Subject: P = V * I
Date: 10/20/2001 11:41:14 PM (GMT-7)
IP Address: 126.96.36.199
In Reply to: What, When, Where, Watts? posted by
Some high-end Fluke meters designed specifically for power system analysis can do that, even measuring current indirectly via an EMF-driven cable clamp. Then the meter outputs all kinds of neat graphs, readings, etc.
But one of those would go haywire inside a computer case, the EMF clamp is extremely sensitive and a computer generates far too much interference.
Software -- A software program might be able report any currents drawn on the motherboard (including peripheral cards and fan headers) *IF* the board has a monitoring chip capable of such a task, but the actual precision would be questionable and fans, drives, etc. tied to the power supply would not be included in the calculation. Software also could not determine the real-time current draw on its own, i.e. apart from on-board monitoring hardware.
Hardware -- Unless a kit already exists, you would probably have to custom design something around either an FPGA or a reasonably self-contained microcontroller (e.g. Motorola HC12), and a fairly complex analog support-circuit (for performing voltage and current measurements). Easily $200 in parts accompanied by a LOT of knowledge and labor. An all-analog system might be possible but it would take up half of the space inside your case
You would also have to measure, and calculate the corresponding power draw, on three different lines -- 3.3, 5, and 12VDC. The 3.3V line, of course, only goes to the mobo and you could read it there; but to get the 5 and 12V feeders before they break up into the various supply lines, you would need to go fairly deep into the power supply.
This may sound excessive but a typical computer fan, depending on size and speed, may add anywhere from 0.05-0.75(+/-)A of current draw and you will need some remarkably precise hardware in order to detect that level of change. Otherwise it would just get lost in the margin of error.
Three easier ways, both much less fun but reasonably accurate, would be as follows:
(1) Pull the power supply from the case and then load it at (for example) 25, 50, and 75% of rated via a high-impedance resisitor bank. Then make all of your necessary power measurements on BOTH sides of the supply, determine the average internal supply loss, and leave a power meter connected to the input-side of the power supply.
You probably won't get precise results because there will likely be at least several percentage points of error due to variations in the supply (i.e. the supply, being real-world and therefore non-ideal, will demonstrate a load/loss curve as opposed to a linear trend), but you can maintain a rough idea which would be at least enough to see -- for example -- how the power draw varies when you add a hard drive, a very large fan, or run a 3D task compared to light Windows work, etc.
If you wanted to get more precise, you could determine the loading at every 10% (or more) and construct the load/loss curve, then use it to correct your reading at any given setting.
(2) Fire up the system with only motherboard, RAM, soundcard, and modem in place, and make current measurements. This will give you a very rough idea what these parts draw. Then go PDF diving and determine what your CPU, video card, hard drive, and CD-ROM will draw at peak. Find the current draw and then look at your fans (which are usually labeled) and determine peak maximum power draw.
(3) Fire up the system, load it to the maximum of your ability (e.g. start RC5 and then launch an online 3D game on top of that), use a standard ammeter to measure current draw on the positive lines of ALL your supply lines (mobo 3.3/5/12, hard drive 5/12, CD-ROM 5/12) and then add/subtract any changes made by swapping fans, etc.
Possibly not the cut-and-dried answer you were after but that's what comes to mind right now. I don't want to discourage you, and who knows, maybe a kit already exists...