Craig405 said:I would say that that plate you have sandwiched between th L shaped metal and the heatsink probably doesnt help.
The more gaps you have to bridge using heat sink compound the more thermal resistance you get, and if your transistors are much hotter than your heatsink then their is definatly a thermal resistance issue.
They should be of a much closer temperature if well mounted.
EchoWars said:You need to reduce the losses at your junctions...ideally you would have the surfaces where they meet machined so that they make as perfect contact as possible, and use heatsink grease at those junctions.
At a minimum, you need to sand the surfaces as best you can, and use heatsink grease, and a slew of bolts to hold the assemblies firmly together (drilled and tapped holes to bolt everything together also need to be ever-so-slightly countersunk to account for the rise in the material caused by drilling the holes).
Here is how I have made it..... everything is o.k. and connected very good with silicon pads between the trans and the L profile
Just a SINGLE bolt? I really hope that you have been lazy in draving, not drilling
BTW. Silicone pads are sub-optimal compared to mica&thermal grease.
sunrise said:I will try to explain.... My mj15003 are connected to the 5mm aluminium L profile. That profile is connected to the 8mm 330x150mm aluminium plate that hold 12 small heatsinks together as one heatsink. picture attached
Waste of time, thermal resistance to top of to-3 is huge compared to even worst L-bracket, there is maybe 0.2mm wall thickess for 5-8mm distance. And TO3 packages are made of steel, invar or something similar lousy conductor(well, some of them are aluminum but I think aluminum cases are more thing from past.)sunrise said:
I would try to put some heatsinks on the mj15003. Maybe that will help...
1.5A and 22V per transistor or is there more of them parallei? that would be around 35W and thermal resistance between junction and to3 case is 0.7C/W so junction temperatures would be 110C at junction. Maximum junction temperature is specified as 175C but this is more like "absolute maximum where device lifetime is equal to zero"