Reverse polarity in shower on a PIR?

[RF Lighting]

I've had RCBOs explode on pressing the test button.

 

[JohnW2]

I've had RCBOs explode on pressing the test button.

As I asked sparkticus, any thoughts about the mechanism of that?

I must look up how these testers actually test RCDs. On the face of it, all they should have to do is create a small L-N current imbalance (at, I would have thought, quite low absolute currents), and it's difficut to see how that could hurt it.

Kind Regards, John.

 

[sparkticus]

As I'm sure you realised, I was thinking more about undoing and re-doing terminations, flexing cables, removing and replacing accessories from back boxes etc. etc., as well, of course, as problems created by carelessness during testing (killing things with IR tests, forgetting to reconnect cables/bonding/whatever etc.etc.)

Oh absolutely I agree (I was just drawn to my RCD testing fear) I have had a couple of call backs regarding faults that I have introduced during a PIR!! I forgot to reconnect a live that I had taken out of an MCB and I unwittingly destroyed a control panel back illumination LED with a 500 volt IR test.

- but what you say is interesting.

I'm actually surprised that RCDs can be killed by 'standard testing'. In my (very limited) experience, I haven't seen that happen. Any idea of the mechanism?

Well, in all cases it was on the first test (the first test that I had performed on that particular RCD) so I suspect it was not the testing but that the RCD had failed sometime before the test but of course I then have to take some remedial action which can be a pain at times when you just want to get the PIR done. Actually with the exception of one which did trip (during an RCD test) but would not reset again. It was an MK 63amp 30mA RCD.

This comes back to a comment I made a while back about not having great faith in RCDs/RCBOs

Just consider yourself lucky that you are not expected to test MCBs!

Yes a water cooled test instrument would need to be developed

 

[JohnW2]

Oh absolutely I agree (I was just drawn to my RCD testing fear) I have had a couple of call backs regarding faults that I have introduced during a PIR!! I forgot to reconnect a live that I had taken out of an MCB and I unwittingly destroyed a control panel back illumination LED with a 500 volt IR test.

It's always refreshing and reassuring to see/hear confirmation that we're all human! I, too, am human, but am probably best advised to keep quite about how I have proved that in relation to matters electrical

Well, in all cases it was on the first test (the first test that I had performed on that particular RCD) so I suspect it was not the testing but that the RCD had failed sometime before the test but of course I then have to take some remedial action which can be a pain at times when you just want to get the PIR done.

Oh, you sound pretty innocent, then. I don't think that's murder - it's stumbling across a corpse!

Just consider yourself lucky that you are not expected to test MCBs!

Yes a water cooled test instrument would need to be developed

Indeed! Seriously, though, it's conceptually a bit strange that there isn't a requirement to test, particularly in situations in which one is reliant on the MCB for L-E fault protection. I guess it's just the (im)practicalities that have spared you that.

Kind Regards, John.

 

[securespark]

I've had RCBOs explode on pressing the test button.

Yay!

I had an old Square D one do that...I sh@t me sen.....

 

[JohnW2]

I've had RCBOs explode on pressing the test button.

Yay! I had an old Square D one do that...I sh@t me sen.....

Same question for you, then - any thoughts about mechanism?

I can but presume that these 'exploding RCDs/RCBOs' must relate to relatively high currents being disconnected when the device operates, with the problem relating to the contacts (e.g.arcing) - since it;s all but impossible to see how the trip mechanism itself could explode.

Kind Regards, John.

 

[bernardgreen]

I'm actually surprised that RCDs can be killed by 'standard testing'. In my (very limited) experience, I haven't seen that happen. Any idea of the mechanism?.

The "common" RCD would need a very large un-balance in the order of many tens of amps before there could be enough energy in the sensor's output to damage the solenoid coil that releases the switch to the OFF position.

It might be that a large DC un-balance could leave the sensor's magnetic core ( of the current transformer ) permanently magnetised and that would then affect its operation.

The "intelligent" RCDs that have electronic circuitry would be far more prone to damage from voltage spikes on the mains.

 

[sparkticus]

The April edition of the Professional Electrician shows a picture of an RCBO that "blew up" when the test button was pressed. Interestingly the picture shows virtually no damage to the ARC chamber but the contact/coil area is devastated. I guess it's is possible that the ARC chamber was not damaged because it did what it is designed to do? (I don't know) The electrician who sent the picture in comments: "On doing so it immediately blew up in my face burning my thumb and the hairs off my hand. If I had not been wearing glasses things could have been a lot worse as the flames hit them full on"

No indication of load conditions at the time of pressing the test button. I have to say, looking at the picture this looks like an internal short. It is hard to imagine a disconnecting load (in a domestic environment) causing this amount of damage.

I remember about two years ago I fitted a new Wylex RCD. When I turned on the power I could immediately smell something burning so I switched off and investigated. The RCD was warm (it had never had a load applied since it was in the off position when I turned power on) I switched power back on, placed the RCD into the on position (no load) and pressed the test button. It worked fine (disconnected) but it failed to disconnect at 1x & 5x test currents!! I returned it under warranty and asked to have the fault investigated but no response to date - I have stopped holding my breath now

 

[JohnW2]

The April edition of the Professional Electrician shows a picture of an RCBO that "blew up" when the test button was pressed. Interestingly the picture shows virtually no damage to the ARC chamber but the contact/coil area is devastated. I guess it's is possible that the ARC chamber was not damaged because it did what it is designed to do? (I don't know) The electrician who sent the picture in comments: "On doing so it immediately blew up in my face burning my thumb and the hairs off my hand. If I had not been wearing glasses things could have been a lot worse as the flames hit them full on"

That's frightening - but we've obiously slid away from the matter of killing RCDs/RCBOs by testing them, since you are clearly talking about a dangerously malfunctioing unit here.

Maybe I'm wrong, but I still find it hard to believe that testing, per se, can do any harm to these devices, since the testing is (I presume) doing no more than emulating a (I guess fairly modest) fault situation.

Kind Regards, John.

 

[JohnW2]

I'm actually surprised that RCDs can be killed by 'standard testing'. In my (very limited) experience, I haven't seen that happen. Any idea of the mechanism?.

The "common" RCD would need a very large un-balance in the order of many tens of amps before there could be enough energy in the sensor's output to damage the solenoid coil that releases the switch to the OFF position. It might be that a large DC un-balance could leave the sensor's magnetic core ( of the current transformer ) permanently magnetised and that would then affect its operation. The "intelligent" RCDs that have electronic circuitry would be far more prone to damage from voltage spikes on the mains.

All good theoretical points, but I find it very hard to believe that 'standard testing', per se, could possibly invoke any of those scenarios.

Kind Regards, John.

 

[sparkticus]

Maybe I'm wrong, but I still find it hard to believe that testing, per se, can do any harm to these devices, since the testing is (I presume) doing no more than emulating a (I guess fairly modest) fault situation.

No I don't think you are wrong. An RCD test instrument (especially a calibrated one) should, within tolerances, initiate only fault currents of 15mA, 30mA and 150mA (assuming a 30mA RCD)

I don't know what the integral RCD test button initiates in terms of test current but I can't imagine it is more than 150mA and quite possibly 30mA.

As bernardgreen indicates, the electronic based RCDs can (presumably) be as susceptible to transients as any other electronic system (which does not fill me with confidence but it is realistic) The non-electronic RCDs are quite simple in design though their potential for mechanical failure and permanently altered core characteristics is ever present I suppose.

I would hope that all these failures (albeit the evidence for such being somewhat anecdotal at the moment ) are properly analysed by the manufacturers.

 

[sparkticus]

This is interesting though it does not discuss RCD failure on/after remote testing. I assume that the manufacturers do use calibrated test currents to test all new production RCDs and perhaps to test the long term effects of frequent and high current (large differential current) on RCDs.

If that is true, and I would be amazed if they did not then perhaps RCD failure as a direct result of in-service-testing is something that is largely unreported/does not happen often.


http://www.era.co.uk/news/pr0709.asp

 

[JohnW2]

No I don't think you are wrong. An RCD test instrument (especially a calibrated one) should, within tolerances, initiate only fault currents of 15mA, 30mA and 150mA (assuming a 30mA RCD). I don't know what the integral RCD test button initiates in terms of test current but I can't imagine it is more than 150mA and quite possibly 30mA.

Exactly, and if any of that could cause an RCD to fail (let alone 'explode'), they would surely be dangerously unfit for purpose.

You quite often talk of your concerns about the reliability of RCDs but, as I hinted at recently, what about MCBs - might they not be at least as unreliable? For what are presumably just practical reasons, we don't test them routinely, and nor are they regularly 'exercised' by a test button (failure to do which is said to be a reason for RCD malfunction).

I suppose that the problem for the manufacturers of MCBs (and the Standards) are that I don't think there is any way of undertaking realistic accellerated long-term reliability tests. In other words, if one wants to determine whether MCBs still operate reliably after 20+ years of never operating, one has to set up the experiment an then wait 20+ years for the answer.

Kind Regards, John.

 

[sparkticus]

Exactly, and if any of that could cause an RCD to fail (let alone 'explode'), they would surely be dangerously unfit for purpose.

Though (I am happy to report) I have not seen an exploding RCBO myself, I have heard of several (probably 6 - 7) including the two mentioned in this thread. My fear re these exploding RCBOs is that it is not the test current but a short between live and neutral! I have two completely unsubstantiated reasons for saying that.

1) I just can not fathom how a test current would exhibit that much energy unless the integral test current limiting device (resister for example) went short circuit which is as likely as being seriously injured by a chocolate hammer.

2) Again, only a shot in the dark, these single channel RCBOs do not have much room in them yet there is a substantial CCC live and neutral in that confined space and frankly the build quality of some of these "modern" devices is far from precision!!

You quite often talk of your concerns about the reliability of RCDs but, as I hinted at recently, what about MCBs - might they not be at least as unreliable? For what are presumably just practical reasons, we don't test them routinely, and nor are they regularly 'exercised' by a test button (failure to do which is said to be a reason for RCD malfunction).

Yes, I tend to agree. In fact, on one hand because of the substantially higher currents (and of course variable currents depending upon rating) required to test disconnection of an MCB, the test may even lead to more failures and/or discovery of more failed devices than that offered by the RCD/RCBO world. On the other hand, the relatively huge motive-force of an MCB test current is likely to do something over and above that of an RCD test current - even if the MCB fails to disconnect there will probably be a noticeable event So I now conclude that we will need not only a water cooled instrument but also significantly longer test leads (and sand bags)

Now of course I am not in anyway suggesting or advocating field testing of MCBs, that would be far too exciting!! Though I have done some "home testing" of my own and came up with variable results. What I did conclude on a 6 amp and 20 amp MCB (quite some time ago) was that they pretty much perform to the published curves for In and I2. I did not test breaking capacity yet

I suppose that the problem for the manufacturers of MCBs (and the Standards) are that I don't think there is any way of undertaking realistic accellerated long-term reliability tests. In other words, if one wants to determine whether MCBs still operate reliably after 20+ years of never operating, one has to set up the experiment an then wait 20+ years for the answer.

Exactly, I'm sure simulations and accelerated testing are done but as I learned from past semiconductor test experience, you can't get 20 years of experience in anything less than 20 years.

 

[JohnW2]

Though (I am happy to report) I have not seen an exploding RCBO myself, I have heard of several (probably 6 - 7) including the two mentioned in this thread. My fear re these exploding RCBOs is that it is not the test current but a short between live and neutral! I have two completely unsubstantiated reasons for saying that...

Maybe 'unsubstantiated', but your first one is compelling. There is no way that the normal test current (via test button or a tester) could cause an explosion. As you say, only a L-N 'event' within the device could really bring that about.

Yes, I tend to agree. In fact, on one hand because of the substantially higher currents (and of course variable currents depending upon rating) required to test disconnection of an MCB, the test may even lead to more failures and/or discovery of more failed devices than that offered by the RCD/RCBO world.

Exactly.

Now of course I am not in anyway suggesting or advocating field testing of MCBs, that would be far too exciting!! Though I have done some "home testing" of my own and came up with variable results. What I did conclude on a 6 amp and 20 amp MCB (quite some time ago) was that they pretty much perform to the published curves for In and I2.

I suspect you mean I1 and I2 - since In is really just part of the specification, not something that can be measured. I, too, have done some bench tests in the distant past, using ELV and, like you, satisfied myself that the published I2 figures seem about right.

Exactly, I'm sure simulations and accelerated testing are done but as I learned from past semiconductor test experience, you can't get 20 years of experience in anything less than 20 years.

Quite so. In some situations, accelerated testing using extremes of environmental conditions (and/or cycling thereof) can be of value - but in the context we're talking about, the interest is in how an MCB performs after 20+ years of 'doing nothing' under normal environmental conditions - and there really is no way of accelerating that. The big problem that creates in this world of fairly rapidly evolving technology is that we often don't have hard long-term reliability data until after a type of product has become obsolete!

Kind Regards, John.