Series arcing causing MCB to trip

[Pretty Mouth]

I recently fixed a fault on a lighting circuit. The fault became apparent after some work I had carried out - I rewired part of the circuit for an ensuite refit, plus some additions to other circuits supplied by the same CU. I needed to provide RCD protection to the circuits that I had worked on, so changed the main switch in the CU for an RCD (I know RCBOs would have been better, but it's an old MK sentry board which only covers part of the installation, and parts are getting hard to come by), and also changed the mis-matched 10A MCB protecting the circuit for a 6A MCB.

The fault started a couple of days later, randomly tripping the MCB, but not the RCD. The clients had noticed that sometimes (but not always) it coincided with them switching lights on or off, but not one specific light. And it would come and go. A day or 2 no problem, then it would return, tripping the breaker every couple of hours. One day it got so that the breaker couldn't be reset, a good thing in a way as it makes fault finding easier.

Eventually I found the fault at a light switch - there was a break in one of the cores of the 4-way switching for the 1st floor landing light. The 2 sides of the broken conductor were just about touching, causing some mild arcing, but this had only slightly damaged the insulation . Repairing the broken core with a wago fixed the problem permanently.

I hadn't worked on this switch, so it must have been damaged for some time, and I suspect the 10A MCB may have been put there by a previous spark to stop the tripping without actually fixing the problem.

What I don't understand is why a fault like this caused the MCB to trip? The circuit, with all lights on, was pulling ~0.75A (measured before the fault got so bad that the MCB wouldn't reset). The break was in series with the load (LED lighting). So where was the extra current, enough to trip the MCB, coming from?

 

[telectrix]

broken conductor coupled with LED inrush, the inrush beig not just on switch on but the break making and breaking interrupting feed to LEDs???

 

[Pretty Mouth]

broken conductor coupled with LED inrush, the inrush beig not just on switch on but the break making and breaking interrupting feed to LEDs???

That would make sense @telectrix , thank you

 

[Lucien Nunes]

Agree with above. The running load current of the affected LED driver was being taken as a string of impulses each time the arc-gap at the faulty connection broke down. If these are greater than the magnetic trip threshold of the MCB, they could progressively accelerate it towards the tripping point even though it withstands one hit of inrush at switch-on. The reservoir capacitor in the LED driver would smooth out the impulses so that the light functioned normally. I would not be surprised if any internal X2 suppression capacitor within the driver now measured completely O/C.

Incandescent lamps offered good real-time feedback of the circuit conditions, in a way that LEDs don't due to the regulating effect of the driver. Highly inductive switched-mode circuits can sometimes still function with serious air-gaps present. One model of CRT monitor comes to mind that used to have dry joints on its S-correction coil in the line output. People would report them as producing smoke or burning smells and we would find a 1/4" gap in the PCB track from where the arcing dry joint had progressed and eaten the track away even while still working.

 

[Pretty Mouth]

Agree with above. The running load current of the affected LED driver was being taken as a string of impulses each time the arc-gap at the faulty connection broke down. If these are greater than the magnetic trip threshold of the MCB, they could progressively accelerate it towards the tripping point even though it withstands one hit of inrush at switch-on. The reservoir capacitor in the LED driver would smooth out the impulses so that the light functioned normally. I would not be surprised if any internal X2 suppression capacitor within the driver now measured completely O/C.

Incandescent lamps offered good real-time feedback of the circuit conditions, in a way that LEDs don't due to the regulating effect of the driver. Highly inductive switched-mode circuits can sometimes still function with serious air-gaps present. One model of CRT monitor comes to mind that used to have dry joints on its S-correction coil in the line output. People would report them as producing smoke or burning smells and we would find a 1/4" gap in the PCB track from where the arcing dry joint had progressed and eaten the track away even while still working.

Thanks for the info @Lucien Nunes . Pardon my ignorance: What purpose does a X2 suppression capacitor have in an LED driver?

 

[telectrix]

Thanks for the info @Lucien Nunes . Pardon my ignorance: What purpose does a X2 suppression capacitor have in an LED driver?

does it do twice as much or twice as fast as a x1? can't do smiley.

 

[Lucien Nunes]

Ah, that was something of a throwaway comment that might not even be relevant. SMPSUs including wall warts, LED drivers etc produce more or less RF interference that has to be filtered and contained to meet regulated limits. One part of that process is usually attenuating RFI that would otherwise be sent back down the supply lead and radiated from it. Hence the delta caps and feedthrough filters often found at the inlets to appliances. Small capacitors in the 1-100nF range can present a low enough reactance at radiatable frequencies that the outgoing connections are effectively commoned, but high enough not to pass significant 50Hz current.

Since suppressors are connected either across the supply or supply to earth they are subject to significant impulse voltages and can sustain destructive power arcs if those impulses cause even a tiny breakdown in the dielectric. In traditional film and foil capacitors such a breakdown will usually cause a short-circuit, which could create a shock hazard in certain applications or a fire hazard in others. Therefore a favoured technology for suppression duties is metallised film or metallised paper, in which the dielectric breakdown arc vapourises the metallisation and condenses it away from the breakdown, stopping the arc and restoring the insulation of the capacitor. They are referred to as self-healing types and classed as suitable for connection across the supply (X), or to earth or ELV (Y) which requires a higher specification. X2-rated caps are often used across L-N on 230V AC equipment and are designed to withstand 2.5kV impulses.

But such transients can occur, and breakdowns can be caused by fast-risetime impulses even when the peak voltage is not excessive. Each breakdown event destroys an area of metallisation so it is not uncommon to find capacitors that have lost much of their capacitance after repeated self-healing. My implication was that the persistent arcing of the faulty connection might have created enough tiny little breakdowns to consume the entire active area of the (probably cheap, underspecced) capacitor.

Where the capacitor's only function is RFI suppression, the user remains unaware and generally unaffected. In other situations, self healing caps can stop the equipment working, e.g. when used as series reactive droppers. When my parents' fridge freezer started icing up I found that the electromechanical defrost timer's motor dropper capacitor was an X2 type that had lost 2/3 of its capacitance (probably due to compressor switching transients) and the timer motor was sometimes stalling. I changed the cap for an identical one, and a few years later the same thing happened. Then I changed it to a 1000V rated polypropylene component and the F/F is still running happily 10 years later.

 

[Pretty Mouth]

Ah, that was something of a throwaway comment that might not even be relevant. SMPSUs including wall warts, LED drivers etc produce more or less RF interference that has to be filtered and contained to meet regulated limits. One part of that process is usually attenuating RFI that would otherwise be sent back down the supply lead and radiated from it. Hence the delta caps and feedthrough filters often found at the inlets to appliances. Small capacitors in the 1-100nF range can present a low enough reactance at radiatable frequencies that the outgoing connections are effectively commoned, but high enough not to pass significant 50Hz current.

Since suppressors are connected either across the supply or supply to earth they are subject to significant impulse voltages and can sustain destructive power arcs if those impulses cause even a tiny breakdown in the dielectric. In traditional film and foil capacitors such a breakdown will usually cause a short-circuit, which could create a shock hazard in certain applications or a fire hazard in others. Therefore a favoured technology for suppression duties is metallised film or metallised paper, in which the dielectric breakdown arc vapourises the metallisation and condenses it away from the breakdown, stopping the arc and restoring the insulation of the capacitor. They are referred to as self-healing types and classed as suitable for connection across the supply (X), or to earth or ELV (Y) which requires a higher specification. X2-rated caps are often used across L-N on 230V AC equipment and are designed to withstand 2.5kV impulses.

But such transients can occur, and breakdowns can be caused by fast-risetime impulses even when the peak voltage is not excessive. Each breakdown event destroys an area of metallisation so it is not uncommon to find capacitors that have lost much of their capacitance after repeated self-healing. My implication was that the persistent arcing of the faulty connection might have created enough tiny little breakdowns to consume the entire active area of the (probably cheap, underspecced) capacitor.

Where the capacitor's only function is RFI suppression, the user remains unaware and generally unaffected. In other situations, self healing caps can stop the equipment working, e.g. when used as series reactive droppers. When my parents' fridge freezer started icing up I found that the electromechanical defrost timer's motor dropper capacitor was an X2 type that had lost 2/3 of its capacitance (probably due to compressor switching transients) and the timer motor was sometimes stalling. I changed the cap for an identical one, and a few years later the same thing happened. Then I changed it to a 1000V rated polypropylene component and the F/F is still running happily 10 years later.

@Lucien Nunes , thanks for that brilliantly detailed reply. I had to read it a few times to get the head around it, but think I understand the principle now.

BTW I recently removed a light switch that may be of interest to you. Not all that old, 90's I'd guess, but unusual as it has lever connector terminals instead of screw terminals.