from +12.12V to +11.88V going from no to full load. That's 2% regulation.

The Wolf 1200W dropped from +12.12V to +11.88V going from no to full load which is also 2% regulation. But the Wolf can do 90A on the +12V and the Spectre can do only 45A. So the Wolf is literally twice as good as the Spectre in terms of voltage regulation.

The Guru 1200W is kind of a crappy 1200W. It drops from +12.12V to +11.88V going from no load to 50% load, so it's just as good as the Spectre in that respect, but it drops from +11.88V to +11.64V. That's 4% regulation. The only thing that really makes it better than the Spectre is the fact that it does actually put out 1200W and it has more connectors, etc.

 

[mpilchfamily]

Well what about overall efficiency? Isn't the lower wattage unit going to offer better efficiency then the 1200W unit? Not because one has a better efficiency rating then the other but because of the overall loads put on each unit.

 

[Phlargo]

So to follow up, Jonny - how much power do we actually need? We have some power supply calculators giving a rough estimate (even if purportedly too high) but, as you pointed out earlier, even if those were correct they assume a single point in time. Do we have histogram information on power supplies expressing their efficiency and output over time?

What is the average "rate of decay" for power supplies? If I need 400 Watts for my system and I buy a power supply now that I want to last for 2 years, what kind of overshoot is necessary to account for that decay?

On another front, did anyone hear about the recent EPA regulation proposal requiring higher efficiency from power supplies (computer included) - how do you think that'll affect the PC power supply industry?

 

[Mem]

Originally posted by: Acanthus
The biggest mistake i see is people grossly overshooting their power requirements on gaming machines.

Oftentimes you see expensive OCZ or Corsair 650w (or more) PSUs on machines with one graphics card and one or two HDs, when a 500w thermaltake for 1/2 the price will run the same machine like a champ.

I always overshoot ,reason being for upgrades and expansions down the road,better safe then sorry,yes I have 700w OCZ with my single 7800GT card(however down the road when I get quad core,DX10 card etc will come in handy ) .

Side note: one thing I have learned is no PSU is perfect,I was reading about 3 guys over at OCUK that had their Corsair PSUs die,another guy had a seasonic blow up within the last few days,goes to show even the best PSUs can become faulty or go out with a bang.


PSUs to me is like ram,you always need more down the road,you can't blame a guy for planning ahead with his PSU ,especially if he can afford a quality high wattage PSU.