The Truth about PSU power ratings

PSUs: The source of DC power for all components in the PC & often a big noise source.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar

Post Reply
MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

The Truth about PSU power ratings

Post by MikeC » Thu Jan 22, 2004 1:02 am

A technical heads up for those who like to pore into PSU details.

A frustrating fact about PSUs and PSU makers is that there does not appear to be a stringent or regulated standard for reporting rated power.

There is a standard for measuring and rating HDD capacity, CPU clock speed, etc, but not one for the simple question of how much power a PSU can deliver. There are so many cases of people with "450W" PSUs having power stability issues running a system that can't possoibly draw more than 150W. And "300W" units that keep running where the "450W" units are faltering. :roll:

It's not just about bad PSUs vs better ones. It's a dumb situation caused by uncontrolled marketing competition. Real regulation would bring PSUs out of snake oil territory and into a more sensible consumer-friendly terrain.

There are many ways that PSU makers fudge to make their units look more powerful on paper.

One of these is out and out lying, it appears. You add up the power on all the lines in many PSUs and they fall short of the rated power by 10, 20 30W or even more. :?

There are more sophisticated ways:

Limit the AC input voltage to a very narrow tolerance. The best PSUs are able to deliver their rated power given a decent range of AC input power, say 90~130V for a 120V unit. It's much more demanding to produce 300W w/90VAC input than with 120VAC, so what some PSU makers will detail in their tech specs (usually not in their consumer brochures) is to specify 115-120VAC for input power. A PSU specified this way will not deliver full power if the AC voltage sags, if there is a brown-out. Surely it causes instability more often than a PSU rated to deliver full power with 90-130VAC. :!:

Specify a low operating temperature for rated output. This is quite common, but again not often seen in consumer brochures, but rather tech spec sheets provided usually only on demand by engineers or corp buyers. A typical PSU operating temp statement is somthing like this:
0ºC ~25ºC for full rating of load, decrease to zero Watts O/P at 70ºC
Examine what that says. Full power (let's say 400W) is available when the unit is at 0ºC ~25ºC. Hmmm. Think about this. :?:

Have you ever felt air blown out of a PSU in a PC running absolutely full tilt (which it would have to do to get anywhere near 400W output) that felt cool to the fingers? 25ºC airflow would feel exactly that: Cool, given that normal body temperature is 37 °C. :!:

So this PSU cannot deliver full rated power when its temperature goes over 25ºC. :shock: OK, what happens to the max power output capacity above that temp? It decreases gradually so that by the time the PSU temp reaches 70ºC, the PSU cannot deliver any power at all. So if you assume that this power drop as temp rises is linear, then max power capacity will drop by ~9W for every degree over 25ºC.

Now having examined as many PSUs as I have over the last 2~3 years, I have to say there's not a single PSU in ANY PC I have ever used or examined that would not measure at least 30~35ºC almost anywhere inside the PSU under almost any kind of load. And if/when it is pushed, 45ºC is nothing at all, especially for or near hot running components like voltage regulators.

So let's say 40ºC is a fairly typical temp inside a PSU. This 400W rated unit would actually be able to deliver a max of just 220W at that temp. Hmmm. Interesting, isn't it? At 50ºC, the available power would drop to just 130W. :shock: No wonder some PSUs have 3 fans each capable of 50 cfm!! :lol:

Here's a simple fact: Really high quality PSUs are actually rated for full power output at as high as 40~50ºC.

The trick is get a hold of the spec sheets that tell such information so you can compare apples to apples. Or ask.
Last edited by MikeC on Sun Jun 06, 2004 12:02 pm, edited 3 times in total.

Pjotor
*Lifetime Patron*
Posts: 485
Joined: Mon Jun 30, 2003 9:06 pm
Location: Uppsala, Sweden

Post by Pjotor » Thu Jan 22, 2004 7:20 am

MikeC, please sticky this and/or put it in the PSU section (outside the forum).

dasman
*Lifetime Patron*
Posts: 485
Joined: Thu Jan 08, 2004 10:59 am
Location: Erie, PA USA

Post by dasman » Thu Jan 22, 2004 7:23 am

Mike,

Very informative :shock: -- you might want to sticky this one...

Dave

halcyon
Patron of SPCR
Posts: 1115
Joined: Wed Mar 26, 2003 3:52 am
Location: EU

Post by halcyon » Thu Jan 22, 2004 10:04 am

Yes, thanks again for very good info, Mike!

JimK
Posts: 345
Joined: Sat Mar 29, 2003 6:02 pm
Location: Hudson Valley, NY, USA

Post by JimK » Thu Jan 22, 2004 10:55 am

A very good read. Thanks for making it a Stickey :)

I opened the PSU forum and it jumped out at me, "Got to read this."

Jim

Got Chow
Posts: 3
Joined: Tue Jan 27, 2004 11:41 pm
Location: Davis, CA
Contact:

Post by Got Chow » Wed Jan 28, 2004 3:03 am

FWIW, I know that PC Power & Cooling rate all their PSUs at 40*C. So a 350W PSU is really 350W or thereabouts at that operating temp. They do make really good PSUs but are probably too noisy for most that roam this forum. But for OCers, these things are extremely stable and super high quality. :) And of course, they tell you the truth when it comes to ratings.

al bundy
Posts: 667
Joined: Thu Feb 20, 2003 5:38 pm
Location: Chicago, IL

Post by al bundy » Thu Jan 29, 2004 1:19 am

Nice write-up on the topic Mike, thanks!

8)

fractal
Posts: 72
Joined: Wed Jun 04, 2003 11:20 am
Location: california
Contact:

Post by fractal » Mon Feb 09, 2004 9:50 pm

Mike

You are, of course, correct that different manufacturers rate their supplies at different temperatures. You are, of course, also correct that different manufacturers produce product that handle a wider range of input voltage while providing full output current.

Where you seem to have missed the boat, and is where 99.99% of all pc enthusiasts fail as well as at least half the pc professionals is. You try to specify the power supply needed to operate your system by "total watts" and fail to understand that you are only addressing one of about five critical ratings. Exceed any of the following and your system will not work.

1 - current rating at +5v
2 - current rating at +3.3v
3 - current rating at +12v
4 - combined 3.3/5 power
5 - combined power

I have seen 300 watt supplies rated as low as 10 amps on the +12 line and as high as 20 amps. The 300 watt supply rated at 20 amps will quite happily drive a modern intel system that derives vcore from +12v along with 4 hard drives. The 300 watt supply rated at 10 amps on the +12 line will NOT drive a modern intel system that derives vcore from +12v along with 4 hard drives, but it WILL run a modern AMD system that derives vcore from +5v along with 4 hard drives.

I understand your frustration with lies and decipt in marketing hype, but please don't fail to grok the reality. There are lies, there are damn'ed lies and there are total watt ratings. Please do not fall into the trap of trying to specify your systems by total watts. You will only be an unhappy camper with unstable systems or a poor camper with an empty wallet, or a lucky camper who shoulda bought a lotto ticket.

The only way to get the right supply for your system is to do the fairly simple analysis to determine the current requirement on each of the rails. With these numbers in hand, you derive the combined 5/3.3 power and total system power. These five numbers will let you select supplies capable of running your system. Only then can you apply other criterion such as how loud the supply operates, what color the fans are, how much you trust the rating by the manufacturer, what the operating temperature range is and other factors. Buying a fine product from a fine manufacturer with a pretty neon blue fan won't do much if it can not supply enough current on the +12v rail..

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Feb 09, 2004 10:34 pm

Fractal, it hardly matters what they tell you about the current available on the individual lines if those are all specified at 25C. If 20A current is available on the 12V line at 25C and output drops to zero at 70C, then the current capacity is dropping by .44A for each degree above 25C. So at 40C, the capacity is 13.4A.

I'd check the temp at which the power is rated before I started digging into details about individual lines, because it's going to be a much bigger factor in determining what the real delivered power is -- whether to the individual line or in aggregate.

I agree with you that in some cases, current capacity on different lines can be the difference between one PSU working well with a system and another similarly rated PSU not working well with the same system. But. to me, this means you're working too close to the rated capacity of the PSU and not giving yourself enough headroom. And in most cases, PSUs have much more current available on one voltage line if there is less demand on another. (Usually 12V vs 5V+3.3V)

Do modern system differ so dramatically in current needs? IE, if my 300W system works well with a P4 system that draws 235W total AC power, then will it not work perfectly well with my XP system that draws 235W total AC power?

This is not exactly a rhetorical question. With the ARM Systems P4-3.2 system recently reviewed still in the lab, I tried this experiment: Replace the Zalman 400W PSU with a Seasonic 300W -- just to see if it would power the thing. I've never had a system in the lab that could draw as much power. Well, the Seasonic works fine, I did not see any evidence of new instability. It did ramp its fan up fairly high, but that's to be expected. The thing I could not do is try this PSU on a 235W AC power draw AMD system. Don't have one that can draw that much.

To move on -- There's not a single "good quality" PSU that has failed because of inadequate current delivery on any system I've built, while the few "poor" quality PSUs that behaved badly did so on ALL the systems they were hooked up to.

I've used 200W SFX PSUs to power several P4 and AMD XP systems. The max AC power draw on all these systems was 130-140W, but one assumes the P4s were pulling on the 12V line much more than the 5V line, and the reverse for the XP systems (which didn't use the 12V plug). The CPU power ratings were all in the 60W range. The current ratings on the PSU were modest:

3.3 + 5V = 110W max
12V = remainder up to 200W total (ie, whatever is not used on the 3.3 +5V). So it would be 90W if 110W was used on the 3.3 + 5V, and 150W if 50W was used on the 3.3 + 5V.

I think I blather... :roll:

Anyway, I think matching current line capacities with actual current draws is not nearly as serious a concern as know what the power testing conditions are.

echoes
Posts: 31
Joined: Fri Jul 18, 2003 10:06 pm
Location: PA, USA

Post by echoes » Fri Feb 13, 2004 4:30 am

fractal wrote:but please don't fail to grok the reality.
Good one. :)

This is very good information here, thanks.

AndrewC
Posts: 174
Joined: Wed Apr 02, 2003 7:16 pm
Location: Mobile, AL

Re: The Truth about PSU power ratings

Post by AndrewC » Thu Mar 04, 2004 7:38 pm

MikeC wrote: Specify a low operating temperature for rated output. This is quite common, but again not often seen in consumer brochures, but rather tech spec sheets provided usually only on demand by engineers or corp buyers. A typical PSU operating temp statement is somthing like this:
0ºC ~25ºC for full rating of load, decrease to zero Watts O/P at 70ºC
I saw this in a Fortron Source PDF
Ambient Temperature: 0 to 25°C
Just wondering if these are two equal statements or if the first quote was specific internal operating temperature.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Mar 04, 2004 7:45 pm

AndrewC --

It's impossible to tell, isn't it? By now you shouldn't be surprised. :lol:

halcyon
Patron of SPCR
Posts: 1115
Joined: Wed Mar 26, 2003 3:52 am
Location: EU

Post by halcyon » Wed Mar 10, 2004 5:31 am

fractal,

what you are describing, is basic knowledge amongst the users of this forum and many others. There are even nice voltage line based load calculators for estimating the load per line (and calculating shared lines across +5/+3.3 circuits).

However, the temperature varience is equally important and not known by most users.

Actually, Zippy/Emacs (who apparently also manufacture innards for PC Power & Cooling) specify their maximum output loads at 40 degrees Celsius. PC Power & Cooling guarantees the maximum output up to 50 degrees Celsius.

While the aforementioned two are not examples of silent PSU makers, they are more honest in their power ratings (both in regards to ambient temp, shared line loads and voltage rail specific loads).

larrymoencurly
Posts: 170
Joined: Tue Dec 03, 2002 11:29 pm

Post by larrymoencurly » Sun Mar 21, 2004 9:51 am

Why does power capacity drop with temperature? Is it mostly because the transformer core gets hot?

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Mar 21, 2004 10:13 am

Efficiency of all electronic devices varies with temperature. Transistors, other semiconductors, caps etc. There appears to be an ideal temp/load where highest efficiency is reached for most electronics devices, but on either side of this ideal range, it falls away, and especially as you go beyond the upper limits, it falls away very steeply.

For example, AC/DC conversion efficiency in most PSUs tested at SPCR is lowest at low power and highest somewhere around 60-80% of max power. At max rated power, it is usually starting to drop.

Hifi audio amplifiers often exhibit similar behavior: At low levels (and temp) they don't sound as good as when much hotter and at higher load. Something to do with non-linearities creeping in at low levels in this case, but probably related to the same temp thing.

canthearyou
Posts: 153
Joined: Mon Oct 27, 2003 5:35 pm
Location: Arlington, Virginia
Contact:

Post by canthearyou » Thu Sep 30, 2004 11:57 am

I am planning to build an A64 system and am choosing the PSU but have my old one around too, which I might want to use until I decide on a quiet model.

I want to understand this tradeoff between the rated total power and the capacity of a particular line.

There is a lot of talk about the 12v line being insufficient in various website forums (one cited an A64 drawing exactly 12A---I don't know how this was measured) and mine is only 12A on the old 330W psu. The mainboard I have for the new system spells out minimums for at least two specific lines, one the 12v line at 8A. So you are saying that if my components do not draw more than 330W I should not be concerned about the 12V line? Is there a way calculate the draw on the 12V line by the cpu? I've read the cpu power dissipation docs but they don't answer my question.

Steve

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu Sep 30, 2004 12:04 pm

canthearyou --

12A is pretty low by today's standards. And if yours is older, then its capacity is probably lower than when it was new -- output capacity drops with age. Most CPUs/motherboards draw heavily on the 12V line these days, so I'd play it safe and go with a new higher power PSU for your A64 system. Most good newer 300W PSUs will give you ~18A on the 12V line.

canthearyou
Posts: 153
Joined: Mon Oct 27, 2003 5:35 pm
Location: Arlington, Virginia
Contact:

Post by canthearyou » Thu Sep 30, 2004 7:35 pm

You're right, I'll just have to be patient.

EDIT: After reading some more I see that the dual rail supplies are capable of meeting the needs of the A64. I gather that the A64 draws about the 8 amps my mainboard suggests as minimum. If one rail supplies the CPU and the second supplies other components there should be enough capacity for all unless the other compents are using a lot of current. A 12A rail dedicated to the CPU would be more than enough.

Steve

Post Reply