Also I'm working at the moment on a external very accurate timebase upon a 4.19 MHz crystal, in Crystal oven.
( Pendulum clocks with invar rods are very accurate, but occasionally may drift a couple of seconds a week. )
4.194.304 MHz divide by 22 gives one Herz.
For a 3/4 seconds pendulum i need 4 complete, or 8 half swings for one impulse per 6 seconds. ( I choose this value because i may use it for 2/3 seconds clocks as well )
I have not got a 1 second pendulum clock at the moment because of the hughe shipping costs to New Zealand $1000 NZ or 700$ US .
I'm thinking along the line of using a CMOS 4017 which counts the 1 seconds impulses from the 4017. decode the 6 seconds, and provide a synchronising impulse to the pendulum.
The 1 second impulse is about 0.2 seconds long.
By AND gating the 0.2 and 6 seconds impulse together to a driver transisitor (2N2222) a impulse can be put into the coil to drive the pendulum.
I haven't got this far yet because of 2 young children at home and long working days, and too much time on the forums i admit, but I will get onto it now sooner because of your interest in this great topic too.
This impulse can't be to strong, because the pendulum is already driven through the graham escapement from the electrically wind clockwork.
What country are you from BTW ?
Regards, Raymond