The inverter inside the Gaash Barak II, also underdrives the lamps, but it is rapidstart (Cathodes heating), and not instant start. Thats why it is good for the lamps.

With emergency lights the tube life is practically never any problem, even when the tubes are cold started and underdriven. You barely need more than few 10's of hours with 20 starts from them over the complete installation life...

The lamps fail already after a couple of battery depletion runs. The ballast can keep powering the EOL lamp at cold cathode and it works (in my home i move EOL lamps from normal ballasts to continue use in an emergency light), but the light output is much lower than intended

If the same lamp is switched between normal and emergency ballasts then it is a problem, the emergency kills the lamp and after a while more it fails on the normal ballast

Then the emergency units have other problem:
Missing battery undervoltage shut down. This should at first prevent really deep battery depletion, so prevent damaging the battery (after 2'nd or 3'rd complete discharge the runtime with the undervoltage present would be larger than without that feature, even when it will look like it will cut out part of the runtime) but as second it prevent the inverter to operate the lamp on the really low power, when the inverter has no chance to warm up the filaments.
And if the undervoltage is missing, then the long lamp life does not make much sense either, when you have to service the unit due to dead battery anyway...

Those emergency inverters get about 120Lm out of an intact 8W T5. The inverters for 36W lamp get about 480Lm (data taken from Gaash, a lantern manufacturer). That mean power input on the order of 2W to the 8W lamp, and 8W to the 36W lamp. It have no chance to warm up the cathodes anyway

The long lamp life does make sense - Why should a lamp be wasted (pollution, cost etc). Even if the maintenance worker is right now up on the ladder, the lamp is still not "free"....

I didn't realize how under driven those are. In theory, could you run the f8 on a f36 emergency inverter since the f36 puts out only 8w?

I can try. I think it will put less than 8W into the 8W lamp due to its lower arc voltage

Ok. Just get back to me when you can, thanks.

The 20..25% of the rated current is still able to warm up the filaments, but it means it should be delivered even in the cold cathode state. And that is the main problem: The power will become temporarily larger during the cold cathode mode (because of the higher voltage drop) and that is the problem with many ballasts: They are just not able to deliver that higher power for the warmup (or better to say the weaker batteries are not able to deliver that power; that is the importance of the undervoltage cutout).

And for the lamp replacements: The problem is, the lamps degrade on their own when not in use (mercury settling, atmosphere poisonning,...; all that is normally supressed by the lamp operation), so it is not able to work correctly after the long sitting time, mainly when the lower power is involved. The main reason to replace the lamps with the battery comes from there...
And the reason for underdriving the lamps is just the efficacy and consequently the environment load as well: For the required 130lm, you may use either a 4W lamp at full power lasting all the 10 years of the fixture life (or even longer), or an 8W one at 2W replaced every 3..4 years. Obviously the later suffices with half of the battery size for the same runtime, so half of the lead or cadmium waste with each battery replacement. So for the environment the extra tube is way less load than the larger battery size...

And then the batteries end up going EOL and ruining the charge circuit due to the lack of under voltage cut out.

With lead acid, the battery EOL is usually either loss of the water due to the constant float charging (mainly when the charging voltage is not made with proper temperature dependence (it should have slightly negative temperature coefficient, so decreasing with increasing temperature). Less likely form of the battery EOL is the electrode sulfatation, but that is usually linked with long time with missing power (a seasonal facility not used off season, with the power switched off completely,...)
Similar problems are with NiCd's.
But the missing water can not cause the charger circuit to overheat or any other condition leading to it's malfunction. For that many products suffice with their own design (e.g. "2W" resistor loaded by the full 2W power dissipation or so; there I do not understand, why for a 2.4V battery they use 9V transformer and dissipate the 7.5V on that resistor).

Only the NiCd may end up short circuit (after cell reversal; but that exactly is supposed to be prevented by the undervoltage cut out), but there just one shortened cell can not cause the charger to overheat, if designed at least reasonably.

I see. Thank you for the explanation. Now do they still use lead and NiCd batteries or have they started using LiPo or Li ion?

The Lion/LiPo are completely unsuitable for such duty, their lifetime would be way shorter than usual for the lead acid or NiCd's.
Although I do believe, many cheepeese makers will use them even there - for the ~1Wh capacities they are becoming the cheapest choice and their cell voltage pretty well matches the LED forward voltage, so even simple linear current regulator means high efficiency, so they look very attractive for someone who wants good runtime of brigh light when showing off at low cost, but don't care that much on the lifetime.

The Lithioum ion based cell performance lies in applications, where they are all the time completely charged/discharged, there they really outperform all other usual chemistries, but not in those stand-by applications.

These Lithium based cells have completely different dominant aging mechanism than all the "classic" chemistries.
With the classic batteries the charging/discharging means the electrodes change their physical size, what leads to fatique cracks, so the dominant aging of the electrodes is cycling.
Then with lead-acid the crystallization of the PbS causes the degradation when discharged, with Ni-based the dendride growth when electrodes are reversed (= cell deep discharged below about 0.6V means one electrode gets already reversed)

But with Lithium similar effects are nearly nonexistent (= they are hidden behind the other degradation effects), maybe except the electrode polarity reversal (deep discharge way below 2.5V).
The dominant wear mechanism is a thin layer growth associated with higher voltage and of course temperature.
This has quite severe consequence for reading the "performance" data:
A classical secondary cell "endurance" test is repetitively charging/discharging the cell, till it looses capacity to certain level. When an NiXX shows 500cycle life, it mean with one cycle/week about 10 year life.
But the fact the LiIon is showing 10k+ cycles in a 3year lasting test does not mean the battery will last 200 years, in reality it will die within the same 3 years period as the test have lasted or even shorter. That is, because the dominant aging mechanism is, how long (accumulatively) the cell hat spend at each given voltage and that was the same with the tests with 10k cycles, as in the real life. And the "shorter" comes from a situation, when the battery stays most of the time fully charged.

So with the emergency lights the batteries are practically 100% of the time fully charged, what means LiIon life shorter than about 2 years...

Provided that i have 8 year laptops which batteries still work (with ~1/2 original capacity), i do think a Li Ion may be a good choice for emergency lighting

If the problem is 100% charged state, why not put a 1.5x capacity battery and keep it charged to 2/3 ?

Plus that would put tons of stress on the lithium batteries. They are prone to explosive failure when overcharged or overheated, which could happen in an emergency fixture.