Looking at the datasheet, there is quite a large error across the voltage/temp range for the 32kHz crystal oscillator. But I'm more interested in the error at a fixed voltage and fixed temperature. In this case, would the error approach that of the crystal tolerance spec? We are using MSP's internal loading caps, but they are not specified with a min/max (only typ) in the datasheet, so will they have their own process variation that would increase the error? If so, any idea how much? We are using a 20ppm tolerance crystal, trying to figure out if our overall error would be near that, or 40ppm, or 100ppm, or ???