Would a standard fluorescent fitting work on DC

The tubes run fine on DC , as will the CFL tubes, if you remove the electronics, and go direct to the tube itself.

The old iron cored ballasts are not happy on DC, or indeed on anything that is not a good 50Hz as they rely on the inductance to provide the dropping impedance.

I have run a 40w tube with a 60W lamp in place of the choke ballast, and it's not quite right, but it runs, resistance ballasting like that is possible, but lossy. (As a child I have seen light weight fittings with a filament lamp ballast, sold for hanging from the ceiling, from the 1960s ' soon power will be too cheap to meter' era.)  Resistance ballast is of course frequency independant, and  therefore will be good from DC to many kHz, though low freq AC tends to lead to flicker and flame-out requiring a re-strike. You can run them from an audio power amplifier via a mains transformer used backwards, but it is worth supplying at least one heater supply that is permanent.

On DC the striking circuit needs more thought, and has more in  common with an electric fence pulser or a car ignition.

May modern electronic 'ballasts' are designed to run on DC anyway,. as the first thing they do is rectify the mains coming in, and then chop it up again at a supersonic frequency that goes to the tube - the advantage is the £2 of electronics uses a far smaller transformer at 30KHz or whatever, compared to the 50Hz part, and the saving on the copper and iron is worth it, as a bonus the design can be made to better adjust to varying mains voltages , and can be more efficient. 

I can well believe that such fittings may well have varying run-done times as the power is removed, and the internal smoothing capacitor goes flat.

A fluorescent light can be used on mains voltage DC and this was done back in the days of DC mains, or on board ships with DC supplies.

The circuit was similar to a standard switchstart  lamp circuit but with the addition of a dropper resistance in series with the choke. This dropper resistance sometimes took the form of a special incandescent lamp, these are no longer manufactured AFAIK.

With DC passed through the lamp, the mercury vapour tended to migrate to one end of the lamp and this resulted in only one end lighting correctly and the total light being much reduced. This had to be counteracted by reversing the supply polarity periodically. Special light switches were manufactured for this purpose. These were rotary switches with two "on" positions and two "off" positions, each of the "on" positions had a different polarity. Since the switch had a ratchet action and could only be turned clockwise, the supply polarity was automatically reversed each time the lights were turned on.


A switchstart fluorescent lamp circuit Without this dropper resistance wont work on DC mains, the current would be excessive and soon destroy the choke or the lamp. It CAN remain lit for a second or two from a charged PFC capacitor.


A more modern fluorescent light would have an electronic ballast, these often work fine on a DC supply and some types can be used to provide emergency lighting from a central battery system. This battery usually needs to have a higher voltage than the AC mains. Often about 300 volts to 350 volts DC is needed for an electronic ballast primarily designed for 230/240 volts AC. Such a light will also stay lit briefly from a charged PFC capacitor.


Alternatively, in the circumstances reported by the O/P the lights might have remained lit for a few seconds due to the presence of an AC supply, this being obtained from one or more induction motors that generated briefly as they ran down after turning off the mains.

I  did the experiment with using a filament lamp in place of the choke it seemed to work ok the starting  time was a little longer and using a 20 watt tube a 100 watt lamp ballast made the tube slightly dimmer than normal but a 150 watt lamp seemed to run it at near full brightness. There was however a noticeable flicker at the tube ends which is much less with the choke ballast in the system. Regarding CFLs I ran out of ne with around 330 volts DC  input and it seemed happy enough switching off the power with around 300 uF of caps gave around 4 or 5 seconds of light until the caps run down

As reported above, a fluorescent lamp can be worked with a filament lamp in series, but the correct wattage/voltage lamps are no longer manufactured.

The closest approach is a 196 volt, 100 watt incandescent lamp in series with the tube. A 4 foot 40 watt lamp in series with a 196 volt 100 watt bulb works fine.

I have successfully run 240v rated CFLs from as low as 72 volts (12 x 6v camera batteries in series, I got a batch of them cheap at an auction, past their expiry date but still fine). Served me as a handy work light for a few years . IIRC I had to try a few different lamps before I found one happy with the (comparatively) low voltage.


Now superseded by the LED floodlight retrofitted with different LEDs and a 12v SLA battery :)


Can't be doing with this newfangled lithium stuff <grin>