From even a cursory look at the components used in most CFLs, it is obvious that the cheapest possible parts have been used, and many of these parts are simply not suited to the voltage, current and temperature to which they will be subjected. The use of 400V DC capacitors across the 230-240V mains is of particular concern, since it is known to a great many technicians and engineers that these capacitors will (not might) fail in this position. This is not a problem with 120V mains, as the capacitor can usually withstand the lower voltage without failure. Since they only have to last for a few thousand hours, the manufacturers obviously think that's enough.

No-one seems to care if the lamps fail with a flourish, but such failures will damage consumer confidence very quickly. Some manufacturers claim that their 400V DC capacitors are rated to 220V AC. Since the nominal mains voltage in Europe is 230V and in Australia it's 240V, even the makers' rather adventurous rating is exceeded anyway. Also, no-one seems to have noticed that using these caps at high frequencies imposes a derating curve from as little as 2kHz. A 33nF 400V Vishay or Philips MKT polyester cap is rated at only 32V AC at 30kHz. As the temperature increases, the voltage rating is reduced even further. These caps are not safe, and should not be used if their voltage rating is exceeded (which it is, in almost all cases). A data sheet for these caps is available from any number of sources. Check for MKT370 data sheet(s), or click here.

Although not actually stated on the specifications, the 220V AC rating is not for continuous use. If it were, why do the same companies make other - and more expensive - capacitors that are designed for connection across the mains? Simple, the 400V DC caps may be used in all sorts of equipment where AC voltages will be present for periods of time, but will not be continuous. These circuit applications will also be relatively high impedance (limiting the maximum current flow), and designed so that a capacitor failure will not cause clouds of smoke. The device will stop working, but nothing else will happen. This is in contrast to the use of the cheapest possible parts where there is little or no limit to the maximum current, other than the house fuse or circuit breaker.

The photos shown here are courtesy of Doug Hembruff's Impact website. The examples are of US or Canadian origin, but the failure modes are universal. There are additional photos on Doug's site, and similar pictures are scattered across the Internet.

According to various industry groups, these failures are considered normal. As noted in the main article page, the CFL is the only product ever offered to the public that includes acrid smoke and severe burning of the outer casing (caused by component failure) as a supposedly normal end-of-life experience for the purchaser. Any other consumer product that failed in this manner would be subjected to an instant suspension of further sales, and a total recall of affected models.

The manufacturers and distributors would also be subjected to fairly intense scrutiny, since the product is obviously faulty. Why is this not the case with CFLs? I cannot understand how a product can fail in this manner, and not only does no-one seem to care, but they don't even think there's something seriously wrong.

In the US, even the Underwriter's Laboratory (UL) claims that smoking and overheating was a common occurrence for this type of lamp at end of life. It beggars belief that anyone, anywhere, would call this normal.

Severe Burning Around Tube Base

The above lamp (Commercial Electric - North America region) overheated and burnt the plastic housing filling the user's bedroom with acrid smoke. The lamp did not shut down and continued to smoke until power was removed. This lamp was directly over the user's bed - very fortunate that he was there to switch it off before anything worse happened. This failure mode seems to be fairly common, and even a quick check will reveal just how hot the filament ends of the tube become. In normal use, the filaments dissipate at least 3W each and are enclosed in the glass tube - they get very hot indeed.

Would any lamp that failed in this way drip burning plastic? Have you ever seen a guarantee on the pack that the lamp will not (and cannot) catch on fire, or drip burning plastic, glue, or anything else?

I know I've never seen any such guarantee. Note too that the neck of the tube got hot enough to crack the glass near the melted area. There is no way that this (or the following) failure can be considered normal - as long as this continues, CFLs are potentially very dangerous products. To allow the general public access to them is crazy - they should be restricted to professionally trained lighting experts, not sold at supermarkets.

The above photo is of another Commercial Electric CFL from Home Depot in the US. In this case the hapless user had no luck for some time when trying to contact the supplier. In more or less the user's own words ... "Commercial Electric was not too helpful, in fact I could tell [the woman on the phone] was reading from a script when I described my trouble. She said it was due to the ballast becoming lose during shipping and normal use. To me that is a defect. I was not that concerned about the warranty but more for safety."

"Normal use" does not cause a hole to be burned right through the casing. The position of the hole is about where I'd expect a fusible resistor to be located, so it is possible that this lamp (and others that have the same problem) drew excessive current - perhaps because a dimmer was in the circuit. Unfortunately, there is no additional information or a photo of the insides, and no way to know for certain.

Since smoke and burnt plastic is apparently "normal", perhaps our legislators will modify existing standards for other products - it could become very exciting if all consumer goods were allowed to fill rooms with smoke or burn holes in the case as a normal way of telling us they no longer work.

In the US, several CFLs actually were subjected to recalls because of overheating and melting/burning plastic. One can only assume that the affected lamps were really bad, because what is shown above obviously wasn't enough.