Power Consumption figures

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Steerpike
Posts: 128
Joined: Sun Aug 11, 2002 3:26 pm
Location: Walnut Creek, CA

Power Consumption figures

Post by Steerpike » Sat Nov 23, 2002 9:36 pm

I just bought a 'Kill A Watt' power meter by P3 technologies for about $50 - followed a link I found in this site. So now I know the 'true' consumption of my systems ...

So far, did the following measurements:
Philips 15" LCD flat screen - no signal - 1 watt
normal use - 15 watt

NEC Multisynch 17" normal use - 60 watt

Pentium III 660, plugged in but power off - 2 watt
on, stable - 42 watt
on, range - 35-55 watt
(1 ATI Rage, 1 Maxtor HD, 2 NICs, generic PSU, etc)

Pentium IV 1600, plugged in power off - 3 watt
on, stable - 80 watt
burning CD - 90 watt
encoding video stream - 115 watt
(1 ATI Rage, 1 ATI A.I.Wonder, 2 NICs, 1 Baraccuda HD, Zalman 300 PSU, etc).

It's tedious having to shut everything down but fun to measure stuff!

MikeC
Site Admin
Posts: 12283
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sat Nov 23, 2002 10:00 pm

It is a fun device, isn't it? 8)

Just remember that for the PCs, what you are measuring is the AC power draw. To get what the PSU is really delivering, multiply those figures by the rated efficiency. If you assume 70% efficiency...

55W AC max for your Pentium III 660: 38.5W delivered DC power

115W AC for your Pentium IV 1600: 80.5W delivered DC power

Now you see why I keep insisting that high power PSUs are more about effective marketing (and exaggerated specs) than real need.

TerryW
Posts: 85
Joined: Sun Aug 11, 2002 3:26 pm
Location: Coquitlam, B.C. Canada

Post by TerryW » Sun Nov 24, 2002 11:13 am

Aw, MikeC. :shock:

I agree there is a lot of hyp WRT hi-power supplies and how they are actually used. Yes it could be a marketing strategy. More Power (as Tim Allen says).

One idea comes to mind though: for electrical approvals like CSA / ULc, etc., manufacturers need to define the maximum wattage the product will consume. The cabinets that my company produces are stamped with the voltage, frequency and wattage (i.e. 7,800W). The wattage is the worstcase scenario and most installations will never use that much.

But don't you think a PSU has a max wattage so users (techo-junkies in the know that is) won't go over that limit? As a techy myself, I would want a PSU with a higher capacity than actually needed.

Kinda like a battery for your car: higher amp-hours / cranking current should have a higher storage capacity and/or longer life. I always buy batteries for my vehicles with a higher capacity than actually needed.

Be kind in your comments. :roll:

Cheerio!

TerryW

MikeC
Site Admin
Posts: 12283
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun Nov 24, 2002 11:58 am

Be kind in your comments.
How am I unkind by pointing out the obvious? cschofie's systems draw at most 40% of the rated power of the Zalman 300W. OK, we need headroom, I agree, but 60% seems like a bit of overkill, don't you think?

I am not suggesting that all 300W PSUs are the same. They obviously are not. Even with only 40% of the rated power being drawn, if most of it is on one of the voltage rails, the PSU could still be working hard. (Although that scenario is pretty difficult to imagine in this case: the Zalman's 3.3V line max power is ~100W; 5.5V is 150W; 3.3+5.5 combined is 180W; and 12V is 216W -- ALL these numbers are considerably above the 80.5W total draw of cschofie's most power-hungry system)

What I AM saying is that PSU requirements are generally overestimated. If you do run into a PSU-related instability despite being run at much lower than rated power, my bet is that one of the voltage rails is close to max, the PSU specis are exaggerated (ie it is NOT a "good" PSU), or the PSU has some kind of defect.

Steerpike
Posts: 128
Joined: Sun Aug 11, 2002 3:26 pm
Location: Walnut Creek, CA

Post by Steerpike » Sun Nov 24, 2002 10:21 pm

Probably should post this in the PSU forum, but I just replaced the stock fan in the Zalman 300 PSU with a Panaflo 12L, and ran the cable outside the case so I could control it with a zalman fanmate controller. While I was in there, I also hot-melt glued a thermistor to the big heatsink (next to the built-in thermistor) and ran a cable from it outside too. Hooked that to my digital doc 5 temp/fan controller. So with the fan on slow (but above minimum), the thermistor is reading 37 C. A second thermistor is measuring the air temp right at the exhaust point of the PSU. It's reading 34 C. My CPU (as indicated by 'hardware sensor monitor' prog) is 36.5, and the mobo is 35. Dtemp is recording the disk temp at 35. Ambient is 23.4 (ambient under the desk, where the PC is).

Summary:
PSU heatsink: 37
PSU exhaust: 34
CPU: 37
Mobo: 35
Dtemp (HD): 35
ambient: 23.4

Dropping the PSU fan to abs. minimum, (inaudible!)
PSU heatsink: 48
PSU exhaust 40
CPU 40
mobo: 38
Dtemp (HD): 38

This may be too hot for the PSU, long term?

I cranked up the power a touch on the PSU fan, to the 'threshhold' where the noise seems to really rise (unfortunately, the panaflo doesn't have an rpm monitor so I can't say what the speed is. Maybe I can now proceed to experiment with heatsinking the PSU!

I'll post these results over in the PSU forum.

gregzeng
Posts: 21
Joined: Tue Dec 03, 2002 7:23 pm
Location: Australian Capital Territory

Post by gregzeng » Fri Dec 06, 2002 4:50 pm

MikeC wrote:
How am I unkind by pointing out the obvious? cschofie's systems draw at most 40% of the rated power of the Zalman 300W. OK, we need headroom, I agree, but 60% seems like a bit of overkill, don't you think?
...
What I AM saying is that PSU requirements are generally overestimated. If you do run into a PSU-related instability despite being run at much lower than rated power, my bet is that one of the voltage rails is close to max, the PSU specis are exaggerated (ie it is NOT a "good" PSU), or the PSU has some kind of defect.
I'm surprised that there is not more discussion on this.
The PSU must assume that all Firewire & USB ports are loaded to their full capacity, with full capacity power-drain, that all possible HARD DISK DRIVES & CD-ROM/ DVD ROM drives are attached,

AND

That all of the above units will be operating as soon as the switch is turn on. The Peak SWITCH-ON power is much greater than the steady-running power consumption.

The manufacturer of the PSU cannot assume that the mains power supply is steady in voltage, nor that all components are safely within the design specifications. I think a 100 % oversupply sounds reasonable to me.
Retired: PC Support (h/w, s/w), Ex-company-director (non-profit sector)

MikeC
Site Admin
Posts: 12283
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Fri Dec 06, 2002 6:16 pm

The PSU must assume that all Firewire & USB ports are loaded to their full capacity, with full capacity power-drain, that all possible HARD DISK DRIVES & CD-ROM/ DVD ROM drives are attached,
I don't know about the MUST. I don't believe makers of PSUs assume the above. They all state categorically that the individual line current ratings not be exceeded and that the max total for the PSU also not be exceed. There is also a finite number of output connectors for PSUs, the smaller ones having fewer and the larger ones having more. This tends to naturally limit the number of components in accordance with power rating.

AMD & Intel have design guides for builders that provide typical power dissipation of components. AMD's approach is -- CPU power + 80% of the total of the other components = the minimum power supply size. For their "High Power System", this came to 242W. For a typical system, they calculated 163W.

But if you think about this, it is really hard for the Windows OS (don't know about others) to utilize 80% of all components simultaneously. If you can show how this can be done, I'd be interested.
And... That all of the above units will be operating as soon as the switch is turn on.
Maybe, maybe not. I don't know how they would all be working upon turn on. Even if they were, if the various line ratings are not exceeded, then it should not be an issue as long as the PSU actually delivers that power.
The Peak SWITCH-ON power is much greater than the steady-running power consumption.
This turn-on headroom is already built in to just about every PSU you can find. In-rush current peak ratings of 60-80A capacity (obviously transient) are not uncommon for PSUs rated even at just 150W. Many also boast over power protection for up to 50% or more.

Steerpike
Posts: 128
Joined: Sun Aug 11, 2002 3:26 pm
Location: Walnut Creek, CA

Post by Steerpike » Sat Dec 07, 2002 9:11 am

Isn't the bigger issue, for us 'silent' types, the fact that the larger the PSU, the larger the heatsinks, the bigger the ratings on the transistors, etc, and thus, can more safely run with reduced airflow.

I guess this goes two ways; if silence is your goal, and you are not interested in 'modding' the PSU, then you should buy a PSU that is as low-powered as you can get away with, since the lower the power, the less fan noise should be generated since the components are not going to get as hot; but if you ARE willing to mod, then you should buy the largest PSU available, and change the fan (and reduce it's voltage).

But, in reality, the noise is going to be governed by the quality of the components, not on raw power ... so all the above is somewhat moot!

As Mike and I concluded in the power supply forum, the Zalman PSU is best served by KEEPING the stock fan, but putting a 25 ohm resistor in series with the fan. The stock fan is quieter than the panaflo at lowered voltage.

MikeC
Site Admin
Posts: 12283
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sat Dec 07, 2002 9:23 am

cschofie,

Your comments bring us back to the noise track, I admit I was wandering off there on my rants about PSU marketing misinformation. :roll: It's a weakness.

Actually, from a quiet POV, a higher power PSU is usually preferable, because you CAN slow the fans down & know that the PSU components can handle the heat better than one which is rated for less total power. This would hold true whether you're modding or not, because, for exmaple, between the Seasonic 300 and 400, there is no difference in airflow, but the compoents in the 400 are rated higher...

Post Reply