RCD tripping. Give me a clue!

[Piratepete]

Hi Guys
A Happy 2014 to you all.
I've recently had two calls to trace the cause of intermittent tripping in domestic installations, and failed miserably!

The latest - 4 way box feeding garage, utility room, garden shed. Customer reckons it sometimes trips when they switch the light on, but can trip without any human intervention.
RCD is fine. Also did a ramp test - 22mA Ok.
IR - 1.47 Meg for L+N of all circuits to Earth with everything plugged in. So it's a bit low but not anywhere near low enough to trip the RCD.
Two fridge freezers - I've been told that these are prone to high earth currents - measured under 1 mA.
Washing machine and tumble drier are usually turned off at socket. Doesn't trip while they're running.

Any bright ideas to track this fault, please?

Cheers!
Pete

 

[telectrix]

once you have eliminated the impossible, whatever remains, however improbable, must be the truth.

 

[Marvo]

1.4 Megs is a bit on the lowside and might not be a stable reading. It would probably at least warrant localising then make a decision if it's worth remedial work.

If you can't see a sufficiently low IR with a Mega tester then use an earth leakage clamp meter around the L&N simultaneously and monitor over time and under different load profiles.

 

[guest119]

22ma?
Did you ramp it with all outgoing disconnected to get the ramp reading for the RCD alone?

Boydy

 

[Octopus]

To properly investigate low IR you need to disconnect EVERTHING and IR all 3 cores.

 

[MDJ]

Could be a collective earth fault. Add up all the leakage from the appliances with a clamp meter and see just how many milliamps are leaking. You may have to split the circuits over different RCBOs/ RCDs

 

[Piratepete]

22ma?
Did you ramp it with all outgoing disconnected to get the ramp reading for the RCD alone?

Boydy

I don't know what you're getting at here. How will disconnecting everything affect the ramp test? The only thing switched on was a light.

 

[Piratepete]

To properly investigate low IR you need to disconnect EVERTHING and IR all 3 cores.

I would've thought that a general IR was adequate. Seperating the circuits will only result in increased IR readings - worst case will be one circuit at 1.47Mohm -This is not low enough to trip an RCD.

 

[Piratepete]

Could be a collective earth fault. Add up all the leakage from the appliances with a clamp meter and see just how many milliamps are leaking. You may have to split the circuits over different RCBOs/ RCDs

Good idea along with Marvo's suggestion,above. I don't currently possess a clampmeter that's sesitive enough. Still saving up for that torque driver that I need to do up CU screws correctly. Could use my inline meter to check overall earth leakage. All depends whether my customer wants to spend the money!

 

[Deleted member 72666]

Why do an IR with everything plugged in?

 

[guest119]

Why do an IR with everything plugged in?

Ditto ramp test!

Boydy

 

[Silly Sausage]

I don't know what you're getting at here. How will disconnecting everything affect the ramp test? The only thing switched on was a light.

You are testing the RCD only.
Testing with circuits connected could result in false readings due to leakage on the circuits.
At the very least, turn all the MCBs fed by the RCD off.

It's GN3 somewhere...I think!

 

[Piratepete]

Why do an IR with everything plugged in?

Because that is the situation when the RCD trips, which could be caused by an appliance. If I'd found an exceptionally low IR, I'd then unplug everything and retest, to see if an appliance was at fault. The fallacy with this is that some faults don't show up unless the appliance is actually running - washing m/c pump motors comes to mind.

 

[Marvo]

Good idea along with Marvo's suggestion,above. I don't currently possess a clampmeter that's sesitive enough. Still saving up for that torque driver that I need to do up CU screws correctly. Could use my inline meter to check overall earth leakage. All depends whether my customer wants to spend the money!

Forget the torque driver, an earth leakage clamp meter would be more important.

You can't use an inline ammeter to check leakage current because it will only tell you what's going down the CPC, not what's being lost down parallel paths. The RCD doesn't monitor the CPC current so the test would be meaningless.

 

[Piratepete]

You are testing the RCD only.
Testing with circuits connected could result in false readings due to leakage on the circuits.
At the very least, turn all the MCBs fed by the RCD off.

It's GN3 somewhere...I think!

Ok. Point taken, but if there is existing leakage, say 5 mA, and the ramp test shows 22 mA, then the tripping current for the RCD is 27 mA, I think.

 

[Marvo]

Yep, but if there happens to be 9mA leakage and your ramp test shows a pass at 22mA then you will leave a non-compliant RCDin service thinking it's okay.

 

[Piratepete]

Forget the torque driver, an earth leakage clamp meter would be more important.

You can't use an inline ammeter to check leakage current because it will only tell you what's going down the CPC, not what's being lost down parallel paths. The RCD doesn't monitor the CPC current so the test would be meaningless.

Good point! Can you recommend a suitable meter.

I accept that the RCD is looking at an imbalance between L and N. Isn't that difference also reflected in the main cpc?

 

[Marvo]

I think many of the UK guys seem to use a Dialog earth leakage clamp meter which I believe is very reasonably priced. I'm sure someone will chip in with the exact model number for you.

Not all fault currents flow down the CPC, for example if an appliance has an IR fault and happens to be in contact with the kitchen sink then the fault current or a good portion of it will flow to earth through the water pipework and/or the earth bonding.

 

[Guest111]

A common one is fridge and freezer compressors,these can test out OK but when they have run for a while and heat up they reveal a fault,that can be a real pain to find.

 

[BigTom83]

Do you not think 1.47 Mohms is a low reading?

Doesn't GN3 state 1Mohm is acceptable reading but anything less than 2Mohms requires further investigation.

Not sure why your testing with everything plugged in? Not sure what this achieves?

The chances are 1.47 Mohms is probably due to it reading through an appliance.

Remove the appliances, lamps, etc. and retest IR on each circuit, see if the readings improve (I guess they will).

If they do try PAT testing each appliance, if one looks dodge exclude it from circuit for few days see if that helps.

Trial & error only way forward if it's an appliance at fault

 

[Piratepete]

Yep, but if there happens to be 9mA leakage and your ramp test shows a pass at 22mA then you will leave a non-compliant RCDin service thinking it's okay.

You have to remember why I did the ramp test, which was to see if it was over sensitive rather than insensitive.
You've made me think though - I'm wondering why my assessors have never pulled me up for not switching off all unecessary circuits (eg.lights) when I do an Auto RCD test with my MFT. I tend to use the cooker circuit and switch off the ring(s) - doesn't do fridges any good! Is it because the standard auto test doesn't include a ramp test? When the tester leaks 30 mA does it take account of existing leakage?

 

[Marvo]

When the tester leaks 30 mA does it take account of existing leakage?

Nope, the tester isn't that clever. If it applies a 30mA leakage test the actual leakage seen by the RCD will be the test leakage plus any other leakage from circuits and appliances.

Ideally you should disconnect all circuits and all loads from the RCD when ramp testing it.

It's good you're beginning to think about the bigger picture, it might help if you draw some circuit sketches of the possible fault paths and testing procedures.

 

[Engineer54]

Good point! Can you recommend a suitable meter.

I accept that the RCD is looking at an imbalance between L and N. Isn't that difference also reflected in the main cpc?

Here you go.....


TM-EL9809 High Resolution AC Earth Leakage Clamp Meter

Both units are rebadged Di-log units, the above is an upgraded unit.
You'll get 5% discount on this model from the forum sponsor test meter too!!


Amprobe AC50A