Hi all, today i got an strange ballast failure.
Some years ago, i took an ballast from an broken 25w cfl (bad cfl, ballast was good) and i put in an blastic box (not sealed, it had a lot of ventilation) and then, i connected an 25w T5 HE tube on it (with 40+40cm wires), it worked perfectly until 3 hours ago, when i tried to light this lamp with result of only glowing electrodes.
Because i wanted to take an photo, i kept this connected for some seconds, then i got an pop sound in the ballast and it failed.

Now, i opened the box, everything is fine, except an 22Ω resistor.
I suspected failure on pre-heat capacitor, but it is good.

Does anyone have ideas about this failure?

Have you checked the tube? MAinly if it ignites and if the electrodes still provide electron emission?
It appear to me the tube didn't want to ignite (air inside?), so the ballast was just keeping the high current and voltages for too long time, till it failed.

The 22Ohm resistors are in the base, so it's failure means the transistors had to fail before (taking the ~1Ohm resistors in their emitters as well).

Hi, thank you Medved for the answer, but i can't test this tube at the time; i don't know if some air leaked in, is there any test i can make without ballast?

A Plasma Ball could be used to test the vacuum.
If that is OK, the lamp should ignite well.

By the way I'm not that sure, whether a 25W CFL ballast woudln't overload the F25T5HE by too high current and the T5HE lamp overload the ballast by a too high arc voltage, the CFL's use to have 0.22A/120V tube, while the T5HE are 0.17A lamps (so the 25W will be about 160V arc voltage lamp)

Unfortunately I have lost my hard drive, so I do not have the OSRAM 2008 lamp catalog anymore with all the parameters listed I used as a reference...

i don't have plasma ball, however, i'll try other methods: despite heavy overloading, can i check with 58W T8 ballast?

That could be again problem for both: The ballast is designed for about 100V arc lamp, so the ~170V will overload it. If the ballast is protected, it will shut down.

And for the lamp, the more than 1A provided by the F58T8 ballast for preheat will vaporize the electrodes instantly...

Filament continuity you may check with just an ohm meter.
Other option would be some small CFL/minimature T5 magnetic ballast in series with about 2uF/450V capacitor, all that connected in place of the ballast in a manual preheat setup (for the preheat use just a jumper wire). If the lamp will flash with a discharge, the vacuum is still OK.

mmm, ok... I'll try 8W T5 magnetic ballast with cap.
I don't know if i can use my mosquito zapper.

edit:
what about using HQI ignitor instead starter? This will become an instant start ballast, right?

I don't think the HID ignitor will work. It provides sufficient voltage for the initial breakdown, but then the mains isn't sufficient to feed sufficient current for glow (cold cathode) to arc (hot cathode) mode transition and the ignitor does not give sufficient current to heat up the electrodes.

The reason is, the cold ignition takes actually three stages and the first two need higher voltage than available with the mains:
1) Initial breakdown of the gas. Need kV range, short time sufficient, so the HID ignitor works well for that part.
2) After the initial breakdown, the high current discharge should be formed. But as the electrodes are cold, extra voltage drop is on the cathode side. As this extra voltage tend to be in the order of 100..200V, it becomes above what the mains could supply as the OCV. The HID's have their electrodes very close to each other, so the voltage across the anode column is rather small, so the mains voltage becomes sufficient for the high current discharge.
And the discharge has to be able to carry significant current, as the cathode fall dissipation is the only heat source to warm up the electrodes. When the current is not sufficient, the lamp does not transition to the hot electrode mode and the electrodes get heavy wear (e.g. on the battery powered fluorescents with flat batteries,...)
3)Once the electrodes warm up, the discharge goes to an "arc mode" - a hot cathode spot is formed, takes over all cathode current and maintained by the focused cathode fall dissipation.

The thing is, with the fluorescents the 2'nd stage would need too high voltage, above the mains. With "preheat" ballasts this phase is practically skipped by heating the filaments before the ignition attempt, the HF electronic ballasts are usually able to deliver sufficient current for the transition, but it means way higher power than the rated one, so all the components are stressed by that. As it takes just a second or so, they could handle it (or preheat the filaments before the ignition attempt).

Hi, i tested lamp with my bug zapper and it works, so it didn't losed vacuum.
Now i think about ballast failure.

The ballast has all the symptoms f being abused by a lamp with too high arc voltage. Normally that uses to happen at the lamp EOL (where the higher cathode drop is responsible for that), here I would even guess the lamp itself is overloading the ballast (I have just checked the Philips web site and it is rated to have 154V arc voltage, the CFL's use to have around 100V).

Expect both transistors and the small resistors in the bases and emitters to be blown off...

First try the ballast with something like F18T8 or so, I would expect to feed that by about 12W, so should work. If the ballast is OK, it should light.
But for the F25T5HE you will likely need something else, either redesign the ballast (add second coil in series with the main ballasting one, increase the DC voltage behind the rectifier to 400V by a PFC stage,...), or find something more suitable.

The resonance step up method is rather tricky, you will have to be careful to not overload anything.

ok, i won't try this: ballast was from an broken cheap 25w cfl, so it wasn't supposed to be used anymore (i have an bag plenty of them).

If you have plenty of the ballasts to do experiments on and won't miss them when they fail, just try another ballast, but monitor the temperature of the transistors.

If you have plenty of the ballasts to do experiments on and won't miss them when they fail, just try another ballast, but monitor the temperature of the transistors.

Ok, i'll try! Thank you!