Induced voltage in lighting

[timhoward]

Yesterday I had a call out to a 3 way switched top landing light which had a bayonet CFL bulb in it. Turn light off at either switch and lamp continued to flash every second or so. The wiring looked ok so I assumed (maybe too soon) an induced current issue but then a family issue meant I had to leave so I wago'd the feed and agreed I'd return today.
Of course today when reconnected it's not doing it, and turning an assortment of other lights and appliances on hasn't recreated the scenario. I'm measuring 4v AC at the fitting when the switches are off. Also noticed 13v AC on another unrelated BC fitting near it.

I've read various previous forum posts about similar happening on both LED lighting and CFL bulbs. The holy-grail post that many of them refer to is now archived and not accessible. I believe I've ruled out any faults in the switches themselves. It's definitely switched on the live side. All cables IR test ok.

Main question - what sort of induced voltage typically starts to trigger this problem? In other words am I barking up the wrong tree entirely?
Might it simply have been be a duff/old bulb with ballast deteriorating in combination with a slightly higher induced load?
Any other thoughts appreciated as I'm scratching head a bit with this one.

 

[Soi disant]

The holy-grail post that many of them refer to is now archived and not accessible.

Have you tried putting the URL of that post into the Wayback Machine?

 

[newfutile]

It’s possible especially with all that 2 way switching cable , did you IR the line and line return as this is where the induced voltage will happen.
the minimum voltage will depend on the lamp, some of the better ones have components to stop this behaviour.

 

[timhoward]

It’s possible especially with all that 2 way switching cable , did you IR the line and line return as this is where the induced voltage will happen.

Yes, all conductors IR tested. No issues found.

the minimum voltage will depend on the lamp, some of the better ones have components to stop this behaviour.

Thank you. As this was an elderly looking CFL bulb this could have been a factor. We'll see if the issue returns.

 

[Lucien Nunes]

Bear in mind that it is not strictly just the voltage that matters. What is happening is that the capacitive reactance ffrom permanent line to switched line is leaking a small and nearly constant current into the switched line, and a different capacitive reactance is leaking current out from switched line to neutral / earth. This forms a capacitive voltage divider that will, in the absence of a lamp, settle at some arbitrary voltage with some arbitrary but very high source impedance. When the lamp is plugged in, both of these parameters, plus the behaviour of the lamp's own internal circuit, will decide whether the lamp flashes.

If the voltage divider settles on a high voltage but the current leaking into the SL is low due to short cables having low capacitance, the lamp might discharge it without flashing. Conversely, if the inward-leaking current is high due to long cables, but the stray capacitance to earth is also high and effectively discharging it, again the lamp might never flash. When the current leaking in is high enough that even with the discharging effect of the stray to earth and any shunt resistance at the input of the lamp, the reservoir capacitor charges up to a high enough voltage for the oscillator to start, then it will flash.

Because the inter-core capacitances of the cables differ between cores, there can be different results according to which strappers are in circuit. If the strappers that are connected to PL and SL at a particular time are adjacent in the cable, there will be more leakage into the SL than if they are spaced either side of the CPC. So you might see flashing with both switches up but not with them both down, or some such effect. The behaviour might also vary with the temperature of the lamp, since the leakage of its internal reservoir capacitor may be significantly temperature dependent.

 

[timhoward]

Bear in mind that it is not strictly just the voltage that matters. What is happening is that the capacitive reactance ffrom permanent line to switched line is leaking a small and nearly constant current into the switched line, and a different capacitive reactance is leaking current out from switched line to neutral / earth. This forms a capacitive voltage divider that will, in the absence of a lamp, settle at some arbitrary voltage with some arbitrary but very high source impedance. When the lamp is plugged in, both of these parameters, plus the behaviour of the lamp's own internal circuit, will decide whether the lamp flashes.

If the voltage divider settles on a high voltage but the current leaking into the SL is low due to short cables having low capacitance, the lamp might discharge it without flashing. Conversely, if the inward-leaking current is high due to long cables, but the stray capacitance to earth is also high and effectively discharging it, again the lamp might never flash. When the current leaking in is high enough that even with the discharging effect of the stray to earth and any shunt resistance at the input of the lamp, the reservoir capacitor charges up to a high enough voltage for the oscillator to start, then it will flash.

Because the inter-core capacitances of the cables differ between cores, there can be different results according to which strappers are in circuit. If the strappers that are connected to PL and SL at a particular time are adjacent in the cable, there will be more leakage into the SL than if they are spaced either side of the CPC. So you might see flashing with both switches up but not with them both down, or some such effect. The behaviour might also vary with the temperature of the lamp, since the leakage of its internal reservoir capacitor may be significantly temperature dependent.

Thank you - I will digest that slowly and carefully later!
By the way, in this case the strappers were both single core double insulated independent cables, as was the outgoing feed to the lamp, and my initial thought was this arrangement would normally be less susceptible than T+E or 3 core due to extra insulation and more variation in spacing over the run.

 

[Lucien Nunes]

I agree that single-core cables are less likely to cause an issue especially if they contain CPCs. The main factor is the distance between the relevant conductors, if PL and SL-tied strappers are a few inches apart the leakage into the SL will be much lower than if they are two cores in one cable. However, if SL is not near a CPC then the leakage out of it will be reduced also. If the cables are widely spaced it is possible that the resistive leakage will dominate. With clean dry fittings and good 3c+E cable, I would expect the capacitive leakage to dominate.

 

[LastManOnline]

Bear in mind that it is not strictly just the voltage that matters.

. I read many posts on here I can contribute to. I read more I can learn from. Every so often I read one that makes me feel "oh boy.. long learning curve ahead". Great post Lucien. I think it's only the above sentence I quoted that I fully understand.