Be careful with Philips I2C RTCs. I don't know how many of them suffer from this "problem," but the one I used last year (PCF85116-3) had an I2C address that was essentially the same as I2C addresses normally used by I2C EEPROM memory. I didn't even realize it until I noticed the clock was going nuts--I then realized that when I wrote to EEPROM I was effectively addressing both the EEPROM and the RTC due to the address clash.

We solved the problem by replacing it with a DS1337. The I2C address was such that there were no conflicts and it worked great. There must be some time drift but I didn't notice anything significant in the month and a half during which the project was under development.

When testing I2C make sure you truly have eliminated all potential I2C problems. Don't just disable your SET_RTC routine but disable ALL I2C routines that write to the I2C bus. If there are bugs in your other I2C routines or you have an I2C address conflict you can end up writing to your RTC when you thought you were writing to an entirely different I2C device (assuming you have multiple I2C devices on the bus).

Regards,
Craig Steiner