I am saying digital as the device is base on an esp32 microcontroller. and usually such a device doesn't contain any silicon/digital logic ( as far as I know).

So simply the device contains an esp32, ads1115 for analog readings and a 2000:1 current transformer.

I pass the neutral and live though the current transformer to measure the difference and detect if some current is flowing through the ground. I am using a 150 burden resistor (as indicated by datasheet) and measuring the voltage on the resistor with the ads1115.

It worked perfectly with resistive loads. The difference is almost zero and when I test with my body to a 240 kohm resistor to the live as this:

live --> 240k resistor --> me --> my feet on the ground

the difference will be 8ma and the esp32 will shut off the main contactor.

I got an accuracy of 1.66 ma or at least 3.222. A sampling rate of 860 SPS (sample per ssecond). I get the max out of 10 measurements ( to be sure I am not measuring while the current is zero anyways) which takes around half-wave (12ms). If this 8ma of ground fault continues for 100ms (I can do less) , the esp32 cuts off the electricity and repower on after some time.

When I started testing with ac-to-dc adapters and motors as loads, there would be difference anyways and the esp32 will cut off the power always without my resistor to live test.

Any ideas while this happens?

I saw a 30ma wrote on a GFCI breaker and adjusted my measurement to this. The esp32 now looks for ground fault bigger than 30ma and waits for continues 250ms of ground fault before cutting power off. If I make it wait for 100ms only the problem is back.

With this it works but I bet it won't at higher loads. I am using small loads (phone chargers. small motors)

Note: The resistor to live test only makes me feel a bit of tingling and 200v goes through the resistor with the rest in me. I have someone with me as backup to cut off electricity manually if things goes bad.