Ever since I saw incandescent ballasts for linear fluorescent tubes, one "dirty" idea crossed my mind.

As I happen to have a crush on (some) incandescent Christmas sets, and buy many used ones to restore or obtain components (especially bimetallic lamps and rare types of electronic controllers), I also happen to end up with loads of working but useless sets or parts (cables, sockets, controllers, discolored bulbs, bulbs with melted-over reflectors, blackened bulbs, "rice" sets which can be fixed electrically but the repair would be obvious, etc). I would like to "recycle" these and create something useful - perhaps a fixture with 1,2,4,6 or even 8 T8 18W fluorescent tubes, whose ballasts would be constructed out of these components basically for free!

Now, about the problem. As I've read, starting an old 20W T12 tube on mains voltage alone (220V) is not that difficult. I even had a ballasted tube that used to start instantly when new. For 18W T8s, however, because of their higher ignition voltages, I need some improvised "starting circuit" to add to my incandescent ballasts - something based on cheap, small electrolytic capacitors, diodes, Zeners, neon-argon starters, thyristors from cheap Chinese 8-function controllers, and the like. Any ideas about this?

I also need some data on T8 lamps - ignition voltage and working current for instance. 36W T8s are interesting for this too, but I fear their ignition voltage might be too high to create with simple circuits. Typical "midget" Christmas lamps are 90-150 mA, while "rice" ones are usually about half of that rating. Would it be possible to ignite several (2, 4, etc) T8 lamps connected in parallel on single, high-current incandescent "ballast" using such single improvised "starting circuits"?

Eager to read and discuss some fresh ideas based on this, you are all welcome.

First is the note regarding one of the last points: If you connect multiple discharges in parallel on any type of ballast, always only one of the discharge will light taking over the full current. So having one higher current ballast feeding multiple lamps in parallel would not work at all.
Parallel connection of lamps is used, but for the different purpose: Because only one lamp strikes, it is used in arrangements requiring backup for a lamp refusing to ignite due to some reason, so the other one takes over. For that purpose are even made special lamps (have seen only HPS) containing two arc tubes. Always only one lights up and the second just stands by.

For operating the fluorescents:
The best candidates for such operation are the "miniature" T5's, so the older types - 4..13W (not the newer T5HE or T5HO). They are all rated at 0.16A. These lamps still ignite from just 230V with preheated electrodes when close to an earthed metal (external field electrode). The smaller ones even instant-start (without preheating).

From other types you may use the F15T8 (0.27A), again two in series.
I have tried the 35cm F10T8 (0.22A) in a series quad with manual preheat.

Practically for any lamp you may use a multiplier-rectifier type of ignitor (something like this, but with additional diodes paralle to C1 and C2, C1 and C2 replaced with 1uF/350V electrolytics in series with 100 Ohm/1W resistors and of course the ballasting bulbs in series with mains input). These work as a multiplier circuits when there is no arc in the lamp, so generate high voltage for ignition.
After the lamp ignites, the series 100Ohm and 1uF still form a weak doubler capable to deliver significant current when the lamp is in a glow discharge mode (to heat up the cathode) and once the cathode heats up and the full arc is established, the circuit is just a plaion rectifier bridge.
It supplies the lamp by a DC, so causes some mercury migration. To combat that, the best thing is to swap the polarity regularly (needs an extra switch) or other option is to place the lamp vertically with the cathode on the top side. With that the cathode becomes significantly warmer during operation, so together with the gravity the distillation effect will cause the mercury to move down, so back to the anode side.
An F40T12 operated horizontally on such ignitor (it had broken filament, but rather intact emission coat) only sufficed with flipping the lamp once a year during fixture cleaning, so I would guess the F18T8 should work in similar manner.

To add the Xmas lamps in series with the FL, you want to know well what is their work current, and how much they can handle in higher current

Lets say it happened so, they can handle both working and preheating current of the FL lamp of your choice - only then you can do anything at all wit them

If it is jsut a few volts worth of lamps in series (1..2 lamps), i guess you can just insert them in series with the F18T8 and its normal magnetic ballast

If it is more incandescents (forming significant resistance), then perhaps use a higher wattage (lower inductance) magnetic ballast, and let more of the ballasting be done by the incandescents. You do need some choke in series to provide a start for the FL lamp (if it is not one that starts on mains voltage easily), but it can be with inductance as low as a HID ballast....



Some problems that may arise :

 - Mismatch between FL lamp current and incandescent lamp current, so one of them is either over or under drived (depending on how much incandescents you put in series)

 - If you use a bimetal flasher in series with the FL and magnetic ballast, then with the kick from the ballast, the gap in the bimetal lamp might lash over as it opens

 - Some Xmas lamps for series circuit have shun in them that shorts out as the lamps eol. Those shunts will effectively increase the power that goes in tnto the FL lamp as well as they go out. Due to the shorter section of string used as ballast, the sensitivvity (change in current from 1 eol shorted lamp) may be more significant than in noreal woth of the Xmas set on mains

Definitely I won't use any shorter sections than corresponds to the full mains voltage. The thing is, with a F18T8 the voltage across a resistive ballast is still nearly 180Vrms, but that difference is better to be utilized for the longer ballast lamp life...

@Medved: I would prefer using 18/36W T8s, rather than T5s. Those give quite a bit more light, cost far less, and are far more common. They can also be conveniently and neatly connected to the circuit with simple terminal blocks.

How many segments of voltage multiplier do you recommend for reliable starting of 18W T8 or 36W T8? Each multiplier segment adds around 300V at peak I presume? Is high voltage from that "ignition circuit" safe for incandescent ballast (Christmas lamps)?
For starting (preheat) needs, a glowbottle starter can also be introduced if you have some idea how it could simplify circuitry and provide automation (switching the starting components off after the lamp ignites). DC mode of tube operation is not a problem as I do not plan to use the circuit regularly.

@Ash: I want a circuit without any magnetic ballast - pure resistive device using only incandescent lamps (with an improvised (electronic/electric) ignition circuit out of necessity).
- I don't plan to use flasher bulbs
- I expect variations in current going through Christmas lamps during tube's starting and running. However, most Christmas lamps I've seen are obviously underdriven, and I expect them to take short current overload well
- safety "shunts" are probably going to be present in Christmas lamps, as most sets do have them. I know and understand they are a safety hazard, as they reduce incandescent ballast's resistance once activated, and multiple burned-out lamps could slowly increase tube's current to dangerous levels (causing explosion in extreme case). I don't have a solution for that problem at this moment (except perhaps an idea of some sort of fuse with value closely above tube's regular running current?).

@Medved: I would prefer using 18/36W T8s, rather than T5s. Those give quite a bit more light, cost far less, and are far more common. They can also be conveniently and neatly connected to the circuit with simple terminal blocks.
How many segments of voltage multiplier do you recommend for reliable starting of 18W T8 or 36W T8? Each multiplier segment adds around 300V at peak I presume? Is high voltage from that "ignition circuit" safe for incandescent ballast (Christmas lamps)?

The F40 used "x4" (it used 4 capacitors and 6 diodes, but the "bug zapper" circuit does the same with just 3 capacitors), I guess it should be sufficient for the F18T8 too...
Actually I do not have any idea, how you can get odd number factor, using single capacitor across one diode with regular Graetz bridge rectifier makes it already "x2"...

For starting (preheat) needs, a glowbottle starter can also be introduced if you have some idea how it could simplify circuitry and provide automation (switching the starting components off after the lamp ignites). DC mode of tube operation is not a problem as I do not plan to use the circuit regularly.

The problem with glowbottle starters is, they are controlled just by the voltage they "see" across the lamp and are rather dumb devices - higher voltage than some limit causes the glow discharge to heat up the bimetal and so the starter to close for some time. The idea is, this threshold would be above the lamp arc voltage, but below the mains peak voltage, so when the lamp is not lit it heats up and closes, when the lamp ignites, the voltage becomes insufficient for the discharge in the glowbottle, so it remain cold.
On an inductive ballasts the lamp current of a lit lamp crosses zero, when there is enough voltage at the mains (close to the peak, due to the phase shift), so the arc reignites immediately after that. That means nearly no gap with no current, so rather high remaining ionization. And that means there is very small voltage overshoot across the lamp, remnaining below the starter threshold.
The problem with the resistive ballasts is, there is rather long gap around the current crossing zero, so the ionization level decreases a lot. That means the reignition voltage becomes higher and so it very frequently false triggers the starter.
With the F18T8 you have some chance to fit the 200V triger ("something..65W", so S10 or equivalent) starter - normally the F18 works with the 100V trigger ("4..22W", so S2 like), so there is some extra margin towards the 200V one. You would have to try.
But generally I would expect better success with plain manual preheat - you see if the lamp lights or not in more reliable way than the glowbottle starter just guessing based on the voltage...
Therefore

@Ash: I want a circuit without any magnetic ballast - pure resistive device using only incandescent lamps (with an improvised (electronic/electric) ignition circuit out of necessity).
- I don't plan to use flasher bulbs
- I expect variations in current going through Christmas lamps during tube's starting and running. However, most Christmas lamps I've seen are obviously underdriven, and I expect them to take short current overload well
- safety "shunts" are probably going to be present in Christmas lamps, as most sets do have them. I know and understand they are a safety hazard, as they reduce incandescent ballast's resistance once activated, and multiple burned-out lamps could slowly increase tube's current to dangerous levels (causing explosion in extreme case). I don't have a solution for that problem at this moment (except perhaps an idea of some sort of fuse with value closely above tube's regular running current?).

The bypass things are usually voltage triggered, so that means it is really bad idea to overdrive the lamps above their voltage rating even for a brief period of time - the overvoltage may false trigger these bypass devices and in fact cause the short circuit even when the filaments are still good...

The bypass things are usually voltage triggered, so that means it is really bad idea to overdrive the lamps above their voltage rating even for a brief period of time - the overvoltage may false trigger these bypass devices and in fact cause the short circuit even when the filaments are still good...

I think we've come to the most important issue to sort out here, thank you! Safety shunts in Christmas lamps are probably voltage activated as you say. I read about a design using plain wire whose special insulation gets burned by very high voltage (full line voltage is present only when a filament breaks), but since I had bulbs going on and off their shunts are obviously different. I also remember once I tested new (high brightness - somewhat overdriven) Christmas set on a 220/220V transformer it's bulbs started burning out shortly after bimetal bulbs activated - probably because of voltage/induction spikes those switches caused in the transformer.

So, the incandescent ballast needs to be protected from high voltage of the starting circuit. Any ideas? Would a simple diode (like 1N4007) in series with the ballast be enough? Or even better idea - to supply the entire ballast/lamp circuit through a diode bridge to avoid flickering a single diode would most likely cause, and provide a barrier for current coming from the voltage multiplier (in reverse direction)?

As i understand from looking at such lamps, the shunt is basically a wire wrapped a couple times around an oxidised section of the lead in wires of the incandescent, So its triggering voltage accuracy is mediocre.....

But to satisfy "not triggering at 8V" and "triggering at 120V" (or 240V, i guess the same incandescents go into sets shipped worldwide, only the wiring of the string so number of lamps in series varies), no accuracy is needed either..

I really doubt they will trigger at any voltage that the filament can handle. But if it fails from overvoltage, then the shunt will trigger. Few might trigger even at normal voltage, if for some reason the oxide layer was pinched and the wire makes contact anyway

Besides, those are not "safety shunts", thry don make the string any safer - they just make it keep lighting on the 1st few lamp eol's instead of going out

I really doubt they will trigger at any voltage that the filament can handle. But if it fails from overvoltage, then the shunt will trigger. Few might trigger even at normal voltage, if for some reason the oxide layer was pinched and the wire makes contact anyway

Shunts in Christmas lamps are simple and effective in theory, but reality is different. Some of my observations:
- a bulb's filament can break and have it's shunt activated, but it can still reconnect again and continue working normally.
- some bulb's filaments "oscillate" in light output, probably because of a shunt. This seems to be affected by bulb temperature and line flicker, but it shows that a shunt can be active even with a working filament.
- the example with my set having multiple burnouts after a minute or two on a 220/220V transformer shows that even (very) short voltage spikes a coil can provide can cause problems in the string. Whether those bulbs burned out because their filaments were on the edge of breaking, or because a few other shunts "triggered" and bypassed several bulbs temporarily, is a question that can't be answered.

Since a bulb-string has an "important duty" in this setup I don't want to risk the shunts mis-behaviour after getting (for example) a 600V surge from the ignition circuit down the string, so I think the semiconductor protection is necessary - diodes are cheap and excellent barriers even for very high voltages.

To protect your light string you could wire an MOV across it between the string and neutral get one rated at say 275 volts and it will protect it from any inductive surges and from any nasties on your mains supply

Dont stuff MOVs into places where they dont belong

They dont provide protection for this sort of stuff (vacuum incandescent lamps dont need any such protection anyway), the list of places where they are of any benefit is very short - nearly only electronics, and only as a last stage of protection after a cascade of bigger stuff through the supply system

But when MOVs break down and fail to reach full current (in series with something else on an open Neutral for example) and are not protected by a thermal cutout, they will catch fire - and set fire to anything that happens to be nearby....

(Thats one reason why most of the "surge protector" strips sold out there are nothing short of an arson device)

I was meaning to put the MOV between the beginning of the Xmas lightstring and neutral to protect them from spikes

@hannahs_light: MOVs are more expensive and harder to find devices than simple diodes, so I would prefer sticking with those as this is a "reuse" project so aimed at keeping things simple and cheap. The ubiquitous 1N4007 (which I can even source from stuff like dead CFLs) has a breakdown voltage of 1000V, and I can look for ones providing even higher protection if necessary. Besides, since I would probably stick a Graetz in the circuit to supply the ballast and lamp by rectified DC to avoid flicker, two diodes in series would effectively raise the inverse-polarity barrier to 2000V if I understand electricity laws well.

A question for Medved: can a single voltage mutiplier circuit somehow be connected to more than one tube to provide ignition? Meaning - each tube would have it's own incandescent ballast separate from the other tubes, so each working as an isolated "entity", just having that one wire bringing high voltage pulse from the common "igniter" to the tube?

A question for Medved: can a single voltage mutiplier circuit somehow be connected to more than one tube to provide ignition? Meaning - each tube would have it's own incandescent ballast separate from the other tubes, so each working as an isolated "entity", just having that one wire bringing high voltage pulse from the common "igniter" to the tube?

It would be more complex than having the multiplier functionality separated for each tube.
The thing is, the multiplier needs only AC to work.
So the idea is to create something to act as the multiplier when the high voltage is necessary (so when the lamp is not ignited, so no current flows; in that case the capacitors are the main components, the diodes parallel to them do not work) and once the current starts flowing, the multiplier becomes overloaded and because of the parallel diodes, it becomes just a bridge rectifier (so the capacitors won't be effectively there).

The 1N4007 needs practically no overvoltage protection on mains, especially when the current upstream is limited (I would recommend to put the rectifier/multiplier between the ballast incandescents and lamp, in that way the capacitors will see less voltage during the lamp operation.

Ok, so I took some free time to reconsider this circuit I had in mind, and draw a schematic. The draft is in an attachment.

The main points are:
- Voltage multiplier gets almost full 220V AC voltage through incandescent ballast (Christmas set) on switch on, because the fluorescent tube is not conducting yet.
- Multiplier's output of approx. 600V is connected to the tube to provide ignition voltage.
- The tube is on a separate branch in parallel to the multiplier, but supplied through Graetz. It's only purpose is to stop high ignition voltage from entering the bulb string (increasing risk for shunt failures).
- The two 100 Ohm resistors for multiplier's condensers can be replaced with just one connected between the two condensers and going to the AC "minus" pole. I would correct it and draw it that way if I didn't use some time-limited online schematics creator.

Now I need your advice on this - is this circuit safe to use the way I drafted it, or are there some sub-circuits created between these components I'm not aware of? I might have oversimplified the voltage multiplier, and I'm not sure I would get the desired effect, but I have no experience with those. I am also not sure if multiplier can be supplied this way (directly through a bulb string) as it will probably continue charging the condensers even after the tube strikes - my idea was that the tube will be a low-resistance path in this arrangement and that the multiplier's branch will get reduced current after the tube strikes.

So, @Medved, how does this draft seem to you? Of course, everyone is welcome to post some thoughts too, I appreciate any help. Thanks!