> Given that not only do these exact frequencies change between systems due to margins of error on clocks, they also change slightly as the system runs (and get

> Given that not only do these exact frequencies change between systems due to margins of error on clocks, they also change slightly as the system runs (and gets warmer, for example.) Maybe it'd be possible to modify some boards to use a CPLD (or an FPGA) for synthesizing those clocks from a common source. This could eliminate most (all?) uncertainty.