Electronic – How to limit current precisely

circuit-protectioncurrentcurrent-limiting

I would like a way to limit the current of a circuit to exactly 100mA (+/- 2-5%). This is on a relatively high voltage circuit (100V to ground), and the current is normally pulsed with pretty fast rise times (let's say 1-2 microseconds) using another circuit (low-side) as a switch and pulse shaper. What I need is a way to limit the current in case the other circuit malfunctions (fails short-circuit, in the worst case). I would like to limit it within also a few microseconds (ie: not more than a few microseconds of overcurrent). I would like this part of the circuit to not use much power when it is not actively limiting the current. In case of overcurrent, the current can either be clamped it at 100mA or possibly disconnected until things reset, but clamp would be ok. What is the best type of circuit to use in this case?

(Background: this is on a very compact board, there are less a few tens of mm^2 to fit this in).

(The average power available is limited to less than 0.3W at the source, even in a short circuit, so power dissipation is not as big of an issue as it might seem; the source will drop pretty quick from 100V to around 3V if there is sustained 100mA, but that takes milliseconds not microseconds)

Best Answer

Below is a high-side constant current circuit using a PNP transistor and op-amp: -

enter image description here

It needs to be modified a little to take one hundred volts though: -

  1. Op-amp needs a top rail of V+ (as shown) and a lower power rail maybe at V+ - 10 volts
  2. PNP transistor needs to be rated for at least 150 volts
  3. Vref needs to be positive rail referenced by using a precision voltage reference such as a shunt reference - it should replace R1.

How it works - lets say Vref is 1 volt below V+. Due to op-amp negative feedback Vref is forced to be across R2. With a voltage of 1V across R2 and a resistance of 10 ohms for R2 it implies that current taken by the load is limited to 100mA.

Having said all of this, the volt drop of R2 (1V) may be too much and a smaller value of Vref adopted. It could be reduced to 100mV and R2 dropped to 1 ohm without much difficulty. An op-amp with output rail-to rail capability is needed as well as the op-amp being able to sense its inputs at the positive rail too. It also needs to be fast and possibly avoid saturation when the load taken is under 100mA - this could be done with extra circuitry.

Anyway, that's the general approach I would consider.

Related Topic