Please forgive what is probably a naive question from this newbie, but a loop filter has considerable delay (half the number of samples for a typical(?) symmetrical FIR or block-averaging filter). This must surely limit the usefulness of very long averaging times? And perhaps it could exacerbate day/night changes by effectively applying a "correction" of the wrong phase if averaging for a whole day was attempted. Presumably you need a "crystal ball" subroutine to try and predict what the trend is going to be in the future in order to get your correction correct for "now"?
Peter _______________________________________________ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts