Cooligy - Microelectromechanical-systems (MEMS) heat exchangers

Judging by this post on the front page, I guess we can all get rid of our rads, tubes, pumps, etc, now.



This does look interesting. Even more interesting will be seeing some real world performance numbers. Near the end of the article they talk about the price being around $30.00 to start out and coming down as manufacturing picks up. ??? That seems awfully cheap for a high-tech gizmo.

Yeah, the price suprised me as well. I guess I'll postpone revamping my watercooling for now

Welcome to the golden age boyz.

Very cool article. Thanks for the link.

I hope it works well. Then we'll have really cheap and good heatsinks.

looks like it should work well.

When is this suppose to come out?

Dang...it seems to good to be true. I remeber about a year or so ago when the TEC controlled by a PCI card thing-a-ma-jig was first reported. The press hawked it as the next big/best cooling solution for the ever advancing CPU speed & temps. It turned out to be a BIG FAT FLOP! I sure hope this thing turns out to be good.

Would be nice to get rid of all the clutter in the case, not to mention the worries re: water, condensation, etc...but for $30, ti does sound to good to be true or their marketing team made some serious errors in developing their pricing scheme. Most companies when looking to launch a product into the market do research on comparitive products, breaking down each to niche markets, mainstream, etc...The price does not bode well for the performance as it appears that the company has determined that this product will compete with the general heatsink market and not the high phase change or water cooling systems. In other words I wouldn't get my hopes up. Still is a cheap alternative to fans, and would be attractive with respect to reducing noise, as the pump design appears to be near silent. Definite cool factor to the set-up also

Ahhhh... here is an interesting quote I missed when I read the story :

This microfluidic system will initially target high-performance CPUs that are used in very restricted spaces in workstations, 1U servers, and small form-factor PCs. Later versions are being planned for other types of ICs, including graphics processors, FPGAs, DSPs, and other dense ICs.

I don't think we are the target market segment just yet.

Looks good. A true commercial application of micro-channels in the true sense of the word. The main issues with true micro-channels always are clogging due to particle matter coming off various inner surfaces of the setup (this does happen over time), clogging due to bacterial growth, and super-low flow rates. I'd be interested in how these issues are addressed, if even they are.

Of interest was this quote, and perhaps very telling:

The pump has a high flow rate of more than 20 ml/minute with a 60-V/mm electric field. It's also silent. Most importantly, it's reliable since it has no moving parts. The flow rates can be scaled up to 200 to 300 ml/minute. The pumping structure has a diameter of 30 mm, is 2-mm thick, and has an effective pore diameter of 1 µm.

Click to expand...

20ml/min is just 0.333ml/sec

0.3ml/sec of water will allow the water to heat up by 1C per 1.4W of heat.

A 100W CPU would cause the water to heat up by ~70C as it flows across the CPU heat source. Given a 30C room temperature, we would now have steam unless the entire system is pressurised.

Also, not condusive to a cool CPU.

At 300ml/minute though, the situation is greatly improved, with the water being warmed by around 5C for a 100W CPU as it flows across the CPU.

This technology is very interesting, but at the sorts of flow rates they're talking about, it is not a high-performance cooling solution for overclockers. In fact, the very heatsinks they lament will still probably be cooling better.

Good for tight spaces though. It's just interesting to read the news blurbs for the information that they don't tell you.

I had come across a reference to this technology a while ago, and I don't think they are going to use water. If memory serves me correctly, I think they are going to use some other combination, and it actually is supposed to turn to steam and then cooled to liquid temps. Its an interesting concept, and will be interesting to see if it works, and how well. Even if it only cools as well as your mid level heatsink fan combo it would be worth it for the absence of noise, especially in the office environment. Still don't see how they could sell it so cheaply.

Another quote regarding the testing done and the fluid involved:

Successful demonstrations of the pump have been performed on CPUs from Intel, Apple Computer, and Hewlett-Packard using buffered de-ionized water as the liquid. The water contains buffer chemicals to prevent growths in the water that could cause blockages.

I think it is basically miniature watercooling. It might work better than watercooling as we know it.

It will still need fans. (or at least work a lot better if you use a fan)

All it does is bring the water much closer to the cpu core so that the water can take away the CPU's heat more effectively. This water still needs to be cooled somewhere else.

It just doesn't need a pump, which is nice.

Exactly how well does that "radiator" work?

no idea how efficient the radiator part of this mini watercooling thing is, but overall, the whole system could end up as good as present watercooling:

the water gets in much closer contact with the core compared to regular water cooling, (1.5mm distance instead of several mm or more). the flow rate is smaller, and the contact area could be bigger.

with this much smaller distance to the core, longer exposure time and larger area, the temperature difference between core and water (after it passes through) will be much smaller than present watercooling.

This would also mean there's a bigger difference in temperature between the water arriving at the radiator and the air cooling the radiator (compared to present watercooling). The bigger temperature difference means that the heat can be transferred more easily to the air.

All of this should mean a much more efficient cooling system than what we have now. Either you'll get much lower temperatures using the same radiator/fan, or you'll get roughly the same temperatures as present watercooling, but using a much less efficient radiator, with much less bulk and cost.

At least that's what I think after reading that document.

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

Example to illustrate what I mean (numbers probably wildly wrong, this is just an illustration of what I meant in all those words):

present watercooling:
CPU at 40C, water at 30C after cooling CPU, passes radiator cooled by 20C air, water cools down to 25C.

new system using same radiator:
CPU at 35C, water at 30C after cooling CPU, passes radiator cooled by 20C air, water cools down to 25C.

new system using much smaller radiator:
CPU at 40C, water at 35C after cooling CPU, passes radiator cooled by 20C air, water cools down to 30C.
(notice that although the radiator cools it down 5C, it is not as good as previous radiator, as the temp difference to air here is much bigger)

The numbers are likely to be all wrong, but that's what I meant in my comments.

It must be said that there are a number of waterblocks out there which already offer even less than 1.5mm of distance from the CPU heat to the water.

Allowing the water to warm up more, by slowing down flow, as it flows over the CPU is not condusive to cooling the CPU well. Really you want around at least an actual 0.5LPM flow rate (500ml/min) to not be in the situation where the rise in temperature of the water as it flows across the water-block is a significant factor in the rise of the temperature of a ~100W CPU. Really, anything at or above 2LPM is much more desirable and brings the water heating aspect into the realms of negligible significance.

It's all going to depend on if this is a two phase system. If it's liquid only then the flow speeds are pretty low, but if it is boiling the fluid then it will be able to handle the heat load. Here is a link to one of the Professor's work, which is for a two phase system.

http://www.stanford.edu/group/microheat/hex.html

I don't think they'd want the water to reach boiling temps, as that would be one hot processor.

Maybe the reason present watercooling doesn't work well with the 1.5mm distance is due to the fact that they can't get the large surface area.

Just my hunch.

EDIT:

that second line does not sound right. I should probably have said:

cooligy MIGHT be able to improve upon the efficiency compared to present waterblocks with a 1.5mm distance to the CPU, by having a much larger surface area.

Since the pipe is closed looped, the pressure inside the pipe can be lower than atmospheric pressure, so the boiling temperature can be adjusted to lower than 100 °C. Probably around 50°C might make sense, since the inside of a computer case near the CPU is allowed to be around 40°C, according to manufacturers specs (42°C in Athlon64 spec.). The specs for cooligy mention something about increasing surface area at least 20X the die size.

I guess they could even choose something with a lower boiling point, so the pressure difference doesn't have to be that great.

Pure alcohol boils at around 70C I think.