B-G431-ESC1 Thermal issue

[Elektromania92]

Dear member of ST community, 

I would ask you about the thermal characteristics of B-G431-ESC1 Motor Control evaluation board.
I purchased it few weeks ago and started to play with six step for beginning and later FOC algorithms. I observed on my board that if power supply V+ 18-20V is active and PWM outputs are switching, of course deadtime is set to ~500ns, board running at 170MHz the PCB temperature reaches  ~50 °C or even more. The temperature is quite uniform. The DC/DC itself produces heat by switching and the power stage too, but for me seems to be a little to much. Is it normal or something is failed on the board? I removed the DC/DC converter and provided the 10V from external supply, the situation is similar.

 PWM outputs disabled. In that case the heat is produced by DC/DC only. Is that normal? I did rough calculation based on datasheet and for me it looks to be too much for roughly 100mA output current.

 Additional heat generated by 3-phase power stage itself when PWM enabled on all phases, no load attached to outputs, deadtime set, PWM freq = 30KHz. Temperature there is also ~50 °C if the DC/DC converter is removed. 
Is that normal or it is corrupted? 

Thank you for your response in advance.

[GMA · Accepted solution]

Hello @Elektromania92,

What you're observing is completely normal given the PCB layout.
For the DC/DC converter, at 100mA on 10V, this results in 1W output. At this current level, the efficiency is approximately 70%, giving a component dissipation of approximately 0.3W. According to the datasheet, we have a roughly 40°C/W thermal resistance from the junction to air, but if we look at the PCB size compared to the datasheet, there's significantly less copper for heat dissipation, and we should be looking at 100°C/W. This gives us a temperature increase of 30°C, or 50°C in total if the ambient temperature is 20°C.
As for the power stage, here too, if we refer to the datasheet, the advertised thermal resistance is 31.3°C/W, but with a copper surface area of 1 square inch and a thickness of 2 oz, or 70 µm. And this is not at all the configuration of this PCB.
Not knowing the current required at the output when you run the power stage in PWM at 30 kHz, it's difficult to answer. Nevertheless, 31.3°C/W will not be achieved with this PCB. If we consider a probable doubling of this value, i.e. 62°C/W, it is enough, under 20V power supply and a peak output current of 5A, for the transistor junction temperature to rise to 55°C for 30 kHz of PWM. Note that the junction temperature is roughly equal of the case because heat has a hard time escaping through the transistor baseplate given the RTHJA value.
This is why forced convection (ventilation) is essential with this board to dissipate heat as effectively as possible.