You can't really fix this - I have the same clock and the problem is that Sony was too cheap to use a proper IC that tracks the time via the AC mains frequency (which power companies go to great pains to ensure averages exactly 50 or 60Hz over time). Here Sony is using a generic 8-bit microcontroller and running the clock off the base time-of-day counter w/out an RTC and w/ the lithium battery as a backup power source. You can find the datasheet for this by searching the sister part number LC87FBG08A (with a different footprint), for example the datasheet is available via Mouser: https://www.mouser.com/ProductDetail/ons... There is an onboard crystal oscillator (part S3X1, X2 on the PCB schematic) supplying the 32.768K oscillations. The best accuracy that can theoretically be achieved is ±20ppm at room temperature (~10.34 minutes gained or lost/year), but with the temperature swings on the board it's going to be worse than that. An option which I haven't tried is to replace the XTAL with a DS3232 TCXO....