how effective is direct metal-to-metal contact for heat conduction? 6

[knowlittle]

I need to cool a 200 mm square aluminum plate, 10 mm thick to 25 deg C. Test duration is 24 hours. This plate is getting constant radiation heat from underneath, about 20W. What I plan to do is to fabricate a cold plate and place it on the subject plate. The cold plate will have copper tubing embedded on the surface, and water will flow through the copper. I will flip it over and place it on the subject plate. Now the subject plate will directly contact copper tubing with water flowing inside.

Is there an easy way to improve metal to metal contact?
Thanks.

 

[mcgyvr]

thermal compound/pads are used to improve metal/metal contact by filling in the inclusions/scratches,etc..

 

[DRWeig]

Thermally conductive compound. It's kinda like grease but usually filled with aluminum. Fills in all the small gaps that will surely be there. It's the gooey stuff you'll find down in thermowells and on top of modern CPUs in personal computers and laptops.

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[DRWeig]

Well, mcgyvr typed faster than I did, so here's another hint:

Put the following phrase in Google -

Thermally conductive compound

The whole first page of hits seems to be manufacturers like Dow Corning. You could find data sheets and instructions that will give you the specs and best performance.

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[renderu]

How thick is the plate with copper and what material? How much does the copper tubing protrude on the surface?

And out of curiousity what is the abiant air temperature? And is the hot plate losing heat from the underside also through convestion?

 

[IRstuff]

You might want to look at mechanically lapping the interfaces, which would increase the contact area. Thermal compounds are not that terribly thermally conductive, so a clean contact area and some amount of compression would probably do better and not have thermal compound getting smeared on everything.


TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[renderu]

This could be just me not being too familiar with cold plates. But I don't think cold plates are meant to have such an imperfect face put against the object surface we want to cool. Sure, thermal compound will help. But I would ask that you also try to flip the cold plate and try it that way also. There's reasons to think it could work better. It's at least worth a try since it is very easy for you to do so. The heat taken away by the fluid is mostly through conduction since air conductivity is very low. What you get is a lower interface resistance with the adjacent plate.

 

[IRstuff]

Oh, I missed that; tubing against the aluminum plate is a bad idea. The point of having the copper plate is to efficient collect the heat from the aluminum plate and divert it into the tubes. Having only the tubes in contact means that the heat must travel through the aluminum instead, and then only have an extremely tiny contact area with the tubes. I can't even imagine why anyone would have picked that option as their first choice.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[knowlittle]

While copying and pasting, I must have deleted a line. The plates sit inside a vacuum chamber. I cannot use anything that will outgas heavily.

I am now convinced direct metal contact is a bad idea (for heat conduction). I am not sure if I can lap 200 mm square, both plates. What about something like soldering wick between the plates? (if they make a mat out of fine metal wires)

I still have the option of thermoelectric cooling.

Thank you for your replies and my apology for not including "vacuum" in the description.

 

[knowlittle]

I forgot to answer how thick the cold plate is. The plate is there to hold the copper tubings flat. That's the only purpose. It can be of any thickness. Basically, nothing's above the subject plate that I need to cool 25 deg C.

 

[IRstuff]

"I am now convinced direct metal contact is a bad idea (for heat conduction). "

No, it's not. It's only bad if the design is bad. All that's required is a sufficiently intimate contact. 200mm^2 is less than 2" square, so how big are these tubes and what's their separation? That's what will determine the thickness require for the copper plate. If you're going to use solder wick, you might as well just go with direct contact. With a sufficiently cold coolant and sufficient flow, 20W is not that big a deal, necessarily.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[DRWeig]

Not that this comment will help a lot, but the processor in my PC dissipates 250W pretty regularly, and it's only 52 mm x 45 mm. That's only about 1/16th the surface area of your warm plate. The motherboard maker's metal-to-package contact is pretty good, but they liberally apply conductive goop too. The coolant pipes are heat pipes. They are completely embedded in a hunk of aluminum. The other end of the heat pipes is a radiator with a fan. Your radiator, of course, has to be outside the vacuum chamber.

The processor gets hot -- 67°C at case surface, 91°C at the internal junctions. 25°C is more of a challenge, but with only 20W I think it can be met.

If you could cast aluminum around your heat exchanger tubes (completely surrounding them) and machine the contact surface nice and flat for good contact with the plate you need to cool, you might get the efficiency you need without too much trouble or any goop at all.



Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[knowlittle]

It's good to hear that it might work. I will give it a try.
1) as you suggested, I will make the mating surfaces as flat as practical.
2) Btw, I meant 200 mm x 200 mm, not 200 sq mm.
3) Yes, I will have a cold water bath sitting outside the vacuum chamber. It will pump cold water through the copper tubing.
4) The reason I don't want to fix copper tubes directly on the warm plate is I need to take it out of the vacuum chamber. The cold plate can be swung up slightly, just enough to make room for the warm plate.
5) My plan: 10 mm dia copper tubes, embedded 50% depth into the cold plate, 50 mm apart,

Any further comments would be appreciated. Thanks,

 

[IRstuff]

so, 20-mm thick copper plate? or 5-mm thick? i.e., 50% of whose depth?

50-mm separation or pitch?



TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[renderu]

I did some checks on some scenerios (difference from top to bottom plate temperature for 20watts heat generation and also minimum flow rate of zero degree C water required to remove that heat). By the numbers I am getting, I think you will succeed even if you screw this up really bad.

 

[zekeman]

My data shows that for an rms surface finish of 50 micro inches flat plate the conductance invacua at a pressure of 2 psi is about
100W/m^2-c
If the cooling area is 200mm*200mm
then for 0 deg C water, I get
100*25*.200*.200= 100 watts of cooling
so the contact solution is plausible, even with less stringent surface finish.

 

[IRstuff]

Since it's only 20W, you could run a glycol mix and get down below -10ºC if needed.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[knowlittle]

Thank you all for checking numbers for me. I will go ahead and fabricate the cold plate. 20 mm thick aluminum, 5 parallel grooves, 10 mm dia copper tubings brazed in the grooves, the other side polished better than 50 micro inch (like 400 grit) for good contact. Thank you again. Will report back but it will be a while.

 

[knowlittle]

Thanks for the conduction number at 2 psi. What about vacuum like 10-5 torr (10-5 mbar, 10-3 pa)? Does it go down by a factor of 2 or 3? Or an order of magnitude?

 

[IRstuff]

If the object are in intimate contact, there is little, if any, reduction in conductivity

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529