JayGilb
Major Contributor
If they can adequately cool the device via the heat sink conductance/pcb connections, that would be a much cleaner design than applying compound to each individual GaN FET. There are many other components in the GaN FET's proximity with varying heights that would make designing as heat sink difficult.Another hypothesis from a US user
"that is a very interesting claim indeed. Air has a much higher thermal impedance than a thermal compound or thermal interface material designed for heat transfer, so one can only wonder what they were doing differently between the two setups to end up having the current solution being more effective at removing heat from the components.
The only thing I can think of is that they may have unintentionally used shorter non-metal standoffs when trying the compound/TIM and that resulted in a higher thermal impedance between the PCB copper and the heatsink, but that would seem like a pretty obvious difference to notice during testing. The GaN FETs have the least thermal impedance through the solder joints and into the PCB copper, which is why pulling heat out of the PCB near the FETs is the best way to cool them down, but surely using the compound/TIM along with metal standoffs between the PCB and heatsink would only further reduce the thermal impedance and improve the thermal transfer. Anyway, this is all speculation."