I need to build 2 active 4-20mA outputs from a 24V supply voltage with unknown load conditions. One basic solution could be like the schematic below. VIn is converted to a current through R1, converted to the "high side" via R2 and then mirrored to an output pin. The load RL can be anywhere between 0 and 1 kOhm, as long as the FET can drive it. Don´t worry about the components being used - it´s just a sketch.


simulate this circuit – Schematic created using CircuitLab

My main problem is the power burned on the PCB in the worst case. Given the maximum current of 20 mA and a load close to 0 Ohm, the whole 24V are shared between the output FET and R3. This leads to 24V*20mA ~ 0.5W of dissipated power. Times 2 channels, we end up with 1W of heat on the PCB which I do not like but can´t get rid of using linear components.

Is anyone aware of integrated solutions using switching regulators to either:

-> step down the 24V if not needed or

-> using a "precision" current source to directly drive the output

  • \$\begingroup\$ What FETs are you actually using? Cooling 1 Watt without a heatsink isn't that big of a problem as long as you can get the power out of the components and onto the PCB efficiently. PCBs make for surprisingly good low-power heatsinks. \$\endgroup\$ Commented Jul 6, 2021 at 16:06
  • \$\begingroup\$ I know that 1 Watt might not sound much but the whole thing is hermetically sealed and should be as small as possible. \$\endgroup\$ Commented Jul 6, 2021 at 16:12
  • \$\begingroup\$ I would expect that circuit to oscillate. Much better idea to use a P Channel FET for M2 (source to R3) and swap the inputs on the op amp (OA4). ie. Take the feedback back to the op amp's inverting input. \$\endgroup\$
    – user173271
    Commented Jul 6, 2021 at 18:27
  • \$\begingroup\$ Can you put numbers to "small as possible"? Because the solution you've sketched there doesn't look very small to me, so I wonder whether our understanding of "small" and your understanding line up! Also, what's the ambient temperature you're dealing with? \$\endgroup\$ Commented Jul 6, 2021 at 23:56

1 Answer 1


Any linear solution will have the same amount of power dissipation, and using a MOSFET or BJT to dissipate the power is easier than using an IC since they typically are happy with a high die temperature.

I don't think dynamically adjusting the voltage of a SMPS would be very good- noise and dynamic issues. It could be done, of course.

Maybe you could live with a 10V compliance and step the 24V down to 12V to halve the dissipation (using a buck converter and filter).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.