Rds-on is so low on my parts, I'd need lots of amps to heat them up.
This might work:
A quick test (using my Fluke on the diode range, 1 mA) suggests that the tempco of this sorta-diode is about -1.9 mV/K, so it's a usable thermometer.
The GaN parts behave like phemts. There's no actual substrate diode, but they turn on like gate-controlled e-fets when the drain goes negative.
I'm making 7ns HV pulses, and the time delay seems stable, to well under 1ns, so it can be removed in the generator.
There's no silk on any of the pads. The BGA fets soldered nicely.
A commonly cited number for the surface of the sun is about 6e7 w/m^2. My fet will only be about 2e6.
Why only 10mA probe current? Even a 10 milliohm FET makes 10mV/amp, so, an amp for probing doesn't seem too bad. Pulsed, of course.
I sent you my hairball.
Cheers, James Arthur
The temperature is inferred by the voltage drop of the pseudo-substrate diode, when the source and gate are both zero volts and we pull the drain gently negative. More current then won't give more temperature signal. The fake-diode tempco is probably highest at low currents.
At 1 mA pulldown current (my Fluke on its diode range) I'm seeing about -1.6 volts on the drain of an EPC8004, and that drop has a tempco of close to 2 mV/K. That's a nice thermometer.
I'll adjust the 5 volt supply to set the ganfet power dissipation to around a watt maybe, and occasionally pulse the p-fet off to scope the temperature. The temp measurement could be very low duty cycle, say a millisecond of measurement every second or so.
Given a full millisecond to make the measurement, I could use 1 mA as the test current. I only need as long as it takes to charge the various capacitances.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required