I would consider 4 things
1) reduce the 1M resistors at the unity gain op-amps, the leak on a humidified PCB can affect resistors of this size.
2) a small capacitor across the feedback resistors for stability
3) replacing D1 with a reversed diaode to Vcc
4) using a 11.0592MHz xtal to hit the bauyd rate on the nose.

measuring AC
as far as I can see you are not actually measuring the AC unless P2.7 is a PWM generated 'DC'.

Erik