TritiLED Lights Up The Night, Doesn’t Make You Glow

Tritium, or 3H is an isotope of hydrogen which has been used as everything from radiolabel in analytical chemistry to a booster to kickstart the chain reaction of nuclear weapons. Lately tritium’s most common use has been in key chains and jewelry. A small amount of tritium is stored in a phosphor coated glass tube. The beta decay of the tritium causes the phosphor to glow. The entire device is called a Gaseous Tritium Light Source (GTLS).

In the USA, GTLS devices are only allowed to be used in specific cases such as watches, compasses, and gun sights (MURICA!). Key chains and jewelry are considered frivolous uses and are prohibited by the nuclear regulatory commission. Of course, you can still order them from overseas websites.

The safety of GLTS devices have been hotly debated on the internet for years. They’re generally safe, unless you break the glass. That said, we’re happy getting our radiation exposure through cool hacks, rather than carrying a low-level source around in our pockets.

Enter [Ted Yapo], an amateur astronomer. After tripping over his telescope tripod one time too many, he decided to take matters into his own hands. He’s designing TritiLED, a dim LED light source which can last for years. [Ted] is using a  Luxeon Z LED, driven with PWM by a PIC 12F508 8 bit microcontroller. Running at 26.3 μA, he estimates about a year of run time on a CR2032 watch battery, or a whopping 15 years on a pair of lithium AA cells. Sure he could have done it with a 555 timer, but using a micro means more features are just a few lines of code away. [Ted] took advantage of this by adding a high brightness mode, blink modes, and an exponential decay mode, which emulates the decay of GLTSs.

Best of all it’s all open source. [Ted] is publishing under the (CC-BY-SA) license on

58 thoughts on “TritiLED Lights Up The Night, Doesn’t Make You Glow

          1. Well you also want to maintain a decent circadian rhythm so that your crew doesn’t feel like shit all the time. White light at night is a great way to screw up that rhythm. I use red lights at home, as well as f.lux/twilight, and my eyes and sleep cycles are so much better off because of it (yes, it’s subjective, but it feels better than staring into blinding white light all the time).

          2. If you ever watched movies with submarines you’d know they can switch to red light for when they surface at night and want to not ruin their night vision or to be easily spotted. It’s not a case of always having red light since that would cause people to go mad I expect.

  1. Wow. This.

    I’m probably going to make a few – one day I hope the design collapses down to the size of a small/medium electrolytic and is powered by one, too :)

    Also – charge by induction and put constellations in the ceiling. Wave your coil and watch them all spring to life!

        1. Worth a try, but astronomy amateurs also like to go out to remote areas to escape light pollution and would be in places with low WiFi presence. But maybe cell tower emission can be tapped? Those are more widespread, albeit not covering all situations.

    1. I like the idea. Guess the only problem is that shrinking it’s size even more and at the same time adding an inductive charger is very hard. I’m not all too deep into RF power transmission, but i would guess the larger your coil surface the easyer it is to get energy out of the field, and of course, the weaker the field (distance to transmitting coil) the bigger you would want your coupling surface.
      You could just use smaller coin cells, like the CR1220, but you will loose quite some run time (CR1220 has around 40mAh, the CR2032 has around 220mAh).
      The idea with supercaps could work pretty well even without the wireless power part, problem is that supercaps are not that magical and still need volume to store energy, and depending on how much run time you want to get from your LED, the cap can quickly get even bigger than your CR2032 and large supercaps get quite expensive very quickly…

      1. Wouldn’t have the same lifetime. Human eyes perceive LED brightness non-intuitively; if you run the LED with PWM you can get the same perceived brightness even though the LED is only on 1% of the time. That’s a big win for battery life over running it continuously at lower current.

        1. You are misunderstanding the human eye visual response (and in fact are getting it opposite to how eyes actually respond to brightness).

          LEDs are non-linear and dimmer than they should be at very low currents because of recombination current that acts like a large resistor in parallel with the device. Run ’em at higher current and a larger fraction of that goes into making light. Nothing to do with your eyes.

    1. Minimum parts count self oscillating flyback converter.
      1x LED obviously
      1x Transistor (selected for specific Base-Emitter leakage capacitance)
      1x torodal transformer with windings designed for specific resistance, inductance & ratio (usually 1:1).
      Some copper tape or similar for RF shielding, or not.

      If designed well, will run LED for years on any single cell of more than 0.7v, with additional feature of telling you when battery is nearly dead by flashing LED as cell voltage dives below BE forward bias and recovers.

      You may add:
      Capacitor if transistor leakage capacitance insufficient
      Variable resistor for frequency / brightness control

      Possible circuit ideas at:

  2. Could he have used a 555? My understanding is they are pretty current-hungry little devices. Set and reset are just voltages on a voltage divider that wastes a fair amount of power..

      1. Yes, it can – or better “could” be and it would still be ~100µA of current. And the classic 555 has a divider of 3*5kOhm. But I would just have used a resistor.

  3. Light switches are an ideal application for this sort of thing. I have a couple of tritium Traser Glowrings I bought some 15 years ago, they’re still glowing but obviously not as bright as they once did. I ended up putting them either side of a light switch so when coming in at night I can always easily find it in the dark.

  4. Even if you break the glass vial inside those Tritium lights, they are not that dangerous. Tritium is a gas whitch dilutes pretty quickly with the air. I’m not saying you should inhale a broken light, but open a window and you’ll probably survive.

    Carrying two of these ringlights in your pockets for 50 years has a similiar dose of radioactivity as one intercontinental flight. I have one on my key chain for about 5 years now and I’m still alive.

    1. Well, that is an understatement. No ionizing radiation will get out of any properly sealed container of Tritium. Tritium emits beta radiation with an energy of 19 keV ( This gives it range of 1 cm in air or less than 0.5 mm in glass ( In fact, most radiation detection instruments can not detect radiation from Tritium as it is stopped in the outer shell of the detector.

      Furthermore, I do not know of any cases where tritium has caused acute radiation sickness (even after ingesting large amounts) as the energy of the radiation is so low. There are far nastier radionuclides.

  5. While it’s a pretty neat light source, I have to ask why this solution to his tripping problems? I do some amateur astronomy and astrophotography and of course have run into similar issues of not wanting to trip over my tripods. My solution was to go to Walmart, buy some glow in the dark duck tape and put it at the top, bottom and adjustment knobs/handle of my tripod. Don’t ever have to worry about charging them since my tripods are almost always stored in my vehicle, too many trees around my house for star gazing. Just seems like way overkill for a very simple problem.

  6. Neat Hack to get use of the µC would be to use the LED as Photodiode in dark-times of PWM, so that it could be switched of on high ambient light. This would further increase run-time to several years if it is stored in daylight.

  7. yes correctly, an LED can be used to measure light level, and this way stop the light in case it is day.
    just too bad the idle current is actually way too high to give anything near the claimed run times,
    I tried exactly the same thing, but took it even futher, only an ultra short blink every 2 sec, but the internal leak current of batteries and just their storage time with nothing connected at all, is about the same, as what such “bugs” draw. I got very disapointing run times, 8 month with 2 AA alcaline cells

    1. Back in the 70s, TI used to make an IC that did exactly this function. It was used in things like emergency flashlights on airplanes so that you could find them in the dark by the blinking LED.

    2. Today’s Alkaline cells have expiration dates at least five years after manufacture, and I would guess this is to about half their capacity, so if your bugs are discharging them in 8 months, it’s not due to self-discharge of the batteries. You need to look at ALL of the current paths in your circuit. Could be that the light level sensing circuitry is taking more current then the LED would!

    1. Because LEDs are really crappy and inefficient at very low currents, behaving like there’s a (large and non-linear) resistor in parallel with the LED: at low currents most of the power is going into that parasitic resistor (recombination current) and not into light. Pulse it at 1% duty and 100x the current (i.e., same average current) and you get FAR more light out. (Note well: the opposite happens at high currents). They have got much better: Cheap (such as they were) LEDs in the 70s barely lit at all at 1 mA, ran well at 10-15 mA (and died at 30 mA…).

      1. I tried exactly this – here’s a picture of two LEDs run with 26uA battery current – one at DC and one with 3us 100mA pulses:

        I actually have to crank the DC current up to 75-80uA before the LEDs “look the same.” @Paul: thanks for the explanation! You can read in that log linked above where I was scratching my head about why the LED is so lousy in the uA current range. You commonly see the top end of the relative brightness vs current curve in the datasheets, where the LED is more efficient at moderate currents than high ones, but I’ve never seen the curve all the way down to zero. Do you have a reference for more details?

        The camera actually doesn’t capture how much brighter the pulsed LED appears to me. I suspect there is a physiological component to this, where the pulsing appears brighter. Here, they say that green LEDs pulsed around 60Hz look 2x brighter to the eye:

        Using the 18ms nominal watchdog timer for pulse generation, you get 55Hz pulses with this circuit. Close on frequency, but with much shorter pulse width than in that reference. More study is required…

        1. I saw that in your log — sorry I neglected to mention it above.
          Here’s an example: Figure 4 shows the low end dropping off very fast below a few mA.

          That Shimizu paper from Nikkei claims the “pulsed LEDs are brighter” effect is due to the Broca-Sulzer effect, which is horseshit — the Broca-Sulzer effect, though totally real, happens only with relatively long pulses (dozens of milliseconds). (see, for example, ) It absolutely isn’t an effect on the sub-millisecond pulse lengths. What Shimizu is seeing is the poor efficiency of the LED driven at low current, and higher efficiency when run at 20x the current but 5% duty.

      1. Well, you could fill out the request form at to find out. Their customer list shows many USA-based customers (I assume you’re in the USA from your question). Or you could just order one off Amazon – they even have some with Prime service — you could have it tomorrow.

        BTW: Truly excellent writeup. Exemplary work. I’m sure I speak for a lot of people here in saying we appreciate that kind of extra effort to make an accessible, accurate and thorough writeup.

  8. If you replace the circuit with a microcontroller, key and other parts with only one resistor, it will be a great hack! If you do the opposite?
    Oh, the author made an excellent trolling for the readers of the HaD. I laughed for a long time!

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.