Neon Lamp Tea Light




Introduction: Neon Lamp Tea Light

About: I like circuits, mechanical gadgets, things that glow, and cats. My dream is to one day own a big Nixie tube.

Updated: Now it's easy to order the PCB for this project directly from PCBWay! Here's the link:

Neon lamps, also known as neon glow lamps, are wonderful little bulbs that glow orange when you give them high voltage and low current. Neon lamps are mostly obsolete now; they belong to an era when many electronic devices were supplied with high voltages in order to drive vacuum tubes, and the voltages required for neon lamps were convenient and accessible. Compared to an LED which might have a voltage drop of 3V, a neon lamp's striking voltage on the order of 100V makes it impractical for use in modern circuits, which use transistors instead of vacuum tubes and require much lower voltages.

However, neon lamps look really cute! They deserve better than to be consigned to the history books, thrown out in favor of surface-mount LED's lacking any sense of personality. With this in mind, I decided to create a small, portable neon bulb light. I wanted my creation to have these properties:

  • Small (as small as possible)
  • Long battery life (1 day at least)
  • Wirelessly rechargeable (compatible with any phone charging pad)

Since I wanted to minimize the size of my circuit, I chose to create my first printed circuit board, using Autodesk Eagle. Initially I wanted to make a necklace with a glowing neon lamp, but the size of the charging coil and the height of the inductor on the PCB made that difficult, so it morphed into a tea light. Maybe I'll try to make an even smaller version in the future!

This project requires medium difficulty electronics (regular soldering, surface-mount soldering with solder paste) and medium difficulty hardware (3D printing, sanding/filing). If you want to skip the theory and jump straight to building it, skip to Step 7: Order the PCB!!!. Also, it may take a lot of experimentation to get the wireless charging to work reliably, because the battery is placed so close to the charging coil that it can absorb a good portion of the energy transmitted by the charging pad. A couple tips to deal with this are detailed in Step 14: Troubleshooting.

Let's get to it!




I strongly recommend ordering this SMD component assortment, since it has nearly all the surface-mount parts you need for this project.

Through-hole Components:

Surface-mount Components:

Miscellaneous Parts:

You'll also need to order a custom printed circuit board for this project.

Step 1: Research

I found a lot of useful information about neon glow lamps from this detailed blog:

Neon glow lamps are a type of gas discharge tube, filled with a mixture of neon and other gases. When there's sufficient voltage across the electrodes, the gas atoms inside become ionized - a neon atom Ne would lose an electron from its valence shell, becoming the positively charged Ne+ and a negatively charged electron. I only sort of know how this works, but I think the positive neon ion is attracted to the negative electrode, where it receives an electron to fill its valence shell and emits light.

Anyways, when you apply direct current (DC) to a neon bulb, you will see an orange glow surrounding the negative electrode. You can also choose to supply alternating current (AC) to the lamp, in which case both electrodes will appear to glow simultaneously. In truth, both electrodes will alternate between glowing and not glowing, but this happens too fast for our eyes to notice when the AC frequency is 60 hertz and above.

Neon lamps last longest when supplied AC with a low peak current (<= 1 mA). However, I chose to supply my glow lamp with DC, because I wanted to make my circuit as small and slim as possible, and I couldn't figure out a convenient way to convert the 3.7V DC from the LiPo battery to approximately 90V AC without using a clunky transformer. This means that only one of the electrodes on my lamp will glow, but I think the effect is still quite good, and maybe I can create a version in the future that uses a DC to AC converter.

The neon lamps I bought are NE-2 indicator lamps, which have a striking voltage around 77V and a maintaining voltage around 57V. This means that the light will start glowing when the voltage across it reaches 77V, and the voltage will immediately fall to 57V. You shouldn't just plug a neon lamp straight into your power supply; it's necessary to use a resistor to limit the current to the lamp to a reasonable value. This resistor will also drop the extra voltage to the lamp - if you supply 80V, then after the lamp strikes, the voltage drop across the lamp will be 57V and the voltage across the resistor will be 23V. Depending on the voltages and current you want, a resistor between 30k and 100k should work.

One more important note is that neon lamps, even of the same type and from the same manufacturer, are not identical. The striking and maintaining voltages of otherwise identical neon lamps can vary by 5 to 10 volts. This is especially the case with indicator lamps, which aren't designed to have precise electrical characteristics. With this in mind, it's important to breadboard the following circuits to make sure that they work with the lamps that you've purchased.

Now that we know a little about our neon lamp friends, let's design a circuit to power them!

Step 2: DC/DC Converter (3.7V to 80V)

I said "design," but our friend Giangrandi already did the work for us. Near the bottom of his site on neon bulbs, he details a safe 1.5V to 90V converter using the LT1073 micropower DC/DC converter chip. This is based on a pulse generator from Linear Technologies Application Note 45, which also includes a number of other interesting circuits.

We want to supply 3.7V instead of 1.5V, since the nominal voltage of the rechargeable LiPo battery we're going to use is 3.7V. Luckily, this circuit is pretty robust, and works great for a range of voltages. I verified this by simulating the circuit using LTspice, a free electronics simulator, and later breadboarding the circuit to test it with my bulbs. The attached file also includes my modifications, which includes low-voltage shutdown to prevent the battery from being discharged too far and also shutting down when the battery is charging.

The LT1073 works by sending low-voltage, high-current pulses to the inductor L1. The maximum current provided to the inductor is determined by the resistor on the Ilim pin, R1. Inductors are interesting because, unlike capacitors, they can create higher voltages than they were initially given. That's because an inductor stores electrical energy in a magnetic field surrounding it, and when this magnetic field collapses it creates a voltage depending on the inductance of the inductor and the speed at which the current is changing. The datasheet for the LT1073 tells us that it sends pulses at a frequency of 19 khz, and this quickly changing current allows the inductor to create a high voltage.

The way the LT1073 makes sure it outputs the correct voltage is by sensing the voltage through the feedback pin FB. If the voltage on FB rises above 0.212V, then the LT1073's oscillator turns off and stops sending pulses to L1, causing the output voltage to fall. If the voltage falls below 0.212V, then the oscillator turns back on, causing the output to rise. This means that the LT1073 is always working to keep the voltage on FB close to 0.212V. If we connect the output voltage to FB through a voltage divider (R2 and R3), then we can regulate the output voltage!

The equation for voltage dividers tells us that Vfb = Vout * R3 / (R2 + R3). Therefore, the output voltage for the depicted circuit should be Vout = 0.212V * (10M + 26.2k)/26.2k = 81V. Running the spice simulation, we can see that the actual voltage bounces between 79V and 83V, which is good enough for our purposes.

One other important part of the main power circuit is the diode voltage multiplier. A diode voltage multiplier can convert some voltage AC to a higher DC voltage. Instead of using just a diode and capacitor to rectify and smooth the voltage coming from the inductor, a diode voltage multiplier (a tripler in this case) gives us triple the voltage at a third of the current. This makes it possible for the LT1073 to provide the high voltages we need for the neon lamp. Without the voltage multiplier, the chip would still try to output 81V thanks to our feedback pin, but the voltage wouldn't be able to rise that high because of the value of our inductor and the 220 ohm resistor tied to Ilim. You can experiment with what happens when you set too high a target voltage by lowering R3 to 10k and watching Vout and the current through L1 in the simulation. There's a nice Wikipedia Article on Voltage Multipliers if you'd like to learn how they work.

That's an overview of how the LT1073 does its thing. The resistor R4 limits the current to the neon lamp - make sure to pick the nearest standard value that supplies your lamp with the current it needs. I made two additions to the circuit to make it work with the other components. The low-voltage shutdown block uses the Set and A0 pins of the LT1073 to monitor the LiPo battery voltage (these pins aren't on my spice model for some reason but I simulated them with that external comparator U3). Set is the non-inverting (+) input of an internal comparator in the LT1073, the inverting input is 0.212V, and the output is the pin A0. Because of our voltage divider with R5 and R6, when the battery voltage is above 3V, A0 is high. When it falls to 3V, A0 goes low, which activates the PNP transistor Q1. This brings the voltage of FB above 0.212V, which turns off the LT1073. R9 adds some hysteresis, so the oscillator won't turn back on until the battery voltage rises to 3.4V. This prevents the neon lamp from flickering wildly when the battery is running low, since there will always be a little noise in the battery voltage.

The other addition is the charging shutdown feature. V4 is the hypothetical charging voltage for the battery, coming from the wireless charging coil when the light is placed on a charging pad. This will be 5V when the battery is charging. I've tied it to FB to turn off the LT1073 when we're charging the light, adding a resistor to limit the current and a diode to prevent voltage seeping into the charging coil when we're not charging.

You can test all these features in the LTspice file I've attached. You can change the battery voltage V1 to different values, or connect the AC voltage V3 in series with V1 and see what might happen when the battery discharges and charges. Once you're happy that the output voltage and current are right for your neon lamps, you can breadboard the circuit and test it in real life.

Step 3: Single-Cell LiPo Charger

There are a lot of lithium polymer battery charging IC's out there, and it took a lot of poking around to find one that was meant for charging a small, single-cell LiPo battery. What I settled on was the MCP73831, which is a small SOT23 surface-mount chip that has a bunch of fancy charging features, including fast charge and a status pin that can turn on an LED while charging. The chip takes an input voltage of 5V, and it's easy to configure the maximum charging current using a resistor connecting PROG to ground.

The LiPo battery I'm using has a capacity of 250mAh. Most small LiPo batteries are designed so the maximum fast-charging current is the same as the capacity (250mA). The MCP73831 Datasheet says that a 4k resistor would limit the charging current to 250mA, so I picked the nearest standard value, 4.7k.

The STAT pin sinks current when the battery is charging, which means I can connect the negative side of an LED to the STAT pin and a current-limiting resistor to the positive side. This will cause the LED to turn on when the battery is charging, and turn off once the battery is fully charged.

The two capacitors are filtering capacitors, which try to get rid of any AC noise on the input and battery voltage.

Overall, this charging chip was very easy to set up and use, and it worked the first time I tried it. However, this is a surface-mount chip, which means that I had to design and order a PCB for the first time ever. There were a couple battery-charging IC's that were in the 8-pin DIP package, but none of them were designed to charge a single-cell LiPo battery without much additional circuitry.

One more note: There are a couple different varieties of this chip, detailed on page 26 of the datasheet, under "Product Identification System." All the varieties are very similar; the things that change are the behavior of the STAT pin and the specific charging voltages, currents and modes. I picked the MCP73831T-2ATI/OT since it seemed to fit the parameters of a single-cell LiPo, but I think the other versions would work just fine.

Step 4: Wireless Charging Module

There are a lot of wireless charging receivers available on Amazon and other places, but I wanted to be able to charge my neon lamp tea light on any regular phone charging pad. Phones use a wireless charging standard called Qi, which allows a charging pad to communicate with receiving devices. This is advantageous because it means a charging pad doesn't need to constantly be wasting power when there's no device on the pad. Instead, every second or so it can send out a small "Anyone there?" pulse. If a Qi wireless receiver sees that pulse, it'll send a reply: "I'm here!" Once that happens, the charging pad and receiver will communicate a bit, and then the pad will start transmitting the full amount of power to the receiving coil. The charging pad can also detect when the receiving coil moves away, and then it will stop transmitting power and go back to looking for new devices.

I bought this little receiving coil, and it worked great. It outputs 5V, and it worked fine with all the wireless charging pads that I have in my house. I suggest testing out the coil's output using a voltmeter, and maybe an ammeter because the battery charging chip will need a lot of current to charge the battery. The coil and chip can get hot pretty quickly when outputting a lot of current, so be careful.

One thing I had some difficulty with was the 5V and ground wires. When I finally tried to solder them into the PCB I ordered, I wanted them to be a lot shorter, but when I cut and stripped the wires halfway down it turned out they were this weird type of braided wire that doesn't have exposed copper in the middle. I ended up desoldering the wires and soldering my own on, which didn't turn out to be too difficult.

Step 5: The Full Circuit

Now we can put the whole circuit together! This is a screenshot of my schematic in Autodesk Eagle, which I downloaded for free with an education license (there's also a free version you can get without an education license or paying). I chose Eagle because I've seen lots of people online use it, and there are a lot of tutorials out there. The only complication I had converting my LTspice schematic into a schematic in Eagle was that I was using some components that didn't appear in the Eagle standard libraries. I found some online using SnapEDA, and I had to create my own libraries for some others. I've uploaded my schematic file here, along with all the libraries.

The only new thing with this schematic is a switch added between the battery and the LT1073 (although the battery is still connected directly to the charging circuitry). As you'll see on the board in the next section, the footprint of this switch isn't an actual switch; it's just two vias (electrically connected holes) next to each other. The idea is that I can solder wires into the holes and connect those to a switch somewhere, or just solder them together if I don't end up needing a switch.

One more thing is that I have two resistors (R3 and R10) in series. This is because I wanted a 26.2kohm resistor in order to get an output of exactly 81V to the neon lamp, but the variety pack of surface mount parts I ordered didn't include resistors with that value (no duh). Each component in a schematic in Eagle represents a physical part which has a physical footprint on the board, so you have to be mindful about the final PCB even when you're designing the schematic.

Step 6: Design the PCB in Eagle

Now we place components and route traces! The wireless charging coil I picked fits within a circle of about diameter 40mm, so I wanted to fit all the circuitry on a circular PCB of that size. That meant I had to use some surface mount resistors and capacitors.

There are three common sizes, or 'packages', for surface-mount resistors and capacitors: 1206, 0805, and 0603, where the digits say how long and wide the packages are. For example, a 1206 resistor is a tiny box 0.12 inches long and 0.06 inches wide - that's tiny! These different packages also have different power ratings (or voltage ratings for capacitors). A 1206 resistor can dissipate 1/4W, an 0805 1/8W, and an 0603 1/10W (here's a handy guide). There are other larger and smaller SMD packages, but these are the most common.

The big advantage to using some surface-mount parts is that it makes my board take up much less space. It also wasn't too difficult to find the parts I needed - this SMD component assortment on Amazon had all the common SMD parts I think I'll ever need, and includes a neat guide to identifying the different components. Of course, we can't naively replace all our components with surface-mount equivalents - in this circuit, L1 needs to be able to handle 1A of current without saturating, D1 D2 and D3 need to be rated for 100V and also be fast-switching diodes, and C1 C2 and C3 need to handle 100V. Maybe there are small, surface-mount components that could meet these requirements, but I couldn't quickly find any, so I stuck with the through-hole versions I already had.

If you look at the image of the board, you'll see that I used 3 through-hole resistors instead of surface mount resistors. These resistors needed to have values that weren't included in the SMD assortment I ordered. In addition, R3 and R10 control the output voltage, so I figured that in the future I could reuse this board as a variable voltage supply by wiring a potentiometer into where R3 is.

The other two notable things are the screw holes in the sides and the four vias at the bottom. These vias are for soldering in the battery and the wireless charging coil. Maybe I should have used a standard connector for the LiPo battery instead, but I thought it would take up too much space and did this.

Oh, also, I opted to use a super tiny surface-mount 0603 red LED as the indicator light for the charge controller. This saved a lot of space, but it was a little tricky to make sure the LED was facing the right way when I assembled it. The SMD component assortment I ordered included multiple colors of 0603, 0805, and 1206 LEDs.

Now that the board is all done, it's time to place our order!

Step 7: Order the PCB!!!

Updated: Now it's easy to order the PCB for this project directly from PCBWay! Here's the link:

Here we go! I was very nervous about this step, since I had never designed and ordered a PCB before, but it all went very smoothly. I used PCBWay, which made the whole process very cheap and easy and felt accessible to someone like me who has no electrical engineering background and had never done something like this before.

All I had to do was get Eagle to generate the drill files and gerber files for my board (through the manufacturing tab and the CAM window), and then upload those to the PCBWay site as a zip archive. PCBWay reads those files and fills in all the complicated parameters automatically! The price for ten boards was $5, and they actually sent me 12 boards - and I used a $5 coupon, so they made my boards for free! The shipping price was a little large since PCBWay is located in China, but my boards arrived in a week and were very high quality. I also ordered a solderpaste stencil, which is optional but makes soldering the surface mount parts a lot easier.

Before you start soldering anything in, make sure to use your multimeter to do continuity tests between important points. Make sure VBAT and GND aren't shorted, since that could make your battery explode and that wouldn't be fun.

Step 8: Surface-Mount Soldering With Solder Paste

Hopefully your SMD Component Assortment and MCP73831 charge controller have arrived by now, because it's time to solder those components in. It's certainly possible to solder surface-mount components using a regular soldering iron with a fine tip, but it's pretty hard, and instead of doing that I chose to use solderpaste.

Solderpaste is like regular solder, but in a clay-like paste form. When it heats up to a certain point, its crystalline structure changes and various things evaporate, and it melts into a much smaller puddle of metal solder. The wonderful thing about solderpaste is that it appears to magically flow to exactly where you want it to be! If you put two dollops of solder paste in the pads for a 1206 resistor and squish the resistor into place, then when you heat up the board to melt the solder paste the solder will pull the resistor onto the centers of its pads. It's a lot of fun to watch. There are tutorials on YouTube that show how to use solderpaste, so I watched a few of those and it didn't end up being too difficult for me.

A stencil can be used to conveniently spread solder paste exactly where it needs to go. The stencil I ordered from PCBWay was a lasercut sheet of metal that had square holes wherever the SMD pads for my parts were. Secure your board so it won't move or twist, then place the stencil on top of the board so you can see all the pads. Then, squeeze out a line of solderpaste on the stencil, and use an old credit card or a squeegee to spread this line back and forth over the holes in the stencil. This deposits just the right amount of solderpaste onto your board in the right spots.

After that's done, use nice tweezers (like the ones from the SMD Component Assortment you ordered) to place the surface-mount components in the right places, one by one. I suggest using the autogenerated part list I've uploaded to guide you (I even printed it out, although the formatting makes that a little difficult). Make sure your diodes are facing the right direction, especially the 0603 LED which you should test with your multimeter to determine the proper orientation. It's easy to have these tiny components go flying if you're not careful.

Once you're done placing the components, it's time to heat the board and melt the solderpaste. This step is the most fun, but it requires some sort of heat source for the board. There are very fancy reflow furnaces that you can program to match the recommended heating profile of your specific solderpaste, and there are people who stick an arduino on their toaster oven and make that work. I didn't have a toaster oven lying around, so I bought this LED Repair Heating Plate on Amazon for less than $15. It doesn't have temperature control, or even a switch; when you plug it in, it gets hot, and that's what counts.

I had great results with this, but it is important to avoid burning yourself or catching anything on fire since the plate does get quite hot and takes a long time to cool down. Once you've made sure your workspace is clean and free from any flammable items, put on some safety glasses, plug in the hot plate and enjoy the show! You can turn off the plate once the solderpaste on every component has melted.

Step 9: Solder the Through-Hole Components

Now that you're done with the surface-mount components, we just need to solder on the non-surface-mount things. This should be pretty self-explanatory, although there are two important points.

Remember the pseudo-switch we added, S1? There isn't actually going to be a switch for the neon tea light, since I decided (quite naively) that it was unnecessary - although mainly there wasn't much space to add one. Therefore, I soldered the two pins of S1 together with a small piece of wire. If, like me, you ordered 10 of this board from PCBWay and have a lot left over, then you can reuse this board for other fun portable neon circuits, and when that time comes you'll probably want a switch.

We haven't designed the neon lamp tea light's enclosure yet, so we don't know how long we want the wires on the battery of the charging coil to be. We also don't know how much we want the neon light to stick up. I suggest soldering wires into the four battery and charging coil vias on the edge of the board, and also the two vias in the center where the neon bulb will go. Then, you can connect these wires to the battery and charging coil with alligator clips and test the board's output voltage, whether it can light a neon lamp, and whether it can charge the battery (and turn on the SMD indicator LED while charging). If all your tests check out and it looks like the board is working properly, then you can desolder these wires and move on to designing the mechanical enclosure.

Step 10: Design a Case

I spent a lot of time agonizing over how to make the case as small as possible, because at this point I still intended to create a neon lamp necklace. I made a multi-layered design in Fusion 360 (free with an education license) where you place the charging coil at the bottom, then place the LiPo battery into its holder and slip that in, then place a thin plate that separates the board from the battery. I used 6mm tall M2 screw inserts that slide into holes in the charging coil holder (orange), and then you screw the board down into those and everything is held in place quite snugly.

The thing is, I did all this hard work on the bottom to make my necklace as small as possible, and then I was thwarted by the height of my PCB! I modeled some of the components on my board and used a convenient feature in Eagle, which lets you turn your board into a 3D model in Fusion 360. At that point, I discovered that the blue inductor was actually really tall, a lot taller than all the other components! This meant I had to make the top part of the case (which screws into the bottom part with 3D printed threads) a lot taller than I had intended. That's the point at which I decided to turn this into a neon lamp tea light. In the future, I think I'll replace this through-hole inductor I chose with a smaller 100uH surface-mount inductor which I recently discovered online. For now, though, this case is perfect for a tea light.

Step 11: 3D Print the Case

Rather than print the parts for my board with a typical FDM printer, I used a resin printer. This was more convenient for me since I have a QIDI TECH Shadow 5.5S at home, but I think you should probably be able to print the case with most printers. The nice thing about resin printers is that they're really precise. The photosensitive resin used by resin printers is a little difficult to deal with, since it's toxic and you need gloves and a ventilator, but the printing process itself is easy and straightforward, and the results are better than anything I've gotten with an FDM printer.

If you do try to use a Makerbot or other FDM printer, you may have to do a lot of filing or sanding away of material. You might have a lot of difficulty getting the 3D printed threads connecting the top cover and the board holder to screw together nicely. If that's the case, and rubbing a little dish soap in the threads to loosen them up doesn't help, then you could try sanding off the threads all the way and turning it into a pressure-fit case. If even that doesn't work, then I suggest throwing away the top cover. I actually think the tea light looks really cool without the top cover - that's why I made a photo without the cover the title photo for this Instructable!

Step 12: Putting It All Together

The final challenge! Congratulations if you've made it this far. There are some tricky pieces to the puzzle here, so don't get complacent, especially when you solder the battery in.

The first thing to do is secure the M2 screw inserts in their holes, in the charging coil holder (the bottom piece). This was easy for me with my resin printer; all I did was dip a paperclip in a bottle of photosensitive resin, spread a little of that resin around in the holes for the inserts, then drop the screw inserts in and cure it with UV light. This worked great for me, but if you used an FDM printer then you can probably put a tiny dab of glue around the walls or at the bottom of the holes and slide your inserts in.

Bend the copper wires that go into the black PCB for the charging coil so the PCB sits on the coil. This will interfere with the wireless power transmission a little, but we'll deal with that later. Place the coil into its indentation at the bottom and squish it down, ensuring the +5V and GND wires are pointing to the side.

Now put the LiPo battery in its holder and slide that down into place, making sure the charging coil's wires fit through the indent in the battery holder. Hopefully this battery holder will slide all the way down against the plastic case; if it doesn't, then the charging coil could be blocking it.

Position the board holder on top of the battery holder, aligning the little caps with the screw inserts and making sure the charging coil and battery wires fit through the slot on the board holder.

Put the board on the board holder. You might find that the board doesn't rest nicely on the board holder; this could be because some of the through-hole components have their leads sticking out too far from the underside of the board, and you can use your wire cutter to fix that. Try screwing the board down into place using two M2 screws - for me, this took a little finicking to get things aligned just right.

Now that you've got the board in place, we've got a problem - the wires aren't soldered in yet! Well, the reason we did all this is to measure how long to make the wires. One by one, grab the middle of the battery and charging coil wires and try folding them into their respective vias. Use a pen to make a mark on the wires where to cut and strip them later, making sure to leave enough length to make assembly possible. Another thing to do now is decide how long to make the leads on your neon lamp - try inserting the lamp into place and holding the top cover next to it to decide how high you want the lamp to stick up. Once you've decided on a length for the leads, fold them at 90 degrees at the point that you'll solder them into the board.

Regretfully, now we have to disassemble the board. Once you've disassembled it, cut the wires at the points you've marked - TAKE CARE NOT TO SHORT YOUR LIPO BATTERY! IF YOU CUT VBAT AND GND AT THE SAME TIME, THE BATTERY WILL SHORT AND BE DESTROYED! I suggest using electrical tape to insulate the LiPo battery wires temporarily, to prevent shorts.

Solder the wires into place, with the wires coming out of the top of the board. BE CAREFUL not to short VBAT and GND while soldering since those vias are right next to each other. If you solder GND first, you can then cover GND in electrical tape when you solder VBAT, and this process is pretty safe. Also, now's the time to solder the neon lamp into place.

Because I decided not to include a switch, your neon lamp tea light probably just turned on. This is good! However, be careful not to touch anything on the board, since some high voltages (~90V) are present. It's not going to hurt you at all since it's high voltage at very low current, but you could certainly damage the LT1073 if you short the wrong two pins.

Now it's time to do the final, real assembly. Follow the same process as last time, and if you made your wires long enough, then it shouldn't be too difficult. If all goes well, you'll have a fully assembled and glowing neon lamp tea light!

Step 13: The Final Result!

Congratulations, you've successfully made your own neon lamp tea light! I had a lot of fun with this project, and I'm really happy with the result. My light has a battery life of about 20 hours, and will charge to full capacity in 4 hours. This pretty much meets my goal of having a battery that lasts for a day, and if you wanted it to last even longer you could change the load resistor R4 to have a value of 47kohm or even 68kohm - I think I'm passing a little too much current to my neon lamp at the moment, and halving that current would probably double the battery life.

If you're happy with your neon lamp tea light, then great! If not, then check out the troubleshooting section below for a couple potential fixes to whatever problems you're having. I hope you vote for this Instructable in the PCB competition!

Step 14: Troubleshooting

I had some difficulty getting the wireless charging receiver to work in the fully assembled light. I disassembled the whole thing and placed the charging coil in the case, and the charging worked fine like that. However, when I placed the battery in the case, I could no longer get the wireless charging to start up.

I think this is because the battery absorbs a lot of the energy broadcast by the transmitter, since the battery is right on top of the charging coil. To fix this, I sanded down the bottom of the case to reduce the distance between the transmitter and the receiver coil, so more energy goes straight to the coil. I also cut some thin strips of tissue and placed them between the battery and the battery holder, which slightly increases the distance between the battery and the battery holder. After that, I was able to get charging to work even with the battery inserted.

If you get to that point and try to screw in the circuit board, only to find that charging has stopped working again, then you can try loosening your screws to increase the distance between the circuit board and the charging coil.

One more trick is to make sure you're placing the receiver coil exactly in the center of the charging pad. I actually traced a circle on the pad once I finally put my neon lamp tea light on a spot on the board where it could happily charge.

If none of these tricks work, then you may have to try multiple charging pads. This charging pad worked well for me, but I had trouble with a fancier brand-name model that can detect when your phone's not in the right spot. If even that doesn't solve your charging issues, then leave a comment and I'll see if I can help.

Thanks for reading this far! This is my first Instructable, so if you liked it then please follow me, leave a comment and vote for me in the PCB competition! I might post more interesting circuits involving neon glow lamps in the future, so stay tuned...

PCB Challenge

First Prize in the
PCB Challenge

Be the First to Share


    • Pi Day Speed Challenge

      Pi Day Speed Challenge
    • Trash to Treasure Contest

      Trash to Treasure Contest
    • Microcontroller Contest

      Microcontroller Contest



    Tip 5 weeks ago on Step 6

    You can reduce the variability between lamps by regulating the current through the lamp instead of the voltage. Swap the neon lamp and the series resistor, so one end of the resistor is grounded. Now you have a voltage proportional to current. You can split this resistor into 2 and pick off the "tap" that regulates the current to what you want. You may be able to reduce the (total, 33K) value of the "ballast" resistor, increasing efficiency. Couple in the original voltage feedback with a diode so if the lamp fails or is absent, the circuit won't "run away" and damage itself.


    Reply 5 weeks ago

    Wow, this is awesome! Your current regulation is actually ideal for powering a neon lamp, since after the lamp strikes at 80V, it only needs 60V to stay lit. I think the value for the resistor(s) can be pretty low - I think 330 ohms would give 0.212V for the feedback at about 600uA? Anyways, thanks! I'm definitely going to try this out next time.

    Edit: I simulated this and it seems pretty good. I didn't think at first that splitting the resistor into two in series was necessary, but it has a couple effects. The bottom (grounded) resistor determines the regulation current, and the top one (on the lamp) doesn't affect the current but "adds extra voltage" to the output since it drops part of the voltage to FB. A big top resistor also reduces the voltage ripple at FB, and output current ripple, which I think is only significant because the 150uH inductor is too big.
    One consideration is that, when the lamp strikes and the lamp voltage drops from 80V to 60V, there's an almost 5V voltage spike at FB (with a low value for the top resistor). This is significant since the maximum voltage allowed for the FB pin is 5V, according to the datasheet. I think there should be an easy way to protect it, but without protection certain bulbs could potentially break the LT1073. Still, this circuit will be a lot more efficient than my original one! (I had about 10% efficiency I think...)


    Reply 5 weeks ago

    Protecting the IC should be easy. First, DO NOT ELIMINATE the "upper" resistor in series with the lamp. There are 2 reasons. First, that resistor will limit the current spike into the IC. 2nd and more important, the lamp's V-I characteristic has "negative resistance" & operating point of the lamp will be unstable.
    The the "positive" resistance in series must be greater than the negative resistance, to make it stable. This is common in most discharge lamps (fluorescent lamp, sodium vapor lamp, mercury vapor lamp, "CFL"). This is why such lamps need a "ballast" to operate.
    Yes, that "extra voltage" is there and contributes to circuit losses, but is a "necessary evil". For expensive high-power things like gas laser tubes, the loss in the ballast resistor is 1/3rd that in the tube! That's why there are patented circuits to "present" high series resistance at high frequencies, but low resistance to DC.
    Look up this patent:
    You could add-in this circuit in place of "top" resistor to see what efficiency you could get. You should be able to lower the ballast voltage drop 10x. It will also reduce the ripple. DO NOT REDUCE the inductor value! Increase the capacitors value in the doubler circuit to reduce voltage ripple.
    For IC protection, add 2 series 1K resistors between top of 330 ohm sense resistor and the IC sense input. At the center tap, put signal diode (1N4148) to the IC's + power rail (anode to resistors, cathode to the rail). The diode & resistors will "do nothing" during operation, but clamp the voltage to the rail during start. You should also have independent voltage control loop "diode'd" in, to prevent voltage runaway if the lamp fails or is removed.


    5 weeks ago

    Wow, this is an excellent project. I’d be interested to see if AC (or almost-AC) could be generated with a tealight sized circuit like this. Seems like I’ve seen a way without using a transformer but I could be wrong. Using more inductors and swinging the voltage up and down with reference to a mid point. Those neon lamps don’t take much current so it might work. I might have to try making my own version someday.

    Anyway, nice job and you have my vote :)


    Reply 5 weeks ago

    Thanks for the support!

    Yeah, there are definitely ways to generate high-voltage AC with capacitors and inductors. Things I've seen so far are RLC tank circuits (oscillator where the current flows back and forth between the inductor and the capacitor), a ZVS driver (uses big mosfets and inductors to create high-voltage, high-current AC like for an induction heater that can melt metal!), and a royer oscillator (don't really understand, but uses a center-tapped transformer and a transistor). The royer oscillator sounds a lot like what you described, since it's basically two inductors in series where the connection is the middle reference point.

    My main difficulty is that I don't really understand oscillators yet hahaha. I think I'll come back to this project in a couple months after I read up on things, and maybe then we can make a tiny AC version. My final goal is neon lamp earrings, but don't hold your breath on that!


    Reply 5 weeks ago

    Stepping up AC is a problem when you need a 1:100 stepup. At low frequencies your inductor or transformer will have a lot of loss due to parasitics. It might be better to make a high frequency oscillator and feed an inductor to make a current generator. The output voltage can be very high. I have one right here hooked up to my computer, feeding a "plasma ball", off of 5 volts.

    Stepping up DC with a voltage multiplier is also a problem with high ratios. You'll need a lot of capacitors and diodes, and the leakage drags you down.


    Reply 5 weeks ago

    A higher oscillator frequency does sound like a good idea - if I remember correctly, the LT1073 is most efficient at a 1:3 stepup ratio (sans voltage multiplier), and it should be sending pulses at around 19khz. I googled "usb plasma ball" and found a lot of useful things, so thanks! I'm a little scared to plug one into my computer directly though, since I think the high-voltage RF could capacitively couple to my computer's ground and fry things if I put it too close.


    Reply 5 weeks ago

    I have one sitting on top of my dell right now. they are pretty well shielded so that RF noise doesn't leak OUT. Kids love to play with it so I gave them one at Christmas. The one I have has a bunch of secondaries in series, stacked on top of each other. That limits the stray capacitance. You'd only need a few secondaries to fire a neon, not the dozen I have.


    Reply 5 weeks ago

    Yeah, that would be awesome! I’ll keep an eye out for if you do :)

    I haven’t played with oscillators much either, but I might have a reference circuit in an old notebook


    Reply 5 weeks ago

    The output of the LT1073 is AC. Otherwise, the voltage multiplier would never work.

    The Orchidomaniac
    The Orchidomaniac

    5 weeks ago

    I probably would have used a transformer, as those can be homemade. You might want to try stepping the voltage up to around 175V or 250V and attach the negative lead to a terminal of a small flashlight lightbulb to make a plasma ball. The bulb should be filled with gas, like the lamp. I'm not sure if this will work, though.
    I have learned yet another way of getting insane voltages from tiny power sources today...


    Reply 5 weeks ago

    Ooh, I didn't know you can make transformers yourself! I have some ferrite cores I salvaged, so maybe I'll try that...
    The lightbulb plasma ball is a good idea, and you can create a similar arcing effect by running a lot of current through a neon lamp - if you don't mind destroying it, that is! I used a small high-voltage module to make a mini jacob's ladder a while ago, but I don't have any plasma balls yet, so maybe I'll give that a try.

    The Orchidomaniac
    The Orchidomaniac

    Reply 5 weeks ago

    There is a great diy plasma ball here on instructables. Use a 555 to drive a mosfet controlling a transformer. It uses around 10,000V, so be careful!
    I put a little bulb at the end of a tesla coil, and as it was a krypton bulb, i got some sick blue-green arcs!
    It is said that if you put a spark gap between the bulb terminal and transformer wire, the plasma streams seem bigger. This is because more charge collects before discharging, though you will smell ozone as soon as the spark starts up!
    BTW, you can make transformers even out of iron nuts or keychain rings as well, though ferrite cores are best.
    May your sparks be great, your winding smooth, and your circuits correct!
    -The Orchidomaniac


    5 weeks ago on Step 14


    But all it does is come on, and stay on until the battery goes flat, right?

    You can easily make a "relaxation oscillator" with a cap, resistor, and a neon bulb. The resistor is in series with the cap and neon bulb, which are in parallel. The resistor charges the capacitor up to the strike voltage, at which point it quickly discharges to below the hold voltage, then it repeats the cycle.

    Because the duty cycle is so small, you could run a half-dozen of these oscillators at or below your energy budget, and if they all have different RC time constants, it will have a flickering effect!


    5 weeks ago

    Very nice project and very well documented. Thanks for sharing.


    6 weeks ago

    Great first project. Thanks for sharing.