Using IoTAWatt to measure residual current

I’m planning to upgrade my ancient Consumer Unit to a modern RCD (residual current device) filled unit. I think they are called GFCIs in the US. I thought I would see if I could measure any existing residual current to see if the upgrade would be problematic. So I put a CT (SCT013) around each of the meter tails and use IoTaWatt to subtract the live and neutral and provide a readout of the residual current to influxdb.

I am seeing a fairly consistent 30mA difference during standard daily loads which should be fine across a dozen breakers, right ?. However, when the Cooker goes on then I’m getting around 100mA which looks to be a problem.

Before I delve further, is this even a viable use-case for the IoTaWatt ? I thought about trying to use a single CT around both tails to get residual directly but dont have anything big enough and suspect anything that big wouldnt be that accurate at a low mA range anyway.

Anyone else tried this ?

OK, starting to answer my own question …

Here, you can see steady state residual current of up to 30mA when total house power is around 500W. You can also see the residual current around 100mA (peaking 150mA) when the cooker is on and using roughly 2000W. This feels possible as the cooker is old, the wiring is dodgy and the power is high. However, there are some negative readings which I dont think should happen when subtracting Neutral current from Live current. It should always be positive right ?

Another graph at a higher resolution with house total power stable at 500W shows that this is problematic as many samples are showing negative values.

Could this be timing errors in the sampling of the two tails or more likely is it just the error margin of slightly unmatched CTs ? The averaging inherent in the lower resolution graph suggest unmatched CTs so I guess I could put them both on the Live tail and see if I can calculate a scaling factor to bring them inline. Interested in views.

I’m amazed that your differential with this technique is so small. But I doubt that is your actual difference. You are probably using 100A CTs. 30mA is 0.03% of scale. The AccuCTs are 0.5%. So that’s not a viable way to measure.

As a practical matter, if you have an excessive ground fault, I would think that’s a compelling reason to go ahead with the installation and fix the problem. There’s the old joke: If you can’t afford the operation, touch up the X-rays.

If you really want to pursue the differential measurement, you can eliminate any differences between two CTs, input channels etc. by using one CT around both the hot and neutral. They will cancel each other out and leave only the differential.