Vampire / Standby power draw …16 W when PSU is OFF and NOTHING connected

Hi everyone,

I get accused of waffling ... so ...

Short version:

• PSU only connected to mains – NO computer components –

o 16 Watts in “OFF” mode (rocker switch at back)

o 32 Watts in “ON” mode

My understanding is that there should be < 1 Watt for a PSU that is EuP/ErP compliant.

Is this wrong or unreasonable?

How does this power draw compare with your:

• vampire (standby load)?

• Idling server (eg drives all spun down but still on)?

• Server immediately after boot?

Longer Version:

I’m currently attempting to build myself another unRAID server as my first just died (MB) … so no data is lost :- )

I decided to build a lowish power system (well it’s yet another Norco 4224 system so it’s not going to be that low) and I made sure I got all the “good stuff”. Ie Motherboard and power supply that are ErP and EuP compliant

http://en.wikipedia.org/wiki/Standby_power

“On 6 January 2010 the European Commission (EC) Regulation No 1275/2008 came into force. The regulations mandate that from 6 January 2010 "off mode" and standby power for electrical and electronic household and of?ce equipment shall not exceed 1W, "standby plus" power (providing information or status display in addition to possible reactivation function) shall not exceed 2W. Equipment must where appropriate provide off mode and/or standby mode when the equipment is connected to the mains power source. These figures are halved on 6 January 2013.[14]

“

So my relevant system components:

• Power supply

o http://www.corsair.com/enthusiast-series-modular-tx850m-80-plus-bronze-certified-750-watt-high-performance-modular-power-supply.html

o "Complies with the European Union ErP Directive (2009/125/EC) so the power consumption of the whole system, in either standby or off mode, can be less than 1W when used with a properly configured, ErP-compliant motherboard"

• Motherboard

o http://www.asrock.com/mb/overview.asp?cat=Specifications&Model=Z77%20Pro4-M

o "Certifications- FCC, CE, WHQL

- ErP/EuP Ready (ErP/EuP ready power supply is required)"

• i5 3570 22nm CPU

• 1 x Seagate 7200 RPM 3 TB drive

• 4 GB RAM

• nothing else load worthy (no other drives, cards etc)

So the system loads were measured as:

•  59-62 W idling in Windows 7

•  32W on suspend to RAM

•  32 W on OFF or Hibernate from within windows (power supply still in the "ON" position at the back of the computer).

Please just ignore the windoze bit …I got the system preassembled … and I needed something to set up the USB stick ;- ) I have also found that it’s easier to verify the power management stuff in windoze than unRAID).

That all seemed rather high for a low power system when off (I’d seen posts on other boards for similar systems idling at 13-17W.

I ended up taking the power supply out of the case. Disconnecting it totally from the PC so that it was only connected to the mains. In the “OFF” switch (rocker on the back of the PSU) it would register 16 Watts!!! Turn it on and it would register 32 Watts.

The power monitor is just a no-name brand but it can measure my phone charger as 2-3 Watts and my tablet charger as about 9-10 watts – so it can measure low values.

So am I missing something? Is there some “get-out-of-jail-free” clause that I am missing with the EuP specification … as 16 >> 1 as is 32.

How does this power supply compare with your vampire load? Sleeping load?

The power prices have recently increased significantly in my area … so I need to keep this as low as is possible.

Any thoughts?

Use a power strip with an off-on switch.

I should have added that the servers are located in the garage.

They all run off the one power board/strip already.

They need to be remotely administerable ... Ie WOL as a minimum ...or IPMI.

So turning off at the powerpoint is not an option.

Thanks for the suggestion though :-)

Something is wrong, either with the PSU or with the measurement.  Clearly, if the power draw was 16W or 32W there would be some detectable amount of heat from the power supply.  Just to be sure - are you measuring Watts or VA?  They can be significantly different, especially with large supplies at very light loading (although supplies with power factor correction should generally take care of that).

If the switch at the back of the PSU is off, then that should be disconnecting the mains feed into the PSU and so the measured consumption should be very close to zero.  A PC that is powered off (not suspended) and that is ErP compliant should be in the sub 1 watt range even if the mains switch on the PSU is on.  Not all home built PCs will necessarily meet that figure but you should still only see 2 or 3 watts. 

Note that having Wake-On-Lan enabled in the BIOS will often break ErP compliance (BIOS settings generally point that out) since the LAN hardware needs power and some other parts as well to allow a power up in response to a WOL packet.  This can take consumption into the 3 - 10 watt region depending on the hardware specifics.

Something is wrong, either with the PSU or with the measurement.  Clearly, if the power draw was 16W or 32W there would be some detectable amount of heat from the power supply.  Just to be sure - are you measuring Watts or VA?  They can be significantly different, especially with large supplies at very light loading (although supplies with power factor correction should generally take care of that).

 

It is measuring Watts.

If the switch at the back of the PSU is off, then that should be disconnecting the mains feed into the PSU and so the measured consumption should be very close to zero.  A PC that is powered off (not suspended) and that is ErP compliant should be in the sub 1 watt range even if the mains switch on the PSU is on.  Not all home built PCs will necessarily meet that figure but you should still only see 2 or 3 watts. 

 

Note that having Wake-On-Lan enabled in the BIOS will often break ErP compliance (BIOS settings generally point that out) since the LAN hardware needs power and some other parts as well to allow a power up in response to a WOL packet.  This can take consumption into the 3 - 10 watt region depending on the hardware specifics.

I am happy for 3-10W with WOL enabled ... which is what I was expecting ... and that was why I was so annoyed when I saw it was drawing Waaay more than that with nothing attached and doing nothing... and why i at first did the power measurements at the various stages of power saving mode in Windoze (sleep .. .hibernate and actual off)

I have been given approval from the store to return the power supplies ... but I just wanted to make sure that I wasn't being unreasonable. I also tried to buy another power supply from a different store and it had the same issue ...admittedly I did want to try a seasonic but they didn't have any in stock so I tried an 850W version of the Corsair TX series that I already had ... in the vain hope it was just a bad batch ... and it had the same issue.

The second store said they were happy to test 1 power supply for me ... but not "keep opening boxes until you find one you like". Hence the post here to see if I was being unreasonable and to see what experiences folks around here had as I have seen quite a few people posting about wanting "low power" systems and the like.

Maybe I'm not following, when you have the power supply connected and the switch 'off' and it reads the 16W, what does it show when you pull the input plug from the power supply?

Maybe I'm not following, when you have the power supply connected and the switch 'off' and it reads the 16W, what does it show when you pull the input plug from the power supply?

Perhaps if I explain it like this. You go to a computer shop and buy a power supply. You take it home. You open the box. Plug the power monitor into the power point. When the power switch on the wall is OFF the monitor reads 0 watts ... As you would expect. You then plug in the standard black power cable into the back of the power supple and connect it to the power point on the wall (going through the monitor).

In this state with the wall switch off it still reads 0.

If I turn on the wall switch to ON but leave the little black rocker switch on the back of the power supply in the OFF position it registers 16W. It should be 0 as th appliance is NOT switched on.

If I then turn on the power supply with the little rocker switch at the back it now registers 32 watts on the monitor despite the fact that the PSUis not actually doing anything or connected to anything to supply power to.

A few watts would be what I would expect if it was connected to a motherboard etc with no WOL etc ... But if its already at 32 watts even when the PC is OFF ... Ie you can use the power on switch at the front of the case to power it up ... It's never going to get below that ... And over the course of a day that's about 1kW a day in just vampire or do nothing mode before I have added all the rest of the drives etc.

Vampire / standby power is a major concern of governments and power companies as it just sucks power for no purpose and runs up your power bill.

So, where is the heat?  Of you are seeing those figures there should be detectable warmth from the supply after a period of time.

Since you say that your power meter reads OK with other devices I am beginning to suspect a faulty outlet.  Have you tried plugging in somewhere else?

Using Occam's Shaving Device, since you're getting the same reading with two power supplies, you've either got a problem with your power meter, or there's something wrong with the motherboard - what happens if you unplug the power connections to the motherboard?

I have tried several different power points in my house ... I have also tried at another location entirely (the other computer store)... All with the same result.

I haven't got the server plugged in currently as I was packing up the PSU for shipping so I haven't noted the temperature ... In my defence it's rather hot at the moment (40 C or 105 F) so I doubt I would actually notice anything much.

I have tried 3 different corsair tx m PSUs.  ... 2x 750 and 1 x 850. All show the same high reading.

I have used this meter to measure lower values of 2-3 W for phone charger on and 0 when off... I have also used it on a Mac mini idling at 13W,28-30W running plex 1080P content....and then 0 when off.

So I'm reasonable certain the power monitor works AOK.

The 16W and 32W IS for when the powersupply is NOT connected to the motherboard.

59-62W is when it's plugged into motherboard and win doze is idling.

32 W when suspended to RAM.

Does anyone have any actual power measurements of their own systems running? Idling? Sleeping? Off?

Used my P4400 "Kill A Watt" meter on my unRaid server and took the following measurements:

unRaid idle (cores @ 800MHz), mechanical drives spun down: 44W

Spin x6 mechanic drives up: 155W for approx 2 seconds

unRaid idle, x6 mechanical drives spun up: 82W

Computer off, power supply switch on: 1W

Power supply switch off: 0W

Black power cord unplugged from meter: 0W

120V, 60Hz

Unraid 5.0-rc8a - Asus M5A78L-MLX Plus - AMD Athlon II X3 450 Rana 3.2GHz - 8GB DDR3 - Antec NEO ECO 620W - Antec Three Hundred Case - 1x Rosewill RC-211 - Parity: 1T Seagate ST1000DM005/HD103SJ - DATA: 3x WD Black 750G - 1x 1T Seagate ST1000DM003 - 1x 500G Seagate ST500DM002, Cache: Intel X25-V SSD 40GB

I have tried several different power points in my house ... I have also tried at another location entirely (the other computer store)... All with the same result.

 

I haven't got the server plugged in currently as I was packing up the PSU for shipping so I haven't noted the temperature ... In my defence it's rather hot at the moment (40 C or 105 F) so I doubt I would actually notice anything much.

 

I have tried 3 different corsair tx m PSUs.  ... 2x 750 and 1 x 850. All show the same high reading.

I have used this meter to measure lower values of 2-3 W for phone charger on and 0 when off... I have also used it on a Mac mini idling at 13W,28-30W running plex 1080P content....and then 0 when off.

 

So I'm reasonable certain the power monitor works AOK.

 

The 16W and 32W IS for when the powersupply is NOT connected to the motherboard.

59-62W is when it's plugged into motherboard and win doze is idling.

32 W when suspended to RAM.

 

Does anyone have any actual power measurements of their own systems running? Idling? Sleeping? Off?

In a previous post I asked if you were measuring watts or VA.  I have now remeasured my system and I am pretty sure that you have been measuring VA.  (I know you said watts, but I think that either you or your meter is mistaken.)

My power meter this evening reports the following...

• Mains voltage 244V (a bit higher than normal for me, but makes little difference to the other readings).
  
• Mains frequency 49.9Hz (very normal)
  
• Meter only, no cable - 2W;  9VA  (clearly a measurement error)
  
• Meter with cable, no PSU - 2W; 9VA  (still clearly a measurement error)
  
• Hard off at PSU switch - 1W; 15VA (this caused by some out of phase current through mains filter components)
  
• PSU switch on, mother board not powered up (no Wake-On-Lan) - 1W; 25VA (out of phase current through filters and very poor power factor near zero power)
  
• Server spinning up all 8 green drives - 118W; 135VA (note power factor much better at normal power levels, so VA only a bit more than watts)
  
• Server running (idle) but all 8 green drives spinning - 75W; 89VA
  
• Server running (idle) with all 8 green drives stopped - 39W, 58VA
  

Your original figures of 16 and 32 are very similar to my VA figures when looking at the power supply with either no load or minimal load, but it is important to note that they are VA figures and not power consumed in watts.  The electricity company won't like a poor power factor loading but it is really only at very low power levels and therefore not that significant.  You will only be paying for the actual watt-hour figure as measured by your utility meter.

 

In a previous post I asked if you were measuring watts or VA.  I have now remeasured my system and I am pretty sure that you have been measuring VA.  (I know you said watts, but I think that either you or your meter is mistaken.)

 

My power meter this evening reports the following...

• Mains voltage 244V (a bit higher than normal for me, but makes little difference to the other readings).
  
• Mains frequency 49.9Hz (very normal)
  
• Meter only, no cable - 2W;  9VA  (clearly a measurement error)
  
• Meter with cable, no PSU - 2W; 9VA  (still clearly a measurement error)
  
• Hard off at PSU switch - 1W; 15VA (this caused by some out of phase current through mains filter components)
  
• PSU switch on, mother board not powered up (no Wake-On-Lan) - 1W; 25VA (out of phase current through filters and very poor power factor near zero power)
  
• Server spinning up all 8 green drives - 118W; 135VA (note power factor much better at normal power levels, so VA only a bit more than watts)
  
• Server running (idle) but all 8 green drives spinning - 75W; 89VA
  
• Server running (idle) with all 8 green drives stopped - 39W, 58VA
  

Your original figures of 16 and 32 are very similar to my VA figures when looking at the power supply with either no load or minimal load, but it is important to note that they are VA figures and not power consumed in watts.  The electricity company won't like a poor power factor loading but it is really only at very low power levels and therefore not that significant.  You will only be paying for the actual watt-hour figure as measured by your utility meter.

Thanks very much for your detailed response - I appreciate the effort it would have taken to obtain all of the above readings :- )

Please excuse my ignorance in this matter as I am not an expert (or even a novice) in this area - i write and build software for a living but not hardware. I can only report what my monitor is telling me .. and it is displaying the units of Watts. It is only a small low cost monitor and not anything fancy. I wanted to get something like a "Kill a Watt" meter ... but I could only find them for 110V and not 240V systems :-( . This is what I have http://www.jaycar.com.au/productView.asp?ID=MS6115&form=CAT2&SUBCATID=1003#3

Your figures give me pause and I'd like to know what sort of tool(s) are necessary to determine the distinct VA and Watt figures. Any suggestion as to where I can buy (online) an appropriate measurement device would be appreciated :- ) I was not even aware that it was possible to get a distinct value for VA vs Watts. I had always thought power was merely Volts * Amps (ie VA). I was not aware of the complexity for Power Correction Factor.

I had a look at the stated specs for my power supply here http://www.corsair.com/power-supply-units/tx-series-power-supply-units/enthusiast-series-modular-tx850m-80-plus-bronze-certified-750-watt-high-performance-modular-power-supply.html and it states

"

Active Power Factor Correction (PFC) with PF value of 0.99

"

I am acutely aware that these numbers should be taken with the proverbial grain of salt ... as the power supplies have varying efficiencies dependent upon the load on the system (usually much better from the halfway max load and up and worst at the lowest power levels).

What I don't really understand is the distinction between VA and power and the power correction factor. I did some quick googling and I couldn't find anything that really explained it well. I saw stuff from UPS vendors and others saying about capacitive normalised power supplies vs the ones in computer PSUs.

I'm just trying to build a system that is as power efficient as is possible that meets my needs (lotsa drives with plenty of CPU ooomph for when I need it) ... but if it's doing nothing I want nothing (or as close to nothing as is reasonable) to be reflected in my electric bill :- )

In your situation ... where you state

PSU switch on, mother board not powered up (no Wake-On-Lan) - 1W; 25VA (out of phase current through filters and very poor power factor near zero power)

If you left it like this for 1 day ...

What would the power company bill you on?

- 1W x 24 hr - 24 Whr

- or 25 VA (pseudo Watts) x 24  - 600Whr

If you left it like this for 1 day ...

 

What would the power company bill you on?

- 1W x 24 hr - 24 Whr

- or 25 VA (pseudo Watts) x 24  - 600Whr

The meter that I have is one of these...  http://www.maplin.co.uk/plug-in-mains-power-and-energy-monitor-38343

It allows you to read watts, VA, and power factor amongst other things.

Last time I checked, domestic utility meters measure watts (not VA) so power factor does not affect your bills (in the question asked, you would pay for the 1W, not the 25VA).  For large industrial users, power factor can be taken into account since it makes a large difference to the efficiency of electricity generation and transmission.    There is an environmental push towards devices with power factor correction for the same reasons.  It is more energy efficient and has cost benefits for those users who are measured for VA-hours rather than watt-hours of energy usage.

For a full discourse on power factor see here...  http://en.wikipedia.org/wiki/Power_factor

If you really want to give yourself something to worry about, you can always get one of those meters that allows you to see the total power draw of your house - then try figuring out where it's all going.  Things like this... http://www.theowl.com/products/owlmicro.php

It's not power correction factor, it's simply power factor. Power factor is simply the ratio of VA to W.

Current can flow into devices like inductors and capacitors. However, these devices don't actually use the current. The current just circulates back and forth between these devices and the utility. So, there is no real power being used meaning these devices are causing a poor power factor because VA is drawn yet little W are drawn. Most households are only charged for power used (watts x hours) so the power factor really doesn't matter.

Most power supplies have some form of line filtering on the incoming power which can consist of capacitors and inductors. So, they can show a VA draw even in the off state.

This is a very interesting discussion folks, enjoy reading it.

...just  a sidenote to this...

If you don't know what portable metering device to be trusted or where to get one...many Utility companies offer "Kill-A-Watts" or similar to "rent" (even for free, if you are a customer).

This is part of their energy efficiency programs they are running with household customers.

Maybe you have a local shop/store of your Utility nearby?

Just for the sake of completeness.

I have decided to purchase one of these meters http://www.ccisa.com.au/shop/index.php?main_page=product_info&cPath=4&products_id=1&zenid=77oakb2surmd6sosejna1g2410

It's a little more expensive but some of the other sites actually use this one as the reference point.

I got this from this page http://steplight.com.au/education/monitor-energy/plug-in-electricity-power-meter/ which has a nice little cheap meter.

But the interesting thing on this page was that it specifically listed the meter I was using as being notorious for being inaccurate. It wasn't the only site that suggested that this was an inaccurate meter (especially for standby loads).

I also had a look at this choice article http://www.choice.com.au/reviews-and-tests/household/energy-and-water/saving-energy/power-meters-review-and-compare/page.aspx and it specifically listed the meter that I decided upon as the second most accurate (the most accurate was another $100 more than this one).

The only issue is that the store is closed until 14th January. Cest la vie.

My thanks to all of you for your patience and education efforts :- )

belorion, did you tried with the new meter? What were the results? I have the same issue with my OCZ ZT550W PSU, which also show the same measures with regular watt-meter: 16W when PSU is OFF using the switch on the back, and 32W when that switch is ON (without anything connected to PSU rails).

Besides, I checked several PSU (like Chieftec GPS-400 AA, or Corsair HX650, and some models whatever I connected) using the same meter and none of them show any draw when switch is OFF, and only several watts, when switch is ON...

As per the posts above - I would strongly suspect that your meter is reading VA rather than watts.  This could account for the figures that you are seeing.  Try to find a meter that allows you to choose between VA and Watt readings to see the difference.