Quote:
Originally posted by Blackeagle
Cathar,
How did you come up with a 40W max. pump size?
I'm not disagreeing with you on this pump. But I am interested in how or why this is what you feel the max size should be.
This question of max pump size is one I've wondered about for some time.
Thanks
|
It's just an arbitrary figure that I've slowly decided upon over much time playing with different pumps.
As you all know, the more powerful the pump, the more heat that is dumped into the water, whether that be through direct pump heat, or frictional heat caused by the impeller spinning trying to push the water. Either way, quite a substantial portion of the pump's heat will make it into the water.
I have dual Nippon-Denso Camry cores (22cm x 14cm x 3.2cm with 24FPI) mounted in a large box I made with a household mains-power exhaust fan pushing through them.
I also have an Iwaki MD-30RZ pump, which draws 80W.
I also run my T'Bred B at 2500MHz/1.95v, and run BurnK7 on it.
I also have an Eheim 1250 which I use for priming the Iwaki pump, and an external reservoir. This is useful because the Eheim draws 28W, and can be used to circulate the water even if the Iwaki is turned off, and it acts as a filter. Typically the Eheim 1250 is turned off.
Now under full CPU load (BurnK7) with the Iwaki pump, my water temps are climbing to 3.1C above the fan intake temperature.
If I turn the computer off and leave the fan/Iwaki running, the water temps stabilise at 1.9C above the fan intake temperature.
If I turn the Iwaki off and the Eheim on, the water temps sit at 1.2C above the fan intake. This 1.2C is due to the heat of the (submsered) Eheim pump, and the heat from the fan's motor (rated at 50W).
Now if we say that the fan + pump = 78W, which equates to a 1.2C rise in water temps, then the radiator setup has a C/W of 0.015.
If the Iwaki + fan = 130W, and equates to a 1.9C water rise, we also have a correlational C/W of 0.0146 (close enough to 0.015 given the margins of error). Note that the Iwaki pump sits in the box, so all of its 80W of energy will either be going into the water or through the radiators via the heat that radiates off its body.
That fan really should be sucking rather than blowing eh?
From this, we can roughly predict the CPU wattage being transferred into the water-cooling loop, and it works out to around 80W, which is about what I'd expect.
So what does all that mess tell us? It tells us that even given a good radiator setup that a powerful pump has a very real impact on the water temperatures, and that a very hot CPU is dumping around 80W of heat into the loop that needs to be cooled by the radiator.
So this is where I get max 40W pump recommendation from. If 40W of heat is coming from the pump, then it's going to be contributing to at least 1/3 of the total system heat. As the pump's power is increased, it can become the major heat source.
Now let's look at some block PQ curves.
The Eheim 1250 will push around 6.5lpm through my setup.
The Iwaki MD-30RZ pushes around 11.0lpm through the setup.
Looking at the flow vs C/W graph of the White Water, this corresponds to about a 0.01C/W difference in terms of waterblock performance. For an 80W CPU, that's 0.8C.
For the extra 52W that the Iwaki uses over the Eheim 1250, the CPU is 0.8C cooler, but the water is 52 * 0.015 = 0.78C warmer.
ie. the very good radiator setup is only
just able to overcome the extra pump heat. A lesser radiator setup (and I won't lie, almost every single radiator setup out there is going to cool way worse than mine) will actually result in higher CPU temps with the pump. I've gained nothing, others will lose.
Given a radiator setup that's cooling 3x worse than mine (ie. something like D-Tek Pro core with 2 x 100+ CFM 12cm fans mounted on either side), then a 80W pump is total overkill. Let's assume the D-Tek setup described will have a C/W of around 0.045, which is probably about right looking at BillA's data in the simulator discussion forum.
Let's assume a 40W pump like an Aussie spec (50Hz) Iwaki MD20-RZ. About the best 40W pump you'll get. It'll push about 9.5lpm through a WW setup like mine, for about a 0.008C/W difference, or about 0.6C difference on the CPU compared to an Eheim 1250.
Given a D-Tek pro core, the extra 12W of the Iwaki MD-20RZ will result in 0.55C warmer water, so it alone can barely give any benefit with the more powerful heat dumping MD-20RZ.
ie. for the typical computer water-cooling user, looking at anything more than a 40W pump is totally futile. Heck, even for me it's basically a no-win situation.
40W maximum, and even then, only if you're buying an MD-20RZ
A 10-20W pump is ideal. That 12V 11W "March" brand pump that was looked at here in another thread would probably be close to "ideal", as it gave a PQ curve about the same as a US spec Iwaki MD-15, but at 11W, it's extremely efficient.
So long as the pump is pushing more than 4lpm in your system and is sucking down less than 40W, you'd got the right pump for you. In fact I'd go so far as to say that the right pump for anyone is one that gives about 6lpm (1.5gpm) for the lowest wattage.
As a general rule for most any waterblock you can buy, don't bother chasing more than 6lpm unless it comes very easy, it's simply not worth it.