Another item of interest might be: http://www.youtube.com/watch?v=3DLEswweMq_tY Which is a very short video where I compare a Rubidium standard to a Fluke 6060A signal generator with the high-accuracy reference oscillator option. I bought that Fluke generator on eBay used in good to fair condition and no doubt it had not been calibrated in years, and yet it was only 30 ppb different from the rubidium standard. I have also compared this same Fluke generator to WWV by listening to the beat note of WWV on 10MHz with the signal from the generator set to 10MHz, on a shortwave receiver. Random atmospheric phase shift variations prevented me from getting an exact difference in frequency but it was less than 1/2 Hz different which is in the same ballpark as the result from the comparison with the Rb standard. Not bad for an ovenized crystal oscillator that is more than 20 years old and probably not adjusted in the last 10 years! Sean On Fri, Jan 4, 2013 at 2:17 AM, Sean Breheny wrote: > As a crystal ages and undergoes temperature variation, it will have a > component of drift which is continuous in one direction as well as a > random walk component which still ends up going in one direction in > the long term but on time scales of days or weeks can start heading > backwards temporarily. Cheap crystals will have a total error of about > 100ppm. Good crystals would be more like 20ppm. If you temperature > compensate them and calibrate them to begin with, then can be about 1 > ppm. If you ovenize them so that they do not see temperature > variations (which cause part of the random walk phenomenon), then you > can get to about 200 ppb. Pre-ageing for hundreds of hours can get you > even further and finally you can do some modeling of the ageing of > your particular crystal and get to something in the 10 ppb ballpark I > think. > > See: http://www.prc68.com/I/Xtal.shtml > > Even back at the 100ppm figure, that is 0.01%, which is 4 minutes in > one month. 10 ppb would take 3 years to accumulate one second of > error. > > Sean > > > On Fri, Jan 4, 2013 at 1:13 AM, William "Chops" Westfield > wrote: >> >>>> I see opinions that solar tracking doesn't need that much accuracy >> >> yes, but the usual behavior you get with a simple microcontroller crysta= l-based timer is a continuous drift in one direction. If I did my math rig= ht, a 1% inaccuracy in the timer makes your time be off by over 7 hours aft= er a month of operation, which is quite bad for solar tracking purposes=85 >> >> BillW >> >> >> -- >> http://www.piclist.com PIC/SX FAQ & list archive >> View/change your membership options at >> http://mailman.mit.edu/mailman/listinfo/piclist --=20 http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .