2009-01-05
Dekatron Emulator
The dekatron is an gas-filled counting tube used before solid-state electronics offered a similar solution. I actually have a few hydrogen-filled units in the junkbox that are well and truly end-of-life and a pair of argon ones, but they need a fairly high supply voltage and don't glow very brightly, no where near as nicely as the Neon filled slower versions that many build into clocks or the ubiquitous "Dekatron Spinner". This circuit is a simple extension of the previous discrete ring counter.
I used green 3 mm LEDs, diffused and coloured encapsulation units of fairly old vintage - not particularly bright, even at 30 mA. The LEDs aren't quite as nice as the warm glow of Neon, but the rotating effect is still interesting to watch.
To create the drilling guide for the 10 holes I wrote a simple PostScript program to place small crosses that indicate the centre of the holes in a circle of suitable diameter.
The code was developed to design the eventual clock (easy to modify should you wish to), and in fact this dekatron emulator is a dry run into building and interfacing the ring counters for a larger scale clock. The printed drill template was cut out and taped to the jiffy box lid and used to guide punching the holes with a 3 mm hand punch. Unfortunately I don't have a 5 mm punch, so for larger LEDs I'll probably have to drill. Punching is much cleaner and easier than drilling IMO, and also much quieter - I was carrying out the construction at 3 am...
Once the lid was punched, the LEDs were installed and the 100 Ohm resistors added. Next the PNP transistors, which were super-glued in place. The NPN transistors followed, then the 10n transfer capacitors, the biasing ring resistors, the grounding ring, the other 10k pull-down resistors, and finally the current-limiting resistor, bias regulator NPN and clocking NPN pull-down.
The circuit was tested and worked fine from sub 1 Hz to well beyond the flicker fusion frequency of my tired eyes... So next a simple oscillator was built of a piece of doughnut board to clock the ring. The oscillator is a very similar circuit, one of the basic ramp generator variety with a buffer NPN to give a nice sharp low-duty cycle pulse. A 2 Meg pot in the capacitor charge path allows adjusting the rate to taste.
My circuit rotates anti-clockwise with clocking, which is the correct rotation sense IMHO. :) Your taste may vary. Solder the transfer capacitors in the opposite way to reverse the direction.
The completed circuit fits with some room to spare inside the small zippy box. An RCA socket is provided to allow connection of the power source. I wired up a USB lead to power the unit. It has basically no practical use beyond a light show - just like a dekatron spinner - but perhaps it could be clocked by the PC and used as an indication of system load or some other "Das Blinken Lighten" application.
Other uses for this topology might be Christmas lights, or similar light-show application. Each stage could be built on a small PCB (or just a lump of components in mid-air) with an ultrabright LED, perhaps inside a ping-pong ball as a diffuser and strung into strings as long as you want. The interconnects need to be four wires (GND, Count/Vcc, Thru, and Bias) unless you arrange each stage to have a capacitor-stabilised bias of its own, in which case perhaps you can get away with three. By redesigning the Vcc/Count line as a voltage source instead of a current limited one you can have multiple LEDs on at the one time (non-adjacent). Each time you clock the ring the pattern of lit leds shifts along. A micro-controller could clock different patterns into the ring and could control resetting it - reset logic is important for using as a real counter and would need to ensure the "0" cell fires alone on reset, in the practical circuit one cell will dominate on first power-up and clocking but it would need to be a formalised reset for counter use. With buffering the adjacent-cell problem could be removed and you'd have a bucket-brigade or delay line memory which might be useful for some applications. Might as well use normal 7400 or 4000 series logic then though.
There has to be at least 5 better ways to implement this, especially if all you want is the visual emulation of a dekatron. The 4017 CMOS decade counter is the most obvious; a single 16 pin device that will do all the hard work for you. The 4017 is also easier to interface for real counting work. I suspect even using Neon bulbs would be easier - component count wise, but they would need to matched fairly well. The Neon would look realistic and could be put inside a glass test tube to look a bit like a real dekatron, and run directly off the mains. The discrete silicon solution is a whole bunch of work to build - a PCB would take the drudgery out of the wiring but it is still a lot of components to stuff. Building it gave me a good idea just how much work is involved in building a clock the same way. It would take at least 46 stages for a seconds counting 24 hour clock. I'm yet to solve the inter-ring coupling problem, but a simple buffer transistor will do the job, maybe capacitor coupling into the pull-down transistor of the next ring will be sufficient, and save an extra buffer transistor while giving edge-trigging at the same time.
Leave a comment on this article.
Attachments
| title | type | size |
|---|---|---|
| Animation Source GIMP File | application/octet-stream | 116.516 kbytes |
| Dekatron Emulator Circuit Source | application/postscript | 16.488 kbytes |