Excellent work, Jack.? ?Thank you for doing this.
The IRFZ24N datasheet from Infineon indicates a thermal resistance from junction to case of 3.3 degrees C/watt? This is a MAXIMUM value; perhaps typical is lower?
(see:??)
The DC power delivered to each device in your test would be approximately 0.5 (one of two MOSFETS)? * 13.5V (assumed) * 8.2 amps (reported) = 55.4 W
The AC output power delivered as RF from one device would be approximately 15.5 watts? (half of 31 watts output, your average output)
The heat to be dissipated by that one MOSFET would then be approximately 40W.? ?This is very close to the manufacturer's limit of 45 W, which is only true at a CASE temperature of 25 deg C and must be derated by 0.3W/degC above that.? ?Your case reached 90 deg C, where the maximum allowable dissipation would be 45 - (0.3*(90-25) = 25.5Watts.? ?At that point you were 15 watts OVER the allowable dissipation by a single device.
The RF efficiency is 15.5/55.4 = 30%, which is pretty low.?
Case to sink temperature measured by the manufacturer was 0.5 degC/watt.? ?Your data show approximately a 30 degree difference between device case temperature and sink temperature, or approximately 0.75 degC/Watt, which might be reasonable since you had a mica insulation, rather than simple greased contact as the manufacturer used in their test.? ? ?Thus your data look fairly reasonable at that point.
JUNCTION TEMPERATURE
With a? (maximum) junction to case thermal resistance of 3.3degC/W and 40W heat dissipation, the junction will be as much as 132 degrees C higher (hotter) than the case.
The maximum junction temperature specified by the manufacturer is 175 deg C.? ?Thus when the CASE gets to 43 degC, under your test conditions, the junction will have reached its maximum allowable temperature.? You cases reached 90 deg C meaning the junction presumably reached 90 + 3.3*40 = 222 degC, which is almost 50 deg above the manufacturers maximum allowable junction temperature of 175 deg C, if you had devices at the worst end of the thermal resistance
I hope I did the above calculations correctly, but if so, it is impressive that you didn't lose a device in this test.? Just eyeballing your data, which started at a case temp above 25 deg C, it looks like in the first 15 seconds or so you would have hit a case temperature of 43 deg C, at which time the junction is at maximum allowable temperature.? ?The main limitation of the device is the junction-to-case thermal resistance of 3.3 degC/watt.? ? It seems to have high peak capabilities, and relative low steady state due to that thermal resistance.? The quoted value of 3.3 degC/watt is a MAX, perhaps you were lucky and your devices had better thermal conductivity from junction to case.
Based on the data, at that level of RF efficiency, it might be advisable to avoid key-down greater than 15 seconds.
If someone sees an error in my thought process here, don't hesitate to point it out!
Gordon KX4Z
