I had some spare time at work, and we don't have an easy to use neon spectral lamp - only one with bare connections you gotta hook up to a multi-kV power supply.

So i decided to pop the cap off a LED lamp, lever out the LED board (which i kept to show the audience what is actually inside a LED lamp), etch a little circuit board that pops right into the place of the old one (keep an eye on isolation distances) and populate it with 5 strings of 2 neon lamps in series with a 82k resistor.
This gives us a current of 0,73mA per lamp.

It works beautifully, but as expected it is a bit dim. I might make another one but with a 2,5mA lamp current. That will dramatically reduce life span but since i can etch and populate a board in maybe 2 hours time in total, that is acceptable.

Side note: i make boards in-house. Getting them made in China is of much higher quality than i could ever do in house, but i am impatient and want my boards the same day lol. Also the trend of import tariffs is also gonna hit us in a month's time, so boards from china will get a 2 euro tariff (fine, it's just 2 euro and it goes into the european tax pot) with 10 euro of handling fees from the carrier (NOT fine, goes into private hands and FFS, 10 euro for allowing me to pay 2 euro? Get shafted, PostNL).

Very very cool! I did a similar thing with 7 of them and I used a sheet of plastic rather than a PCB but it did work! Like you say there isn't much light emitted by these but it is cool to see. It's long since been disassembled so I can't take a picture of it but maybe that's for the better since mine was made very very cruddily.

Sure your ballast resistors (crammed to the inside of the board) get enough cooling? You are designing a compact lamp after all, and are responsible for thermal mode

Yes, if i run it for a while, then rapidly take it out of the fixture and pop the cover off, my highly calibrated finger thermometer says the temperature remains at or below 60 degrees C. If there wouldn't be mains voltage in there, i could just about continuously hold my finger onto the components. It runs cooler than the LED lamp did before the conversion.
Just out of an abundance of caution i also drilled some small holes right on top and around the lamps themselves.

The resistors dissipate considerably less power than normal because i run 2 lamps in series. I have a whole bunch of them including the resistors from the store, but a single lamp with resistor at 1mA gets considerably hotter, to the degree that i have to redesign a tube shaped lamp i made with 20 of them in there.


I do think re-making this with 20 lamps (or tripling the current and accepting faster wear of the neons) will make me run into thermal issues. I'm gonna dig through the recycle bins and see if i can find a large globe shaped one for a bigger version, where only the resistors are on the board itself and the indicator lights on their own (shrink tube isolated) legs. Just ordered 200 of them before the tariffs kick in, might as well...

Reduced the series resistance to 47k for 2 lamps in series on 230v. Current is now 2,3mA per lamp and brightness is a lot higher.
Heat production is significnatly higher too - with an IR thermometer the hottest bit is at 107 degrees C, which is acceptable considering the life time of this thing is allowed to be just 1000 hours or so. I think it is the resistor that is so hot, but i can't really tell. Either way, it is right at the bleeding edge of its power rating at a calculated 248mW for a 250mW resistor. I'm gonna publish a little article about making a spectral lamp like this for schools that don't have a high physics budget, of course 1/2w resistors will be specced along with a better circuit board layout that allows to drill extra holes in the board itself. Depending on the light bulb used, you might also be able to drill holes in the base to really get a good convection current flowing.

I also tested the current with the stock resistors (120k for 1 lamp on 230v) and that results in a current of 1,5mA. At that current they will still last thousands of hours, so i think i will definitely get 1000 hours of use out of my neonified led lamp. I care more about brightness than about longevity in this case, 200 (at the cost of a whopping 5 cents a piece!) more neon lamps are on their way

I will likely copy this project at some point but with a lot more neon lamps. maybe arranged in more of a shape that conforms to the globe.... a project that will be a while as I got other stuff I am working on currently.

Too bad you can't easily get these glow lamps with hydrogen, helium, argon, krypton and xenon too. They would be some amazing spectral lamps.

Although, the lack of positive column discharge probalby influences the spectrum. In fluorescent lamps with thin enough phosphor to be somewhat transparent, the cathode glow color is noticeably different from the positive column color.

I will likely copy this project at some point but with a lot more neon lamps. maybe arranged in more of a shape that conforms to the globe.... a project that will be a while as I got other stuff I am working on currently.

You'll have to keep heat production into account. The original LED board is aluminium substrate, thermally coupled to the plastic coated aluminium heat sink hidden in the base. That way it can dissipate around 10w, heating up the "plastic" base to around 55 degrees C.
You're best off with a much larger light bulb, i have some CFLs with 15cm decorative sphere on them. Those would be ideal and allow you to easily stuff in 30 neon indicators with no thermal issues at all. But i'm not gonna destoy those. Just patiently waiting until i find something suitable in the recycle bin somewhere.

The neon lamps and resistors can not (easily) be coupled thermally to the heat sink, so at more than say 4 or 5w you'll run into problems keeping everything cool enough. Maybe lamps without the aluminium heat sink exist, so you can drill a bunch of holes in both the plastic base, the bulb, and the PCB to get a convection draft going.

The green and blue "neon" indicators use a different kind of gas. Also easily available. But keep in mind the green ones must be run at well under 1mA to not screw up the phosphor. They're a lot more delicate than the standard ones or the blue ones.