Posts: 1,162
Threads: 26
Joined: Aug 2005
The 98035 is one strange design!. There are really 2 clocks in there. One is done in software on the 'nanocontroller' (the 1820-1691 IC), and is used while the machine is powered up. This program also controls the outputs, etc.
There is a separate battery-backed clock used when the machine is powered down. This is in the 24 pin IC on one of the PCBs (not the ROM chip, OK ;-)). THis chip seems to be a digital watch IC, repackaged. The outputs are the drive lines for a 3 digit 7 segment display (with annuciators, that's all you need for a 12 hour display and US date format). The control inputs are essentially the buttons you'd find on an LED watch.
Softwre on the nanocontroller (again) reads/decodes the 7 segment data (I am sure this what the outputs are, I've driven a display from the chip in one of my 98035s...) and simulates 'pressing the buttons' to set/read the clock.
Alas, since it's a 3 digit display, the crystal is not the normal digital watch crystal -- there's a divide-by -3 in the counter chain (the normal watch crystal is 32768 Hz -- divide by 2^15 to get 1Hz).
FWIW, the oscillator testpoint mentioned in the manual is one of the display scan lines (the fact that it's a multiple of 3 Hz, and the fact it's a 2:1 mark-space ratio, is what put me onto the fact tht it's a 3 digit multiplexed display output...)