Tiny neon bulb driver circuit

[prosper]

I'm trying to build the smallest possible circuit to light a neon bulb. Ideally, I could fabricate some round PCBs about 6mm (or less) in diameter, and incorporate them into a bulb holder or bezel of some sort. The end goal is to be able to use neon NE-2 bulbs in a circuit the same way as you would use a LED, for a retro aesthetic. The main difficulty is that I want to light them with AC so that both electrodes glow; which eliminates a DCDC boost circuit. Power Efficiency isn't a primary concern, but I'd like it to keep the input current requirement to about 20mA or so. The NE-2 needs about 100V to strike, and maybe 1/2mA at 80V continuous. Right now, I'm working with the circuit attached below.

I have two questions:
1. What are my options for a transformer that's both tiny and has a high ratio of turns? Right now I've hand wound on an EE5 core which really sucks as it needs a LOT of windings, and it isn't as small as I'd like.
2. Are there any circuit improvements you can think of or alternate circuits to consider?

[PCB.Wiz]

Getting that step-up from a LVC gate might be a big ask?

I did see these go past in recent news, in SOT26
https://www.nexperia.com/products/analog-logic-ics/isolation/transformer-drivers

Those would allow a less-ideal transformer.

[prosper]

oooo! that's a cute part!

Yeah, the LVC seems to mostly work, but I need to do some more testing and simulations to see if it'll be reliable enough; it's definitely not designed for this purpose.

[David Hess]

The output from a flyback converter may be AC coupled through a capacitor to drive a neon bulb.  This would allow the use of existing flyback transformers and integrated flyback switching regulators.

[prosper]

you know, I tried that but I gave up on it pretty quickly... it probably bears going back to it and playing around some more. Added advantage is that I could achieve high boost ratios without as many turns. If I size the capacitor carefully, I bet I could run in open loop, too, reducing the size a bit

[PCB.Wiz]

oooo! that's a cute part!

Yeah, the LVC seems to mostly work, but I need to do some more testing and simulations to see if it'll be reliable enough; it's definitely not designed for this purpose.

Another part could be the Nexperia 74LVC1G19, which gives two push-pull outputs for twice the swing ?

lcsc have MX6505T - similar to the Nexperia part above, tho with higher FET rating, so can power the coil from 12V for less of a leap.
(still needs 5V for the core)

[Zero999]

How about the SN6501?
https://www.ti.com/lit/ds/symlink/sn6501.pdf

[RoGeorge]

I would use a 555 + a current transformer, or for more dramatic effect a miniature Tesla-coil.

[prosper]

you know, I'd wondered if a current transformer might work. I'll probably give it a shot. They're available in really small packages with high ratios. What's stopped me so far is the really tiny primary inductance, and liklihood of saturation problems. They're not characterized for this sort of operation, so the datasheets are of limited use.

[TurboTom]

As the transformer, you may experiment with trigger coils for Xenon flash tubes. Here are some examples shown. They are meant to be powered by a capacitor charged to a few hundred volts and then generate several kilovolts at their secondary, but driven with a high frequency low voltage signal, they will be very well suited to power your Neon indicator.

[prosper]

Interesting idea. I've also experimented by winding a new primary onto a drum inductor. Works OK in flyback or flyboost mode. Not so great as a resonant converter.


Also, the attached works; a classic 'joule thief' type oscillator. Various transformers work with it; the smallest one I've found was to wind a few dozen turns around the body of one of those 'resistor-style' axial inductors. (specifically, 50 turns with a tap at 25t over a 10mH '1W' inductor) Surprisingly, efficiency is actually quite decent; lighting a neon bulb at about 30mA/1.25V. Stability is poor though, and it's very picky about supply voltage.

[amyk]

What input voltage are you looking for?

[Analog Kid]

Surprisingly, efficiency is actually quite decent; lighting a neon bulb at about 30mA/1.25V.

Wait, what? 1.25V?
No neon bulb is going to light at that voltage. Is that a typo?
I've built lots of joule thieves that run on 1.5 volt cells and will brightly light a high-brightness LED, but certainly not a neon bulb.

[squadchannel]

If it were 1.25V, it would just be a light bulb.
Neon requires about 150V.

[prosper]

the joule thief posted above takes in 30mA from a 1.25v source, and boosts it to the point a neon bulb lights up.

[Analog Kid]

Now that's believable.

[Zero999]

Interesting idea. I've also experimented by winding a new primary onto a drum inductor. Works OK in flyback or flyboost mode. Not so great as a resonant converter.


Also, the attached works; a classic 'joule thief' type oscillator. Various transformers work with it; the smallest one I've found was to wind a few dozen turns around the body of one of those 'resistor-style' axial inductors. (specifically, 50 turns with a tap at 25t over a 10mH '1W' inductor) Surprisingly, efficiency is actually quite decent; lighting a neon bulb at about 30mA/1.25V. Stability is poor though, and it's very picky about supply voltage.

Does it light both electrodes?

I was going to suggest a Joule thief circuit, but decided against it because the output waveform isn't symmetrical, therefore I would expect it to light one electrode more than the other.

[mikeselectricstuff]

You could probably do it without a double-wound transformer - should be possible with a simple flyback output stage - transistor (e.g. MPSA42)  with inductor ( low millihenries) to positive with the neon across it, driven by a squarewave.

 

[prosper]

Does it light both electrodes?

I was going to suggest a Joule thief circuit, but decided against it because the output waveform isn't symmetrical, therefore I would expect it to light one electrode more than the other.

It does, but not very well/evenly, as you noted. Capacitively coupling the output also makes the circuit that much more finicky, too

[Wolfram]

The ATB322515/322524 come to mind, as compact step-up transformers with a useful ratio. They are made for flyback operation so primary inductance is pretty low, and wiring them for self-oscillation might be tricky on account of them only having a single primary, but it's something to look into. Flyback only provides unipolar pulses, but a capacitor in series with the neon lamp, as shown in prosper's diagram, solves this.

[mikeselectricstuff]

There are some chips designed to drive EL lamps that may be of interest, e.g. MIC4826 - these are typically flyback boost followed by an H-bridge - could be good if you care about lighting both electrodes.
It is is also not too hard to use a similar architecture with discrete parts, though finding small, high-value inductors may be tricky

[prosper]

There are some chips designed to drive EL lamps that may be of interest, e.g. MIC4826 - these are typically flyback boost followed by an H-bridge - could be good if you care about lighting both electrodes.
It is is also not too hard to use a similar architecture with discrete parts, though finding small, high-value inductors may be tricky

100%. Building a circuit that will light a neon bulb is trivial. But finding a tiny inductor is trickier. I have some parts on order to try... I think I'm gonna revisit a flyback topology with an ac coupling cap again, driven by a simple pulse generator without feedback

[Langdon]

There are neon bulbs that can be powered by 120VAC, directly from the mains (perhaps through a resistor of some sort).

Why do you need a transformer?

[Analog Kid]

100%. Building a circuit that will light a neon bulb is trivial. But finding a tiny inductor is trickier.

How about winding your own? I've done that for low-voltage applications (Joule thieves) and it has always worked well. I just use small pieces of ferrite rod and magnet wire.

[prosper]

Doable, but it's a pain to put a hundred plus turns onto a 4mm core