Timing accuracy of the Apollo Guidance Computer?

Question

The AGC timing reference came from a 2.048 MHz crystal clock. But how accurate was this clock circuit? Note 3 of this schematic: "The value of the follwing components to be determined at unit test" is an indication for trimming the frequency by selecting the value of C2.
The thermistors RT1, RT2 and RT3 look like a temperature compensated DC level. The temperature coefficient of the XTAL might be compensated by the reverse coefficent of selected capacitors.

But how accurate was the derived real time clock of 100 Hz? An error of 10 ppm would be 0.864 seconds a day.

An accurate real time clock was not only important for the timing of course corrections but also for astro navigation using the sextant and telescope.

Answer 1

The Apollo 8 onboard clock's measured drift was $0.953\ \text{ms/hr}\ (0.26\ \text{ppm})$, while the specified limit is $7.2\ \text{ms/hr}\ (2\ \text{ppm})$. They also mention that there were "periodic clock updates from the ground" indicating that they accounted for the drift and over the mission time of 143 hours.

Text transcription:

    Computer.- Computer operation was nominal throughout the mission. A computer clock drift of 0.953 msec/hr was measured prior to and during the flight. A drift of this magnitude is well within the specification limit of 7.2 msec/hr and was accounted for by periodic clock updates from the ground. One restart occurred during the mission when the crew attempted an illegal exit from the landmark tracking program (program 22). The insertion of a verb 34, which requests termination of a function when the program is requesting an optics mark, causes random transfers which, in this case, resulted in a parity fail when an unused memory location was read. Table 6.9-VI is a list of the computer programs used during the flight.

Source: Page 101 of the Apollo 8 Mission report