Light Pipe Clock Display

by Piffpaffpoltrie in Circuits > Clocks

457 Views, 5 Favorites, 0 Comments

Light Pipe Clock Display

Front view.jpg
KGM-M3 how it works.jpg
KGM-M3 side view.jpg

Several years ago I found four cute light pipe displays with 20 mm high numerals for my vintage display collection. I got them from an ebay seller located in Poland. There is no manufacturer's label on them, unfortunately, but they look almost identical to the British-made KGM-M3 displays from which I was allowed to show you two internal views (see above).

Light-pipe displays date from about 1960, and I found an English patent specification for them (PDF attached below). The two smaller pictures above, courtesy of Udo Radtke, http://www.tubecollection.de/ura/lichtleiterdisplays.htm, illustrate the functional principle and are worth at least 1000 words each. These display units contain twelve small incandescent bulbs, also referred to as 'pea bulbs' - ten for the numerals 0 to 9, and two for the left- and right-hand decimal points.

The light emitted from the light bulb currently active is injected into the edge of a transparent, plastic light guide. Twelve of these guides are stacked behind each other. Little pits in the shape of numerals and decimal points are provided in each of the light guides. When the light from the pea bulb hits these pits it becomes scattered, and the numeral (4 in the example above) lights up when observed from the front. Thanks to total internal reflection within the light guides, almost no stray light can be seen except the one from the numeral currently illuminated.

BTW, the same principle of conducting light from a light bulb through a transparent light guide can be found e.g. in earlier model train engines.

With some patience and some luck, you might still find these displays on ebay from time to time, but not really cheap, unfortunately.

I thought it would be nice to build a digital clock using these beauties. And in a first go, I produced a quick and dirty prototype that works nicely. This prototype is described below for you.

Downloads

The Concept

Bulb module with bulbs.jpg
Rear view without contact block.jpg

My idea has some major drawbacks, right from the beginning.

  • The pea bulbs contained in the displays (see above) are rather power hungry (1.2 W at 12 V, resulting in a current of 100 mA per bulb).
  • But the more convincing argument against the incandescent bulbs is that they have a limited lifetime - compared with, e.g., LEDs.
  • And the most convincing argument against the whole project is that I'm lazy - and so was reluctant to design the clock electronics required to drive the displays.

The first two points can be solved by using LEDs instead of the pea bulbs. LEDs only have advantages compared with incandescent bulbs - they are very bright, they consume a lot less power, and they live for a very long time. Apart from that, they are much more easily available nowadays than incandescent bulbs.

However, LEDs are totally different - not only electrically, but also mechanically - from incandescent bulbs. Due to this fact I had to replace the display's original contact blocks with PCBs (printed circuit boards) to which the LEDs and the connecting wires are soldered. Although the colour of 'warm-white' LEDs isn't exactly identical to the one of the pea bulbs, they are rather ok if you don't compare them side by side.

The LED PCBs

Bulb module without bulbs.jpg
LED PCB.jpg

I recommend high-efficiency, high-brightness, warm-white LEDs with 5 mm diameter for this project, specified for a current of about 20 mA. They are installed on custom-made PCBs of my own design that mechanically match the displays.

In order to find the exact locations for the LEDs I scanned the bulb module shown above. I placed this scan into a vector graphics software and could, then, use this scan as a guideline for designung, on a different drawing layer, a PCB having the donuts for the LEDs at exactly the right positions (see the PDF file below). Using the toner transfer method with my laminator I could deposit the etch resist onto some copper-clad PCB material (see my instructable 'https://www.instructables.com/Modified-Laminator-for-PCB-Toner-Transfer-Revisite/'). Don't forget that you need the mirror image of the artwork for correct transfer! Towards the end of the instructable mentioned here you will also find some information on etching and what materials I recommend to do so. The PCBs needed some drilling, too, and I did that with the PCB drill press described in my instructable 'https://www.instructables.com/PCB-Drill-Press-With-Improved-Aim/'.

Once that all was done, I had to populate the PCBs with the LEDs (2nd picture). Not every digit needs all twelve LEDs - for a 24-hours clock, the leftmost digit (tens of hours) only needs three (0, 1, 2) or even only two (1, 2) LEDs, and the 3rd digit from the left (tens of minutes) only needs 6 LEDs (0-5); I also omitted all the decimal point LEDs - so I used only 29 LEDs in total rather than 48.

To do so, I recommend the following procedure: Insert all the required LEDs at their locations on the PCB (there is one per digit) without soldering right now. Make sure that their polarity matches the driver you use - in my case, all the anodes are connected together and, via a current limiting resistor (about 100 to 150 Ω for approx. 20 mA), to the 5 V supply; the individual cathodes go to the clock PCB. Then thread this assembly into the display unit. Push the LEDs forward so that they are at the most forward location within the display. Then solder them to the PCB, cut the protruding wires, and install the resistor on the solder side. Only one common resistor is required per display, since only one of the LEDs is on at a time.

Downloads

The Clock PCB

Clock PCB component side.jpg
Driver chips.jpg
Clock PCB solder side.jpg
Proto.jpg
Where to find 5VDC.jpg

As I already mentioned, I'm sort of lazy. So I was glad to find a four-digit Nixie tube clock PCB, already populated and soldered, from an ebay seller in China - for less than about CHF/EUR/USD 25, shipping inclusive (!). It is designed for the Russian-made, front-view IN-12 Nixie tubes (ИН-12 in Cyrillic script), but these are not included. It operates with 5 V DC from e.g. a USB charger.

If you shouldn't know what the heck a Nixie tube might be (these display tubes are about of the same vintage as my light-pipe displays), check 'Nixie Tube' on Wikipedia - you will love it!

Nixie tubes require a rather high operating voltage (150-180 V DC) but very low currents (around 2 mA or even less). When checking the PCB, I realized that ULN2003A driver chips are used there (2nd picture), operating very close to their voltage limit. But they don't have any problem at all when driving LEDs with 20 mA at only 5 V instead, so substituting the Nixies by my light pipe displays was easy-peasy. The only trouble was finding the correct connecting points on the PCB (3rd picture above). Google is your friend, it also provides a data sheet for the IN-12, see below :-) It is in Russian and in Cyrillic script and so requires a moderate amount of (educated) guesswork.

As you can see from the 4th picture above, my assembly is rather haywire for the time being (and as is well known, too, nothing is as durable as a provisional setup). You can see there also some light spill from the rear of the LEDs shining through the PCB material.

The point where the +5 V can be stolen from - close to the USB socket - is shown in the 5th picture above.

This clock PCB also contained four RGB LEDs for illuminating the Nixies from the rear, which is imoho totally unnecessary per se. It makes even less sense when abusing the clock PCB for my light pipe display. I unsoldered them and put them in my junk box - for later use.

Really nice, however, with this clock PCB is the fact that it contains an RTC chip (a very accurate real-time clock) as well as a battery power reserve - once you have set the clock, you might also disconnect it for, e.g., carrying it to a different location, without it losing its time setting.

The (somewhat low) life span of a Nixie tube can be increased when all its ten numerals are switched on for a moment in certain intervals, one after the other. This doesn't in fact make sense for LED-illuminated displays. But as I didn't want (being honest: as I wasn't able) to dive into the controller's programming, I left this feature as is - you can see it in the attached video of my prototype. You can as well see there that I only installed the LEDs really necessary for displaying the time, as already outlined in Step 2 above.

A word of CAUTION: The clock PCB still contains the original DC/DC converter from 5 V to approx. 180 V, although this supply voltage isn't used in the current configuration. It cannot deliver a lot of current, so it isn't extremely dangerous. However, when you are not careful when handling the PCB while it's powered up you might get seriously zapped!

WYSIWYG

Front view.jpg

... or almost, at least. Unfortunately (but fortunately for me), the displays look better by far, and have much higher contrast, in reality than in the pictures. Alas, my camera cannot catch that, so you have to make do with what I'm able to show you here.

And what you cannot see here as well: At more or less the same brightness, the whole 4-digit display now only consumes 4 x 20 mA x 5 V = 0.4 W instead of 4 x 100 mA x 12 V = 4.8 W, which is, as a matter of fact, a reduction of some 90% - thanks to the use of LEDs!