Digital clocks
that time forgot (or that forgot the time)

see copyright notice. Page created 21-May-2001 updated 2-Nov-2019. Use the button groups above to navigate quickly around the site.

This was my first horological effort. I made it circa 1976 from a motley collection of end-view "Nixie tubes", 5490 counters and unmarked (7441?) decoder/drivers. The ICs were "surplus" stock, and I think the tubes came from an amateur radio junk sale.

The power supply has long since been cannibalised but I remember being determined to spend as little as possible on the project, so the HT was rectified straight from the mains (no transformer). I grudgingly put a transformer in for the 5V logic supply, but because the logic current draw was more-or-less constant, I got away with a large wirewound resistor instead of the conventional voltage regulator :)

The rightmost (seconds) display was a genuine B-5092-A Nixie Tube, made by Burroughs in the USA. The rest were type GN-4A made by STC/ITT in England. I must have had some spares because I dissected one to yield the set of cathodes pictured here.

I designed the software and circuitry (excluding the receiver module) for the Rewbichron (1982) and Rewbichron 2 (1983) MSF Clock kits, sold by Ambit International (later Cirkit) and described in the magazine Radio & Electronics World. The latter design was always intended to support more elegant displays than the original 7-segment LEDs, but here's one that never made it to market! It's based around an NEC Fluorescent Indicator Panel type FIP60B30T, which was essentially a valve with 60 radial anodes and two annular mesh grids, such that you could make a long or short illuminated "hand" appear at a given place by pulsing everything as necessary. The challenge was to find a sane way of multiplexing the electrodes to make three hands go round as expected.

Various driver chips were said to exist. I couldn't locate supplies, but I did have a box full of HV PNP switching transistors. So I opted for a transistor per anode, driven at their bases and emitters by bitmaps decoded from the Rewbichron's CPU. Beneath the bottom row of transistors you can see part of the decoder, i.e. several square inches of veroboard taken up by hard wiring!

Finally, here's what I claim to be a prototype of various "mechanically-scanned LED" devices that later proliferated as both domestic clocks and advertising displays. I made this contraption, christened The Scandulum, in 1986. It comprises a Z80 CPU clocked at 32.768kHz [sic] which "kicks" a 6V model motor once per second. Attached to the motor is a 10-inch pendulum, carefully trimmed to resonate at 1Hz. As the pendulum swings L to R, the CPU pulses seven red LEDs with a dot matrix pattern that's timed to produce a "digital" time display, in much the same fashion as a dot matrix printer (remember them?). It relies on persistence of vision. Exhaustive tests showed that only about 50% of observers could actually read the time; the trick was to stare at the wall behind the pendulum, and not try to follow its movement.

Of particular note is the method used to set the time. I've always distrusted mechanical devices such as switches, so this design went to extremes and limited the controls to one microswitch (on the left in the picture). Once the pendulum is swinging like a good 'un, you press the switch once; the hours then start to increment at each swing. When you see the correct hour, press again; then the minutes start to count up. A third press of the switch holds the seconds count at zero; release the switch at the correct time. Simples...

top ▲