AWOL:
Try this
A very interesting read. I took a look at the Shortt clock and am rather impressed. I did like his code an may barrow it. All I want from the GPS runs is cycles per PPS signal, temperature of the crystal, and number of satellites. I will throw out any 3 satellite fixes.
Mouser has a 16hz crystal with 10 ppm variance for less than a dollar. I think with temperature compensation and averaging that I could get to 1ppm.
A lot of the proof will be in the pudding. Maybe I will make an UNO clock and see how well it does over a year. (I got a lot of UNOs when radio shack went out of business).