I have several fluorescent lamps, both T12 and T8, that will not light in any fixture except one of my many 4xF32T8 instant start troffers due to broken electrodes. I have been using these lamps in said fixtures for a while with no failures of lamps or ballasts. My question is, am I doing my ballasts a disservice by running lamps with broken cathodes on them? I know it's wearing out the lamps, but I'm fine with that because they're pretty much dead anyway. I'm really just trying to squeeze more life out of otherwise dead lamps. I just don't want to shorten the life of the ballasts, even though they're just modern electronic Chinese junk.

The EOL T12 and T8 lamps will only light for several mins before losing vaccum.

Generally, the failing lamp is a stress on virtually any ballast, larger (HF electronic) or smaller (series choke, when not counting the starter wear). So any ballast would be at risk on EOL lamps. And mainly the cheepeese ones, where the maker could cut corners on such "unimportant things" like failure modes.

What I do not understand as much is the reasoning, why to "squeeze the rest from their life": If the lamp did it's job and so is worn out, it could do some hours on the IS, but it would for sure die for good quite soon and that point could take the ballast with it. So you may get few hours from the dead lamp, but fry out the ballast on it...

My experience with cheepeese ballasts is, they tend to serve quite faithfully, but only till the first lamp problem, then they usually die. So to get maximum light from such fixtures the only thing you may do is to use really good quality long life rated lamps instead of the OEM one - the ballast and so the fixture would continue to work as long as that first lamp in it...



What is important on IS ballast, in what condition is the emission coat on the filaments, if they are broken or not does not matter by itself. The emission coat is normally consumed over the lamp life and when it is missing, the affected cathode does not perform it's function well and that mean the arc either die out or the ballast get stressed (their output resonance circuit).

Filament breakage is primarily not the normal (wear out) failure mode, so it would be important to know, what broke them.

Some ballasts (all the cheap electronic connecting all 4 lamp pins; that mean most of the instantly starting ballasts sold in 230V areas) have the filaments wired ion series with the resonance circuit, so once is there either no arc, or insufficient cathode emission their filament heating power is supposed to blow out the filaments (and protect the ballast in that way). So in such case operating such lamps on the IS circuit (where the resonant circuit is closed internally, inside the ballast box) could yield to ballast overloading.

If these lamps were operated on "classic" RS circuit, the low heating voltage can not melt the filament even when it looses all the emission, so the filament for had to sure broke for another reason than the normal wear (so it could be structural fault in the material, mechanical shock,...), so the emission layer would be still likely in a working order, so good enough for the IS ballast.

Preheat circuits tend to evaporate the filament when it is heated for too long time by the ballast short circuit current, what happen when the lamp does not want to strike and the starter heat up the filament again and again, so I would not torture the IS ballast by those lamps either.

The lamps in question are not black at the ends but at least one cathode reads open. My guess as to why this would be is mechanical shock, though they must have been pretty fragile since I try to handle lamps gently. I assume there is still emissive substance on the broken filaments because the lamps operate quite normally. These ballasts only have one wire going to each lamp end, so whether or not the filament is broken should make no difference as long as there is emissive substance left. I haven't had a lamp go EOL yet in these fixtures, so it'll be interesting to see what happens when one does.

yeah some lamps will run for years on IS if not switched frequently even if the cathode is broken on one end. the ballast doesn't see it as a higher load unless the lamp starts flashing orange at the ends and flickering then you might want to remove the offending lamps, most decent quality ballasts survive multiple EOL lamps, matter of fact most electronic ballasts i change are just randomly dead (there was no lamp EOL as they are always group relamped) i probably change 60 or so ballasts a week of various brands being a lighting technician, most are 4 lamp F32T8, then probably 8ft T8 the other ballasts are longer lasting, i have found a big killer is the 25W 4FT T8 lamps that replace the 32W lamps, they have a lower arc voltage and draw more current from the ballast overheating them

None of the lamps flash or do anything weird, so I'll probably leave them in for now.

The older T12 (here, in Europe), tend to be indeed very fragile - only a small kick and the filament break. The newer T8 (in Europe they are compatible on the most common preheat gear) were way more robust...

If you have the possibility to measure the arc voltage (you need a meter rated to at least 10x the ballast operating frequency, so for most HF ballast till 1MHz and/or battery powered oscilloscope - it have to be really not connected to the ground, even the "Y" capacitors in it's supply are NO-NO).
Normally the rms voltage at HF should be about 10% above the specified voltage for the mains frequency, peak value ~60% higher. So F40T12 should read about 115Vrms/160Vpeak.
If the voltage is within +/-20%, I would consider that as a good lamp (normal tolerance range), but if it goes to 50% or more above the expected value, such lamp is likely dead.

Good way, mainly if you are not sure about the lamp specification or your meter have not sufficient frequency range (at rated maximum frequency the error could be +/-40%) is to compare the voltage of multiple lamps of the same type. If one's voltage  is way above others, it is likely really dead.
Other signature is the DC voltage component (use a DC meter switched to the range corresponding to the AC voltage): The lamps nearly never age symmetrically, so if you see the DC voltage component more than ~10% of the AC voltage, the lamp is likely dying. If the DC component is way less, the lamp is still good...

If you have high voltage neon transformer like 6,000 - 15,000 volts and less than 30ma... It will lit up any EOL tubes as long the gas/vacuum still intact while even inside broke electrodes. 

The catch you have to use special wire rated for very high voltage and custom Insulator bulb holder/socket. Regular sockets only goes good up to 600's volts typical. So careful and even little 20ma pass you could kill you if mishandle with high voltage transformer.   

I was told it could last 15,20,30, or even 50 years? of these EOL tubes hook up with neon transformer, maybe who knows...    


            

@AZTECH:
But the 30mA would be about 1/10 of the rated light output, while it would run the cathodes in cold mode, so with about 1/2..2/3 of the total lamp power ending there... Cold cathode tubes are usually designed with way higher voltage drop both per unity length, as well as total, on the positive column of the discharge, so both their brightness, as well as the efficacy, is then way higher then with the originally hot electrode lamp run on the same gear.


And even most IS ballasts are able to somehow operate the tubes, their ~1..1.5kV ignition voltage and ~600V running OCV is usually sufficient even for cold cathode mode. But as the voltage drop across the tube is way higher than designed, the real power transferred to the tube is high as well (so the ballast have to handle higher power than it is designed for for longer than few 100ms for the startup). Moreover the high impedance of the tube make the ballast output stage to operate in a phase regime not really convenient for the output transistors (hard switching). Moth could yield worse ballast reliability...
And other aspect: The extra power end up heating the electrode with the bad emission, what overheat the tube there. This overheating could be so severe, the socket holding the lamp could melt...

When I lived in West Columbia, SC I had a grow light ( a 2 lamp 40 watt rapid start fixture) and in the lag side I put in a Philips F32T8 (CW color) with one cathode broken as someone bumped it. It always lit instantly. It ran fine. Those 2 fixtures were destroyed in my 2006 fire, but the lamp had put 15,000 timed hours on it and showed no aging.