I'm working on 2000W furnace controlling with Atmega32a and an optocoupler with BTA40600 triac, my program is designed to heat up the furnace to 850 degree Celsius in one shot (always ON state til temperature reaches) and wait on 850 for 40 minutes, then heat up to 920 degree Celsius. everything is fine but after 10 cycles of working, triac eventually burns out and the furnace is lost control (after burning out, the triac is short-circuited and the furnace is connected directly to the main). I think the reason for triac fault is to remains ON state for 40 minutes at least, is it true? the alternative is to control the triac with slow PWM which means tells it to stay ON for 2 seconds and after that OFF for 500 ms.
if you have a different experience please share with me
by the way I use a large heatsink for cooling purpose of the triac
IGPIR.jpg
Actually, the best way to get over this problem to replace the triac and relative component with industrial SSR, I think 40 to 50 amps SSR is enough !!
Best Answer
The triac will be dissipating in the neighborhood of 10W at the specified current (read your triac datasheet to get the exact value, but it won't be much less).
So you need to get rid of 10W for the heating cycle. You can pretty much ignore heat capacity and assume steady state. Suppose you want the junction to get no hotter than 110°C, with an ambient of 40°C maximum (make sure your oven is not heating the triac itself), so the rise is 70°C. Thermal resistance junction to case is 0.9°C/W so the case rise must be less than about 60°C, and the heatsink performance should be less than 6°C/W, which is a fairly modest requirement. A bigger heatsink will give better reliability.