I am looking to design a constant current source that can give between 30A and 60A used for electroplating. Obviously, an LM317 isn't going to cut it for this project. Does anyone have any thoughts for a design that would allow for 10A steps (5A preferably).

My power source is an HP Model PS-3701-1 725w 60A PSU.


  • 1
    \$\begingroup\$ what is your output voltage range? What is your max input voltage? How fast do you have to react to load changes? Are you even sure that the power source you have can supply the amount of power you need (it just seems to be some kind of PC PSU, which definitely isn't meant for very heavy loads on individual rails, I guess) \$\endgroup\$ Mar 27 '17 at 16:55
  • 2
    \$\begingroup\$ @mkeith 60A LED source?! Woaaah! \$\endgroup\$ Mar 27 '17 at 17:05
  • 1
    \$\begingroup\$ That would be handy, indeed! I was also thinking about simply recommending small arc welding equipment. I really don't know the effective resistance / forward voltage of electroplating in action, however! \$\endgroup\$ Mar 27 '17 at 17:09
  • 1
    \$\begingroup\$ assuming output voltage is very low, I'd tend to use a reactive primary-ballasted low voltage transformer with schottky bridge output rectifier connected directly to the load. But that's just me. \$\endgroup\$
    – Neil_UK
    Mar 27 '17 at 17:13
  • 1
    \$\begingroup\$ LED supplies are constant-current. Correct me if I'm wrong but it seems like they'd parallel just fine. Get a 25, 20, 10, and 5 amp supply, and you should be able to turn on any combination you want. \$\endgroup\$ Mar 27 '17 at 17:27

There is an old saying from Clausewitz: "Everything in war is simple, but the simplest thing is difficult." So there is a very simple solution to your problem, but you may have some difficulties with it. The basic concept is


simulate this circuit – Schematic created using CircuitLab

but of course God (or the Devil, depending on your source) is in the details.

First and foremost, the circuit is a lie, in that two FETs will not remotely do the job. I'd recommend about 20. This will allow 3 amps per FET, and assuming about 10 volts across the FETs, a power dissipation of 30 watts per FET. Both of these are perfectly reasonable mumbers for TO220 packaged FETs, and almost any n-type power FET will do.

Of course, if you did a quick bit of math in your head, you also realized that the total FET power dissipation will be in the 600 watt ballpark, so you're going to need a really hefty heat sink. This actually isn't all that hard or even all that expensive, particularly if you pay attention to forced air cooling - but you do have to pay attention.

Finally, that innocuous current sense resistor is going to cost you as well. 10 mohms at 60 A will give 0.6 volts, which explains the pot/resistor values, but it also implies 36 watts in the resistor. You'll want a non-inductive resistor for the job, and while they're available they're not cheap. Here's an example of what you might want, available from Digikey for 30 bucks and capable of handling 100 watts. With a proper heat sink, of course. Or you could simply run 10, 0.1 ohm, 10 watt resistors in parallel making sure to use a Kelvin connection. Again, managing heat will be something of a problem. And using a smaller resistor will reduce the heat, but an LM324 has 2-3 mV of offset, so you really don't want to go too much lower or you'll lose accuracy on your current set. Oh yes, and don't forget to take into account the resistance of your wiring. It doesn't take much at 60 amps to make a difference.

But it really does look simple, doesn't it?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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