"Wim E. van Bemmel" writes:
>The engine drives an AC generator through a V-belt.

        That would seem to be the best hope if I was doing this project.
Someone has already done the mechanics for you so all you need to do is
what somebody else already suggested and read the ripple frequency.  It should
give a very high resolution since the alternator is usually spun at several
times the engine speed in order to allow charging at idle.  I can't imagine
that the system is so free of ripple that a few millivolts can't be
capacitively coupled in to an amplifier to be digitally sampled.

        You asked for no untested ideas and I can state categorically that
alternator ripple is one of the biggest headaches in mobile communications
electronics.  You actually should have about two volts (on a 12-volt system)
of alternator signal when the engine is running so your tachometer just
needs a power lead from the electrical system and you should be ready for
speed measurements.  The only thing that is peculiar to the setup is to
establish a relationship between engine RPM and alternator output frequency
and that is a one-time linear constant.

        One cautionary thought comes to mind.  This might seem like the
basis for an automated throttle control system and it could be except that one
would definitely need to have some sort of fail safe system to guard against
the run-away condition that would result from a broken or badly slipping
alternator belt or failed alternator, itself.

Martin McCormick