Fluorescent start voltage measurement.

Problem: Too many 40 watt fluorescent fixtures with questionable ballasts, sockets, and bulbs.

Theory: Measure the striking voltage of some fluorescent tubes. This would allow discarding tubes that are actually bad, and proof of good tubes to test the ballasts with.

Available equipment: Fluke 27. It has min/max hold, auto range, manual range set, and a limit of 1000 volts. The 1000 volt limit worries me, and the time the meter needs to acquire a valid measurement.

I expect to buy a ballast for this test jig. That allows the option of using a fast switcher or a tradidional 60Hz magnetic ballast. Using a double wide ballast (2 tubes) would be good because I can use it as a replacement ballast after the testing is finished. If using a single tube ballast would make the measurements easier, I can afford that.

Ideas about maybe a "peak sample and hold" for high voltage AC? Other ideas are welcome!

Ballasts/Bulbs

I've been trying to find a ballast mfg. that I could modify to run at 38.5 KHZ. I have programs for IR and ST Micro. If you could post any brand name that uses either of these chips I would appreciate it. But on to your problem. Which type of lamp? T5, T8 or ??. Seems like you could just use a high voltage diode and a cap to do a peak detector. Maybe just reset it with a push button. Failure to strike may be close to your meter depending on which bulbs and ballasts you are using -- but probably ok. 800 volts for T8's. Edit - no problem thats peak to peak.

You can do a peak detector with a rectifier and a cap. Be careful the cap is not too big or it will effect the boost operation of the starter.

Fluorescent light modelling is quite complicated. The strike voltage is dependent on temperature (including how long since turned off) and aging of bulb.

Run current versus lumens is a function of excitation frequency. Optimum is 20-30 kHz. At 32 watt T8 running from a 25 kHz electronic ballast gives as much light as a 40 watt T12 on a 60 Hz mag ballast.

You can easily change out 40w T12 with 60 hz ballast to 32w T8's with electronic ballast.
https://1000bulbs.com/product/2204/BG-GE232N120.html

The lamps are F40 T-12. (four foot, 40 watt)
The ballasts that I have autopsied were full of tar. Nearly impossible to dissect without wrecking them, unless one would want to start with a bucket of solvent, and that would probably soften the enamel on the magnet wire in the transformer(s). I have no idea if they had integrated chips in them.
I do appreciate the link about frequencies. I'll go look at that.

Right now I'm thinking: a quad of 1n4007's with a .047 uf cap trying to form enough time constant with the 10 meg input impedance of my meter, so the meter has enough time to set a max-hold reading. If the 1n4007's smoke, I can buy 2000 volt, fast diodes.

I am assuming that the striking voltage is higher than the running voltage.

Ballast inductor value for 60 hz are set to run bulb at about 60 vac.

For a 40 watt T12, inductor would be calculated as followed:

Bulb run current = 40watt/60 vac = 667 mA.

Inductor reactance = (120vac - 60vac) / .667 Amps = +j 90 ohms

Inductor value = 90 ohms / (2*pi*60 hz) = 0.24 Henries.

A 14.7 uF AC cap across 120vac input will power factor correct light.

There are various circuits for starter. Typical in small bi-metalic contact with parallel cap. The bi-metalic contactor is typical item to go bad in ballast assembly.

Electronic ballast have no starter. They generate a much higher start voltage that starts the ionization without preheating end filaments (which lowers the strike voltage)

and a serious thanks, again.

You keep typing and I keep picking up facts I did not know.

The ballasts I do know (F40-T12 dual bulb) don't have a bi-metal starter. That's why I am expecting a striking voltage to happen. I am happy to find that the electronic ballasts do not depend on the filaments. I bought 2 of them last December and they start very quickly.

Is strike voltage the only way to measure a bulbs condition? Stead state current and voltage have to have something to do with it. I have plenty of bulbs at work that will still strike at the same voltage and then flicker for the next 40 years.

To test the "quality" of a 20 or 40 watt tube, you can try them on the following circuit:
**broken link removed**

It gives a strike voltage of 1,000v and you can see how they react and glow on a much-more-sensitive supply. I use it to sort-out the good from the bad.

Another round of gratefulness....

I am beginning to think my idea was a bit half-baked because of the good ideas comng in here.
Umm...yes, bulbs can start well and flicker (or be ugly in other ways), but they are easy to sort. If they are ugly, they go in the trash. The first part I want to sort is, "Are these not starting because of the bulb or the ballast?"

Colin...I have no idea how I would make that transformer. I would need to know the core and where to get it, along with the wire size. The 12 volts and the 3055 are no trouble.

Here is the link to the Fluorescent Inverter project:
Fluorescent Inverter

I can work with that.

Now, where is that radio my nephew left here for me to repair....

As to strike and run, I define strike as a successful sustained start up. Yes, there is degradation in ionization for both initial ionization and sustained run ionization. It all depends on mercury vapor concentration. Plating out of mercury to glass and ends evenually reduces the mercury vapor so the bulb is hard to ionize and also reduces the UV energy emitted. (it is UV energy the excites the phosphor on inside of glass that give off the visible light).

Heating of the filaments lowers the ionization voltage.

The filaments are also heated during normal run operation. The bulb run current is flowing through the filiments so they heat during sustained run mode. When replacing with an electronic ballast you want to tie the two filament connections together so there is balanced run current flow across the filament.

Once ignited, the ionized bulb has a negative resistance profile, meaning as more current is put through the bulb, the lower its effective load resistance gets. As result, bulbs must be driven by a controlled current source. Driving a fluorescent bulb from a voltage source will quickly spiral into a run-away situation causing destruction of the bulb.

For longest bulb life, the current must be controlled at the bulb rating. Another degradation is AC duty cycle. You want equal positive and negative going current flow or more mercury will plate out at one end of bulb. This is the problem with the simple inverter circuits shown earlier. As with incandescent bulbs, filament tungsten is also ejected to inside of glass at ends of bulb.

Electronic ballast converts AC mains to a DC voltage. The DC is chopped with two stacked MOSFET's to create a high voltage squarewave at 25-40 kHz. The bulb is fed through a series inductor just like a 60 Hz ballast, only the inductor is a smaller ferrite core inductor to give the right reactance ohms at the switching frequency.

If the developed DC is high enough voltage, no preheater circuitry is necessary. CFL on 120vac typically just fullwave rectify the 120vac so their is only 160-170 vdc created. This requires a preheater starter circuitry which is a weak link in the design, causing premature failures. They also have a poor power factor with current supplied in short spikes at AC main voltage peaks.

The colder the bulb the higher the ignition voltage required. (surrounding temperature environment.)

The color temp is controlled mainly by phosphor mix coating the inside glass surface. Bulbs can be from 2700 deg K to 6400 deg K color temp. Lower deg K is has more relative red color emission (actually blue end attenuated more). Typical 'soft white' is 2700 deg K. Daylight is 6400 deg K producing a very harsh blue-white light. Since excitation source is UV plasma, the lumens is greater for the higher deg K bulbs but it is more light output at the blue end of spectrum. Most folks are used to soft white incandescent bulb so softer, lower deg K color fluorescent bulbs are more popular.