Here on LG i see lot of complaints about "safety" of integrated CFL's (those "energy saving incandescent retrofits"), but all i've seen originated from US and/or Canada, so 120V areas.
While my "most violent EOL" experience with cheap, crappy CFL's is small "PUF" and tripping circuit breaker (no smoke, no fire,...), while cheap incandescents do "BANG", shatter glass everywhere and trip the CB as well. Actually my CFL's always died more "peacefully" then incandescent. And to note, i'm living in Europe, so 230V area.

For me it's clear the difference in incandescent behavior (larger voltage => arc forms more likely and last longer), but it is a mystery, why should be here any difference between CFL's.

So the question i ask to the "public" are:

Are my observation representative?

Do you have any explanation, why 120V CFL's tend to die so dangerously compare to 230V ones? (even if no real fire, the smoke itself i do consider as the toxicity danger)

Or here described behavior is here only concentrated from large usage (as people here do often maintain lighting for whole wide family and all their friends or even professionally), so the thread is in reality not as severe?

I have been thinking same thing. I myself have never experienced any dangerous with CFL. Some cases they only say puf and some smoke will come out and maybe the fuse will trip but usually CFL just stop working and thats it. I too live in a country where mains voltage is 230V.

Yeah, the few that died in our house died silently. Actually I have a "dead" one but when I tap it, it usually comes on...

The short-life and crapiness complaints are a bit weird to me, we bought some cheap Philips lamps about a 5 years ago and ALL of them are still going today. Few of them burn daily for a few hours. The one that died is a "noname" one but it still lasted about 3 years.

There's an electrolytic cap in the lamps, which can explode with quite a bang when it dries out. Most people throw the lamps away but this is a simple to fix problem.

Maybe the 120V ones die so loud because the short-circuit current involved is higher, and that they require bigger transistors (which of course go with a loouder bang) to switch the higher operating current?

CFL short lifetime is mostly caused by high ballast temperature. And while the optimal position for an incandescent is vertical, base-up, lot of luminaires are build to fulfill this (to get longest bulb life and lumen maintenance), plus there is an enclosure diffusing light. But for CFL this is the deadliest combination - enclosing prevent good ventilation and base-up causes the ballast is in the hottest spot.

But the question was not why CFL's die too soon, but what cause the difference of the EOL failure mode between 120V and 230V versions.

My colleague put an hypothesis, then the reason is the voltage doubler in 120V versions: When transistors die by overheat (due to e.g. hard switching at lamp EOL), they short the DC bus after the rectifier. On 230V versions, as on the input is a bridge rectifier, this short causes very huge current from the mains, so the fuse (or CB) trip and disconnect the power before some bigger component overheat to cause fire. But when doubler used, the short current is limited by doubler's capacitors impedance, so the current is low to trip the fusing, but large to gradually overheat them and set the fire.
But this does explain only, when the smoking does originate from capacitors, not the tube.

So still, how to explain this ? Where the severely overheated component are the tube ends and not capacitors in the doubler? This failure mode seems to be quite common in 120V area (repeat's many times on LG), but i have never seen any in 230V area. In Europe we have only hard-shorts (and on better quality pieces the internal fuse react quicker, the CB trip)

Just a poor guess, but maybe the capacitor between the filaments shorts and causes the full load current to go only through the filaments, overheating them and melting the plastic around them?

Just a poor guess, but maybe the capacitor between the filaments shorts and causes the full load current to go only through the filaments, overheating them and melting the plastic around them?

This short might be present there forever and does not cause any significant overheat, as the current is not bigger then the operating one - the ballast act as a current source, so the short is the minimum wattage transfer. The huge electrode loss is caused by increased cathode fall, when the electrode looses it's emitting properties. When good electrode has about 9V fall (so about 1.5W dissipation), the worn-out electrode might have easily 100V fall, so 16W dissipation. And the filament resistance causes drop about 10V, so again very small 1.5W. But when filament wear away, transistors start to overheat much quicker then the lamp (as are smaller) and in 230V versions break much (so short, so fuse acting) sooner then the lamp overheat.