> On this latter point, let's say that a clock driven by the mains is > correct to within 10 seconds/day (about 1e-4), but in the longer term > accuracy is improved. To reach even 1ppm level, you need to accumulate > time for 100 days. What chance is there of accumulating for 100 days > without a power failure?
There are other issues besides power. Will your OS stay up that long? How many other things are you running on that box? Will you get impatient with the software? Will you need to rearrange that corner of your lab? I've got one box that's been up for 50 days since my last power failure. I'm pretty sure I could keep it up for a long time as long as I didn't want to do much else on it. That last power problem was out for several hours. Most of my problems are minor glitches. A UPS could easily ride over them. I think the interesting problem area would be filtering the input signal. Mechanical clocks have a flywheel. We'd need an analog filter on the front end and/or a software filter. What's the Q of a typical mains powered clock? What is the Q of a (small) motor/generator set with a big flywheel? Does anybody have any scope traces of a typical power glitch? How many cycles get trashed? What's the typical duration? Does the phase of the before-glitch signal match well with the phase of the after-glitch signal? -- These are my opinions, not necessarily my employer's. I hate spam. _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.