The only controlled experiment I am aware of was run by Mark Sims in March 2009 and he measured a 100% difference in PPS change per hour between the old chip and the new chip in holdover in the same Thunderbolt. Big enough to be statistically significant.
febo.com/pipermail/time-nuts/2009-March/036831.html Didier On Mon, Dec 9, 2019, 1:45 AM Charles Steinmetz <[email protected]> wrote: > Didier and Bob both suggested that perhaps the Tbolt's Kalman filter is > better served by the "other corner of the board" temperature data > supplied by the existing DS1620, and was designed to operate that way. > > If this were true, we would expect Tbolts with newer DS1620 chips > (chunky 1C temp data) or defective DS1620s (e.g., constant minimum > temperature reporting, such as -55C) to perform *VERY* differently in > holdover than Tbolts with properly operating, mC resolution DS1620s. > But all indications are that there are no such differences (and there > are reams and reams of historical data on this). > > So, I remain convinced that the Tbolt's Kalman filter operates as I > described in my last message (12/4). > > Best regards, > > Charles > > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
