Earth leakage fault - consider using an whole lab insolation traffo - discuss

[MadScientist]

Hi, IN my home lab/office I have the usual test gear , small CNC mills etc

Annoying my house which i bought three years ago , occasionally tripos its RCBO for no apparent reason , typically flicking the RCBO on again ,restores the status quo

Tracking down the issue is exceeding difficult , the RCBO protects essentially two mains rings and is not on the lighting circuits ( The house was built in 1995 and is coded correctly for the time  ).  This year in 12 months I have three-four trips

I will replace the RCBO soon  and Im considering putting a separate one on each main ring to start the process of segregating the problem

In the mean time I have two choices , Isolate  the whole lab on a 230-230 isolating traffo, ( to eliminate leakage currents from the lab  ) or putting the computing resources on a UPS ( which is what I am doing first ) The UPS isnt really a solution long term , the long term solution is no trips !

The isolation traffo isnt a long term idea, but would at least eliminate over 12 months is anything in the lab generating earth leakage currents, but I have had the RBCO trips when nothing was powered on in the lab . I am aware of the common mode issues with isolation , but putting the whole lab on one would generally eliminate issues of inadvertent grounding

After that Im not sure how to track down this elusive fault , due to the infrequency of its occurrence

ideas ?


Thanks

Dave

[vk6zgo]

Some heavy domestic appliances have a bit too much leakage by design to be used on an RCBO, & in Australia are allowed to be driven from an unprotected source.
Maybe they aren't in your country-----where are you?

In this country, things like refrigerators, washing machines & dryers are all used from switched power sockets, & are used RCD/RCBO protected.

Hot water systems, stoves, etc are hardwired, & some of those may be in the group which will be marginal with an RCD/RCBO.

The problem could be a faulty RCBO though---- such things do fail, eventually, just as conventional circuit breakers do.

[madires]

An isolation transformer for the whole lab is a bad idea because you wouldn't notice a bad device besides other issues. I'd install a dedicated RCD (plus mains) for the lab, one (plus mains) for the data center, and also one for each mains ring (UK?). This way you would have only a partial blackout if one of the RCDs is triggered.

[MadScientist]

My next step is to consider putting the lab on a seperate RCBO , its not " too " awkward as , my lab is close enough to the main distribution board , but as always trying to run twin +earth wires through an existing decorated house is a PITA.

Im in ireland , and the practice here is as per the UK.  Heavy loads like Cookers etc are not on a RCBO , ( in 1995) , newer houses now have all house RCBOs however

Ill replace the RCBO as a first step , thats the easiest and a have a 500W UPS coming to at least protect the computers

its a bugger of a problem to track down

[MadScientist]

because you wouldn't notice a bad device besides other issues

There is nothing to stop me fitting a local RCBO to my safety transformer , which would detect local  leakages to the PE wire on equipment connected

[Ian.M]

An isolation transformer, will prevent *ANY* upstream RCBO tripping (short of primary insulation breakdown) and if you ground its secondary side neutral, you can use RCBO protection down stream of it as-if it was an independent supply from the utility co. or a generator etc.   Of course that removes the isolation 'safety' if you touch a single conductor, but IMHO 'safety' that doesn't trip anything when you touch one conductor then electrocutes you when you touch the other one is no safety at all.

There are considerable advantages in running an isolation transformer for your bench like this.  Stray leakage currents all return via the local ground-neutral bond on the secondary side, and you can be sure the neutral is near ground potential, and if you add a heavy duty mains rated  bidirectional TVS diode across the primary, the supply on the secondary side will be a lot cleaner.  Also , if you use separate secondary side RCBOs for bench lighting, test equipment, IT equipment and mains sockets used for devices under test, a D.U.T with a ground fault wont trip more than the one RCBO, wont cut power on stuff that should be shut down cleanly and wont cause a safety issue by killing the room and bench lighting.

To find the intermittent fault without removing the 'master' RCBO, you'll need to get a pair of RCDs/RCBOs with guaranteed discrimination, one with a slower time characteristic as the master and the other faster one you can put on the suspect circuit.  The other option would be to gut the old  RCBO for its balanced current sensing transformer, and use it to build an isolated residual current   sensor you can connect to a data logger, and put in series with both poles of each suspect circuit in turn.

If you are in N.I. you will probably fall foul of the building regs part 'P' competent person requirements, and a full reinspection would be needed on anything building wiring etc. you've worked on (and always work as if the most pissed off bureaucratic inspector possible has to pass the work).  Therefore this is a 'purely theoretical exercise' for any part of the above that cant be implemented by plugging in to an existing 13A socket,  and I do hope your posting history and profile are anonymous enough to protect your real identity and fine location! 

[Zero999]

An isolation transformer, will prevent *ANY* upstream RCBO tripping (short of primary insulation breakdown) and if you ground its secondary side neutral, you can use RCBO protection down stream of it as-if it was an independent supply from the utility co. or a generator etc.   Of course that removes the isolation 'safety' if you touch a single conductor, but IMHO 'safety' that doesn't trip anything when you touch one conductor then electrocutes you when you touch the other one is no safety at all.

It's actually perfectly safe and legal and is known as an IT earthing system. The most important thing is an insulation monitoring device must be fitted to give a warning if either side of the supply is connected to earth.

https://www.bender-uk.com/know-how/technology/it-system
http://www.studiecd.dk/cahiers_techniques/The_IT_system_earthing_in_LV.pdf
https://en.wikipedia.org/wiki/Insulation_monitoring_device

[ArthurDent]

The 'fault' may be leakage that is just slightly below the trigger point and occasionally cause a trip or it could be something like dust or hair in a dryer touching a live terminal. If it is some foreign matter touching a line I'd expect it to gradually get worse. If it is leakage caused by capacitive coupling in wiring or components I'd expect it to be somewhat consistent timewise between trips and somewhat random.

Have you tried unplugging each device in turn and using a multimeter on a high resistance range to measure from each line to the ground/earth at the plug? That might show the problem if it is consistent leakage although it might not because the multimeter uses a low voltage for resistance measurements and this could influence the testing.

Some time ago in the forums there was a case where a poster had a long run to a light using 3-way switches and the capacitance between the 3rd wire that was connected at just one end and the other wires cause the GFCI protector to trip.

[SeanB]

Simplest is to replace the breakers for the ring main circuits with combination RCD and breaker units, as these are common enough ( and also mandatory in the UK anyway for new builds) and this will mean all the stuff on a single ring has it's own breaker. You just have to find the neutrals of each ring and connect them to the requisite breaker, instead of them all being common to a single RCD unit.

[David Hess]

An isolation transformer for the whole lab is a bad idea because you wouldn't notice a bad device besides other issues. I'd install a dedicated RCD (plus mains) for the lab, one (plus mains) for the data center, and also one for each mains ring (UK?). This way you would have only a partial blackout if one of the RCDs is triggered.

If this was a consideration, then the neutral on the isolated side would be connected to ground, which is what happens at the distribution box, and an RCD could then be used.  So the isolated side is isolated from hot and neutral of the AC power line but not from ground on the isolated side.  Leakage goes between hot or neutral and ground so with the ground connected, the RCD will then work by detecting the current difference in only hot and neutral.

The only reason I would do this is to get a cleaner lab supply without all kinds of noise on the neutral from other circuits and it is particularly useful in an industrial environment with large motors.  But this would also work for a separate earth ground although that can be a problem if devices on the isolated side have connections to devices on the non-isolated side like between buildings.

[MadScientist]

Thanks so far guys , some stuff here to chew on , the idea of adding individual Rcbos to the rings is a good one , but some to ensure diversity in the Rcbos is an issue

[rhb]

I had a similar problem which I ultimately traced to a buried wire feeding a yard light.  It was *very* time consuming to sort out. The wire was proper wire for direct burial.  If you suspect a particular circuit, patch in a ground fault interrupter on that circuit alone.    Stick it in a handy box with enough wire so you can easily connect it up  to the appropriate circuit and mount it on the wall next to the panel.  Do a neat job so the wires from the handy box go into the wall and through a panel knockout in proper fashion.

I'm assuming you are experienced and comfortable working with mains wiring and understand about cable clamps, insulating bushes and such.   If not, please disregard.

[MadScientist]

I had a similar problem which I ultimately traced to a buried wire feeding a yard light.  It was *very* time consuming to sort out. The wire was proper wire for direct burial.  If you suspect a particular circuit, patch in a ground fault interrupter on that circuit alone.    Stick it in a handy box with enough wire so you can easily connect it up  to the appropriate circuit and mount it on the wall next to the panel.  Do a neat job so the wires from the handy box go into the wall and through a panel knockout in proper fashion.

I'm assuming you are experienced and comfortable working with mains wiring and understand about cable clamps, insulating bushes and such.   If not, please disregard.

Yes very comfortable woth mains wiring

The problem with series RCBO  is that you can’t be sure if the “test “ RCBO will trip instead of the House one

[jpb]

I had this fault on my present house. When we moved in we kept having the mains trip.
I found that this was because they had not changed the consumer box when they put a new kitchen in and so everything was going through 1 rcd. All electrical tests seemed fine but every few days the mains would trip.

I had the box changed for one with a separate rcd on each circuit and not had a problem since.
The issue is modern appliances often contain filters with capacitance to ground and the cumulative ground current is enough to exceed the 25mA limit or be very close to it. Very hard to trace because it is a different thing each time that causes the trip and it happens at random times.

[HB9EVI]

When I reconstructed our mains from the 70' here in the house, I first planned separate RCDs for each kitchen, washmachine and lighting, but in the end I installed a single three phase 30mA model for all circuits and I had no issues, though there are likely many devices with potential ground leakage over Y-caps.

So before setting up an isolation, I'd rather change the tripping RCD first