The different circuits described in the past few days all had problems (the first ACS Hall-effect hookup worked reasonably well, but was not sensitive enough for my purposes).
Here’s the circuit as it is now:
(a slightly different layout would have avoided the overhang, had I known the complete setup in advance)
There are 3 independent circuits on there connected in series (so the same current passes through each of ‘em):
As you can see in the first image, there’s a little screw-less terminal block on the board. It lets me short out any combination of these setups – this was useful until everything had been built up, and also lets me rule out interference while focusing on a specific setup. Note also that this is an epoxy-based board, not the pressed-cardboard type which breaks easily and isolates less.
Let me just repeat that messing with AC mains voltages like this can be extremely dangerous. This circuit has all the wires completely exposed, and that’s intentional: I prefer to be reminded of the risks, instead of getting a false sense of security because it all looks ok! One way to deal with these risks is to completely stand back from the entire hookup and only transfer measurement data over wireless whenever applying power to this stuff.
Tomorrow I’ll describe a simpler setup to get to the bottom of the measurement anomalies I’ve been seeing so far. With many thanks for all your comments – there’s clearly a lot more to it than with simple logic level signals!