The power dissipation for the 10K anode load is 4 Watts  (20mA @ 200V) and the design uses four 5 Watt surface mount resistors to accommodate this, to give a combined thermal capacity of 20 Watts.  However, even testing in a cool workshop (~18 DegC), the resistor surface temperature was 122 DegC, and after a longer test session, I had seen 132 DegC.

This is serious as the glass temperature for the FR4 PCB is 130 DegC which means that these resistors are running way too hot for the PCB.  Assuming ambient temperatures up to 50DegC, and also a 20DegC safety margin, would mean that this design is running at least 50DegC over temperature.

There are a number of options that I could pursue,

• Increasing the number of SMT resistors (4 -> 8), but this is already getting to be a significant area of PCB.

• Investigate modifying the PCB to add thermal vias and/or add some PCB holes to promote airflow around the components, but this is still a lot of heat to dissipate in the PCB.

• Move to a higher temperature FR4 PCB, or to a new type of material, like ceramic.  (Online PCB manufacturers do offer these materials at a premium.)

Having thought about it a little, I think that I will use through hole component (THC) resistors next time, and make sure that they are off the PCB, and with some PCB holes / cuts to promote airflow.

ToDo: Resistor datasheet does not require derating at these power levels - but how could this work for a SMT PCB??