You know the thing, when the bogeyman comes the fluorescent lights starts to blink...

I've been thinking about building this special effect fluorescent fixture that imitates the one with eol lamps blinking continously.

It could be probably done with all-electronic with right schematics but that might be to hard for me. The idea is to get reasonable life time with both lamps and gear.

I was thinking if I could use separate cathode transformer to keep the cathodes heated properly all the time and use separated circuitry to turn on and off the arc voltage from magnetic ballast. Ofcourse, it would not most likely be exact similar as the real eol lamp but I would guess pretty close. Getting random on off signal to the lamp is probably easy and there is ready circuits for that but I think the problem is how to control the lamp with it. Relay would be easy solution but Im little skeptical how long can relay take that kind of use.

Any thoughs about such setup and estime would it give reasonable lamp life or if i'm completely lost with my idea? The idea was to use 40w t12 lamps and I live in 230v area

There are some ways i can think off..

Use actual EOL lamp with good starter. This will destroy the starter in not long time

Use good lamp and St151 starter. The starter won't come off when the lamp starts and so it will keep flashing. This will destroy the lamp and starter in not long time

Use good lamp and starter, with a relay that momentarily interrupts the Phase once in a while, requiring a restart. 40W lamp may restart from such interruption without needing the starter, but 36W is less likely to. This is a bit better on the lamp and starter if you dont fire it all the time, but still will destroy the lamp and starter

Use lamp with relay in place of the starter. In this case when the relay is ON the lamp is preheating, and when it switches OFF the lamp is supposed to start. With long enough preheats (2+ sec) the lamp wear will be relatively low and it will start on the 1st attempts most of the time. This will wear the relay somewhat. You can add small (few nF's to few 10's nF's, high voltage... basically few capacitors from starters in parallel, each of them is about 5 nF) in parallel to the relay contact to lower the wear but it won't eliminate it completely

Use a relay to click in/out an St151 starter. The St151 will start the lamp and keep flashing, but if you disconnect it with the relay (at a moment when the lamp lights) it will remain to work normally. Or try to connect St111 and St151 in parallel and disconnect the St151 with the relay. This way if the erlay removed the starter when the lamp was still starting, the St111 will complete the starting normally

Use a rectifier with GTO in place of the starter, in a similar way to how Electronic starter works. By timing the gate voltage for the GTO you can keep the FL lamp on or off, preheat, start, extinguish and so on. You'll need a GTO that can withstand impulse voltages in the kV range, and a similar voltage grade optocoupler to isolate it from your control circuit (or make your own optocoupler with big air distance or optoconductor)



Most of those setups will cause short circuit of the ballast output for long periods (intermittently when the lamp is flashing or continuously when the lamp is glowing in the ends). This will overheat and damage most Fluorescent ballasts. To make such setups without damaging ballasts, connect them in 2x2 series/parallel array. This way the current through the ballasts is lower in half, and they can be left shorted for unlimited time without overheating

As "random signal" for electronics audio input can be used. To make "intermittent triggering" compare the audio signal to a preset DC level using an operational amplifier, this sill provide an impulse once in a while when there is some peak in the audio



Some more things :

Make a lamp that glows in the ends... This may not look like much, but we lighting folks know that the ballast inside is burning up and this is after all a FL lamp fault mode. Use a low voltage supply or ballast array to make the effect without overheating the ballast

Make Hg migration by adding a rectifier bridge of suitable voltage between the choke and the lamp (AC on choke and DC on lamp/starter) (some starters will work properly on DC only in one polarity. If the starter only glows and does not do anything else then it will overheat, you have to flip its polarity)

Make a lamp that glows dim Blue in one end. Just use a lamp wih broken cathode on stuck starter. The current is mimimal and it does not heat up the lamp or the ballast

Power the lamps with a CCFL inverter to make them light dimly. Most simplest CCFL inverters (with the few simple components and no control chip) are also sensitive to the supply voltage and component parameters, so if you can supply unstable DC voltage to it or play with the value of some feedback resistor and such, it will vary the lamp brightness accordingly. Do that in a cycle pattern with a 555 timer or with "random signal" audio input. The underpowered lamp will look like emergency inverter powered lamp in a power outage, and if the brightness is going up and down (555 timer), it will look like the battery is about to go out, adding to the creepy effect. You can keep this to a later stage of your show, as in - first failing lamps on 230V, then the power went out and only a few on emergency inverters are left. Imitate the batteries going out and switch off the lamps gradually

There used to be common type of luminaires (known here as Ice Packs, but they may be called something else in other countries) that are made of flat sheet metal tray, with the lamps, gear and everything on it, and covered with a Plastic wrap around cover. As the gear and wiring is directly exposed to UV from the lamps, the isolation on it decays over time and it becomes a dangerous and scary looking mess of wires with cracking isolation and falling apart terminals. If EOL lamps were left in, then also the ballasts and the wiring near them are burned. If you have such luminaires, they are great exactly for something like this..

 - Bypass all chokes, put in lamps and stuck starters, connect low voltage supply so the lamps glow in the ends

 - Connect the output of CCFL inverter to the old wiring to power the lamps dimly. (effect would look best if you have an ice pack with actual emergency gear in it, or just use the actual pack as inverter). The CCFL drivers have limited output current, so they won't cause big sparking if the destroyed wiring shorts out, allthough they are capable of burning up the remains of the isolation on the wiring. Fairly safe as long as not used near anything flammable

 - Connect new hidden wiring to the sockets, and connect any of the haunted 230V schemes (with hidden or remote ballast), but make it look as if the destroyed wiring and ballasts are connected



As kid i made once (in the early 00s) a Fluorescent lightNING simulation for decor for a show. This was supposed to simulate natural lightning, not to show the Fluorescent lighting itself as the source of horror. The setup was like :

Timed 230V pulse generator. I used a staircase lighting timer (heater type) with power connected to the switch input permanently, so every few minutes it clicks the heater on for a couple seconds and back off. Those timers are no longer made and i dont know if there is anything similar in other countries, but an equivalent can be made with 555 timer + relay

The 230V pulse feeds a few ordinary 36W Fluorescents with fast starters (i found the glow starters that start flicking the lamp the fastest by experimenting, comes out one of the brands of cheap noname starters were the most flashy). So its like - Power applied, few flashes, darkness

The 230V pulse feeds a step down transformer and a relay. (if you use electronic timing circuit, then just control the relay with it directly). The relay is connected across the "play/skip track" button on an old PC CD ROM drive

In the drive is a CD with tracks like :

[track 01] THUNDER SOUND EFFECT __long silence________________________________________________________
[track 02] THUNDER SOUND EFFECT __long silence________________________________________________________
[track 03] THUNDER SOUND EFFECT __long silence________________________________________________________
..
..

The CD drive audio output was connected to the main speakers

So every time the lamps flash, and then the CD is advanced to the next track and plays the thunder sound, then nothing (keeps playing the silence, untill it is advanced again by the next pulse)

(on a separate note, i messed up with school time table once with a similar CD with tracks of silence)



Is the same lamp required to go between Normal and Haunted modes ?

How much switching operations you expect in the Haunted mode ?

Does the lamp have to be 40W ? What about 36W ?

Are relay clicks to be avoided ? (it can be wired to a remote location)

The sole purpose of 40w lamp is that I have some spare cathode heating transformers for old fl dimmer system t12 can also start in rapid start mode with simple choke ballast. That way I could keep cathodes in right operarion temperature all the time and switching lamp on and off by cutting the arc voltage (while lefting cathodes heated) would not effect the lamp life. Also, 2x40w is not too bright when it's on and dual lamp gives more of the effect (with different time blinking). While not so much necessarity, this would also give me a possibility of selecting wether to use spooky mode or normal rapid start mode or combination (one lamp  normal, one lamp blinking) if I could control the arc voltage with relay. I guess this kind of system would be easy to the ballast as there is no bigger than rated current going thru.

I would say something like 300-400 hrs life per lamp would be ok.

I think relay click is not a problem. The real things in the movies ofcourse have those metal starter sounds like every other old fl luminaire that are meant to look old.

Then you can go either with relay, or with normal (not GTO) thyristor behind a rectifier bridge, similar to whats in a dimmer, and control it with optocoupler. Something like bt138 may be ok for a start i think..

Si I guess lamp and ballast can handle this frequent "rapid" starting quite well?

I think this is the way to go to not make it too complicated.

For really long life you may mimic the fluorescent using a string of LED's (put them into a glass tube from an old gutted fluorescent tube).
A 24V regulated DC supply plus ballasting resistor for 6-LED long strings, switched by some MOSFET.
In the "spooky mode" you may use rather high power (up to the full LED rating), but for steady light I would reduce the brightness a bit (e.g. using 50% duty ratio at about 1kHz to eliminate flicker)
The advantage is the full setup being SELV...

For real fluorescents the T12 would be the easiest to work with (low ignition voltage...; as described by others above), but they are pretty rare and I wouldn't want to destroy them for such thing.

The T8's would be still a good material given the availability, but they need higher ignition voltage, what will complicate the controller.
For any fluorescent you may adopt a decent programmed start electronic ballast: The controller chips usually allow digital signal ON/OFF control, as well as timer capacitor controlling the preheat time.
A pulse on the Enable will always reset the controller, clamping the voltage on the preheat timer capacitor will keep the controller in the preheat mode for all the time when the lamp is supposed to be off, released only when teh lamp is supposed to light (even for a short pulse, when mimicking an ignition flash).
You may need to modify the preheat timing capacitor for way shorter time (and then take over the preheat time control by the microcontroller, even for normal light use) and the preheat operating frequency so the heating current/voltage would be low enough so the filament won't overheat even when long time in that mode. And of course that mean the preheat time after power ON should be made way longer than the original (at least 2 seconds; it is in the effect control microcontroller)
Of course, both "Enable", as well as the preheat timer clamp signals should go via optocouplers to provide insulation between the mains in the ballast and the effect control microcontroller.

T12 is not a problem in that sense, I'm only going to make 1 or 2 units of theese and have lot's  of modern narva T12 cw lamps to use. Also I really dont think theese going to have so much use.

But speaking of t8, a special electronic dimmer ballast with 1-10v control could also work. Those usually won't let you dim it completely off but enought to work it out. I would just to have to find ou one with fast enought respond for the dimming setting. I guess building a control circuit for such would not be that difficult but I have no experience how such electronic ballast would tolerate such jumping.

I think you could make it quite elegant with different "programs" if electronic dimmer ballast and simple microcontroller is used (or raspberry pi etc.) But I don't know any programming so I guess that's too technical to me.

T8 is not a problem either, all you have to do to run T8 in your circuit is momentarily short across the lamp for starting

I too think a 1..10V ballast may not like being fed with audio..

To get a realistic effect, i think it is really needed to imitate the starting/fault modes of a real Fluorescent light and not try to suffice with "flashing lamp" because it just won't look the same. Its like trying to imitate a flame with LEDs...

Standard dimmable ballasts are designed to provid stable and predictable light output at any dimmer setting. That means they have to use some sort of closed loop feedback system for the arc current control. And that has its consequence of the system becoming slow to respond on the dimmer setting. Normally that is rather desired feature, but won't work for the spooky effect.
You really need very fast response and that restrict you to just an open loop control (you do not need it that accurate) and that goes in completely different direction than the dimmable ballasts are designed.

The momentary shorting on magnetic T8 is rather hard to implement: You will have difficult time to find suitable switching components for that...

So I guess I stick with T12 then.

This spooky thing did bring my mind one onteresting implementation of lighting upgrade. I was in a basement of certain old big apartment complex. The basement had long corridors featuring different kind of service areas and storage spaces and I was on a job searching for telecom room. The lighting in the corridors was upgrade at some point and used led fixtures with each having it's own motion detector (lighting only that particular fixture). The off delay was pretty short, around 30 sec. I stopped at one point to look the blueprints trying to find if Im even close to that place. After a short while I noticed how the lights behind me started to turn off one after another until only two closest of me was on, otherwise it was dark corridor. I can believe that was pretty energy saving setup but I can't say thats really a wise implementation of motion detectors.

I built a strobe with nothing but array of 8 18W T8's, each on one 36/40W choke (overpowered but not too bad for such use), and just relay contacts instead of starters, all relay coils controlled by 555 timer through a transistor. It worked acceptably well. Not all lamps were striking on all flashes, but most of the times they striked. In the case here this does not matter (and it did not matter in my appliction for which i made the strobe either)

Sounds like it could be a cool/interesting project

But its also got me thinking....
I might have to try something like that for Halloween this year, I'm thinking I could use either of 2 things I already have:
* A 2' LED retrofit tube - the type that runs directly off mains, and is dimmable.
* A F40T12 and dimmable magnetic ballast.

In either case I'd use one of my christmas-light controllers.
I'm thinking that "candle flicker" mode might make a nice creepy bad-fluorescent look LOL
Have no idea if constantly varying the output level at a relatively speedy rate like that would hurt the ballast, but I don't see how it could do any harm to the LED tube.

Dimmable how ?

If there are any mechanics involved, definitely they are not meant for all the time variations so this ends the plan with a big Dont

If the control is electronic it does not care too much, however the LED tube might depending on its design. And if the LED tube EOLs it may start rectifying (as in half the input diode bridge burned out, not as in EOL Fluorescent) and that will be very hard and damaging on the ballast

Dimmable how ?

If there are any mechanics involved, definitely they are not meant for all the time variations so this ends the plan with a big Dont

If the control is electronic it does not care too much, however the LED tube might depending on its design. And if the LED tube EOLs it may start rectifying (as in half the input diode bridge burned out, not as in EOL Fluorescent) and that will be very hard and damaging on the ballast

I actually meant it as 2 separate things where I'd use one or the other, not both together..

The LED tubes connect directly to 120v mains, no ballast needed. So setting one up would be very easy.

The dimmable magnetic ballasts have a neutral, and 2 hot/live connections - 1 for filament heating, 1 for the ballast. In an actual case you'd use a special fluorescent dimmer designed for that series of ballasts...but a standard (fluorescent rated ofcourse) dimmer will work as far as dimming goes, but it won't have a power (on/off only) connector for the filament heating circuit on the ballast.

Instead of a dimmerswitch I'd be using the controller connected to the ballast input (and filament heating would simply remain on).
I know that much will work because I've done it in the past (different actual controller) but only for simple fade on/off, and that worked fine...but such a setup (I assume atleast) would be a whole lot more gentle on things than a rapidly changing random flicker. (but note: I can control the variation percentage - how much between brightest and dimmest it goes, as well as the actual minimum/maximum levels)

I don't know about the florescent idea, it should work, but for LED tubes, just get a non-dimmable tube and use a leading or trailing edge dimmer and they will flash and flicker like a (...) no fancy controller needed just a $1 restore rotary dimmer, or you could just run a dimmable LED in series with a florescent starter, I do this with low wattage gls and they flicker and flash in a truly random way, and most last for several seasons of Halloween and an LED should pull enough watts to make it work, I also pull the covers off them so you can see them flash and I stick them in a clear plastic pill bottle tied to the wire so it really looks like something is shorting out...

Make sure you post the video I want to see them when your done, I like Halloween decor as well..