Circuit for earth conductor presence detection
Reading a Tesla Wall Adapter installation manual , their brand of an AC EV charger, chapter about “Ground Assurance” and here is the quote from there:
Although the equipment can work with L1 and L2 connected only, because of the safety reason I assume they require Earth terminal of the charger to be grounded as well with the ability to monitor it(apparently).
Mostly referring to the US power grid where L1 and L2 are taken out from the split phase transformer with neutral grounded at one place in the system.
How the circuitry for something like that conceptually look like?
I was thinking about inducing DC current into the ground, small enough in value to not trip the RCD and measure the voltage drop in the ground/earth itself(I guess??) - current source and shunt for voltage sensing? Wondering if inducing any additional resistance(shunt) in series with earth is legal?
For what is worth their input power terminals are: L1, L2, and PE - there is no additional I/O solely for this function to work.
2 answers
The section you quoted says: "by injecting a small amount of current into the ground".
With a solid ground connection, the voltage on their ground pin shouldn't change no matter what current is injected onto it. If they inject a current and see a resulting voltage change, then they know the ground connection is not solid.
The question then becomes voltage change relative to what? I have no special knowledge about their circuit, but if I was faced with this, I'd probably generated an "average" voltage from the two hot phases. I'd filter that to get rid of the power line frequency and give it time to settle. Now you have an internal voltage that shouldn't be changing relative to the ground connection. You can inject short pulses of a few mA and a few 10s or 100s of µs and verify that you don't get corresponding changes in the difference between your internal reference voltage and the ground connection.
I expect a lot of noise, so I'd probably do these pulses regularly and synchronously measure the voltage difference. I'd then get a filtered value of difference with pulse on and difference with pulse off, and compare the two. If you continually do pulses, like 10 µs on, 10 µs off, then after a few line cycles I'd expect you get a good enough signal.
Again, this is all speculation. I don't know how they are actually doing it, and I haven't tried this. There are probably some gotchas you discover as you try to implement something like this.
"You can inject short pulses of a few mA and a few 10s or 100s of µs and verify that you don't get corresponding changes in the difference between your internal reference voltage and the ground connection."From the "recreated neutral" point?
No, specifically NOT that. That generated neutral is a signal that has very limited current capability. And, in the ideal case, it has the same potential as ground, so you can't easily cause it to dump current onto the ground.
Surely there is already rectified power from the two line phases available in the system somewhere. I'd use that to source the few mA to inject onto the ground.
Fault Codes (Page 38)
PE ground is mandatory to be functional and is tested. Typical standards expect < 25 Ohms for residential zones.
RCD's and similar testers use as test resistor to force an unbalance current using well-balanced common mode latching relay. The imbalance creates a differential current to trip at line voltages and would be in low mW levels.
Every region or country has its own standards for safety using RCD's and this also depends on the power category for PE grounded equipment.
PE is used to shunt noise (EMI or RF pulse coupling) in switching transformers which provide isolation for line frequency at high voltage. The interwinding capacitance leakage is generally small so the "Y" film caps only need to be say 100 times bigger to shunt the voltage to PE Gnd with the help of a high inductance BALUN or CM power choke.
There are many test methods for the presence of PE gnd and depends on the purpose of the test. For self testing I would measure AC voltage, then DC voltage then constant current diode test.
I imagine Tesla would do something similar with and measure the low frequency AC voltage drop after any other primary safety test. Measuring with high frequency could be problematic. I would have included an RCD-like functional test into the charger product that does not require human intervention.
Keep in mind RCD's and GFCI's use a test resistor to simulate leakage but actual force a latching relay using a balanced CM coil to detect the residual different in L1-L2 or L-N currents.
0 comment threads
0 comment threads