Hi guys, I have often wondered this.  At work and in my garage with my Lithonia electronic ballasted T8's (both running Sylvania Octron Ecologic T8's they just shut off like an incandescent at EOL.... I just had one prematurely shut off in my garage yesterday, after only 6 months use, the rest are still going. The bulb refused to do anything in another fixture and a replacement bulb works perfectly.  Our old T12's never did this, what is the purpose and mechanism of this boring EOL....seems like even in proper circumstance it will prevent me from "getting every last bit" out of the bulbs...

Thanks,

Mike

Try to zap it with a piezo element - Any flashes ? If no then vacuum loss

Check continuity in the electrodes - is one broken ? If yes, possibly it snapped from mechanical tension (lamp manufacture defect) or maybe ballast killed it prematurely as part of its own EOL protection (ballast had a glitch and it overloaded the electrode ?) - Try to power it on a magnetic and see what happens

May be totally mechanical stuff like snapped wire in the end cap (from mechanical stress, thermal expansion etc) too

It is not property of the tubes, but of the ballast.
When the tubes are dying, their voltage drop increases.
Simple unprotected ballasts (or the magnetic ones) continue to drive the lamp, so it still light somehow (so it it is getting "the last bit of the lamp"), but such state is dangerous both for the ballast, as well as the lampholders.
As the electronic ballasts use a resonant output circuit, the increasing arc voltage mean the curents in the resonant circuit rise as well, causing an overload of the components.
The excessive voltage drop is in fact only on the cathode side, what mean there is an excessive power dissipation as well. And that mean the lamp body start to overheat there. As the tube is already at it's EOL, the damage to the tube does not matter, but the heat spread to the sockets as well and so these could get damaged (even so, they release the lamp and let it fall on the floor).
Magnetic ballasts are not as dangerous, as they are usually unable to hold the arc for so long, but the HF arc become very stable, so the danger become real problem.
Therefore the HF ballasts shall incorporate some sort of shut down, what should turn OFF the ballast before the EOL lamp could cause any damage.
Many ballast designs rely on the lamp filaments acting as a fuse - they pass the resonant current through them, so when the current increase (aided with the heat from the ion bombardment), the filament fuses, breaking the resonant circuit and so usually the ballast oscillator feedback. This approach have one weak point: When the filament is too strong (it could be when the tube looses vacuum - the air cool the filament so, it could hold the elevated current), it may remain intact till something else die (usuallythe ballast transistors or inductor).
Therefore the ballasts aiming for at least some reliability have to contain explicit EOL detection and shut down circuit, so the shutdown won't depend on the questionable tube quality.

And is it not a waste to not operate such lamp?
I don't think it make any sense to try to get the few hours till the lamp really die completely when it mean risking the lamphorder and/or ballast. Moreover the lamp loose it's efficacy dramatically in such state - the extra cathode drop does not generate any light, but it consume power, so it is way cheaper to replace the lamp beforeiteach such state.

The EOL may look boring, but the aim in light design is to provide a comfortable light, so if something is about to fail, it shhould be as least disturbing as possible, effects like flicker, flashing, or so are treated as really unwanted, so the best in class systems are on purpose designed so, to not create any color show or so...

I think that what MikeT1982 encountered, may be a scenario that I'm feared about it in the past regarding to CFLs and cheap 4 wires electronic ballasts, that have nothing besides lamp filament fusing: When the lamp filament fusing operates, the lamp continues to light until the filament fuses, then the lamp extingush. What if the lamp filament fusing operated before the lamp actually reached EOL, and so killed a good lamp prematurely?

The filament fusing can not operate prematurely. Either they are too weak to operate (that is fault of the lamp, as they would break on any other ballast),
or the high current is passingthrough them for too long, so they overheat. But that can not happen on a good lamp, as the filaments warm up rather gradually (over a second or so) and when they reach the temperature to support the arc, the lamp get ignited and that cause the high voltage and high current to disappear - virtually within half of the ballast cycle, but as the filaments approach the operating temperature, the discharge build up - all that happen in 10's of miliseconds. That mechanism prevent the good filaments from reaching any too high temperature, their material need certain heat energy for that and to get it there need the time.
Fromthe perspective of the filament overheating, the preheat ballasts are ofway higher risk for the lamps, as the preheating is not linked to the real filament temperature at all, while on the resonant ballast the link is direct: Hot filaments mean discharge and that mean damping the LC so, all the energy is swallowed into the discharge.


So if the filament break, the cause could be only within the lamp.

Cant some "self healing event" in the starting capacitor cause a glitch in the ballast that would make it blow the lamp, then continue working (with the next lamp and "healed" capacitor) like nothing happened ? Thats what i sspected in the 1st place

Cant some "self healing event" in the starting capacitor cause a glitch in the ballast that would make it blow the lamp, then continue working (with the next lamp and "healed" capacitor) like nothing happened ? Thats what i sspected in the 1st place

What may happen is either open, short circuit, or drop in capacitance.
Open cause to break the circuit, so both the current, as well as the voltage disappear, so the ballast stop working.
Shorted capacitor cause the resonator to disappear (there is no capacitance anymore to keep the resonator), so the ballast start to operate in an inductive mode, so the current is only slightly higher than the normal lamp run current. So the filaments would become only slightly heated, very close to a stuck starter on a preheat ballast.
The drop in capacitance would cause the voltage to rise and current to drop, as the characteristic impedance of the resonator is increased. But that is rather hypothetical scenario, as that would be possible only when the capacitor self-heal. But for that there is not enough energy in the capacitor and there is no other source in the circuit to feed sufficient current for the "recovery" fuse (in the capacitor structure) to open. These capacitors most frequently fail into a short circuit, but while once they do so, it is for good...

Physically the resonator can not deliver more energy than stored within it's components. And because all this stuff is sized for high frequency operation, although the apparent power in the resonator is rather high, due to the high frequency the energy stored in it's components is still very low, even during the ignition attempt (when the currents could reach ~5..10x the lamp rated current when the lamp fail to ignite) way lower than in the magnetic ballast at nominal lamp operating current.

What happens in the exvent when the ballast blows a lamp filament (for aging lamp) ?

What happens in the exvent when the ballast blows a lamp filament (for aging lamp) ?

The inverter is build so, it feed the circuit at (or very close to) the series LC resonance frequency and/or given minimum frequency (when the resonance disappear).

Now I would start with starting of the lamp:
You turn ON the ballast. At that time, there is no discharge, so the series LC circuit is nor damped, so have high Q.
As the inverter feed it at the resonance frequency. That mean the energy is build up there, so both the voltage, as well as the current amplitudes rise. The current is responsible for cathode heating, while the voltage for eventually igniting the arc.
Now when the arc appear, it become quite a low ohmic resistance parallel to the capacitor, so damp the LC circuit so, the Q drop (close to, or below 1). As the quality factor have dropped, the inverter is not able to excite neither high voltage, nor high current anymore, so the thing continue in the non-resonant run mode.
Now when the emission mix wear out, it cause the voltage drop on the cathode side. This drop is nothing else than an extra series resistance with the arc, so the total resistance parallel to the capacitor rise, so does the Q of the LC resonator. And as the wear progresses, the Q rises further.
This cause the inverter to again become steered by the resonant frequency, so attain a maximum boost due to the higher Q, so both voltages, as well as the current goes up.
And as the wear progress, so does the current rise, so does the heat delivered to the filament, eventually reaching it's melting point and so the filament fuses and so break. Note, than the elevated cathode drop heat the filament as well, so even lower current become sufficient for the filament to break.

 So during the normal start, as the filament temperature rises, the filament emission capability rises with it. And because the temperature for a good emission with a good filament is on rather low temperatures (for the tungsten to remain strong), the good lamp always reach the good emission temperature before the destruction temperature. And as reaching the emission temperature mean the resonance get lost, the high heating current disappear exactly at that moment, so such filament never reach the destruction temperature.
But as the filament get worn out, the emission temperature rise, so does the cathode heating. In fact the ballast keep the cathode temperature just so hot, so the system operate exactly with currents necessary to keep the cathode on such temperature. So as the wear progresses, the filament temperature rises, till it reaches the point of structural failure, so it break.

MikeT1982, I think your lamp has probably lost vacuum, my guess is most likely due to cathode break on an instant start ballast. I have 2 sylvania octron T8's that have lost vacuum prematurely, they will not do anything no matter what fixture I try them in.

Thanks so much guys.  This is very interesting.  I too suspected vacuum loss but i put the lamp in a bag and smashed one end carefully with a hammer (had to crush it to fit it for disposal although bad idea i agree :-/) and it made the very loud pop of implosion.  Possibly partial vacuum loss? It could be some other early end of life too as the other fellas were saying.  Very interesting!  I just remembered i had smashed it and figured i would mention about the loud pop.

Are you sure the "pop" was from an implosion and not only from the cracks spreading along the tube (if there was a tension in the glass from the not completely even cooling during tube manufacture)

Did your lamps have any sort of blackening at the ends?, I was confused about mine at first because they have very little end blackening. Mine also don't have any clear spots that you would normally see when the air rushes in so it would have to be a slow leakage.

Medved, That's very possible! I didn't think of that you're probably right! Lighting guy, I bet you guys are right about vacuum loss, especially since you have 2 that did the same! Thanks guys. All the others are still going strong.  My first thought was that my ballast has gone bad and I was worried but relieved that when I put in one of my backup tubes it works fine! Yes the ends were starting to get slightly blackened, all the rest of them are too, there were no clear spots I could notice as in sudden inrush of air blowing off phosphors from a spot on the glass so I bet a slow leak nailed it on the head, as these bulbs are rarely used more than five minutes at a time unless I am out working in the garage. They truly don't have very much runtime on them. I was hoping for them to last a year or two or three more. It is very interesting to know the method of failure now:-) well at least to know what it most likely was! :-)