Gypsy Caravan

[SimonH2]

That's theory. In practice when a worker put a nail in the wall of a porta cabin to hang his coat on and it hit the cable
the 30ma trip went out in cabin.
the 100ma type S went out is distribution board.
the settable RCD set at 500ma and 30 seconds delay went out in the companies main distribution board.
the settable RCD set at 1A and 60 seconds delay went out in the supplies distribution board.

N-E fault ?
Not disconnected by any single pole RCDs.

 

[JohnW2]

N-E fault ? Not disconnected by any single pole RCDs.

It sounded very much (to me) as if eric was talking about "RCDs" in the usual sense (what some people call RCCBs, but I'm not sure that's really any less 'general') - in which case I have to say that I don't think I've ever seen a single-pole one. The only SP RCDevices of which I'm aware are RCBOs.

Kind Regards, John

 

[JohnW2]

That's theory. In practice when a worker put a nail in the wall of a porta cabin to hang his coat on and it hit the cable
the 30ma trip went out in cabin.
the 100ma type S went out is distribution board.
the settable RCD set at 500ma and 30 seconds delay went out in the companies main distribution board.
the settable RCD set at 1A and 60 seconds delay went out in the supplies distribution board.

As Simon has just written, that could surely only happen if it were a N-E fault and both the '30mA trip' and the '100mA Type S' were both single-pole devices - something I've never heard of (unless they were RCBOs)? ...

... if a DP downstream 30mA RCD operated, a properly functioning upstream Type S shouldn't operate (that's why we use Type S ones). If either the 30mA or 100mA operated and was DP, there's no way that upstream RCDs with 30 or 60 second delays could operate unless they were seriously faulty (delay not working). ... or am I missing something?

Kind Regards, John

 

[Lectrician]

6mm 2 Core SWA (3 core if you wish, but 2 core would be cheaper).

32amp MCB at origin (will discriminate for overload with 16amp).

16amp RCBO hookup at caravan with rod, isolating the PME sheath.

The Rolec hookups are cheaper than some brands of RCBO on their own.

2.5mm Flex looping around sockets in van, directly to a blue plug into the hookup. It's not really a caravan, so no need for DP iso etc in the van.

 

[SimonH2]

Or, the fault was such as to overload the devices and result in them tripping after the fault was cleared (poor design) or damage them (poor design).

 

[ericmark]

The RCD were all all pole switching.
First was in the consumer unit double pole.
Second was in a small distribution unit 4 pole.
3rd and 4th were current transformers connected to a control board with moulded breakers so like a RCBO the overload current needed would have been around 300A and 500A so unlikely they tripped on overload.
We had trips open like this many times and it was really a problem having a concrete pump trip out because of a fault in a cabin. OK this was back in 1992 and trips may have improved now but in 2004 in Heathrow Airport terminal 5 we had a 10ma RCD socket on our bench and pressing the test button would trip the 100ma in the main board.
Theory it should not happen in practice it does. In my own house two 1992 RCD's alongside each other feeding two old consumer units original re-wireable type and if I don't switch off all the MCB's before resetting the RCD often on resetting it will trip the one next to it. Theory they are not connected to each other practice the spike caused resetting will trip the other one. As will electric storms.

 

[JohnW2]

The RCD were all all pole switching.
First was in the consumer unit double pole.
Second was in a small distribution unit 4 pole.
3rd and 4th were current transformers connected to a control board with moulded breakers so like a RCBO the overload current needed would have been around 300A and 500A so unlikely they tripped on overload.

I really don't understand, then, unless there were multiple faulty devices. If the first (30mA) RCD tripped in the required time, the second (Type S) one would not have tripped (as a result of the fault) if it were working within spec. More to the point, if either or both of those first two resulted in all-pole disconnection (necessarily within well under 1 second) it makes no sense that upstream devices with 30 and 60 second delays should operate (as a result of the fault) unless they were seriously faulty - from what you describe, it sounds as if these (long) time-delays simply didn't work. Despite what you say, the most credible explanation would seem to be that the two upstream ones tripped on overload, not L-N imbalance - was the overload function not time-delayed?

... we had a 10ma RCD socket on our bench and pressing the test button would trip the 100ma in the main board.

That's perhaps not quite so surprising (dependent upon what test current the button was creating). The problem with the the scenario you describe above is that none of the time-delays appeared toi work as intended.

Kind Regards, John

 

[ericmark]

I would agree John it should not happen but it did. It was not the only surprise we had. For me it was first job using so many RCD's and single pole switching was a major problem. To take out two then OK but 4 seemed strange specially when set for such a long delay. I have a feeling that with a leakage of over a set limit the delay was overridden. But was 1992 so can't remember all details. However this tripping of multi RCDs was common with a major fault.

In the main the test button goes line to neutral and not a clue why the 10ma trip took out the 100ma one but it did every time. However the 30ma trips of same make did not. To comply with non auto restart of items like grinders we had used both RCD FCU and RCD sockets active type cheaper than using proper contactors and these when tripped did not trip the 100ma trips.

 

[JohnW2]

I would agree John it should not happen but it did.

I obviously can't argue with that. However, there is obviously something very wrong if an RCD trips 29+ or 59+ seconds earlier than it should. However, the answer may well be ...

I have a feeling that with a leakage of over a set limit the delay was overridden.

... which could, of course, mean that the behaviour you saw was actually 'correct', if the upstream RCDs were no longer behaving as time-delayed ones.

In the main the test button goes line to neutral ....

In the absence of an earth connection, that's really all they can do - introduce a path from L on one side of the RCD to N on the other side, thereby creating a current imbalance (regardless of load).

Kind Regards, John

 

[Martynrosscampbell]

Thanks Lectrician,

A good straight answer from a fellow Devonian!

Thanks everyone else as well.

It may seem bad to you that I don't know some of this stuff but at least I'm here asking advice rather going for the 'that'll do' approach. Everyday's a school day and all that!