A while back I decided I needed to make my own dc load bank. Using mosfets as the power element seemed the most natural, since they are what I have the most experience with and can scale to very large powers.

However. most of the transfer characteristic charts in the data sheets do not have much detail at low currents, and are plotted against log(current) which makes it hard to see just how sensitive they are at the low gate voltage region.

I used my SDG1025 arbitrary waveform generator to give me a controlled gate voltage staircase up, while two BK393 datalogging multimeters measured the gate voltage (VGS) and the drain current (IDS). It is a little bit hard to correlate VGS and IDS in this setup since both meters only log at ~2Hz and are not synchronized, so each gate voltage is held for about 10 seconds.

Test Waveform in Easywave

The main goal of this to determine what tolerances would be typically required of a opamp / current / power sensor feedback circuit to keep good power or current control, that is

Here is my first data, logging the threshold region of an International Rectifiers IRFP150MPBF 100V, 42A N-channel mosfet. The data is for gate voltages ~3 – ~4.3, with VDS ~= 5V0. This test has a R2A-CT4-38E heatsink with thermal pad in free air, I will repeat this later with forced air to limit the temp rise, and maybe go to higher currents depending on how hot it gets. This relatively short test at ~10W already got up to 70degC. I might modify the sdg1025 waveform to allow for some cool-down time between samples. I also might need to split the test into 2 parts, to allow for me to switch current ranges on the DMM. I might need to get another multimeter so I can log temperature….

Somewhat annoyingly, my BK393 logs voltage at 2Hz, but current at 600ms intervals. I used octave’s interp1 function to resample the voltage reading to the current reading’s time base using linear interpolation, which seems to work pretty well.