Measuring a Power supply

[ehcaes]

I was getting bored today and was thinking how people always say cheap PC power supplies never put out their rated wattage so thought I might do a dance with the devil and measure the current one of my many PC El cheapo power supplys puts out , as far as im aware you splice the live wire of the Ac cord and connect the meter probes on the same wire. Is this correct? Im getting the same reading with it off as oppopsed to when its on. I have the meter connected to ma. Figured thats right because the fuse box trips at 15ma. I followed this link as a guide http://www.twistedmods.com/article.php?artid=136 although it is american and there on 110 Vac should be the same procedure apart from the voltage here being 240 right?

[david77]

The principle is right. Try using the "10A" or "20A" range your meter offers not the mA range.
What reading do you get? It should be different when the PSU is on or off.

This measurment is highly dangerous! 240VAC can kill you!

What's more it doesn't tell you much about the max. power the supply can put out or the stability of the secondary voltages.
It only tells you the power consumption of the PSU in standby/off mode, no-load mode or with a load connected (e.g. you PC).

To get a picture of the quality of your supply you'd need to take some measurements of the secondaries under rated load.
In this case you'd need something to sink the current into, a DMM and a scope to get some idea of what is going on  with your PSU.

Take care!

[ejeffrey]

If you want to measure the power draw of your computer, get a kill-a-watt style device.  They are cheap, safe, and accurate.  Measuring the current draw will give the wrong answer due to load factor.

None of this addresses the issue of whether the supply is capable of delivering rated power.  What you will measure is the power consumption of your computer plus the losses due to the power supply inefficiency.  In order to see if the rating is honest you need a high power adjustable load (really, multiple loads for multiple voltage rails) that you can dial up to the rated current and see if the supply holds up.  You don't need to do anything at all with the mains power side unless you also want to measure the power supply efficiency vs. load.

[Zero999]

What's more it doesn't tell you much about the max. power the supply can put out or the stability of the secondary voltages.

I think that's what he wants to know.

In order to do that you need a dummy load. It's certainly true that power supply manufacturers often mislead people by specifying the maximum peak output rather than continuous output. This doesn't matter as long as one is aware of it because a PC will never draw the maximum power rating contentiously.

[alm]

Static load tests are of limited value to simulate something like a PC with very aggressive power management. If the power supply is overkill for the load, a power supply with a lower power rating would have been more efficient. Of course this would require the power supply to actually meet its specs.

[Zero999]

You're right, you need an oscilloscope to do it properly.

[kaptain_zero]

I'll second ejeffrey's suggestion of using a kill-a-watt style device.  I purchased one for myself and it has saved me far more than I ever paid for the device. Unlike most other means of measuring, it will keep track of consumption and give you the stats whenever you query the device.

It will give you amps drawn, power factor, watts consumed over time etc.  I got quite the surprise when I plugged my ancient deep freeze into it and left it on for a month. Coming back I checked and discovered that it was much more efficient than I had thought.... the power hungry culprit turned out to be my upstairs computer... an older unit, but even in sleep mode, it was consuming several hundred watts of power! It now only gets turned one when I HAVE to use it, and it gets turned back off as soon as I'm done. 

One word of advice when using a Kill-A-Watt type device... there is no storage of data.... if you pull the plug the data is gone. If there is a brief power outage... the data is gone. It's hard to read when plugged into the wall so I just added a short extension cord between the wall socket and the Kill-A-Watt meter then the device's cord plugs into the meter. This let me pick up the Kill-A-Watt and read the display without disturbing the ongoing measurements or crawling around on my hands and knees.

Regards

Christian 

[david77]

The better energy monitors or Kill-A-Watt devices often have a small battery to buffer the memory.
The one I've got doesn't loose data it continues to work even when unplugged. It is not the cheapest
one available, though.

[saturation]

I concur fully.  The basic KAW in the US is down to $15-20, but if you get the memory version, its 2x that.  You don't need to run the thing for weeks to estimate your consumption, and the likelihood of a power failure during your test is fairly small.   It has certainly paid for itself many times over and in the US, its even leased through libraries for those who only need to use it rarely.  I can estimate my electric bill to within $5, and I cut my bill over 30% by simply eliminating waste, not impacting my lifestyle at all.
 

I'll second ejeffrey's suggestion of using a kill-a-watt style device.  I purchased one for myself and it has saved me far more than I ever paid for the device. Unlike most other means of measuring, it will keep track of consumption and give you the stats whenever you query the device.

It will give you amps drawn, power factor, watts consumed over time etc.  I got quite the surprise when I plugged my ancient deep freeze into it and left it on for a month. Coming back I checked and discovered that it was much more efficient than I had thought.... the power hungry culprit turned out to be my upstairs computer... an older unit, but even in sleep mode, it was consuming several hundred watts of power! It now only gets turned one when I HAVE to use it, and it gets turned back off as soon as I'm done. 

One word of advice when using a Kill-A-Watt type device... there is no storage of data.... if you pull the plug the data is gone. If there is a brief power outage... the data is gone. It's hard to read when plugged into the wall so I just added a short extension cord between the wall socket and the Kill-A-Watt meter then the device's cord plugs into the meter. This let me pick up the Kill-A-Watt and read the display without disturbing the ongoing measurements or crawling around on my hands and knees.

Regards

Christian 

[kaptain_zero]

It looks like they have upgraded the KAW units then. I don't recall one with a memory being available when I purchased mine. And yes, estimates can be made over shorter periods but that deep freeze gets opened so inconsistently, I left it for a month to get a more accurate idea of it's yearly consumption.

Christian

[saturation]

The basic KAW, Model 4400, no memory:




The memory KAW "EZ", Model 4460:

[ArtemisGoldfish]

At work, we use a large DC load (6kW), AC power supply (these are very, very heavy, but are very useful for testing, as they let you generate 50Hz and 60Hz at multiple voltages), oscilloscopes for noise/ripple measurements, multimeters for more accurate voltage measurements, and current clamps for measuring current (in the case of the power supplies we test, this lets you check load sharing between multiple supplies.) Websites like Jonnyguru.com and hardwaresecrets.com do proper power supply testing for the most part, checking voltage regulation at multiple load levels and noise measurements. However, it'd be nice if they checked transient response, though I guess you can't have everything.

[ehcaes]

Must look into getting one of those KAWS thanks peoples, came to mind during my experimenting that I wont get max wattage when power supply not under full load good for a general idea though tested many other things during experimenting, fridge, bedside light wow those energy savers really save energy about 4 times less consumption, figure replacing all the others that dont have them Can save quiet a bit on elec bill, 4 out of 5 energy star rated fridge drew quiet a bit of current when cooler kicks in but was really suprised how little when it cut out

The principle is right. Try using the "10A" or "20A" range your meter offers not the mA range.
What reading do you get?

yep 10A range worked don't I feel like a goofball lol ,was going to measure a cheap codegen rated at 450 watts didnt bother taking it out and hooking up to a pc with some load so just measured my its got an own PC which has a antec 80 plus certified eps 1000 watt and worked it out at 120 watts when running about 5 instances of firefox with 47 background processes 25% load on a Q9550 CPU

if you want more specs please ask, Have trended all the readings I took, Will post if anyone would like to see.

[saturation]

Nice, gear like that can characterize and quantify the qualities of a PSU very nicely.  The KAW only really does watts x time, the other measurements it shows are just byproducts of its need to calculate watts.

At work, we use a large DC load (6kW), AC power supply (these are very, very heavy, but are very useful for testing, as they let you generate 50Hz and 60Hz at multiple voltages), oscilloscopes for noise/ripple measurements, multimeters for more accurate voltage measurements, and current clamps for measuring current (in the case of the power supplies we test, this lets you check load sharing between multiple supplies.) Websites like Jonnyguru.com and hardwaresecrets.com do proper power supply testing for the most part, checking voltage regulation at multiple load levels and noise measurements. However, it'd be nice if they checked transient response, though I guess you can't have everything.

[AlsInd]

wow... i have talked about my old professional pc days on three different places now! well... in those days i used to engineer our pcs by checking components with a scope. the main emphasis was given to the psu because a crappy psu will cause all kinds of freaky problems with pcs and we wanted to avoid this because we had the best warranty in the industry. i used a dummy load that i had built and as i recall dave did a blog on building a load so that should do. if you have a lot of old hard drives, you can string a bunch up and create your own load but that can get a little messy and it is difficult to work around. you will definitely need a scope if you want to check the quality of our power and it would have to be an analog scope which you can get used on ebay for not much at all. in fact just about any analog scope is good enough for this work as long as it works :-)
cheers:-)
al...