Neon flicker flame lamps are lamps that have two flame shaped electrodes placed in parallel inside a neon gas filled glass globe. A resistor is usually mounted in the base to limit the current through the discharge. Without it the neon would pass far to much current and the lamp would fail. Typically a neon discharge lamp strikes on each half cycle at about 90V and then sits down at 50V across the electrodes. On the UK mains supply which is typically 240V (supposedly European 230V), the resistor inside the base of the lamp dissipates much more power than the glow discharge. To get decent brightness the resistor has to pass enough current, and that means it runs very hot. Most failures I've seen in these lamps in the UK have been the resistor failing due to thermal stress over time.
I decided to experiment with adding an external resistor to limit the current further, and also experimentally used a 100nF 400V series capacitor. Both techniques worked well in reducing the power dissipated by the lamp, without affecting the intensity too much. My preference was for the capacitor though. It limits the energy on each half wave of the supply and is a robust and cool running way to take the load off the lamp.
One downside of using the series resistor is that if the lamp or base fails short circuit, then the resistor will be applied directly across the mains and will get very hot, possibly combusting, and worse still, potentially igniting the plastic candle shaped base.
Neon discharge lamps tend to give an area of discharge proportional to the current flow, so if you limit it too far it will result in a less active glowing area in the lamp.
A capacitor can't be used alone in series with a neon lamp as it would cause a series of high current spikes that would damage the electrodes and blacken the lamp (sputtering). In this application the resistor inside the lamp base limits the current to an acceptable level.
I added a 1 megohm resistor across the capacitor to discharge it when the power is removed. Otherwise it will hold a charge and could give a slight tingle from the plug pins.
Lamp dissipation test results. Not taking cathode drop and capacitive limiting behaviour into account.
Bare lamp. 240V @ 6.2mA = 1.49W
10K resistor dropped 46V, so 240 - 46 = 194 @ 4.6mA = 0.89W
100nF capacitor "dropped" 145V so 240 - 145 = 95 @ 4.16mA = 0.4W
Measurements across resistor and capacitor were totally non sinusoidal and quite frankly just rough values for reference.
Capacitor type was metallised polyester and not mylar. 'cos they're cheap!