Hi. I was wondering what causes a basic core-and-coil ballast to fail? I am thinking that such a device should have an unlimited lifespan - barring an overheat condition caused by a  lamp (or starter) malfunction? In other words, what actually wears out in a simple ballast? I understand in a lamp there are electrodes that lose emissive material, but under ideal conditions what (if anything) is used up in a simple ballast to cause it to fail?

Based on some of what I've read in other posts I know I'm asking this question in the right place. 

Corrosion of both the insulation and metal parts, material fatigue cracks by vibration (hum) yielding shorts - first between core foils, increasing their eddy current losses and overall vibrations, increased vibrations cause winding impregnation to crack, letting wires move towards each other, so rub off the insulation,...
Generally apply the rule "10degC higher working temperature => half of the life", but on good quality products we are talking about decades.
But most ballast failures i've met were due to water leaking into the lantern, the rust really kill the ballast...

Temperature sure has big effect for life of a ballast. based on what I have seen modern days "plastic fixtures" usually fries ballast faster that those that are fully metal. I think older magnetic ballast last much longer than newer one. I know couple of store that uses over 40 years old shoplight fixtures (mostly 2x36w with heavy and big preheat ballasts) and most of them still had original ballast. one electrician of hospital gave me dozen of 20 years old wet space fixtures that were used in loading area of hospital, most of the ballast looked pretty crispy and the electrician said that many ballast has allready once replaced. Partially in later case I blame the plastic housing of the fixture for frieng those ballast.

Edit: how much would be ideal running temperature of choke-type preheat ballast?  

@Roi: The colder the better, ideal is "room temperature".
Older ballasts themselves last longer, because they were physically larger and often had lower losses, what both yield to lower operating temperature.
But there is downside of it: They were heavier, so very often deform the thin sheet metal lanterns were made of, yielding failure of water sealing, so the water might pass in and then it kill all electrical equipment inside, including the ballast.
The best you might do is to mount the ballast so, the lantern body serve as an heatsink, so with good thermal contact with some large and robust piece of metal. In plastic housings is good practice to use large aluminium plate as base of all internals, it would serve as good heatsink for ballast and at the same time good heater for lamps (it take away heat from the ballast and give it to lamp tubes; especially helpful when cold)

Thanks for the educational responses, guys! I never thought about the effect of hum on ballast life.

i would suspect long term overheating to be the big killer.most coils are dunked in varnish and vacuum impregnated which should relieve much of the vibration stress.also gives protection from water.

I have a really big old ballast that had several decades of use before going into storage, still goes strong now with no noise. That must be because it's so big and is cased pretty thick metal which takes the heat away.

The ballasts most at risk are simple series chokes in a switchstart circuit because they can be left for weeks, months or even years before someone replaces the tube and stuck starter. Yes I have seen one that was switched on for about 8 hours per day, every day for 3 years with a dead tube and stuck starter and still worked after but I shouldn't think it had much longer to go.

@Medved, I agree with you on water coming into contact with ballasts, they just rust really quickly and leave the internal coil exposed to water which doesn't last too long. I got a lamp with a ballast in which had got wet when the roof leaked and it's now a lot noisier than it used to be.

From experience I've seen that a good quality ballast that is looked after can last a lifetime.

Sometimes ballasts fail due to poor quality, old age and power surges.

Bigger ballasts with bigger coils last longer than the more modern smaller ballast with smaller coils but operates the same type of lamps.....

I have also noticed that much more simpler ballast will outlast a more advanced ballast with capacitors and some extra components which has more parts to fail and possiblity damage the ballast....

Mercury vapor ballasts witout a capacitor (like HX) will last through no lamp while CWA ballasts will easilly get killed by no lamp.

IMO its luck of the draw. I have found certain brands to be more reliable. Some will just last for ever well near enough. Others will only last a few years. I have repaird Magnetic ballast fittings where one magnetic ballast has failed very quickly and the one next to it has simply not failed ever at all! The old Atlas and Thorn ballasts in every type just didnt seem to ever fail and were very very good! I think the late 70s and Early 80s was a very good time for magnetic ballasts in the UK well switch start anyway!

Older ballasts here (Israel) were open-core (ie. no closed magnetic path in the core), made like this :

Bunch of core plates --> wrap in cardboard --> coil --> more cardboard --> some more core plates --> enclosed in sheet metal wrap

The cardboard layers (and all spaces between them) are open to the outside (insides of the light fixture), so water / dust / insects etc can get in

If water gets in and soaks thwe cardboard (esp. in old ballasts) it can leak current from the coil to the outside metal and eventually flash over



New ballasts sometimes have open contact where the coil wire is connected to the terminal - The terminal connection depends on pressing a sharp metal edge against the enameled wire to cut through the enamel, but there are sometimes manufacture defects where it was not pressed hard enough, or when tightening a screw when installing causes the sharp edge to pull out a bit



Without stuck starters the ballast will usually live for very long time

In my experience the water is mostly damaging to the core: It is not as well protected as the winding, so it soak in the water, start to rust, this rust cause the core to expand, expanded core stress the winding, till the insulation layer on the wire crack and you have flash-over.

Freezing of the water in the winter add to this as well...

@Ash: The "more core plates" is what make the path to close the magnetic loop.
If I understand well, similar design were ballasts I've seen here some time ago (pre-1989), their plates orientation was so, there could easily form eddy currents, with consequent quite high losses...
Open magnetic circuit designs would be huge, heavy and too lossy...

I have some (2 of them are installed with F36 T8s around the home), will post pics tomorrow

I recently purchased a 70's 14 watt desk lamp, only to get it home and find out it would'nt light. Sure enough, the simple little ballast had burned up. In the case of this particular fixture, I wonder if the previous owner didn't accidently leave something heavy on the switch, which would have duplicated the effect of a stuck starter.