Quote:
Originally Posted by thingstodo
How many 1/4" copper channels are in parallel?
I'm told that the thickness of the aluminum - shortest path between the IGBTs and the copper channels - is the greatest resistance to heat flow. The rest of the aluminum is cooled, but the important part seems to be the path between the heat source (IGBT base) and the cooling (Glycol)
I (since it's Paul's design but it's my heat sink) have 4 passes of 3/8" wide by 3/8" deep (I think) x 10 inches long in series. Most of the area covered is beneath the IGBTs. The cross sectional area works out to about the inside of a 1/2 inch copper pipe, but the machining is not quite as smooth.
The rest of the piping - radiator to pump, pump to heat sink, heat sink to radiator - is 3/4 inch home plumbing flexible plastic pipe that likely should not be in a car. But when you hit it with a heat gun and form it, it holds it's shape quite well.
My circulating pump is rated for about 4 gpm at a couple of PSI head.
|
I think if the aluminium is cooled then providing you have good thermal conductivity between that and the IGBTs then the block will draw the heat from them.
As far as I know the idea is to maximise the surface area of the aluminium to water contact so that the water can draw as much heat out of the block as possible while it is in contact. Ideally I guess would be a multi finned heatsink immersed in the water flow, but sealing etc is more problematic.
I only used one pass in my prototype and it hasn't been tested, but I tried to maximise surface area while maintaining material for the MOSFET screws to secure without protruding into the water channel.
Machining roughness should improve surface area slightly providing it doesn't cause cavitation issues, but obviously the IGBT surface should be a smooth as possible. I put a fly cutter over mine and it's reasonable but will probably get it smoother before final assembly.
My channels are milled with an 8mm cutter and 15mm deep. Material thickness to the MOSFET surface is 3mm. Sounds similar to what you have done in terms of cross sectional area.
I'm hoping the water will fill the channels properly too. I might put a clear perspex cover on for initial water testing so I can see any trapped air or circulation issues. Maybe introducing some dye once it's running would show up any dead spots.