I'm trying to control a high voltage device using a microcontroller but I'm having interference problems.

(Basically I am feeding a control signal to this PWM so I can switch it on and off. The PWM is driving an ignition coil.)

It is running from a battery so there is no proper earth point. If I do connect the battery negative to earth, everything is fine, but if not earthed, the microcontroller will misbehave (triggering inputs) as soon as the high voltage circuit comes on.

I have lots of decoupling capacitance near the microcontroller, and a choke + regulator between it and the supply.

I assume there is a lot of noise on the -ve (GND) rail which is common to both the HV and the controller circuit.

Without actually earthing the circuit, is there a way to keep the noise away from it while still using a single battery supply?

(sorry for typo in title: It should be "Isolating")

 

Can you post your schematic?

 

If I do connect the battery negative to earth, everything is fine...

That's a bit puzzling. As suggested, a schematic would help.

 

I would say if you are not using an opto-isolator, you should.

 

I would say if you are not using an opto-isolator, you should.

If both circuits are powered from the same battery then an isolator won't really help since the ground for the input and output must still be tied together.

 

I think there's still some value in using the isolator. The ground and signal wires can form a loop which acts as an antenna for electromagnetic radiation. If there's an isolator it can be wired with two closely twisted conductors which carry equal currents in opposite directions, and thus wouldn't be subject to much induction. But if the load really is creating hostile radiation, I'd try to float the electronics relative to it, and have only optical signals cross the barrier. If you really get desperate (I never have) you could have an "extended isolator" in the form of fiber-optic cable to carry the signal!

 

Thanks, please see attached diagram.

I've managed to get rid of unwanted triggering by connecting capacitors between the input pins and GND.

The problem that remains is that when the HV coil is running, every part of the circuit seems to have a little HV on it. If I touch a pot for example, it gives me a little zap. Would I be right in thinking that there's nothing I can do about that (except for insulating it) since there is no real earth reference point?

 

Forgot to add picture! doh!