I'm wondering how fluorescent tubes lose vacuum on IS ballasts. I see that when the filament braks, the support wires slowly melt down to the base of the bulb but how and why does this happen? Does it have to do with the fact that these lamps are running cold cathode (even thought Teh electrode stems have melted and are red hot) or just depleted emission mix on the filament that causes breakage. I've also seen tiny arcs between the tube seal and the electrode stems. Why does this phenomenon occur as well?

Electrons in the outmost shells of atoms in metals are free to move inside the volume of the metal part, but when you try to pull the electron completely out, it pulls back : The electron is negatively charged, the rest of the metal part (which now lacks an electron) is positive and they pull each other

To overcome the pull force and get the electron out anyway, you put in energy. The energy required depends on the material and temperature (higher temperature - lower additional energy required). If you try to pull the electron out by electric field then the energy you put in depends on the voltage you apply and the charge of the particle you pull. The charge is 1 electron so it is allways the same, then the energy provided depends ultimately on the voltage

The emitter coating on the cathode is a material designed to release electrons with the least possible energy requirement when hot. So a hot cathode lamp with emitter coating will take the lowest possible voltage

The overall lamp voltage depend on the lamp : the voltage drop on the cathode (thats the part that make the cathode losses) + the voltage on the length of atmosphere between the ends of the tube (thats the part that is doing usefull work for making light)

The current (set by the ballast) * voltage drop on the cathode = the power dissipated on the cathode, so the heat. If the voltage is high, the power dissipation will be high and the cathode will heat up, lowering the voltage. If it is low, the cathode will cool down and the voltage will get higher. It balances itself at the point where the temperature match the voltage



When the emitter is depleted, the arc goes from the tungsten or from the support wires, but for them the voltage drop is way higher than for the emitter, so the point of balance between voltage and temperature is way higher as well. The temperature is higher then the melting point of the metal then it melts

Metal is good heat conductor. As the arc point melts close to the stem, the heat also goes to the section of the wire that is fused through the glass. That expends, untill the glass cannot handle the tesion and pops

I am not sure about the arcs to the glass stem but i can think of possibilities :

The metal from the wires evaporates, then it settles on nearby surfaces including the stem, and including the area in contact with where the wire comes out, so the film of deposit is connected to the input voltage and arcs

Glass at high temperatures becomes electrical conductor, then maybe when the point of arc gets really clkose to it the glass itself too starts conducting something

And the problem occurs mainly on the IS ballasts, as they are designed to feed the load by the high (about normal arc) current even when the lamp voltage is way higher than usual. The original motivation is to heat up (by the cathode drop times the arc current) the filaments as quickly as possible, so they reach the emission temperature and so reduce the cathode fall voltage, so stop the sputtering. (the sputtering happens, because the voltage drop there actually accelerates the ions, so higher the voltage difference, faster the ions, so they hit the cathode surface by way greater force and just sandblast the surface off).
The other electronic ballast designs usually have either the filaments in series with the resonator, so once the excessive heat up happens and the tungsten breaks, it means the circuit stops generating the high voltage, so the arc get extinguished. Or the circuit is made so, when there is the high voltage for way too long (longer than some seconds), it shuts down the inverter. The reason is, the operation with such excessive voltage boos causes quite high stress on the circuit, so the intention is to prevent such loading when the lamp is actually failing.

And because these ballasts have this capability, they are able to provide that huge heat even when the emission mix fails.
The classic RS or preheat just do not have such reserve, so once the cathode fall increases, the ballast becomes unable to hold the arc and it extinguish, or the starter triggers and starts to "preheat" again (EOL preheat flashing).

I've seen that glass becomes electrically conductive at high temps (I used a vacuum stem from a dead incandescent bulb and heated it with a torch, then I used an IS ballast for the power.) I have also seen the bulbs get so hot that a pinhole or crack forms in the glass. I've also heard of broken filaments arcing during preheat EOL and heating the glass enough to reach incandescence and electrical conductivity, and thus vacuum loss. Although it's rare, it can occur and I've seen some post a picture of it here before.

I have seen on a couple of occasions a dead tube where the white coating has been destroyed and there is a couple of inches of clear glass its usually very streaky. Would this be caused by loss of vacuum when the electrodes melted? I only saw this effect a few time and not for many years

Once the vacuum get lost, there would be no traces of discharge at all.
The clear glass at the ends, together with vacuum loss, usually means the glass cracked quite abruptly, so forming quite strong air blow, blowing off the phosphor dust.

Thanks I've not seen it for ages so I guess that tube makers have fixed that problem

It have little to do with tube makers, but rather with the gear used

SRS is more prone to it than switching based gear (Switch Start, Perfect Start, Vivatronic, ....), but now SRS is slowly and steadily becoming rare since it does not work well with 36W / 58W T8 lamps, while the switching ones stay in use as they work well with all lamps

Electronic gear used in Europe is usually EOL protected, so again it would more often cut out before the lamp reach the condition where the stem lead  wires burn to the glass. In the US it is opposite, they use a lot of the 2-wire Instant Start system where the vacuum loss EOL is pretty much the normal mode of EOL

I second that, Ash. I have seen examples of violent eol with SRS gear on here before, where the cathode blew out and the tube lost vacuum. I have heard of preheat tubes losing vacuum (albeit rare), due to a busted filament arcing on the inside. I wasn't sure if this was on 120 or 240 gear with the preheat failure like that. I constantly see eol tubes at the grocery store where the tube is black from three inches to the end on one side, and missing phosphor on the other, signifying a violent eol. I have also had a vacuum loss on rapid start, but it was on an electronic rapid start instead of magnetic.
On another note, I have also see. Rare instances of mercs losing arctube pressure, due to shoddy construction (tiny electrodes) and sometimes bad moly seals from water left inside the outer bulb.