Fluorescent tubes near AC

[Oliver]

I remember about 7 years ago, going to Poundland and seeing the T5 fittings near the air conditioner. The tubes were right next to the air outlet. The tubes looked like they were shaking up and down really fast. It was quite a strange effect. Anyone ever heard of this or seen it?

[Ash]

The cold temperature makes some of the mercury condense, so the tubes become mercury starved. Combined with some other factors some tubes might react to this by striating (arc swirling around the tube), which is what you seen

Other odd phenomena (but this one seen only on HF ballasts) is standing waves, this looks like a striped pattern showing up along the tube, sometimes it is still, other times it would appear to be running along, or emerging/converging to the center of the tube

[Oliver]

I remember when we had a long power cut here and the 8w tube in one of the emergency lights was going for about 2.5 hours. When the power came back on and tube was at full power, the tube was doing this strange effect. It was like the arc was going up and down. The "line" was horizontal and it almost looked like the tube was moving up and down. It was a strange effect but it settled after a few minutes. It might have been a similar effect to what I was seeing in Poundland all those years ago. It was very fast though, looked at the speed of rectifying. It really did look like the tubes were rattling and shaking up and down however, of course that isn't possible so it must have been some sort of effect from the AC blowing onto the tubes.

[Ash]

The effect is that the arc takes shape of a thin line (instead of a wide diffuse area as normal FL discharge is), then it is moving around the tube perimeter like a jump rope, due to electromagnetic, thermal and pressure wave effects

Those effects are most common with HF ballasts, as the HF driving may awaken a standing pressure wave in the tube (same reason why HF cannot be used for HID), or amplify other effects which alone are usually too subtle

Most EM inverters are based on a very simple and not very stable circuit, which frequency is susceptible to shift due to changes in supply voltage (the battery discharging) and tube performance. It is quite possible for such circuit to end up in a positive feedback loop with the tube, chagning its frequency in a cycle, which affects the tube performance, and so on. It tends to happen at specific levels of battery discharging too