Test equipment to identify cause of RCD trips

[Caecilius]

The RCD in my split-load CU trips out occasionally, something like every other month. No obvious cause, and it generally happens at night. I want to identify the cause, or at least narrow it down, but the random nature plus the low frequency is making this quite difficult.

I've got a multimeter, and have made up some "leakage plugs" with resistors from live to earth, but I think I need some more specialist test equipment, specifically an insulation tester to check each of the RCD-protected circuits and an RCD tester to check the RCD device (and test the new one if it needs replacing).

This equipment is quite expensive, and it's not something I'm likely to use often. Does anyone know if it's possible to hire this sort of equipment for a reasonable cost? Any pointers to good hire shops that might carry it?

Alternatively, are there any sources of cheap test equipment for this sort of thing that would be suitable for light DIY use, or are these things always several hundred pounds?

I've got a split-load CU with an 80A / 30mA RCD protecting all the power circuits. The lights and immersion heater are not RCD protected.

[John Rumm]

It is available for hire. Or alternatively buy something suitable on ebay and then flog it again after.

They are usually fairly pricey. Individual testers on ebay might be had for £50 though.

Where abouts in the country are you?

Does your CU take single module RCBOs? You could move once circuit to the non RCD side on a RCBO and see if the trip moves, stays, or vanishes.

[Caecilius]

Medway towns in Kent.

I'm not sure, but it takes standard MCBs. It's a Tenby C Series CU.

There are no free slots left in the CU, but I guess I could get another neutral bar and change the split to give one more non-RCD slot at the expense of an RCD-protected slot.

[John Rumm]

Or change the RCD to an switch and use three RCBOs in place of the MCBs? (or just do away with the split an used RCBOs where required).

[Caecilius]

I'm not keen on changing the RCD for a switch to bridge the two sides, as I think it would be non-standard and may look a bit strange. I wouldn't want it to look like a bodge.

Doing away with the split and going to RCBOs instead of MCBs might work. I thought it would be too expensive, but it seems RCBOs have come down in price since I last looked a couple of years ago, and they now seem to be around GBP 35 each. I need five, so around GBP 175 plus the cost of a new bus bar.

My tests with leakage plugs indicate that the RCD always trips at 16mA (15K resistor) on every circuit. Some circuits will trip at 11mA (22K resistor), probably due to a few mA leakage from the appliances. So I suspect that the RCD is tripping at quite a low leakage current and that plus the number of circuits on the RCD mean that I don't have a lot of headroom left.

So RCBOs may well cure the problem, and at the worst they would indicate which circuit needed further investigation. I'll need to check if my CU will take single-way RCBOs though.

[ARWadsworth]

I am not sure that you will get very accurate results that way. What is the voltage at your CU (it looks like you have assumed 240V) and what tolerance is there on the resistors? +/-5% is quite normal for resistors.

Cheers

Adam

[Caecilius]

Yes, the resistors are 5% tolerance (gold band), and as you suspected I'm assuming the supply voltage is 240V. I've made up three leakage plugs: 47K (5mA @240V), 22K (11mA @240V) and 15K (16mA @240V).

There's bound to be some error, but I don't think it's enough to be of concern, as I'm only trying to find out roughly what leakage currents will trip the RCD. I'm not really bothered by a 10% error (between 1 and 2 mA with the quantities I'm dealing with).

I guess I could measure the mains voltage and resistance values with my multimeter, but I don't know the tolerance of that so I'm not sure if I'd be any better off.

[js.b1]

1 - Dump the resisors.

2 - Visual inspection of accessible every wiring accessories. Switches, sockets, switch fused connection units, ceiling rose, immersion terminals & isolators, shower terminals, anything outside, consumer unit terminals etc.

Realise a bad connection in a jn-box under a floor board, now getting warm, can melt insulation enough for a N-E short. This is why I prefer jn-box if used to be under the landing stair area (can't block with furniture, easy to lift carpet or laminate).

3 - Disconnect all but necessary. Unplug everything. Turn everything off. See if the problem re-occurs.

4 - Remove the RCD, fit individual RCBO and use a longer busbar That is to say, convert split-load into non-split load, buying RCBO for as many circuits as necessary. This way you pin the problem down to a particular circuit AND improve electrical discrimination re current 314.1 17th regulation.

If you baulk at the cost of 2-3-4 RCBO and busbar (=A34) consider...

- Insulation Resistance tester is =A3100-120 on Ebay and may not find the problem

- Spark will charge =A380/hr+ for as much as 10hrs because intermittent faults are hard to track down

IR tester may not find the fault because the short may be due to vibration, moisture, heating (load) and when it occurs there is sufficient current to carbonise the problem away. For example, consider 60yr old Tough Rubber Sheathed (TRS) cable rotting at the ends such that the rubber is at the bottom of the accessory box. The IR test results were perfect, because air is a perfect insulator :-) That is why IR tests are best done after visual.

So do the visual check first, look for N-E problems such as a wiring accessory screw (earthed) trapping a neutral wire.

[NT]

Sounds like you need a =A31 megger:

you dont have a megger, its still worth checking appliance insulation with a multimeter, often this will pick up the offender, but not consistently.

Do appreciate the safety issues with using meggers.

NT

[Ron Lowe]

Depends how D-I-Y you want to get...

I faced a similar problem, and built a D-I-Y leakage meter.

I took an old 2-pole RCD ( Merlin Gerin, as it happened ), and carefully opened it.

I removed all the mechanical switching parts, leaving only the current transformer with it's 2 heavy windings passing L and N, and the sense coil which I wired to a pair of external jacks.

A 10k resistor was placed as a load across the sense coil, to provide a voltage output rather than a current output. This is important, otherwise the open-cct output voltage from a current transformer can posentially get hazardously high.

I then performed a calibration with a 12vAC supply, some test resistors, and a pair of flukes set up to measure mV out versus mA leakage. I calibrated in 1mA steps from 1 to 10mA, then in 5mA steps up to 50mA. The calibration was outstandingly linear, and I printed up a table of mV to mA. I got 45mV / mA.

Now, by wiring this in-circuit with the troublesome circuit ( both L and N, obviously ) I was able to make actual measurements and isolated several faults.

I plan to write this up with photos in due course.