[Since87]

Originally posted by Alchemy
Thanks for the advice, though I don't quite understand entirely how it makes any difference how close the junction of the two PS leads and the pelt lead are to the pelt, given the fact that the resistance of the leads shouldn't be much at all. But I'll try it.

At the high currents these pelt run at, the resistance of the wiring becomes a fairly significant issue. I've heard people tell of the insulation on the pelt leads melting with these pelts before.

My 172 Watt Pelt appears to have 20 gauge leads. 20 gauge wire will dissipate 4.2W/foot at 20A. Also, the voltage drop across the wire will be fairly substantial at 0.21V/foot. (From the Becooling picture, it looks like the 220 Watt pelt they sell has thicker wire, but I can't say for sure.)

Here is a link showing resistance at different wire gauges.

Originally posted by Alchemy
If I can get a hold of a multimeter (I really should!) I'll check the current through each PSU.

Yes, I'd definitely recommend getting ahold of a voltmeter. I wouldn't use the current measurement mode of a multimeter though. Most handheld multimeters are going to be fused at 10 Amps, and the fuse will blow if you try to measure the full 20A. As you can see from the numbers I mentioned above, the voltage drop across a known length of wire will be easily within the measurement range of any decent voltmeter. Knowing wire length, gauge and voltage drop gives you the current. (V=I*R)Also, the added resistance of the current input to the multimeter will probably be large enough to throw off any measurement of how well the two supplies share the load.

Originally posted by Alchemy
If I check the voltages and find one PSU is giving me more than another, I should be able to correct it by inserting an appropriately sized resistor, correct? And how large a discreprency would you worry about? One volt? Half a volt?

I would be seriously worried about a 0.2V difference between the two supplies if the measurements were done at 10 Amp load. I would suggest doing something like this:

Get the pelt setup with watercooling running on the hotside.

Connect both supplies to the pelt with 10 foot of 16 gauge wire between each supply and the pelt.

Connect the voltmeter between the +12V outputs of each supply.

Turn the supplies on as close to simultaneously as you can manage. Note the voltage reading and turn the supplies off quickly.

If the load is balanced between the supplies, the voltage reading will be 0V. If one supply is carrying the full load, the voltage reading will be 0.84V. You are most likely going to see something in between these two readings.

Suppose you target a maximum imbalance of 12Amps-8Amps; the voltage measured between the two supply outputs should be less than 0.17 Volts. If the voltage difference isn't much greater than that, you can probably get the load balanced adequately just by using different wire lengths between the supplies. If you need help doing this let me know.

Originally posted by Alchemy
Would it be foolish to try to wire a 120mm case fan parallel to the pelt? It would only draw off half an amp or so.

The only concern I would have here are, "What are the consequences if the power supply fails and thus the fan does also." Unless, the fan continuing to run, could protect some hardware in the event of a supply failure, I wouldn't worry about it.

Originally posted by Alchemy
Also with the PSU, to get an ATX running without a mobo I need a resistor (10 Ohms? Higher?) on the 5V line and a short on the 5V standby, correct? I remember seeing the exact instructions somewhere but I can't seem to remember them, and don't want to try anything until I'm sure.

I believe you need to short the "Power Good" line of the power supplies to ground to get them to start. It would probably be a good idea to connect both Power Good lines through the same switch so that the supplies start as simultaneously as possible.

I don't recall what you need as a load resistor, but 10 Ohms sounds about right. You may need a lower resistance but certainly not higher. Keep in mind the power dissipated in the resistor. (P=V^2/R) And definitely search around for more information. Hopefully the supplies will come with documentation telling what they need.

Originally posted by Alchemy
Finally, putting the more powerful pump on the hotside is a very good idea, but it's not feasible right now since I know I'll have most of the equipment in a wooden box nearly 3 feet below the CPU, and I really think my cheaper pump doesn't have the strength to get the water up there. I'll try to switch them out just to see, though.

It's my understanding that if it is a closed loop, the additional three feet doesn't matter much. I could be wrong though. Because you don't have a very large difference between your pumps, the temperature difference due to pump heat would probably be very small. I just wanted to toss that thought out there. Other issues very likely will be more significant than the difference in pump heat.