Page 1 of 1

How can I cut down the power automatically when PC is in S5?

Posted: Tue Jul 01, 2008 7:30 am
by wdekler
When I let my PC run a task at night (for example defragging) I always schedule a shutdown for when it finished.

The next day, if I won't use it I switch of the power completly.

I measured that my PC uses up to 30watt in S5 state (shutdown) so is there a way to automatically cut the power when shutdown has completed?

The so called standby killers for TVs don't work as you re-activate them via your remote.

The other devices which just shutdown the rest of the devices when the PC goes out aren't very helpfull because they don't use much energy in my case. And, when I tried them they reacted to line noise and started switching at the same time that my energy meter switches to a different rate. Not very healthy at all for the involved devices...


thanks!

Posted: Tue Jul 01, 2008 12:52 pm
by jessekopelman
Something is wrong here. There is no way your PC should be using 30W when off. Anything higher than 5W is extremely abnormal. Did you measure this with all peripherals disconnected?

Posted: Tue Jul 01, 2008 2:08 pm
by wdekler
I measured this with all devices connected. This includes a router, switch, water cooler pump and an AV receiver in standby. I still suspect the PC to be the main culprit here but I'm not sure how much a typical PC consumes in standby

I'll measure just the PC this weekend to make sure what is happening here.

Posted: Tue Jul 01, 2008 6:15 pm
by jhhoffma
Well, the next step would be to disconnect all the other devices. I would suggest that the router and switch will draw the most power when in standby, followed by the receiver in "standby".

It's unlikely that your PC is drawing more than those devices when OFF.

Posted: Tue Jul 01, 2008 7:09 pm
by suzyj
jessekopelman wrote:Something is wrong here. There is no way your PC should be using 30W when off. Anything higher than 5W is extremely abnormal. Did you measure this with all peripherals disconnected?
Most of the PC's I've measured are in the 10-20W range in shutdown. Even really efficient ones.

But then we have 240V AC power here, so the quiescent power loss is higher.

Posted: Tue Jul 01, 2008 7:37 pm
by suzyj
Here's something I've been putting together that will accomplish this. It went on hold while I bought a Sparkle Power SPI220LE to power the PC, but I've found that it's standby power consumption is actually worse than my old power supply, so this is back on the boil:

Image

Basically, I use a small 10W 5V switcher to supply the 5VSB power. The enable line from the motherboard to the existing power supply gets cut, and this gets inserted in the way. When the PC is in standby, the main power supply is shut off via a relay in its AC line connection.

I intend on connecting a couple of IEC outlets to the switched line, so that my audio amp and monitor get switched off when the PC is in shutdown.

The data for the switcher doesn't give idle power, but it's supposed to be 75% efficient, so at worst I figure it'll dissipate 2.5W at idle.

The whole lot goes on a PCB that's 1" wide by 3.5" long:

Image

But it could just as easily be built with Veroboard or similar.

In my application, I have the PC (12W standby), my audio power amp (10W standby), and monitor (10W standby). Hopefully this'll reduce that 32W to 2ish.

Posted: Tue Jul 01, 2008 10:48 pm
by jessekopelman
suzyj wrote:
jessekopelman wrote:Something is wrong here. There is no way your PC should be using 30W when off. Anything higher than 5W is extremely abnormal. Did you measure this with all peripherals disconnected?
Most of the PC's I've measured are in the 10-20W range in shutdown. Even really efficient ones.

But then we have 240V AC power here, so the quiescent power loss is higher.
Yes, 240V seems to make a huge difference. > 5W is very rare for 120V. Maybe 50Hz vs 60Hz also makes a difference. Anyway, as we agree, even 240V would not explain 30W standby/off.

Posted: Wed Jul 02, 2008 5:53 am
by CA_Steve
suzyj wrote:
jessekopelman wrote:Something is wrong here. There is no way your PC should be using 30W when off. Anything higher than 5W is extremely abnormal. Did you measure this with all peripherals disconnected?
Most of the PC's I've measured are in the 10-20W range in shutdown. Even really efficient ones.

But then we have 240V AC power here, so the quiescent power loss is higher.
Some comparables: My 120V system with Corsair 520 draws 10W in shutdown. The router draws 2W. Stereo receiver (off) draws 2W.

Posted: Wed Jul 02, 2008 11:22 am
by BillyBuerger
Hmmm, maybe my Kill-A-Watt is off because every PC that I've tested with it so far has been less than 5W when off on 120V. Even a beastly Prescott P4 with a crap PSU only measured 3W. My AMD 4050e, Gigabyte 780g and Seasonic SS-300SFD was the lowest measuring at just 1W. It seemed that putting these PCs into an S3 sleep state also gave very similar power draws compared to completely off. At least with the better PSUs they did.

Posted: Wed Jul 02, 2008 1:24 pm
by jessekopelman
BillyBuerger wrote:Hmmm, maybe my Kill-A-Watt is off because every PC that I've tested with it so far has been less than 5W when off on 120V. Even a beastly Prescott P4 with a crap PSU only measured 3W. My AMD 4050e, Gigabyte 780g and Seasonic SS-300SFD was the lowest measuring at just 1W. It seemed that putting these PCs into an S3 sleep state also gave very similar power draws compared to completely off. At least with the better PSUs they did.
Same here. I've tested three different systems and never gotten more than 3W at standby. Some PSU (like the Corsair HX520 apparently) do have a higher standby draw than others. S3 and Off should always be very similar as the MB should only need a Watt or two to maintain this mode. By the way, I've used my Kill-O-Watt to measure other things, like lamps, and the measurements have always lined up with expectations so I have no reason to doubt its accuracy (within a Watt or two).

Posted: Wed Jul 02, 2008 2:15 pm
by suzyj
I'm starting to doubt the accuracy of my kill-o-watt too.

The current draw from a switch-mode power supply isn't constant. From what I've read, the better ones go into a discontinuous mode of operation when they've got a light (or no) load, where they switch on for a few cycles to charge the caps, then switch back off for a while.

I wonder if the kill-o-watt (I've got an Australian equivalent) isn't confused by this behaviour, and rather than properly integrating the instantaneous power, reports either the effective maximum or something else.

In any case, I know mine doesn't do power factor on light loads properly. There's no way my PC power supply has 100% power factor on standby.

I think I'll try an experiment to calibrate my kill-o-watt. I've got an SPI220LE power supply. This model was reported here to have a standby power of 2.2W (presumably on 110V 60Hz). I'll capture the actual current and voltage waveforms using an oscilloscope (we have a nice Fluke battery powered one at work that's made for working on mains stuff, so has very good isolation), then integrate them manually to find the actual power and power factor.

Posted: Thu Jul 03, 2008 3:27 am
by wdekler
I've now measured the PC without any other device being connected to the meter and the power consumption turns out to be just 6watt (230v). There must be another device which is responsible...

edit: it seems that my EKM 265 energy meter has been misbehaving and gave me the wrong results. That seems to explain the much to high power consumption which I previously measured.

I have also disabled the WOL function but I cannot imagine that this would make up for the up for the "missing" amount of standby power (I'll measure this later though to make sure).

I'm still interested in cutting down the power when in standby as this is the green computing forum :D

That design to accomplish this looks interesting suzyj... ! Just curious: why the two invertors after each other to T7?

Also, this design introduces a (small) risk of sending 230v to my mainboard.. I don't think that I trust my electronics skills enough to build it :wink:

Posted: Thu Jul 03, 2008 4:33 am
by suzyj
wdekler wrote:That design to accomplish this looks interesting suzyj... ! Just curious: why the two invertors after each other to T7?

Also, this design introduces a (small) risk of sending 230v to my mainboard.. I don't think that I trust my electronics skills enough to build it :wink:
I was concerned that when the main supply was switched off, as its 5VSB circuit is at 0V, the pullup on it's enable input would instead pull down to ground (and trigger the enable on this board). So the two inverters just isolate the enable for the main supply so that can't happen.

Posted: Fri Jul 04, 2008 12:09 pm
by suzyj
I've verified that my Australian kill-o-watt is very inaccurate when measuring standby power.

It appears that it doesn't work properly when the power factor is lots less than 1. I took voltage and current waveforms from my SPI220LE power supply, and manually summed the instantaneous power, giving me just 0.6W shutdown. The kill-o-watt reports it as 15W, which is actually the apparent power (assuming it's got a power factor of 1).

Here's a screen-grab from my oscilloscope. The power (channel 1, 1V/A) is close to 90 degrees ahead of the voltage, giving it a power factor of ~0.

Image

Posted: Fri Jul 04, 2008 12:54 pm
by Stravos
suzyj wrote:The kill-o-watt reports it as 15W, which is actually the apparent power (assuming it's got a power factor of 1).
From the P3 Kill A Watt usage guide:
The Watt/VA Key is a toggle function key. Press the Watt/VA key once to display Watt meter, then press
key to display VA meter. The LCD will display Watts as the active power, where VA is the apparent
Power. (VA=Vrms Arms)
I think you may have confirmed that the device is doing what it was designed to do. :)