Cables are always an issue. In this application I need four wires: GND, VDD (+5V), Data A and Data B. I thought standard USB cables would be interesting because they are easy to get and very cheap. My prototype has a connector with two USB sockets so the bus can be passed through to the next controller.

In this application I chose to put the configuration of the signal controller into hardware. Resistors are used to set analog inputs of the microcontroller to defined voltages with 8 possible levels for each input. That way no download and storage into EEPROM of the address of the individual controller is necessary. The controller reads theanalog inputs at startup to learn what signal type it is controlling: Vorsignal Blocksignal Einfahrsignal Ausfahrsignal Blocksignal mit Vorsignal Einfahrsignal mit Vorsignal Ausfahrsignal mit Vorsignal Gleisssperrsignal Sorry I do not know the English terms! It is about different types of signals! Two more analog inputs are used to define the address of the controller (64 address is the limit).

@Lucky:  I think the GPS module is not performing a conversion to UTC from GPS time. This requires knowledge of the number of leap seconds since GPS time started (and other minor correction terms). The relevant GPS to UTC conversion parameters are transmitted periodically (every 12.5 minutes) by GPS satellites, so if we can get the datasheet for the GPS receiver, we might be able to receive the exact correction factor, or even get the receiver to perform the correction. I used a u-Blox 7  receiver on a clock  I made myself and the time is correct, as this receiver takes care of it.   I didn't find the detail protocol for the module used in the clock kit though...... My clock design was inspired by another Elektor article about using LED segments out of light bulbs.  I used a PIC24HJ128GP502 processor, which has on-chip RTCC, and a chain of several MIC5482 serial-to-parallel high voltage drivers.  You can see it from a long way off if you turn the PWM dimming off! Code in C using XC16 compiler.  

@F1Andy: thanks for the code! With all due respect, I'd rather find the source of this issue but I really don't have a clue what could go wrong in this clock design. I agree with Paul: I wouldn't be surprised if it was a bit late, but ahead....

May be my old Artob Projekt from the mid 90 's is interesting also. German : http://buene.muenster.de/darc/artob/homebal.htm English : http://buene.muenster.de/darc/artob/homebale.htm Regards Armin