Mini Pump

[punctured]

Anybody know of any miniature water cooling attempts? I'm looking for a tiny water pump and radiator. Any help would be appreciated. Thanks ahead of time!

[BrianW]

What do u plan on cooling? A tiny heat source?

BrianW

[punctured]

Just trying something completely insane. Trying to cool a mATX desktop case barely larger than a VCR.

[koslov]

I'd look for really small aquarium pumps:
http://www.wcaquatics.com/cgi-bin/it...aqua_pump.html
http://www.sicce.co.uk/nova.htm

Finding a small pump shouldn't be a problem, but make sure it has at least 3ft of head. Check ebay.

The heat exchanger could be difficult. You'd need something with extremely low liquid pressure drop.

Something like this: http://becooling.safeshopper.com/17/49.htm?742

I don't know what the pressure drop is on the BI Micro radiators, but it's worth checking out.

You might also want to make your own, and contour it to the case interior. Just get some flexible copper tubing and solder fins to it.

For your waterblock, make sure it has a very low pressure drop, maybe just an open chamber such as the Swiftech blocks. -edit: Wrong (read below)

Overall, I doubt you could get temps better than aircooling within such a limited space. The problem is any pump that will fit will have a very poor pressure rating. And to compensate, you'll have to use a low pressure drop radiator... which likely means you'll have a huge deltaT between your coolant and the air. -edit: Wrong (read below)

But people have put WCing in those Shuttle XPCs, so it is definitely doable.

edit:
Here are some links:
http://www.overclockers.com/tips1073/
http://www.barbones.com/index.php?page=stlwc
http://www.pc-max.de/shuttle.htm
http://forums.sudhian.com/messagevie...threadid=31551

Some photos / exact dimensions of the case would be nice.

[Cathar]

Koslov, why would a "low pressure drop" block be better?

[koslov]

If he gets a pump that will actually fit, he won't have a whole lot of pump pressure to work with. If he wants to maximize radiator efficiency, he would do well to minimize overall pressure drop.

I know some restrictive blocks like WW work very well even with low flow, but you also have to consider how well radiators work at low flow.

As you can see with this chart,



thermal resistance starts to go up dramatically with flow under 2 LPM. And this is for a fairly large radiator, I bet the effect would be even more pronounced with smaller rads.

edit: img tags

edit2: I suppose the chart better shows that air flow has more of an influence on rad performance than liquid flow. But if you are forced to use a very weak pump and the flow turns out to be less than 1 LPM, you can see where problems with rad efficiency start to arise.

[Cathar]

Okay, here's where I'm coming from.

Let's assume:

An Eheim 1046 (5lpm/4' head height pump) or equivalent
2 x Black Ice Micro's in series (assume that pressure drop is about the same as a 4-pass Black Ice Classic)

and

1) White Water
or
2) MCW5000A

Let's say that at 5lpm that the pump's fittings and system tubing has a total pressure drop of 0.2mH2O, which is probably pretty close given 3/8" pump fittings and tubing.

At 5lpm, the radiators have a pressure drop of 2.0mH2O

At 5lpm, the WW has a pressure drop of about 0.8mH2O

At 5lpm, the MCW5000A has a PD of 0.25mH2O

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

So total PD for WW system at 5lpm is 3.0mH2O

Total PD for MCW5000A system is 2.45mH2O

The radiators are the dominant part of this equation.

For the WW setup we're looking at about 2.6lpm

For the MCW5000A setup we're looking at about 2.8lpm

ie. At this level any particular waterblock's resistance is not really a major factor to contend with.

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

If we stuck the radiators in parallel then at 5lpm we'd have 0.5mH2O PD through them.

Total for PD for WW system is then 1.5mH2O

Total PD for MCW5000A system is then 0.95mH2O

Predicted flow rate for WW is ~3.2lpm

For MCW5000A is ~3.5lpm

Now take a look at the cooling performance for both blocks at those flow rates and whether or not the waterblock is going have a significant impact on the flowrates to tip radiator performance "over the edge".

[koslov]

OK, that makes sense. However if you assume a heat exchanger where the pressure drop is less than that of the tubing (such as that BECooling fin and tube radiator), the flow rates would be very different. But you're right, even in that case, the difference in flow wouldn't be great enough to change the WB performance by more than .03°C/W.

Err, just noticed: Did you mean Eheim 1048, because the 1046 deadheads at 1.2 mH2O? (http://forums.procooling.com/vbb/att...=&postid=44918)

Let me understand your PD estimation:

1. "Ballpark" final flow rate for given pump

2. Calculate PD for every component at "ballparked" flow

3. Add PD for all components

4. Match PD against PQ curve of pump to get final flow rate

Is this right?

So, let me create a new case, one with a much smaller pump:

Sicce Micra modded to run inline
Tube-and-fin radiator, such as the BECooling 5x5 ( tested by Bill here: http://overclockers.com/articles481/index04.asp ).

PQ curve of the Micra:


PD graph of the BECooling 5x5:


PD graph of the WW:


Bill's 462-B,


edit: Removed calculation, so no one else has to muddle through the mis-conversions. See Cathar's post below.

[koslov]

Ah screw it, I'm calling it a night. I think if I redid the PD calcs, you would be right anyway. Thanks for correcting me.

[Cathar]

Err, just noticed: Did you mean Eheim 1048, because the 1046 deadheads at 1.2 mH2O?

I meant the 1046. 4' = 1.2m deadheaded. See my post above.

As for the rest:

You need to normalise the pressure drops at a certain flow rate that allows for a lowish margin of error first, and then map it back to the pump's PQ curve.

Remember pressure drop is proportional to the flow rate squared.

So revisiting your scenario, being the micra, be-cooling rad, and WW or MCW462B, we find that at a normalised 5lpm the pressure drops for each system is:

WW -> 0.7mH2O (rad), 0.2mH2O (tubing/pump fittings), 0.8mH2O (block) -> Total of 1.7mH2O
MCW462B -> 0.7mH2O (rad), 0.2mH2O (tubing/ pump fittings), 0.4m H2O (block) -> Total of 1.3mH2O

Now let's map that back to the Micra PQ curve.

WW system @ 2.2lpm => PD = 0.33m - and this matches up with the PQ curve for the Micra - predict 2.2lpm therefore

Modded MCW462B system @ 2.4lpm => 0.30m - and this matches up with the PQ curve for the Micra -> predict 2.4lpm

Looking at the C/W curves for the WW and the modded MCW462B at 2.2/2.4lpm respectively, and we find ~0.205 and ~0.235 respectively.

The WW is still clearly the better choice.

I think you need to check the axes on the Micra graph. They are litres per hour, and I think in a number of places you misconverted.

[koslov]

How do you figure out the intersection between the pump's PQ curve and the calculated PD vs flowrate? I understand that 1.7 : 5² = .33 : 2.2², but how did you get the 2.2?

[Cathar]

Originally posted by koslov
Ah screw it, I'm calling it a night. I think if I redid the PD calcs, you would be right anyway. Thanks for correcting me.

Hmmm, I feel like a bastard now.

The main point I wanted to make is that the pressure drop alone of a block doesn't mean that much, it's what the block is capable of doing with at that pressure drop.

The nature of centrifugal pump PQ curves is such that even fairly large differences in block pressure resistances (+100%/-50%) result in fairly small differences in the final flow rate (-20%/+20%).

[koslov]

Of course this whole argument wrt the WW is moot because

1. I doubt you'd be able to fit the Y inside a VCR-like case without seriously kinking the tubing (unless you used 90° fittings on the block... which means more PD calcs).

2. 1/2" tubing isn't really ideal for such a confined case. Swapping out barbs is easy enough though.

2. It's not really available yet!

I see now your right about the pump pressure thing... now I don't feel so good about buying a 28ft head pump for $120 USD! Hey, there's always microjetting experiments...

[Talik]

I was thinking about making a small water cooling setup like this for a movie/music box a little while ago. It got delayed due to lack of time and funding.
I'm not sure what your end goal is of this tiny system, but I presume it's something similar to my concept. If so, then your reason for going water cooling is so you can run a silent system, rather than overclocking.
Where I decided to go with my mini-system, should I pick it back up, would be to convert the top of the case into a heat plate and let the heat disperse into that so you wouldn't have to use any fans on your radiator. This method also takes up very little room(1/4" added height to your case)

[punctured]

That's actually pretty funny. I was originally making this as a media box. Bare minimum for inards... hoping there would be more room for water cooling. Of course, like you said, not thinking of this for overclocking. Also thinking that the box would get too hot with limited airflow beside the TV. I would love to cut out all fans completely... but that would be a little difficult. Still thinking.

[shroomer]

This may sound stupid, but the Thermaltake Aquarius system is pretty small. I mean, if you're already looking at lesser parts to begin with you may as well look into that.