1. What is better for a fluorescent - a magnetic ballast + starter or an electronic ballast? (Officially, the electronic ballast should be less harmful in case of often starts but I heard that it's not completely true.)
2. Some starters make very fast interruptions, the other less frequent, stronger pulses. The later (slower interruptions) surprisingly starts the fluorescent faster - there is just one unsuccessful flash and then it starts while the fast starter makes it fight for 4 or 5 seconds. Does it make any difference in the lamp life?
3. What is the main cause for the tube end blackening - long burning hours or frequent starts?

Thanks.

1. Depends on the type of electronic ballast. A lamp on magnetic preheat will usually consistently outlast a lamp on electronic rapid start or instant start. The only exception is 24/7 service, in which case the electronic may do better due to the higher OCV. However, a lamp on an electronic programmed start ballast might possibly outlive a lamp on magnetic preheat. It's hard to say since nobody here's done a comparison between the two yet.

2. Slow starters are better for the lamp than fast, blinky starters.

3. Blackening is simply a result of the lamp wearing out. Every fluorescent lamp will develop blackening upon undergoing EOL, regardless of the conditions it was run under. To partially answer your question, a frequently started lamp will not last as long as a lamp operated continuously with few starts.

Magnetic ballasts lasts longer than electronic ballasts and aren't sensitive to voltage surges and much less sensitive to open neutral than electronic ballasts. Also, a magnetic ballast with an electronic starter or even a thermal starter, can have much longer lamp life than even electronic programmed start ballast, because they preheats the lamp correctly, compared to electronic programmed start ballasts, which their preheating reliability is questionable (In Carmel hospital, I've seen numerous german Osram FH 28W/8xx HE T5 lamps that failed within several months on Helver programmed start electronic ballasts. On the other hand, we have US Philips F40T12/D on Shaingout magnetic ballasts and Arlen Pulsestarter EFS600 (Electronic starters), and Eltam Perfekt-Start R1 (Thermal starter), which lasts more than TEN years and are still looks like new.

@TheMaritimeMan, @dor123: Thanks.

1. OK, we have to count here all premature lamp deaths due to electronic ballast failures. I've heard that when the el. ballast dies, it usually destroys the lamp, too.
I was curious because Osram specify their lamp life (e.g. Osram Lumilux = 20,000 hrs. *with an electronic preheat). I think they mean the best quality EP. What is the expected life of the same lamp with the magnetic ballast + starter?

2. So it's better to use lower-range starters (e.g. 25-65W) than high-range ones (4-65W) because the high-range starters that must work even with a 4W fluorescent make the interruptions in case of 65W fluorescent so frequent that the lamp has to struggle blinking for many seconds.

3. I saw very old T12s like this with tens of thousands of starts (magnetic ballast + starter) and thousands/tens of thousands of burning hours and there is no blackening visible. (It could be because of the tube thickness - it's probably more visible with thin tubes.)
But it seems that the blackening is more connected with the EP than with a magnetic ballast.

Keep in mind that Philips and GE have been known to use cathode shields which help prevent visible end blackening until EoL while Sylvania doesn't use em which is why you see them start to get dark early but continue on a normal life.


http://en.wikipedia.org/wiki/Fluorescent_lamp#mediaviewer/File:Tanninglampend.jpg


There is some good reading here regarding the ballasts and what not.
http://en.wikipedia.org/wiki/Fluorescent_lamp

1) The damage goes mainly in the other direction: Failing lamp kills the ballast (overheats it's output stage). Truth is, with many protection designs the ballast component wear is, what makes the ballast protection not working, so one could argue if that wasn't ballast failure as the prime cause.
And to "wipe out the traces", the failing tube still seems to be working on another ballast, but in fact it is just frying it out as well (and that includes damaging the starter on a magnetic ballast - the starter does not have to really start to flash, just the heat from the weak discharge is sufficient to weaken the starter electrode material, damage the capacitor,...)
But indeed, very frequently the cold starting of cheap electronic ballasts is the factor causing the lamp to die rather early, if frequently switched...

2) The "wattage range" is not really related to the starter's timing properties (controlling how long it preheats,...), but it is important from the other side: When you operate some lower voltage lamp, a starter rated for the higher wattage ones ("..65W" or so) may stop preheating when the lamp ignites the discharge on not yet warm electrodes (glowbottles very rarely preheat the electrodes enough by the first cycle) and let the electrodes to warm on the cathode dissipation only. That mean the cathodes are for quite long time exposed to cold electrode discharge and that is not what they like as much... So for low voltage lamps (usually goes together with low wattages) always use the "4..25W" starter.
The starter timing is given by the hysteresis build in the starter and that parameter tend to vary more than a decade, so even when one starter shows good timing behavior, there is no guarantee another piece of the same model won't flicker like hell... Very frequently the "good" starter was flashing like hell when new too. The reason is, the only hysteresis is very frequently just formed by the micro welds between the electrodes and these are extremely unpredictable.
3) To suppress the darkening, some lamp makers used "cathode guards" with their products, metal "bracelets" around the cathodes, intended (beside other functions) to block the sputtered material from reaching the glass. With T12 it was quite common, because the large diameter caused the affected section of the tube was quite long. In the narrower T8 the sputtering affected section is way shorter (and the used fill gas provide quite some protection), these guards are not used anymore, so the blackening become visible.

Thanks.

@RyanF40T12: I read these Wikipedia articles some time ago and I hope I understand (at least partially) pros and cons of fluorescent starting types.
But (AFAIK), we don't have anything like instant, rapid or quick start here (if we are speaking about standard fluorescent T12..T8 tubes). These types of starts are rather used in the U.S...? In most cases, we use a preheat (either with magnetic or electronic ballast) here - that's why I asked how to do it best for the lamp.
I don't know, what way are most CFLs started - the best ones seem to use a programmed start.

I searched the Internet and it seems that electronic starters should double the lamp life (in comparison to the classical "thermal" starter). So are they the best choice? (All the electronics is prone to failures due to voltage surges etc.)

Here in North America, resonant start is the primary CFL starting method. Before a few years ago Programmed start (which Philips employed) had a fair bit of market share, but I don't think that's the case today. Some CFLs today might use straight instant start, but I don't know.

Regretfully we have a lot of "instant start" and a few "Rapid Start" T8 ballasts in many of our T8 fixtures with very few programed/delay start, however I am strongly encouraging the folks who order new ballasts for my church group to go with programmed start the next time they need to order some. 

I'm still pondering over electronic starters. Do they really extend lamp life "up to 2 - 3 times" or is it just another marketing BS?

OK, with a blinking glow bottle starter, there's probably a lot of "sputtering". But with a good one, that preheats for a while and then starts the lamp at once? Isn't it the same as the electronic one? Or, does the electronic starter moderate the input current in some way to prevent sputtering even more?

The problem with glowbottles is generally quite poor control over the preheat time.
It is controlled by the cooling rate of the electrodes (that is controlled pretty well - the mass, thermal resistance to the ambient and the closing temperature are well known), but over the thermal difference given by the starter mechanism hysteresis. And that is the weak point in the design.
Usually it leads to too low temperature difference, so the usual method for bimetal thermal switches by it's shaping to implement the snap action is not usable.
Usually the hystereis come from the micro welds, formed betwen the contact surfaces when the bimetal closes. The force required to break that micro weld is then the hysteresis. And from it's nature, this tend to be very unstable, so vary a lot (not only batch to batch, but piece by piece and even start per start of a single starter), so the resulting preheat time does vary a lot as well.
At some time back some makers were using permanent magnets: As the contacts closed, the magnetic force attracting one to each other formed that hysteresis. But this was always "on top" of the "micro weld" one (so as the welds were getting stronger overt the life, they lead to stuck starter problems anyway) and the magnets tend to loose their force when the starter become heated at the end of lamp life (and together with the related higher cost), it was abandoned.

Other factor limiting the time is the requirement to successfully start the lamp within few seconds upon power ON, in fact this was the primary performance criteria (2 second is the usual limit for well operating starter).
As the starter has to first heat up to the closing temperature before it starts the real preheat and the time required for that is way longer than the real preheat (about 3..5x), majority of that starting time "budget" is consumed for the initial heat up and then the cycling is made fast in order to start the lamp as quickly as possible. The resulting preheat time is then about 0.2..0.5s, sometimes enough to allow the arc build up with the first attempt, sometimes requiring multiple cycles, it any case the arc starts in a cold electrode mode and only after that heats them up to the operating temperature. And that means a sputter, which consumes about the same portion of the lifetime as about 3 hours of steady burn time.

In contrast to this, the electronic starter is turned ON and so starts preheating immediately at power ON, so nearly the complete "starting time budget" (so usually about 2s) could be devoted for the lamp preheating.
And the electronic allows quite accurate time control (+/-30%, so about factor of 2 from min to max case; compare to factor of 5..10 for the microwelds in the glowbottles), the 2s typical means about 1.5..3s before the first ignition attempt, the system could guarantee at least 1.5s preheat time still with the maximum 3s of the starting time.
That 1.5s means the electrodes are practically sufficiently heated, so the ignition take place without any sputter wear at all.

So when the glowbottle consumes an equivalent of about 3 hours life per start and the electronic none, with the 3hour ON (/15min OFF) per start test cycle it means about double life time.

I'm still pondering over electronic starters. Do they really extend lamp life "up to 2 - 3 times" or is it just another marketing BS?

To add to Medved's explanation, see FrontSideBus's test. Other members have also done tests that prove that electronic starters (or at least, Arlen Pulsestarters) indeed extend lamp life.

Great explanation, Medved, thanks! I didn't realize that the preheat time is so limited and how volatile glow bottle starter functionality is.

Thanks for the link, Trent. I like the test FSB has done. It's open and shut case even though those 15 minute intervals are a bit synthetic.

It's strange then, why electronic starters are rather not so easy to get (at least here). For example Osram don't make them at all, only glow bottle ones.
I think it could be that in its late hours (+10,000 or +12,000), the lamp efficiency is affected and there's a visible lumen drop. That might be due to phosphor changes as it is exposed to the UV light for too long - don't know. So even if the electronically started lamp survives, it doesn't perform so well at the end.

There's no large lamp preheat left here in the USA.  Rapid start has been around for many decades.  Electronic ballast with T8 lamps are rapidly replacing magnetic ballasts as they are no longer being made.  Of USA made lamps Philips is the only one that uses cathode shields but now only on 1 end. GE experimented with cathode shields many many decades ago. 

RS on 24 / 7 does well.  So does preheat on 24/7 service. Electronic instant start works well only with infrequent restarts. 


Powell

I'll tell you in no uncertain terms my fluorescent likes. Ahem...

TEEE-TWELVE!  Thank you.