SMD mosfet heat dissipation vs through hole help

[Electro nS]

for purpose of 12v 60A switching power supply (half bridge or full bridge ) which is a better package option SMD mosfet (D2PAK 7) or through hole (247-3) ? (assuming both have 3mohm Rds)

i have used through hole to247 for a some time and they are really good and can be heatsinked easily . But for size and other issues i looking forward for using the strongest SMD package (d2pak 7) and i donot know how to keep it cool ( without a FAN) , using aluminum and copper heatsinks or bus bars only , does it heat up my PCB traces and how can i remove that heat out ?? (approximately 60A peak for 10s and 40 Amp continuous )

please any ideas , thoughts and advices , somewhere to start even a picture of an industrial product that uses such mosfets !!

thanks in advance guys

 

[mtwieg]

First you should carefully estimate power dissipation, then determine what your required thermal impedance is, based on the max ambient temperature.

SMT packages can't dissipate much heat through the copper (unless you have extremely thick copper), but there are special heatsinks that are small and easy to mount which can improve thermal performance: https://www.aavid.com/sites/default/files/literature/Aavid-Board-Level-Heatsinks-Catalog.pdf#page=25

Those heatsinks help, but you're still much worse off than with a TO247. Also one downside of those heatsinks for SMPS is that they will radiate switching emi like crazy.

 

[FvM]

Good solderable D2PAK heatsinks have downto 12 K/W thermal resistance with natural convection (Aavid Thermalloy 7109). Probably still too much for your design.

 

[Electro nS]

thanks guys i was also skeptical as you about SMD mosfets ( i will investigate that further )

just another question what if i used the to247 in special mounting to be like SMD see figure 1,2 :



does it work this way ( tab on heatsink and PCB appling pressure on the otherside ) or would this create a non uniform pressure on all mosfets ?? the thing taht makes me think of this is this image where each mosfet is screwed to heatsink , rather than depending on pcb for pressure :

 

[FvM]

I won't consider the flipped TO-247 as true SMT solution. Equal height of all devices might be an inappropriate assumption looking at package dimension tolerances. So you either need a thick gap filler compound with respective higher thermal resistance, or screw the transistor individually to the heatsink.

Chipscale packages like IRF DirectFet might be another option http://www.irf.com/technical-info/appnotes/an-1035.pdf

 

[mtwieg]

Yeah you won't get the advertised performance out of TO247 unless each one is fastened to the heatsink with a screw or clip. You would need a thick conformal TIM to accommodate any unsmooth topography, and the PCB itself would be under a great deal of stress... your third photo looks like a good solution though.

I've used the DirectFet family in a design before, and it did work quite well. However one issue with them is that the packages have such a low profile it really restricts what other components can be put nearby on the same layer without being crushed by the heat spreader.

 

[Electro nS]

thanks guys for the reply i will use screwed mosfets to247 for the safest operation.

i have also heared that when using SMD mosfets an effective way of cooling would be an aluminum substrate PCB for the Power section . but sadly i donot have acess to such PCB process and i think i would be expensive .