A simple active DC dummy load for PSU testing

[megajocke]

You should use one opamp per FET to get equal current through each.
Hexfets aren't that good at sharing current when put in parallell. OTOH, with 0.47 ohm source resistors and for relatively low input voltages and high currents the balance is probably pretty good.

 

[wine&dine]

Hexfets aren't that good at sharing current when put in parallell. OTOH, with 0.47 ohm source resistors and for relatively low input voltages and high currents the balance is probably pretty good.

That's what I thought when I built my own load a while ago. Suffice it to say the re-built version now uses one opamp per FET. Dual supplies and/or offset diodes are not needed, 12V will do fine - just use opamps with 0V capable inputs - output is not critical as FET conduction starts around 4V only.

I built mine into a PC power supply case stripped of the electronics, keeping the fan to cool the FETs - which I mounted on a spare passive Pentium-III type heatsink. When I rebuilt it with per-FET opamps, I added a fan speed regulator too.

The possible FET oscillation mentioned before (at RF frequencies) is killed by the gate resistors - 1K seems awfully high as it will add too much delay together with the FET gate capacitance - 100R is more like it with one opamp per FET.

Your 2.2nF will likely stop the entire control loop from oscillating - in practice, just try out the value (lower is better, too low will oscillate).

Don't put a cap on the test voltage input.

 

[wine&dine]

One more observation -

While the IRFP460 is rated for 13A continuous / 280 Watts, it will be hard to get more than 60-70 Watts out of it on a good heat sink.

At the 24V you show that's some 3A - and at that current, your .47 ohm source resistors will dissipate less than 5 Watts each.

On the other hand, at the 10 amps your 50 watt resistors could take, you already drop 4.7V over the resistor - the load voltage regulation range is reduced to 7.5V - 12V (lower limit by Rds+Rs, upper by Pmax)

In my design, I therefore decided to limit the voltage on the source resistors to about 0.35V at Imax.

As you noted, DC-coupling the input requires very careful setup of your function generator for step tests. Why not AC-couple it instead?

Attached is my circuit. Two ranges - 0.3 and 3A full scale; AC in (50 Ohms impedance) 1V = 1A.