The fan is triggered by the voltage across the LED. This does not fall to zero as it would for a filament lamp, as there is a very small capacitance in the switch wiring ,and the LED draws no current at all when voltage falls below some threshold. result the switch line drifts to some tens of volts AC. If you have a high impedance meter you can probably detect this, though the exact value will change as you try and measure it.

On 50Hz mains the standard fix is a mains rated capacitor of ~ 0.1uF across the light (or in the fan between trigger live and neutral) - now to minimize switch arc, an RC series combo (example) is probably more universal, and is designed for mains use though in most cases a simple cap (example) with enough voltage rating would do.

Mike.

Thanks for the info. I was sure that there was a better solution than increasing the power demand.

How is the light switched?

Standard SP pull cord. I did try disconnecting the switch live from the switch to ensure complete disconnection.

You could wire in a momentary pull-cord switch to trigger the fan separately to the light.

Even if you separate fan and LED circuits you will still get the LED lights flickering  if the switch line is too long and introduces enough capacitance to half heartedly start the LED electronics some of the time, as it seems to be the case here. Use a shunt C R  or C to suppress the picked up voltage and and fix it properly.

M.

The light was installed last year and didn't appear to have a problem on it's own. The issue only arose when the fan was fitted this week, which was why I assumed the problem was caused by an interaction between light and fan. The switch wire is only about 1M long so any pickup would be very small, although all of the connections are in the fitting so that won't help. I had considered using a two pole switch and using one pole to turn the light on and the other pole for the fan. This isn't such a tidy solution but the materials are readily to hand. I would have specially order a suppressor. It does beg the question that if there is an inherent issue with the light why the manufacturers don't include something. The light was chosen and installed by a local sparky (the property was tenanted at the time) so presumably it isn't top quality.  

well the fan trigger line (switch live) is in parallel with the existing switch and my increase the coupling. 3 and earth cable has one core near earth and the outerjacket, and one core near another live core. You may win a little by using the core that only has earth as the neighbour as the trigger (switched live) but I'd be very wary of saying that will be enough.

I will note that my toolbox has a few CR  units both for ghost voltages and for ‘declicking’ switches with those funniness where the security lights come on when the motor starts that sort of thing, a few surgegards for limiting inrush and some clip on ferrite rings to go RFI hunting as things I can go to and try in a few mins, and rather like the ferrule crimpers, its not often needed, but when it is,  just the ticket. It depends rather what sort of thing you do for a living though.

Thanks for the help. Have ordered a suppressor so hopefully that will solve the issue. (2 pole 6A switches aren't as readily available as I thought they were?)

I can understand how capacitive coupling would cause intermittent flickering of an LED (the LED's circuity presumably includes a capacitor which charges up via the leakage current and when high enough, causes the LED to light briefly). But a fan is just a big coil of wire with a low static resistance. Surely there couldn't be enough leakage current to power the fan for a second?