Successfully dissipating 1.8kW of heat

[Meathead]

It's a research project on increasing efficiency on the power side of a hybrid type vehicle. I'm super excited about it but can never talk about it until its finished and demoed. Talk about a buzz kill.

 

[RnRollie]

I've found places that make square copper tube with a round hole bored through the middle. If you can find that stuff in the proper size it would probably work pretty nicely. The ends could be threaded for fittings, and you're set.

I think I'd try flaring the pipe on the blocks you already have first though. That's cheap and may be quite effective.

What is this thing you're cooling, anyway?

^^^this

as for going from pipe to hose, it depends on space
3/8 ID copper tube will be at least 1/2 OD, but probably 5/8 OD, which is of course larger as 3/8 ID hose.
However, boiling water and submerging the hose in it will soften the hose enough for it to be forced over the (copper) pipe. My guess is that inserting the pipe for a length of 1 or 2" into the hose and finishing off with jubilee clips might work. Especially if you roughen up the pipe end a bit.
This is assuming you have 1 to 2" extended pipe end available... i wouldn't try it with less.

Of course the threaded square pipe would be better... and fittings are a dime a dozen from any plumbing supply or DIY store

 

[SF101]

If its going on a vehicle it would make sense to just develop it with a vehicle rad wouldn't it?

then you could use moving air and air scoop to cool it also.

 

[Bobnova]

Vehicle cooling systems run in the 180-210f range, too hot for MOSFET cooling. Even the "cold" side of the radiator is 140f and up in operation.

 

[SF101]

Yes but it doesn't have to be looped into the same system as the motor im just thinking a system that uses rads the company likely already has would be easier to pitch and cheaper to implement in the future.

 

[GTXJackBauer]

Its probably the computer system of a vehicle, not part of the engine if you will. The allowance of high temps would be way different.

 

[SF101]

I understand lol but if your going to include radiators why not use something likely to be available. + with that much heat your going to need something serious to make it work.

 

[Meathead]

If its going on a vehicle it would make sense to just develop it with a vehicle rad wouldn't it?

then you could use moving air and air scoop to cool it also.

Thats what I was thinking. I'm just doing whatever is easiest and will prevent the chips from burning up. I'm using pumps and radiators similar to a car only for the simplicity and reliability of a basic setup. The cooling system is not the part of the project its just a byproduct of having a total power output of 60kW and having triple digit amps running through parts of these boards.

 

[Ben333]

Just thought I'd add, I got that eBay rad the other day. It's brand new, and a great size. Picture three 320.120 rads, next to each other. The kicker, it's aluminium... and was advertised as copper. But... AL conducts heat better than brass, and I'm impressed with the fin density, etc. Can't complain for the money.

If you have multiple blocks, one thing you may want to do would be add a T on each end of your rad, and then have one pump feed half your blocks, and the other the rest. Like two loops with one rad, should increase flow and overall circulation in the radiator. And yep... the swiftech pumps are hard to beat! I bought one new in 08 and never looked back. Still using it.

 

[Ben333]

Any updates? I got my setup going on that geo metro radiator. With two waterblocks, the MCP655 still moves water like a beast through this big thing. Before it was blead, the tubing would actually pulse slightly, which tells me the pressure / flow is just fine.

My rad all dressed up /w two sets of reducers. 1" --> 3/4" & 3/4" --> 1/2"

 

[Bobnova]

Beautiful, that's pretty much exactly what I was thinking.

 

[Meathead]

I like the rad nice job. Especially the connectors. Looks pro Ben.

For myself, I just went with what was familiar for me. I'm a bit pissed I ran to a local Microcenter as they had a nice deal on some fans and half the fans I grabbed are led fans. Oh well. The picture is testing on our potted inductor. Will set it up more permanently later on.

The temperature from coldplate to aluminum case is like a 6C delta but the inductor is like a 70C - 80C delta in some portions. We are having issues though finding a potting compound that does not have metal particles causing losses yet still has a high thermal conductivity and is not a gigantic PITA to work with. The inductor is being potted into an aluminum case and that is what is strapped to the coldplate on the left. Potting compound is 0.682 W/m*K epoxy type which I'm sure is a huge bottleneck.

 

[Ben333]

Oh, wow, so a single 120x3 did the trick just fine? That's awesome.

 

[Meathead]

Oh, wow, so a single 120x3 did the trick just fine? That's awesome.

No that is a test setup and that is a 120x4. I have 2 120x4 and 2 pumps when all 12-16 blocks are in use. That was just for monthly updates meeting. It is only connected to one block currently.

The real test setup is near 8 ft long and 2 ft wide lol. It will eventually be consolidated to one board though so this will only be used for a month tops for testing.

I'll have pics in a week or two.

*EDIT* I meant I'll have pictures in a month or two lol. We ran into some issues that are almost finished up being fixed.