Terry Pinnell said:
i built a circuit to do this about 20 years ago and will try to find
its schematic when i have time later today but right now i recall it
used to appear entirely random but my priority first is to go out
shopping for a few more packs of periods
Well, I found the full stops, but not the circuit. Best I can do is a
description.
A low frequency astable with a pot to vary its period typically has a
low of 10 ms and high 200mS. After inversion/buffering, each low gates
a TTL HF oscillator (about 24 MHz). Output from that in turn goes to 6
cascaded TTL flip-flops (binary dividers). So during the 10 mS low
time, about 24*.01*10^6 pulses are counted by the dividers. Q1 on
average changes about 120,000 times, Q2 60,000...and Q6 4,000 times.
In practice there is a random variation in the successive periods of
the astable, so the combination of 6 Q outputs effectively changes at
random, freezing when oscillation stops, then resuming again a short
time later. These drive coloured lamps via transistors.
Note that during oscillation the divider outputs are still changing,
but the time is so short this is not visible.
Additional divider outputs could be added, of course.
I also added a timer operated by a push button, so that instead of
constantly changing, the lamps changed to a new pattern only when
button was pressed. (It was a novelty toy for one of my two sons.)