NXP manufacture a 'chip' , (PCF2127) the package being embedded with a watch type quarts crystal resonator which claims typical time keeping accuracy of +/- 3ppm (I.e. 6 parts per million) quoted over a temperature range of -20degC to +70degC. Placing in a stabilised temperature 'oven' would yield substantially better accuracy. The crystal resonator is either 'in lock' at the fundamental if set up properly, or it is not...and drifts continuously as has been said occurs sometimes with a 'common hobbyists application'. where the potential accuracy / stability of the fundamental is not realised. The worse the layout and component selection, the futher the resonance is pulled away from the crystal fundamental, which is why it may be a good idea to get a factory calibrated and trimmed device if such accuracy is required. On Fri, Jan 4, 2013 at 5:21 AM, RussellMc wrote: > > I'm not saying you can't implement an accurate clock with a crystal, I'= m > just saying that it frequently doesn't seem to happen. I'm pretty sure > I've seen a fair number of crystal-controller micros that have quite > noticeable drift when used as clocks (which is, after all, a pretty commo= n > hobbyists application=85) > > If you "oven" a 32 kHz watch type crystal with an adjacent tightly > coupled 38.x degrees C wrist, and then trim it manually for notional > zero drift, you can get a good wristwatch accurate to about 1 second > per week*, in my experience and AFAIR. That's about 1.5 ppm over 1 > week. ie long ago I monitored my wrist watch's drift and at a battery > change had the tech adjust the trimmer by the amount I suggested and > thereafter it was essentially 'spot on'. I wore it on the inside of my > wrist (note to self: ensure non radioactive dial) and never took it > off ("100 metre" waterproof essential). I think that was a metal cased > Pulsar - which is a Seiko made watch in a case cheap enough that > Seiko don't want to let you know that they make it. > > * I originally wrote "one second per month" but I'm not 100% sure that > was correct. (It may have been 30+ years ago). I know it was very > accurate indeed once trimmed. The tech was surprised when I told him > how much adjustment I wanted. He had a frequency meter calibrated in > seconds per xxx error and my claimed drift matched what he measured > almost exactly. As you'd expect - but he can't have met many engineers > :-). > > > > Russell > > -- > http://www.piclist.com PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > --=20 Perry Curling-Hope Research and Development --=20 http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .