RCD testing electronics

[fcb]

I'm designing an RCD tester into a piece of kit I'm working on.  I thought I'd see if anyone here has either designed one before (and is willing to share details), or has the schematics of a commercial unit.  Here's my issue:

To test an 30mA RCD I need to apply the following currents and starting phase positions.

15mA (half trip) 0 degrees start (must not trip in 2s)
15mA (half trip) 180 degrees start (must not trip 2s)
30mA (trip) 0 degrees start (must trip <300ms or <2s for Selective type RCD)
30mA (trip) 180 degrees start (must trip (<300ms or <2s for Selective type RCD)
150mA (5x trip) 0 degrees start (must trip in <40ms or <150ms for Selective type RCD )
150mA (5x trip) 180 degrees start (must trip in < 40ms or <150ms for Selective type RCD)

The start phase will be easy to control (micro-controller) and I'm thinking of implementing the simulated earth leak current using a constant current source (with the reference fed from a divided down version of the input voltage, probably use a digital pot - so the current tracks linearly with the input voltage) - this should look like a resistor to the input supply.

The rules on testing look like the test shouldn't ever run to more than 2 second at 1x current (14.4J, assuming 240V x 0.03A x 2s) and 5.4J at 5x current (240V x 0.15A x 0.15s), so I expect I'll have to use a fairly large power transistor with a small heatsink inside the unit.  Perhaps something like the BUL38D or BUX85G.  From memory, the current has to be +/- 5%.

I'm a little concerned that I'm overthinking/engineering on this thing, or missing some industry standard way of doing this.

[SeanB]

Just use some 2% 5W resistors, and a set of small triacs to switch them as required.  Much easier to do the zero cross switching, and you can easily use MOC3023 alone to do the switching, just use resistors to give 15ma, 15mA and 120mA, so you use 1, 2 or all 3 to do the testing.  Overdriving it a bit on the last test, but it should survive, or just use 2 60mA sink resistors. For safety use a 70C thermal fuse sandwiched between the 4/5 resistors to provide thermal protection. 4 or 5 resistors, as you will need one driving another simple opto transistor to provide a zero cross interrupt for your microcontroller that is doing the display and such work.  Powerisolation so that a simpler interface is available, or so you can have USB connection and a local lithium cell for power.

[fcb]

Thanks Sean,

I'd like to be more precise about the current, as the mains voltage in these applications varies by more than +/- 10%, and need to hit a current accuracy of +/-5%.

See attached PDF, R3 would be replaced by some sort of variable circuit (opamp buffer feeding a digital pot), and the transistors and opamp part numbers are all wrong.

[SeanB]

Well, you do not need the -12V rail, and can use a LM324 as opamp. Make R4 so it has the near full dissipation at 150mA, keeps the transistor Vce low at high current, and shifts the dissipation to a component that can handle it better. Still not happy with current flow to ground pin, I would be much happier with a separate DPST relay that isolates the RCD test circuit totally until needed, just in case you have a failed earth connection, which would float the entire case of the device to mains potential.

How are you doing correct connection testing, you need to check for either missing or miswires on all leads, so you can get all the mistakes, and which can find the most common faults. A plug in RCD tester is quite common, it will test the RCD at various currents, and will also do checking using 3 neons for all faults. The one I have also has a floating earth detection using a small LCD display with a warning triangle, using the tiny current that otherwise flows via a correct earth to flow via your body capacitance  and a blocking capacitor and resistor to tell you the earth line is at a voltage over 30VAC above true ground.

[ajb]

Sounds like you want the current waveform to be sinusoidal, right?  How are the 180deg tests meant to work?  Is it really meant to test starting on the positive and negative half-cycles?  The shortest 'must trip' time is 2+ full cycles, so surely that doesn't matter, does it?  Or is it supposed to be 90deg, meaning a sine wave shifted by a quarter cycle? 

If it's the latter, then using the divided-down line voltage as the reference to a current source isn't going to work for those tests; you'll need to come up with another way of producing a sine wave that is synchronized to but out of phase with the line voltage.    Also note that if it's the current that's specified and not the equivalent resistance, then you don't need to current to be proportional to the line voltage at all, because that would imply that the current would vary as the line voltage does.  You want the (RMS?) current to be fixed, regardless of line voltage, and you want the current waveform to be synchronized to the line in terms of frequency and phase.

There are certainly ways of doing that in, but it's also tempting to just use an MCU to generate a sine wave that's synchronized with the line voltage waveform with the appropriate phasing, if you want to be really precise about it. 

Regarding the earth current, if you have access to both upstream and downstream of the RCD, then you don't need to run the test current into the equipment ground.  Instead you can run the current from the downstream line to the upstream neutral, or vice versa. 

[fcb]

Well, you do not need the -12V rail, and can use a LM324 as opamp. Make R4 so it has the near full dissipation at 150mA, keeps the transistor Vce low at high current, and shifts the dissipation to a component that can handle it better. Still not happy with current flow to ground pin, I would be much happier with a separate DPST relay that isolates the RCD test circuit totally until needed, just in case you have a failed earth connection, which would float the entire case of the device to mains potential.

How are you doing correct connection testing, you need to check for either missing or miswires on all leads, so you can get all the mistakes, and which can find the most common faults. A plug in RCD tester is quite common, it will test the RCD at various currents, and will also do checking using 3 neons for all faults. The one I have also has a floating earth detection using a small LCD display with a warning triangle, using the tiny current that otherwise flows via a correct earth to flow via your body capacitance  and a blocking capacitor and resistor to tell you the earth line is at a voltage over 30VAC above true ground.

Hi Sean,

The schematic published is just a quick test in LTSpice - the final opamp/transistor choice is away off, as is the PSU choice. I agree with your point regarding R4.

As there is allsorts of extra test functions in the unit (checking correct wiring, monitoring voltages, phases, N-PE resistance etc..), there will be plenty of chance for isolation relays. I would hope that no one builds the circuit as is and expects it to work.

[fcb]

Sounds like you want the current waveform to be sinusoidal, right?  How are the 180deg tests meant to work?  Is it really meant to test starting on the positive and negative half-cycles?  The shortest 'must trip' time is 2+ full cycles, so surely that doesn't matter, does it?  Or is it supposed to be 90deg, meaning a sine wave shifted by a quarter cycle? 

If it's the latter, then using the divided-down line voltage as the reference to a current source isn't going to work for those tests; you'll need to come up with another way of producing a sine wave that is synchronized to but out of phase with the line voltage.    Also note that if it's the current that's specified and not the equivalent resistance, then you don't need to current to be proportional to the line voltage at all, because that would imply that the current would vary as the line voltage does.  You want the (RMS?) current to be fixed, regardless of line voltage, and you want the current waveform to be synchronized to the line in terms of frequency and phase.

There are certainly ways of doing that in, but it's also tempting to just use an MCU to generate a sine wave that's synchronized with the line voltage waveform with the appropriate phasing, if you want to be really precise about it. 

Regarding the earth current, if you have access to both upstream and downstream of the RCD, then you don't need to run the test current into the equipment ground.  Instead you can run the current from the downstream line to the upstream neutral, or vice versa.

Thanks for your reply AJB,

Not really worried about the current being sinusoidal, as the mains itself might not be truly sinusoidal (in fact it rarely is).  A micro-controller and some sort of switch circuitry on the around the opamp +IN will handle 0/180 syncing.

I agree it's tempting to use a micro to generate the waveform driving the constant current circuit - i'll have to work out how much distortion of the waveform can be coped with before I'm out of my +/-5% tolerance.

The existing scheme would have the micro-controller calculating RMS and then steering the DAC/digital potentiometer (R2/R3 potential divider) to give the required RMS constant current.

Also, only have access to the outputs of the RCD, no downstream access.