The antique clock from my grandmother has a unique sound. Its bells are resonating through the house.
Unfortunately, not all residents can enjoy the sound of clock bells during the night, so normally the clock is stopped in the evening and started (theoretically at exactly the same time 12 hours later) in the morning. Since the latter was often forgotten, the clock was not always running or it was necessary to fast-forward to the correct time because we didn't restart it on time. Luckily, the design of the clock allowed for a rig that could automate the start and stop procedure.
At this point I should add that I have experimented with dampers between the hammers and the bells and/or other dampening tactics on the bells, but those rigs would still leave the clock to generate a lot of 'awkward' hammering noise. Taking the clock apart was far too invasive and risky and therefore not an option.
Of course not all clocks are suitable for this kind of rig. In this case, the pendulum swings quite close to the bottom of the clock casing, thus making it possible to create a rig that sits on the bottom but stays out of sight.
The control and timing is performed by an AVR Atmega88. It might be a bit overqualified for the job, but it has the advantage of having a timer that can be clocked from an external crystal, so it is very easy to make it an accurate clock using a simple watch-crystal (32.768kHz). Also, the mega88 works on voltage below 5V, so it can be powered by 4 rechargable batteries (4.8V). Since the stepper only moves about 40 seconds each day I suspect the batteries won't need to be charged that much.
The rig contains a stepper motor to start and stop the pendulum from the clock. This actual stepper is a 5V type from Ebay, so it will work with the battery power of 4.8V. It also has a 1/64 gear so it's not really fast, but the gear compensates in sturdyness to absorb the pendulum impacts. On each side there is a switch to detect the rest position of the pendulum start/stop arm. Apart from the AVR there is only a ULN2003 to drive the stepper, a watch crystal and a LED, so it's only a 4 parts circuit (7 if you count the programming header and the microswitches). The circuit is in this diagram. Actually, I think I haven't even included the decoupling capacitors in the power supply lines that are drawn in the circuit.
The AVR program is very simple. The watch crystal triggers an interrupt that counts seconds. When 12*3600 seconds have passed the clock is started. After again 12*3600 seconds the clock stops. This cycle repeats forever. A led light blinks each second so you can establish the startstopper is still working. Stopping the pendulum is accomplished by moving the stepper arm in vertical position very very slowly. After the pendulum has come to a full stop, the arm is returned to rest position. Starting the clock simply flips the arm over. Here is the AVR Code.
Here is a demo vid from the first prototype that shows the actual physical action: Pendulum StartStopper PrototypeAfter fixing the last bugs in the AVR code the clockstopper is airbrushed in matte black to become almost invisible and now fully functional:
In this case, the rig is set to a startstop timing of 40 seconds to demonstrate and keep the video short at the same time.
This concept is universally applicable to other pendulum clocks, but the mechanics and timing must be customized to each specific clock. It's still a nice hack though.
Update: I've replaced the 4 NiMh cells with 1 3.7V LiOn cell, which should last a lot longer. Because the stepper motor is formally a 5V model, the code has been modified to 8-step mode to keep the stepper motor working.