Why did my oscilloscope hookup trip my RCD?

Question

I was attempting to investigate an SPI interface on a power line meter (typical model that measures Voltage, Frequency, Amps, Watts).

I opened the module and found the SPI pins broken out. So I plugged in the meter and the oscilloscope, and connected the oscilliscope probe to the CLK pin, and the probe ground to the GND pin.

A second or two after connecting the GND pin, the device IC caps blew off, and the RCD for the property blew and had to be reset.

What did I do wrong? How did this blow the RCD?

Answer 1

The oscilloscope's probe ground is connected to the earth (0V). It's likely that the "ground" of the power meter is not really ground. It's likely that the board's "ground" is actually at the neutral or line voltage, so it would create a circuit between neutral and ground or line and ground, which the RCD detected. But, without a board schematic, it's difficult to debug.

To debug the circuit, a differential voltage probe would be best. Otherwise, the probe ground could be connected to the earth prong in the power meter.

Answer 2

GND in the context of an electronic circuit usually reffers to the reference rail against which things in that circuit are measured. It may be floating and isolated, it may be tied to mains earth, or if you are unlucky it may be tied somewhere else. GND on external connectors will nearly always either be floating or tied to mains earth but once you get out your screwdriver and start connecting to ports that were only meant to be used for factory program/test/debug or connecting internal modules then all bets are off.

If you are building a power meter with no external data connections then it's generally easiest to tie your "circuit ground" to mains live. That way you just need a simple capacitor based transformerless power supply for powering the circuit (I expect thats what thosebig caps are for). A series resistor for measuring current and a resistive divider going off to the neutral for measuring voltage.

Even if I did have a data connection I'd be tempted to do it this way and then optoisolate the data connection.

Scopes usually tie the ground of their inputs to mains earth. You can get scopes with floating inputs but they are uncommon and expensive. You can also get isolated probes but again expensive.

Circuit ground tied to mains live, scope ground to mains earth, the result is a short circuit from mains earth to mains live. BANG.

I expect the way this device was debugged during developement was to feed it from a floating output isolating transformer. Once that was done the circuit ground could be connected to the scope without causing a short circuit.

Answer 3

Small AC-powered devices can use the live or neutral for GND (DC ground).

Your AC wiring probably has the live, neutral and protective earth (PE) wires. Neutral may not be at the same potential as PE (live obviously is not).

The RCD compares the current flowing in the live wire with the current flowing in the neutral wire (they should be identical), if they don't match (which means that some current is flowing into PE) the RCD will trip.

Your oscilloscope probably has the DC GND connected to PE through its power supply (there are also special fully-isolated scopes available). When you connect scope GND to GND of your meter, you probably connect PE to live or neutral, which trips the RCD.

Answer 4

The photo reveals that the device uses a transformerless power supply design - fairly obviously from the lack of a transformer, and additionally the readily recognisable yellow "X2" capacitors plus the group of components on the right-hand side of the top PCB which are representative of those typically used in that type of power supply. If the device doesn't offer any (non-isolated) external electrical connections to its internal circuitry, the "ground" rail of the power supply does not need to be at Earth/Neutral potential and it would appear that this device indeed uses such a "floating" ground rail.

As is the norm the oscilloscope input "ground" is connected to Earth and when you attached the "ground" of your oscilloscope probe to the "ground" on the device you were effectively short-circuiting that "floating" rail to Earth/Neutral.

Had you monitored your scope after you attached your probe and before you connected your ground lead you may well have noticed what a large voltage was present on the probe.

What isn't immediately clear without knowing the specific design of the power supply is why the ICs were damaged. Presumably this is not a case of the circuit "ground" simply being tied to the Active line. It would make an interesting exercise to trace out the power supply circuit and see exactly how your ground connection led to the outcome you experienced.

It is a clear and important lesson not to make assumptions about what a "GND" is! That applies in any circumstance, but is all the more important when this type of power supply is in play. While not an exhaustive check, some simple meter checks (continuity check while disconnected, and both AC and DC voltage while powered) are always worthwhile between a known ground and a "suspected" ground. And if you've gone ahead and connected your oscilloscope's probe anyway, it's also worthwhile to check that there's not anything unexpected going on there.