Higher power consumption in idle than load

florinp3 · Post by **florinp3** » Mon Sep 15, 2008 1:54 pm

I finally built my NAS/browsing/music/seed box. It's made of:

Noah 3988 (80W PSU + power brick) case, Jetway NC62K, AMD 4450e + fanless stock cooler, 1 TB WD GP drive (3 platter), Prodigy HD2 sound card (tight fit, the case doesn't officially support 3,5" HDD + PCI card + riser).

Everything is working fine, the problem is with measured power consumption. I'm using this device, a cheap (~30$) power meter. It seems accurate enough, I measured a 60W light bulb and it showed 57-58W.

Ok, so measuring the power consumption gives me strange figures:

0. Everything stock: 4450e @ 2.3 Ghz and 1.25V, C'n'Q disabled, AMD Driver installed:

idle (2.3Ghz @ 1.25V): 29W
playing HD movie (2.3Ghz @ 1.25V): 38-44 W

1. Everything stock: 4450e @ 2.3 Ghz and 1.25V, C'n'Q enabled, AMD Driver installed:

idle (1.0Ghz @ 1.1V because of C'n'Q): 44 W
playing HD movie (2.3Ghz @ 1.25V): 32-38 W

2. Undervolting in BIOS: 4450e @ 2.3 Ghz and 1.15V, C'n'Q enabled, AMD Driver installed:

idle (1.0Ghz @ 1.0V because of C'n'Q): 42 W
playing HD movie (2.3Ghz @ 1.15V): 27-32 W

3. Underclocking and undervolting in BIOS: 4450e @ 1.0 Ghz and 0.800, C'n'Q enabled/disabled, AMD Driver installed:

idle: 40 W
playing HD movie (1.0 Ghz @ 1.0V): 24-28 W - slideshow

* playing HD movie = Coral Reef Adventure trailer (used in SPCR reviews)

I've tried the same settings (1-3) with CrystalCPUID and I get the same results. RMClock crashes after a few seconds, but seems to cause the same behavior.
While playing around with CrystalCPUID I got at one point a normal behavior: 20-22W in idle, 25-30 during HD playback, but after a restart, using the same settings I was back to 40W in idle.

Voltages are confirmed by CPUID and AMD Hardware Monitor. All measurement are made with the chipset and RAM undervolted to the lowest values available in BIOS.

Later edit: I've sent an e-mail to Jetway technical support and got a surprisingly well documented e-mail (some much better regarded brands have failed me with standard responses - "undervolting is bad, mkay?").

I had use the watt meter to measure the NC62K power consumption .
I am sorry , I really can't see your problem in here .
The Vcore 1.1 V and 1.25V seem has not apparent different .

this is test environment and result refer to you .
Athlon64X2 4450E / A03 BIOS / NC62KR20 mainboard / ASUS DVD-RW
/ Seagate 3.5 inch 80G HDD / 400W power supply / Adata DDR2-800 512M X1

CPU Vcore measured Window Vista Idle power consumption
1.25V (3dmark2006) 1.26V 58W max (heavy loading)
1.25V (CnQ) 1V 27~28W
1.1V (3dmark2006) 1.1V 55W max (heavy loading)
1.1V (CnQ) 0.85V 27W
1.25V (S1) 1.26V 23W
1.25V (S3) 0V Under 1W
Power off 0V Under 1W

Attached our Power consumption report refer to you .

Any ideas? Could the MB be faulty?

wywywywy · Post by **wywywywy** » Wed Sep 17, 2008 12:25 am

Hi. I once had a similar problem (but I was worried about heat rather than power at that moment), and in the end I found that it was caused by indexing. So make sure indexing service is switched off. Thanks.

jaganath · Post by **jaganath** » Wed Sep 17, 2008 9:11 am

Everything is working fine, the problem is with measured power consumption. I'm using this device, a cheap (~30$) power meter. It seems accurate enough, I measured a 60W light bulb and it showed 57-58W.

This does not mean it is accurate for complex loads like a PC. A light bulb is a simple resistive load with a power factor of 1. a lot of cheap power meters have trouble with devices that have a PF of between 0 and 1, which a PC usually is.

Arvo · Post by **Arvo** » Wed Sep 17, 2008 1:30 pm

From technical specification of this device:

* Leistung: +/-5% des gemessenen Wertes, +/- 10VA
* Power +/-5% of measured value, +/- 10VA
:::
* Power Factor > 0,80 cos Phi

Apparently it doesn't measure W (active), but VA (reactive - much easier to measure digitally). Your PSU seems to have PF much less than 80% while idle, thereby you just can't measure real wattage of your system with this device.

florinp3 · Post by **florinp3** » Mon Sep 22, 2008 12:56 am

Yes, that's probably it - I noticed that with lite use I get 21-25W with a PF=0.48 - 0.65, but when it jumps to 40+W the PF goes to 1.00. So probably the power meter can't measure the power factor below a certain limit and defaults to 1.00, miscalculating the power consumption. So can I assume that my idle power consumption is around 20W?