HP48G Crystal / Timer Accuracy

[TomC]

As has been well documented here, the HP48 timekeeping is 'not too hot'.

As we know, the crystal is a 'standard' 32768Hz crystal that is multipiied up to get the calculators '4MHz' operating frequency. (I wonder if this multiplication varies with OS version? I'm currently studying a ver R and a ver L.)

Has anyone studied or actually executed a replacement of the crystal?
- The physical package is an industry standard type and quite large, so it should be easy mechanically.
- In selecting a replacement, I suspect series resistance, capacitance are critical, but I'm no expert here.
- What was the spec of the original crystal?

I will pursue finding a more accurate replacement (assuming one exists).

TomC

[Steve Simpkin]

I used to have a small scheduled program on my HP-48SX that I wrote which would automatically run once every 3 days at 2:00 am. It would change the calculator time by a small fixed amount to correct the error drift of the crystal and then turn off the calculator. The result was that overall clock drift was less than a second a month making it the most accurate timepiece I owned (before radio controlled clocks). I determined the correction factor initially through measurement and then by making small adjustments. The effect on battery life was negligible. This sure beats opening the calculator and replacing the crystal. It should be more accurate as well.

[Bill Duncan]

I think this is the time adjustment program I use. I set it up to run at 4am every morning. It's an excellent system to keep the time pretty accurate.

https://www.hpcalc.org/details/2730