A Similar Project
I have a similar project using a Standard Electrical Time (SET) slave clock made in the USA. (My interest in these clocks developed when I read about the British Synchronome system and the Shortt astronomical clock.)
I considered running my master off a 32768hz watch crystal, but decided in favor of a 4 mhz time base. There are 86,400 seconds in a day so a 1 hz error in the watch crystal translates into a three second error daily.
My controller is based on a crystal controlled PIC microcontroller. Each instruction cycle takes 1 microsecond so I use a standard 60,000,000 instructions per minute for the base pulse. I modify the pulse count to achieve sync with WWV, our time standard. In my present controller a pulse count of 60,000,620 provides an error of within a second over the course of a week.
This first controller used a PIC 12F508. Two pins operate the STOP and ADVANCE switches. One pin sends a seconds pulse out to LED or a piezo speaker. One pin operates the TIP120 switching transistor that drives the clock. The final two pins drive the crystal.
Delay loops perform the timing and the MPLAB simulator allows precise testing of the loop periods. No interrupts are needed.
A new controller uses a 14 pin PIC. Because my slave clock requires 24 volts at 100ma I decided to use a wall transformer to drive the clock and controller. There is no battery backup. So an LED is programmed to light if we've suffered a power failure.
Two additional switches allow the clock to advance or retard its seconds pulses against the WWV standard. This allows feeding the WWV and clock pulses into a scope to precisely sync the two.
Its been an interesting project. I've been impressed with the consistency of a regular microprocessor crystal at room temperature. In my first controller I unsuccessfully tried to vary the crystal frequency by using a variable capacitor in the crystal line. I've since decided that changing the delay period in software is a better approach.