measuring watts with a multimeter - help?

i know it's not what they're for, but the xbitlabs article that tested vga card power with one piqued my interest (see here and here).

now i understand the theory - find the current by the voltage drop across the (known resistance) resistor, multiply that by the voltage coming out of the resistor and you have the power used by whatever you're testing. what i don't understand is the method used in the article above, as i want to duplicate it for my own purposes but am not smart enough to understand the choices made:

- why four 0.12 ohm resistors in parallel? why not a single 0.03 ohm resistor? was it to iron out the variance of the cheap-o resistors themselves? why didn't he just test the exact resistance of a single one?
- why did he insulate the wires into/out of the resistors? wouldn't it have been easier to test with a multimeter if they'd stay bare and wrapped?
- do i sound too inept to try this myself? i won't be testing vga cards of course, as i'm not going to solder anything onto a card (for starters). just seems like a fun way to get my feet wet.

just something i've been wondering. i'd appreciate any thoughts/flames/ramblings.

Picture. Click to enlarge.

Some basics Ohm's Law.

The formula that we/he is using for power will be P=I*E. Power= Amps * Voltage.

--------------

Circuit #1 gets amperage by finding the voltage drop of a known value resistor and using the formula I= V/R.

In circuit # 1 you will not get an accurate voltage reading from meters V5 or V12. The voltage will be thrown off by the added power consumption and "voltage drop" of the resistors. The Amperage reading also gets thrown off by rounding the value of the resistors. they are known resistance of .12Ohms, but they have a tolerance, so they could be anywhere from .114 - .126 ohms each. Then, when you do the division, you round some more to get your current.

As you can see, this example is pretty flawed. There are a lot of places to round off, and lose important decimal places. The biggest factors are voltmeters v5 and v12. They are reading lower than normal, because of the added resistors.

It appears that the circuit was done because the user didnt know how to manage an ammeter or didnt have one available.

---------------

In circuit #2, we take the amperage reading directly. No messing around with resistors or any fancy math. Measuring it this way not only gives us a more accurate amperage reading, but also eliminates a few steps. The more complicated things get, the easier it is to make a rounding error, or leave something out. Doing it this way also eliminates any added resistance of the circuit from the un-needed resistors for measuring amperage.

Now that you have your amperage and voltage readings, all you have to do is multiply them together to get your wattage (P).

It is fairly straight forward. Just multiply V1*A1 to get the power of the 5v circuit. Then multiply V2*A2 to get the power of the 12v circuit. Now add the results together to get your total power consumption.





I did this all pretty quick, so I will go over it later to see if I missed anything.

one thing i liked about the xbitlabs method was how cheap it was but if an ammeter or 3 will help my accuracy, then sure. would this or this suit my needs alright?

also, do you have any better idea for testing agp/pci-e card power than xbitlabs' insulating the card pins then soldering power lines directly to the card? i suppose i could do it the wattmeter way - get a board that supports integrated and agp/pci-e graphics simultaneously, get overall system readings when running integrated-only, then whatever differences we find with a card added are solely the card's fault. of course the motherboard chipset might draw a bit more power just because an agp/pci-e card is present, but i'm not sure how to eliminate that.

Any cheapo multimeter should do it just fine.
Something like this






As for the range on it, in Xbit's test they never went over 4amps on any rails, so you should be plenty safe with a 0-10amp meter like the ones you pictured.

As far as metering it with three meters. You dont even have to do it all at once, depending on how accurate you want to be. You can do the 12v rail first, then unplug it and meter the 5v rail.


The only thing about using a cheap multimeter is that you may lose decimal places if the display only goes to .xx or somthing. If you are doing any real work with meters, get a true RMS meter that has a decent decimal place to it. I recommend Fluke meters. Especially the 170 Series for people just starting off.



As far as the second part of your post...I dont understand what you wrote. I read it like 3 times.

my bad. in order to test the true power consumption of agp/pci-e video cards, you can't just rely on the auxillary molex power connector as the cards themselves draw power through the slot - there are multiple 12v and 3.3v pins there.

the xbitlabs author insulated the power pins on the card with some strips of insulating tape, then to power the card (since its power source was cut off) he had to solder 12v and 3.3v wires directly from the psu, through his shunt and onto the card itself - he had to figure out where on the card to solder these lines so it actually got the power it needed.

i don't want to do that so i figured you could get idle/load ac consumption figures for a configuration that used onboard graphics (i already have a kill-a-watt meter), then by adding the agp/pci-e card you're interested in (while keeping the onboard graphics enabled), the difference in idle/load ac consumption figures will be the draw of the card you added. or to remove psu efficiency issues, you could stick ammeters on the lines coming out of the psu and measure the difference between no-card and with-card configurations from there.

sorry i make no sense, i know i must be abusing terminology as i haven't dealt with electronics for quite a few years.

Ok, I think I get what you are saying.

What you have suggested is a decent approach, imho.

Take a power reading without the card plugged in, then plug it in and see how much it changes.

It is good, but not terribly accurate. It should work for all intents and purposes though, just remember to round up.

when measuring system ac power, dont forget to take psu efficiency into account .. some power is lost during conversion. what is a better approach is to measure current drawn into the mobo via atx / 12v power connector