I could not resist it... I bought this little and quite modern hand held 4W black light flash light. Cheap, it was 5 Euros and it worked. But when I looked into the black light tube, I saw a bit orange glow. Strange... So I thought, lets put in a UV-C tube, to see what is happening. It is a shitty armature with no preheat and the heaters are shorted.

But I've never seen this before... The heater on the left doesn't do anything, the current goes trough both pins. The heater on the right has a super bright light around it. But the beam inside the tube is also behaving very strange with an open part not lighting up.

After less then minute to take the pictures, my nice and shiny UC-C tube got grey at one side. After 24 hours the tube is clean again... So is was probably some mercury vapour on the glass...

I've not made any measurements because this thing is eating tubes for lunch...

Has anyone seen this before and have an explanation for it? I can do measurements, but then I need some extra tubes I guess...

Regards, Roland

Part of the tube not lighting up on the right side is probably something to do with standing waves generated by the electronic ballast.

 I had one of these before.   These are defo tube eaters.  This cause it cheaply made light with one end heated and a simple
  Inverter circuit.  It because it taking Dc power and chopping the DC up into transformer.  As result tube being heated at one end and cold cathode on other tube.  It act a recitfer and cold cathode end start sputter off the emmision material from cathode cause it never got to heated up correctly.  The only way to fix this is having a dual battery supply  +6 and -6 volt for the cathode to work right in a DC chopper circuit

That's crappy single ended blocking oscillator circuit, employing one transistor. It certainly underdrives the tube, preventing the filament to heat up and transfer to an arc discharge mode. So a tube operates in a glow mode, high cathode drop and high sputtering.

In a glow discharge, cathode part may have a slightly different color. Also sputtering may be so high, that cathode activation coating atoms may start to color the discharge themselves.
 

In thermoionic mode (a proper mode of operation of a fluorescent tube) electrons needed for the discharge are peacefully emitted by hot a cathode activation coating, in a great excess. So, cathode drop is low, gas ionization is also low, and cathode part is rather dim.

In a glow mode, emission is more violent, electrons are emitted mainly by accelerated ions bombarding the cathode, cathode voltage drop is high, gas gets more strongly ionized and the cathode spot glows brightly.

The main problem is, the 4x R6 cells is just too little battery to allow proper tube operation. The borderline cold cathode operation is a way to somehow optimize te total running cost (batteries usually cost more than the consumed tubes here), it minimizes the power spent on heating the cathode.
So yes, they are tube eaters, but not unjustified for the time they were made...

The ballast drives the tube at DC (the tube itself is the rectifier), that is why the different pattern, which helps a bit (concentrates all the cathode heating to just one electrode; the mercury migration uses to not be a problem with such short tubes), but for a "4W" tube you would still need about 2..3W power feed, way above what those batteries allow.

There are OSRAM flashlights, I had one too, also some recent pictures here. Runs off 4x AA alcalines or NiMH. There was a properly designed inverter, no heavy sputtering, kinda hot start. Using base-less 5W PL tube. The problem is to found the designers having straight hands) OSRAM surely insisted on this

Though, later CCFL flashights were better in this application. No limits on ON/OFF cycles, no wear when run on a flat battery.
 

The lamp wear on weak battery is a problem that could be solved by proper ballast design (even with those single transistor circuits - starts with using a proper input filter, not exposing the vbattery to the HF current ripple), but the main issue is the limited energy available in the batteries vs the power the lamp need.
I'm speaking mainly about the FxT5 era, so no high capacity NiMh, so practically carbon-zinc, maximum alkalines, nothing else (no LiIon, no NiMh, only very expensive and low capacity and limited power NiCd, so not practically usable either). And CCFL did not exist (as a commercially viable option) either.
With higher power or extra cathode heating the lamp won't wear out that fast, but the batteries will last very short so the running cost of the lantern would be very high because of all the batteries it needs.
Or you may drive the lamp at lower power and no supplemental cathode heating and sacrifice the lamp life, but get longer runtime from the battery. Yes, it will eat lamps for lunch (lamp life was designed around few 10's of hours, replacing the lamp after 10 sets of batteries was not so uncommon), but the total running cost would still be lower (cathode heating would mean spending 15..20sets of batteries instead of 10, that was more expensive than a new lamp). Moreover the lower efficacy would mean there would be no point of using a fluorescent and go to incandescents as a light source in the first place.

Later came the generally more efficient PL lamps, where it started to make sense to spend some power on the supplemental cathode heating (the PL needs lower power for the filament), plus the high capacity NiMH and generally dropping prices of the batteries made the electricity cost for these lanterns significantly lower, justifying spending a bit more power to boost the lamp life. But that was past the F4T5 era.

With the BLB running on R6's the expected usage was so low operating hours with so great gaps between the batteries would degrade so they won't be able to drive the extra heating. Lanterns for more serious uses were designed with R20 and even 6 or 8 cells, so there was no problem with the power anymore. But the ones designed for R6 were meant more as either very occasional use or even just as a novelty (where the short life was not a big deal)

For Osram flashlight, let's assume 3W at slightly underdriven tube, 80% converter efficiency, ~5V average battery voltage. That means about 0.75A from battery. That results 3.3 hours from a set of modern alkaline AA's or NiMH's. Not bad. I actually used that Osram lantern for the intended purpose (light whlile going asleep in a tent) and it did not suffer from a short battery life.
 

It is a joke to run a fluorescent flashlight on carbon zinc 'heavy duty' batteries

It may be 3W for the fresh batteries (so 5V), but becomes barely a watt when the batteries gets exhausted (nearing the final 0.8V/cell, so 3.2V). As the battery gets weaker, not only the inverter drives at lower power, but its efficiency drops as well (the power dissipation going mainly into the battery internal resistance, as not that many ballasts use multi-mF low ESR capacitors on the battery input, so expose the battery to the high ripple current of the primary).
The tubes get eaten mainly when the batteries get weaker, not that much when new.
Plus the problem is not that much the 2 vs 3W when the lamp is running (in fact for a DC operation as low as 1.5W would be OK), but the transition between glow to hot cathode is problematic. There the inverter efficiency gets very poor (too high voltages, high switching losses, reverse current through transistor so high primary AC ripple current), so unable to deliver enough power to warm up the cathode hot spot. So the lamp stays in the cold cathode mode for way too long.

Plus with 0.75A I doubt the 3h runtime on LR6. According to typical LR6 charts, even at 0.5A load the rated capacity is a bit above 1Ah when assuming discharge down to the 0.8V, so the maximum usable current for the 3h runtime is around 350mA. So with 4.8V about 1.7W in average, so 1.5W for the tube. But at final 0.8V/cell (so 3.2V) this becomes barely 1W for the tube (assume typical behavior of a nearly constant input current with the simple single transistor inverters), definitely not enough to transition when started with battery such weak.

With 0.75A I would guess for barely a hour (~1Ah).

Well, may be 3 hours on AA alkaline is a bit optimistic. BTW AA Energizer specs looks a bit weak with up to 300mOhms ESR. We also should note that in its intended application, a flashlight is used in short cycles, that way allowing batteries to depolarize, resulting a better performance than in a single cycle discharge from datasheet.

Still, I believe it will get firm 3+ hours per charge on good quality 2500mAh NiMHs.

Depolarization: If you compare them with the "3.9 Ohm flash light" standardized test (so 4mins ON, 56min off) chart, it is not that much different, so either the overall chart counts on depolarization gaps to happen or the polarization does not have that much effect here.

The modern NiMH's are of course way different story, even though on a "2250mAh" cells I would not expect that much more than about the 1.5Ah for the 0.5..0.75A load range.

There is also other subjective factor: What seems to like "lamp lasts forever" in such high power flashlights is in reality quite surprisingly few hours: Common bulb ratings for "dry"/alkaline cells is 25 hours the most, with about 1.1..1.2V/cell nominal voltage rating (so being overdriven with fresh batteries), but stil it seems to be lasting "for ages" and most of the flashlights tend to fail mechanically or get smashed way befprethe lamp burns out...
With fluorescents when the lamp life is in the 100hour ballpark, the lantern rarely approaches the end of the lamp life before something else fails (usually corrosion from leaked batteries). So even when 50..100x shorter than on mains ballast, still sufficient for the intended purpose.

But it is true there were real garbage lanterns on the market where the lamp life was barely a hour or so, so really short even for a flashlight standard, but I do not talk about those...