Mains voltage LED string power supply experimentation.

I've made a custom set of lights for my brothers mantlepiece, since a commercial set I'd got him were too ferociously bright looking. So this set started off as a 100 lamp tungsten set of lights that I modified a capacitive dropper PSU for, and then swapped all the original tungsten lamps for nice fat diffused 10mm LEDs inserted in the correct polarity along the string.
To get the visual effect I required, I first upgraded the PSU (power supply unit) with new improved component ratings, and then initially put in a 100K resistor where the main dropper capacitor goes. The 100K resistor was partly experimental to see how much current it would pass, and partly so I could dab some different capacitors onto its exposed leads.

At this point I should mention that I was indeed dabbing components onto live electrical connections at full mains voltage (240V). I wouldn't really recommend this sort of practice unless you know exactly the routes to ground that electrical current could take.

So here are the results:-

100K resistor alone = 0.5mA LED current.
100K bridged with 22nF = 0.75mA
100K bridged with 47nF = 1.25mA
100K bridged with 100nF = 2.6mA
100K bridged with 220nF = 5.6mA
100K bridged with 330nF = 8mA

I chose the 100nF capacitor in the end and the lights look great.
Power consumption of the whole string is around half a watt, giving a running cost of just 50 cents/pence a year for continuous 24/7 operation.

One of the most surprising results was that the 100K resistor alone passed enough current to make the LEDs glow at a very nice intensity. The power dissipation across the resistor was an incredibly low 0.025W which is one tenth the resistors power rating.
With the 0.5mA current the lights would have operated at 1/10th of a watt and cost just 10 cents/pence a year to run 24/7. That's for the whole string of 100 lights!