Gas discharge lamps consist of effectively 4 main components:
Each of the components has a specific function, and premature failure of each is can, at times, be attributable to some power related cause (but not always, so please don't hop on any bandwagons!).
Please note we did say premature failure! There are some who claim a lamp or component failure at or around the projected life is a premature failure. A 1000 hour life means even down to 750 hours is still acceptable!
Before we enter the depths of fluorescent lamp failures, we thought it may be advantageous to explain how fluorescent lamps work. Below is the circuit of a standard lamp fitting. As these are still the primary type in use we'll only cover these, although there are 'fast start' and 'electronic ballast' in use.
When power is first applied, the mains voltage is effectively found across the starter. This warms up and bends a bi-metal strip causing it to short itself out. The filaments then warm up and, in turn, warm the gas in the tube. The starter cools (as it has shorted itself out) and breaks contact. As the current was taken through the ballast there is, as the starter breaks, a huge 'back-EMF' spike which attempts to cause the gas within the tube to 'strike'.
If the gas was not warmed enough, the starter will repeat the performance of shorting and breaking. If the gas is warmed sufficiently, the current through the tube increases the losses through the ballast which, in turn, reduces the voltage across the tube and consequently the starter. This, therefore, reduces the heat available for bending the bimetal strip and the starter then ceases with attempting to short out.
Ok, sounds simple, so there is really very little that can go wrong. This is true, except when things do go wrong, they can range from merely irritating through to spectacular.
Now onto the failures; Each component is described in detail below with possible failures.
"The heart of the system"
There are different gases used for different applications. Common ones are mercury, sodium, argon, and xenon. Often a 'starter gas' is added with a lower 'strike voltage'. Neon-Sodium is a common mix. The starter gas conducts when the lamp is first turned on and heats the main gas which then starts to conduct as the heat 'loosens' the electrons within the molecules.
"Applying make-up to a very ugly load!"
Correcting this inefficiency (power factor) is done with the aid of a capacitor whose 'opposite effect' cancels out the current requirements of the predominantly inductive properties of the ballast. The intended result, according to the source, is the lamp fitting to appear less inductive and look more like its incandescent friend (i.e. resistive).
The power factor correction capacitor does not affect the actual operation of a gas discharge tube in any manner, shape, or form. The current through the ballast and lamp remains unchanged!
Having the capacitor, however, drastically reduces the feed current to the fitting as a whole - sometimes by as much as half. This reduction means the sizing of the wiring, including fuses and circuit breakers, can be reduced too making the existence of the PFC cap more than worthwhile. Should a cap fail (and be disconnected), the feed current to the fitting would rise. It is, therefore, imperative that failed capacitors are replaced.
Force the dielectric to conduct a large chuck of current, and it acts like a fuse - it burns. Most times it just simply vaporises and becomes inert, but other times it chars and becomes conductive. If a 'hole' develops in the dielectric then the insulating properties of the dielectric cease and standard ionisation breakdown can occur - as happens when two contacts are two close and a spark 'jumps across'.
There is a related enemy being voltage, but it is actually not the voltage that kills a capacitor. Although it is breakdown that occurs, the current is what does the damage. There is a strange issue with a capacitor; It is itself that delivers the initial damaging current! As the capacitor is strapped across the mains, the mains usually assists the spark and delivers the 'killer blow'.
Heat does not help the situation as this usually causes the innards to distort, often reducing the insulating properties of the dielectric as well as causing the conductive parts to move together thus further reducing the voltage handling capabilities. Capacitors are rated with maximum temperatures, and good ones show how far to derate the voltage vs. working temperature.
PFC caps also have a maximum VAR rating. By acting as a power factor correction capacitor, energy is transferred into and out of the capacitor, and this is not a lossless action and heat is a result. Working a capacitor at its maximum VAR rating will again leave little headroom for what real life will throw at it.
While on this point, it may prove prudent to jump ahead and read up on "harmonics and PFC capacitors". Harmonic content on the supply voltage will cause higher currents though the capacitor (i.e. the dielectric) leading to premature failure.
There are cheaper, non-hermetically sealed constructions allowing moisture ingress to seriously challenge the breakdown voltage of the capacitor. Also, the cheaper ones are not particular about the discharge resistor and have this touching the capacitor innards thus heating the sensitive dielectric to well above working temperature.
And, finally, if there is one thing that can even damage a really good quality capacitor, it is switch-on inrush. The capacitor is usually completely discharged (by the internal resistor), and switch-on has a knack of occurring close to the cycle peak. This results in a huge current being dissipated through the dielectric seaking out all the weak spots.
This situation is exacerbated when the installation has lower feed impedances (the installer uses thicker cables to limit voltage drop) causing even higher inrush currents. Multiple luminaire installations are most vulnerable especially when wired 3-phase delta. A cure for this is NTCRs, and is covered in this page.