Q on grounds..

You don't get "static" from AC.

Both of their earth grounds should be bonded to the earth ground at the entrance panel, and the neutral should be bonded there, and nowhere else. This provides a path for any AC-induced leakage currents, and _should_ be a very low resistance, so a few milliamps won't generate enough volts to upset anything.

There may be a small differential voltage for various reasons, but that's what the optoisolators are for.

Hope This Helps! Rich

• Perhaps, but if the "black box" controller is left floating, it is possible for static build-up between it and the AC line via induction to the inputs.

• "Both" earth grounds? There is only one - from the input AC power (Line, Neutral and Earth for single or 2-phase power; L1, L2, L3, Earth for 3-phase delta). The "black box" has no "earth" ground - it only has an internal common line (which according to convention only is called "ground") for its own signals. What i have seen is that "black box" common ("ground") tied directly to Neutral (one phase power, but presumedly ditto for 2-phase and

Rich gave you the correct answer.

The National Electrical Code *requires* that the neutral be bonded to the earth ground at the service panel, and that it (the neutral) not be connected to ground elsewhere.

The setup you have seen - black box common tied to neutral and earth ground bonded to its metal enclosure - violates the code.

Ed

So there are at least stray capacitances from the reed coil, from the optoisolated triac and from the logic power supply to the black box signal ground.

That is the easy part to fix. Use a mains transformer with a static shield (connected to the AC ground) to significantly reduce the stray capacitance between the mains live connector and black box signal ground.

If this does not fix the problem, you have to analyze the currents that are flowing from one stray capacitance to an other stray capacitance and make sure that this does not flow through any critical signal ground paths on the PCB.

If the signal ground is connected to AC ground, you should analyze the current paths from the stray capacitances through the signal ground and into the wire going to AC ground. Then select the point on the PCB for the connection to the AC ground wire, in order to avoid stray currents flowing all over the PCB grounds.

If the inputs and outputs are isolated and if the subassembly is in an isolated box, is there really a need for grounding the signal ground ?

For EMC requirement some capacitive coupling to AC ground may be needed. Also to prevent static build up, a large resistor (say 100 kohm) may be needed between the signal ground and AC ground, if the direct connection of signal ground to AC ground is not desirable.

Paul

Thanks.

Thanks.

its=20

and=20

power=20

an=20

transformer=20

=20

these=20

other=20

the

else.

be

that's

AC neutral is bonded to earth ground at the service entrance, for services that have neutrals (3 phase delta may not). Per Code. And no where else, also per code.

its=20

=20

to=20

the

else.

_should_ be

that's

=20

common=20

own=20

=20

(if=20

It violates NEC only if the neutral gets connected to the earthed enclosure at that point. What may be a problem is the isolated parts of the system.