Andrew Warren wrote:

> Steve:  I was imagining a system in which continuous reception of
> the signal was necessary in order for the thing to keep moving.
> To stop it, you'd simply stop sending the signal.

  That's not too far out either.  I answered direct to Jose, but it
might JUST bear repeating, that for the remote control function, it
might be worth considering analogue systems that have been in the past,
or are presently used for ... remote control of vehicles/ models.

  As I see it, the "original" dual-channel analogue coding method was
called "simple-simul" and used a mechanical escapement to generate on-
off switching whose duty cycle coded one variable and repetition rate
coded the other, such as steering and speed.  You simply integrated the
output to determine the first variable, and used a "tachometer" circuit
(monostable triggered by the signal, then integrated) to determine the
other.

  This can be performed digitally quite easily; count up while the
carrier is on, count down while it is off, next time it goes on, the
cumulative result is your first parameter value, count the period from
one like transition to the next to derive the other parameter.

  This method can be made to contain redundancy quite easily, and fail-
safe (no carrier for too long; it stops) likewise.

  And the obvious alternative, a "digital" verson of which has been
mooted in this thread, is the pulse-width servo signalling as almost
universally used now for R/C models.

  Cheers,
        Paul B.