Great points! Do take a look at the papers linked by me and @MCH in posts above. Also see the V2 changelist posted above. These PCBs are already 2oz, it's needed just to carry the current, never mind the thermals! In terms of "carrying the heat down" do review the papers, the thermal modeling is not quite that simple for SMD components. While heat _transfer_ mostly happens though the pads, the highest temps remain in the core of the resistors. Thermal pads and vias would be a definite win in V2 below the resistors because there's no airflow down there and moving the heat to the backside is the only possibility. It's not so clear that adding heatsinking to the topside would help much, but I'd welcome a stronger analysis! The thing is that these 2512 2W chips are quite thick, so cooling from one side only is probably not optimal.
You mean for the Fosi? Yes, full suite on the way, I'm just settling the dummy load in for now.
Has that actually been tested or is that just a guess at a likely scenario? Would surface tension of the melted solder hold them in place?
You wouldn't ask that if you'd. ever done manual SMD reflow or rework! Bloody things fall off at the drop of a hat unless the board is perfectly level.
That's 200W for the four boards, I assume. 200W/(40K*4*0.03m^2)=40W/m^2K, which roughly agrees with expectations:
Exactly so. And the graph you shared is nice, showing that increased airflow will only get you so far. So yeah, I'm sticking with 400W continuous (100W per board) as the upper limit for this version. We could probably double that with a version 2 that has better heat transfer front to back and some heatsinks to provide more surface area. But then it stops being a silly project and starts getting all serious.
If you heat the legs of most soldered components and flip the board upside down, they’ll usually drop right out.
I was not, so thank you! He went to 200W continuous I think? I tested my setup today after getting some 3D printing done by my son to make a decent enclosure, and adding heatsinks to the chip side. I was able to run at 400W (so 100W per board) with heatsink temps at 75C, but the backs of the boards got to 115C. I've ordered some more heatsinks to attach to the back of the boards, which I think would let me get to 800W continuous.
My prediction was broadly right when using a big pedestal fan, but that's not so helpful when trying to make a useful device. Good airflow and chip-side heatsinks are getting me to 100W/board. I think adding back-side heatsinks will get me to 200W/board or 800W total, at which point I'll run out of heat capacity in the airflow around 50cfm/85m^3/hr
I think you and I have different ideas about what might be "most simple". Unless you are joking in which case well played! Immersion cooling is messy, requires careful handling, and eventually the heat capacity of the coolant runs out so it's not truly continuous in its power handling.
I do not know exactly.
That looks really-really nice,congrats to your son!Similar one can be used to cool amps for extreme performance I guess.
That is a huge difference. Are you sure the thermal coupling with the heat sink is good and/or your IR measurements from the heat sink are accurate?
