|The electrode structure (still images)||.avi file (238k)||
Can anyone explain the physics of this flickering? The lamp is operated from 230v 50Hz mains via a BC base. It has two apparently identical droplet or flame shaped plate electrodes which are separated by less than 1 mm. It must also have a series resistor built into the base. It seems to me that the inner surfaces must have an insulating coating, otherwise the glow would be between the plates. The lamp does not appear to be a precision assembly and there is no apparent movement. Placing a damp finger on the bulb causes the flicker rate to increase dramatically.
My speculation that the inner surfaces are insulated was wrong. At constant pressure the ignition voltage (and maintaining voltage) versus electrode separation characteristic for glow discharge tubes is a bathtub shaped curve:-
Diagram from Reich, H.J., "Applications of Electron Tubes", McGraw-Hill, 1944, 2nd. ed., 2nd. imp., p.428
The lowest voltage striking path length, at the minimum of the curve, is is such that the glow forms on the outer surface of the electrodes.
It had been suggested by Colin, G3YHV, that the behaviour might be different under D.C. conditions. This is not the case, apart from the obvious difference that only one electrode glows. The tube strikes at 140 volts and extinguishes a 117 volts. At low voltages the glow is confined to a small area at the base of one of the electrodes and the flickering takes place to and from a similar sized adjacent area. As the voltage is increased the size of the glowing area increases and flicker rate increases slightly.
It seems possible that the flickering is caused by the build up of charge on the glass envelope. The increase in rate caused by the damp finger points to this.
Reich (op. cit. p. 473) referring to Thyratrons (grid controlled arc rectifiers) says:-
"It is of interest to note that the design of the grids .... is such as to shield the other electrodes from charges that collect on the inner surface of the envelope. If this were not so, the control action would be erratic"
which suggests that surface charges on the envelope may be responsible for such effects.
"I think it's the heating of the electrodes. The neon gas glowing produces some heat (not a lot, but some) and the electrodes are designed to bend and twist a little when they do get warm. When they do bend, another area of the electrode gets heated and the first area cools a bit. Thus a repetitive cycle."
This flicker is at ~ 1 Hz. I can't see any bending by eye and I don't have any facilities for high speed macro photography.
"More likely, it's a small magnet that's "guiding" the discharge. This may in turn make the gas move in the bulb, but unless it's a god-awful large bulb, I wouldn't expect that to be a consideration."
It isn't a magnetic effect. I've tried interfering with it with some pretty hefty magnets (ex speakers and dot matrix print heads) and there is no visible effect on the glow or the flicker frequency. The envelope of the bulb is 70 mm (2 3/4") high and 32 mm (1 1/4") in maximum diameter, not including the BC base. The electrode plates are approximately 27 mm (1 1/8") high by 13 mm (1/2") wide.
"As far as the answer to your question goes, heat generated by the neon plasma rises, reducing the ionization potential in the 'hot patch', causing the plasma to follow the convection currents until it rises high enough above the peaks of the electrodes that the ionization potential gets to be too high to maintain the arc. When that happens the plasma will quench and a new one will start at the base of the electrodes, which are probably closer to each other than the tips, a la Jacob's Ladder."
This is a glow discharge not an arc. There is not a direct relationship between ionisation potential and ignition/extinction voltages for glow discharges, they are very many times as great as the ionisation potential. The ionisation potential of Neon is 21.56 eV, the lamp's ignition/extinction voltages are 140/117 V. In order to sustain a glow discharge, the mean free path of an electron moving in the field between the plates must be long enough for it to achieve an energy of 21.56 eV. It must also have a significant probability of striking a Neon atom rather than the anode.
I will investigate the possibility that convection in the gas is causing the effect. This convection effect has also been suggested by Brian Reay, G8OSN on uk.radio.amateur.
Here are two new .mpg files of the lamp operating in other orientations. This is an attempt to investigate the possibility that convection of the gas in the bulb may be all or part of the reason for the flickering. I think the results disprove this hypothesis. Sorry about the quality.
This is the result with the lamp rotated 90 degrees in the vertical plane. This is the result with the plane of the electrodes horizontal
I was unable to observe any change in behaviour with change in the orientation of the lamp, despite waiting for about 30 minutes in each position to allow a 'steady state' to develop.
"I think you've discovered a phenomenon called a 'relaxation oscillator', in a crude form."
A few reasons why I don't think a relaxation oscillation is the whole story.
This thing runs on UK 230 V 50 Hz mains. I can't immediately see how a relaxation oscillator can run with a 1 - 5 Hz range frequency on 50 Hz supply without a bunch of steering diodes and a fairly big capacitor. The whole thing fits into a standard U.K. BC base, so there isn't much room for anything more than a resistor in there.
The flicker rate seems a bit too random and there is the question of why the glow moves from one part to another of the electrodes.
If there are diodes in there then it must be full wave as the thing runs just the same on DC in either direction.
The flicker rate doesn't seem to be voltage dependant as you would expect for a relaxation oscillator.
I have seen Nixies which run relaxation oscillators on the self capacitance of the electrodes when using a high impedance supply just around the striking voltage, but they were in the low HF region, certainly not visible.
I'll try to put a resistor in series with it an scope the current. That may give some useful info.
All the images and motion shots on this page were generated by me with a Canon Powershot A40. Some shots used a homebrew close-up achromatic doublet of about 5 dioptres. All may be copied for any non-commercial use.