I have a RTC ds1307 mounted on a prototype board ( the one with 3 pin stripped ), and I'm experiencing a delay of about 5 sec/day, far too big compared to some research I did. Can that delay be caused from extra capacitance added by the prototyping board or is just the crystal? This is the crystal I'm using.
Well, I solved the drift. After checking the crystal capacity and seeing it is correct as the frequency, I decided to power the Arduino power from an external power supply, my DS1307 uses the +5v from the Arduino board. With the new power supply the clock is accurate ( I can't see a drift of second in 24 h ). I don't know if it was a problem in having a barely enough voltage with the USB or if that supply was "dirty", but in any case the single cause was the power supply.