knudsen@ihnss.UUCP (11/02/83)
I'm amazed at how rapidly the time-of-day clock in my Coco OS-9 loses time. I'd say it loses a minute per hour, or one part in 60!! No, I wasn't running any interrupt stuff or even hitting the disk much -- mostly just perusing the manual and trying out commands. Can the 3.58 MC colorburst crystal really be that far off and still work on a color TV? More important, if the timing is that screwy, how much can we believe the disk RPM program so nicely supplied by the Shack? One part in 60 equals plus/minus 5 RPM, slightly greater than the allowable speed range. What am I overlooking? mike k
sdyer@bbncca.ARPA (Steve Dyer) (11/02/83)
I found that my OS-9 loses time, too. The key here is NOT that the clock is wrong, but that many CoCo OS-9 operations (like disk I/O) require interrupts to be turned off. Naturally, when this happens, the system 'loses time', since its conception of time-of-day is driven off of the 60hz clock interrupt. Every 60 clock ticks, the time is incremented by one second. My own feeling is that the line clock is suitable for scheduling, but a true time-of-day clock card is needed if you are looking for accuracy. What I would love to see is a multi-function board containing 2 serial ports, one parallel port, and such a clock. /Steve Dyer decvax!bbncca!sdyer