Power supplies drawing power when they shouldn't

[Virkash]

Hi all,
given that i want to set-up my computer for wake-on-lan, i wanted to check how much power it draws when it is shut down. So i bought this energy logger:
Avidsen 103755

Both my power supply (cooler master Silent Pro M2 850W) and my motherboard (asrock Z68 Extreme7 Gen3) are ErP/EuP compliant, so i expected something around 5W.

Instead it draws about 28W at the wall, even disconnecting every USB peripheral.

And now begins the interesting part:

I tried turning off the power supply (with its physical switch in the "0" position) and it STILL draws 20W!

So i contacted Cooler Master and i'm about to send it to RMA.

But i need my computer, so i bought a replacement that i plan to sell when i get back my main power supply.

An EVGA 500W (100-W1-0500-KR) which is also ErP compliant.

First thing when i got home has been measuring its power draw and, lo and behold, it draws 9W. When switched off and connected to nothing.

I contacted EVGA but i still have to hear from them.

How is this behaviour even possible? Isn't the switch supposed to be hard-wired to the 220V line?






P.S.
I don't think it is a problem of the energy logger, i tried with various other appliances and it always gave me a reasonable reading.

P.P.S.
Sorry for my english...

 

[coffeejunkee]

Both my power supply (cooler master Silent Pro M2 850W) and my motherboard (asrock Z68 Extreme7 Gen3) are ErP/EuP compliant, so i expected something around 5W.

EuP limit is 0,5W so 5W would already be ridiculously high. I think your powerlogger is defective or simply not very good.

I use a Voltcraft and it shows 0,0W when I power off my pc.

http://www.techpowerup.com/reviews/CoolerMaster/Silent_Pro_M2_850W/5.html

Review shows 0,45W standby use for your psu. 20W would mean a defective unit. But since the EVGA psu shows 9W which is also way too high I think defective powerlogger.

 

[SOFTengCOMPelec]

P.S.
I don't think it is a problem of the energy logger, i tried with various other appliances and it always gave me a reasonable reading.

I suspect it is giving the wrong (overall) reading, because it has been confused by the input filtering (Capacitors etc). Which are probably still connected, even if the switch (at the back of the psu), is off.

The filter is NOT really using much/any power (OVERALL). But it can be confusing your energy meter.

tl;dr
The filter can cause currents to flow, in and out of your power supply (even when off). But OVERALL, it is not really using energy (or much). Because it returns the current back into the circuit, for other things on your mains, to use.

I'm not 100% sure, but think a decent/expensive/quality power meter, would be able to differentiate between actual power consumption, and apparent current flows.

 

[Ketchup]

I suspect the capacitors are giving you these numbers. They still hold energy, even when the unit is switched off. The capacitors in power supplies are designed to hold much more power than appliances around the home (if the appliance you tested even has a capacitor).

See here:
http://superuser.com/questions/344811/safety-of-opening-a-psu
and here (warning, sad story):
http://www.kctv5.com/story/19767067/teen-electrocuted-while

 

[Virkash]

EuP limit is 0,5W so 5W would already be ridiculously high. I think your powerlogger is defective or simply not very good.

The usb port always on, and i have a bunch of peripherals connected, i expected those to draw some power

I suspect it is giving the wrong (overall) reading, because it has been confused by the input filtering (Capacitors etc). Which are probably still connected, even if the switch (at the back of the psu), is off.

The filter is NOT really using much/any power (OVERALL). But it can be confusing your energy meter.

tl;dr
The filter can cause currents to flow, in and out of your power supply (even when off). But OVERALL, it is not really using energy (or much). Because it returns the current back into the circuit, for other things on your mains, to use.

I'm not 100% sure, but think a decent/expensive/quality power meter, would be able to differentiate between actual power consumption, and apparent current flows.

So neither the power supply nor the energy logger are faulty, it is just that the latter is not good enough to distinguish between a power draw and a capacitor constantly charging and discharging?



But why are the capacitors before the power switch?

 

[Ketchup]

This is how I understand it, so someone feel free to correct if you know I am wrong:

The switch on the power supply only turns off the ability of the unit to actively accept power from the exterior plug. It does not affect the capacitors that are already holding power. A capacitor only knows how to accept the amount of power it can hold, and discharge when something is actively requesting power from it. If you have your motherboard set to WOL or USB power 24x7, your meter would still show trickle power being delivered from the capacitors for a period, if no external power is being delivered to the unit.

 

[SOFTengCOMPelec]

The usb port always on, and i have a bunch of peripherals connected, i expected those to draw some power




So neither the power supply nor the energy logger are faulty, it is just that the latter is not good enough to distinguish between a power draw and a capacitor constantly charging and discharging?



But why are the capacitors before the power switch?

I'd love to know the complete answer, myself, so if anyone can add to my answer, please do.

I did partially, know the answer, before looking it up.

But apparently, it is because the computer power supply (SMPS, = Switched Mode Power Supply), can produce huge amounts of electrical interference. This is NOT allowed, in modern EMC regulations (Electro Magnetic Compliance). The exact EMC rules, varies by country. Better regulated countries, have rules which make the components (type) BEFORE the on/off switch, reasonably fire proof/safe, even though they are before the on/off switch and/or fuse.

So if you had the mains coming in, straight into the switch (ONLY), and then wired into the power supply. It could be a significant source of EMC, because the electrical interference (that the SMPS) generates, can get to the switch, via inductive (etc) paths, and then onto the mains, and mess with other things in your home/business environment.

So the initial electrical filter (which absorbs and reduces the electrical interference, both from leaving the power supply, AND from entering the power supply), is often/sometimes BEFORE the on/off switch.

Also the initial mains filtering, is NOT necessarily made/done by the power supply manufacturer. It may be a bought in, complete module. So it would not have been easy to put the switch within its path, anyway. As it might connect, straight across the input socket, or be built into the socket, etc.

Don't forget, many/most modern power supplies, these days, DON'T even have on/off switches on the back of the computer (power supply).

EDIT:
This google books link, added to my existing knowledge about it.

https://books.google.co.uk/books?id=tpwrKfeyEV8C&lpg=PA367&ots=Sv-wFxiun8&dq=why%20are%20filter%20components%20before%20on%20off%20switch&pg=PA367#v=onepage&q=why%20are%20filter%20components%20before%20on%20off%20switch&f=false

 

[SOFTengCOMPelec]

So neither the power supply nor the energy logger are faulty, it is just that the latter is not good enough to distinguish between a power draw and a capacitor constantly charging and discharging?

Yes, that is basically it.

This is how I understand it, so someone feel free to correct if you know I am wrong:

The switch on the power supply only turns off the ability of the unit to actively accept power from the exterior plug. It does not affect the capacitors that are already holding power. A capacitor only knows how to accept the amount of power it can hold, and discharge when something is actively requesting power from it. If you have your motherboard set to WOL or USB power 24x7, your meter would still show trickle power being delivered from the capacitors for a period, if no external power is being delivered to the unit.

Hopefully the master on/off switch on the back of the power supply, will either FULLY 100% turn the entire power supply off (0 power used), or switch off (isolate), everything, except for the initial mains input filter. Which is BEFORE the power supply itself, and is basically an interference suppressor.

Your reply, seems to be talking more about the standby capability of modern power supplies, and/or the bigger capacitors, which are in the main parts of the power supply, hopefully AFTER the on/off switch. i.e. they would be off, when the on/off switch is off.

It gets confusing to describe it in words. Circuit diagrams, are so much better.

 

[pcgeek11]

If the main switch is on the power lead it shouldn't be using any power. None.

 

[Mr Evil]

It sounds like you have it set to display VA (Volt-Amps) instead of W (Watts).

 

[Red Squirrel]

Open it up and see if the switch is REALLY a hard switch, I've seen devices where it's actually a logical switch that basically tells the micro controller to shut power, so there is always some standby power usage. We have a projector at Church that does this, the reasoning is that it needs to keep the fans going if someone hits that switch to prevent killing the bulb. Not sure why a computer PSU would be this way though. It might also act as a reset if it got shorted out and went into over current protection shut down or something.

28w does seem kinda high for idle/power off but honestly I never tested any so maybe it's actually normal.

 

[SOFTengCOMPelec]

It sounds like you have it set to display VA (Volt-Amps) instead of W (Watts).

A very good idea/point.
+1.

This article, explains why (W) is what is being actually REALLY used, electrically speaking. Also the uses for the VA are also explained. It's rather technical, sorry.

 

[piasabird]

What about wake on lan sensors and sensors for for keyboard and mouse. There are probably some circuits that have power on the motherboard that power is sent through. Even though you unplug USB there is still probably something going on to sense when a USB port is plugged into unless maybe you can disable the port in the bios.

 

[pcgeek11]

If the circuit isn't complete you cannot have current flow. If there is no current flow there is no power being used. This equals zero watts.

 

[ArtForz]

+1 for "is this measuring VA and calling it W?".

Look at the IEC of a CM Silent Pro M2 850W: http://www.techpowerup.com/reviews/CoolerMaster/Silent_Pro_M2_850W/images/in_transa.jpg

That's a 1uF X cap || ? X cap (guessing 0.68uF or so, the green one) right across the input terminals.

Impedance = 1/(2*pi*f*C) -> 1.68uF has ~ 1895 Ohm impedance at 50Hz.

230V RMS / 1895R = ~ 0.121A RMS.

230V * 0.121A = ~ 27.8VA.

Suspiciously close to the measured 28 "W", isn't it?

 

[PhIlLy ChEeSe]

Hate to tell ya, but even if it's shut off the plug will still draw on the wall if it's plugged in.

 

[Yuriman]

My rig draws about that much when idle on the desktop.

 

[Virkash]

Ok, i may have arranged to make some measurements in a lab in a few days, but i fully expect that you are right, and that that power actually is not consumed. I will keep you posted.

 

[Virkash]

I did manage to check with a professional power meter, and the problem was exactly what many of you fathomed, as in, there is no problem besides my power meter calling Watts what are actually just volt-ampere. Thank you very much to all of you.

 

[SOFTengCOMPelec]

I did manage to check with a professional power meter, and the problem was exactly what many of you fathomed, as in, there is no problem besides my power meter calling Watts what are actually just volt-ampere. Thank you very much to all of you.

Great!

Thanks for letting us know.

I'm glad that many of us, were apparently proved right in the end.