[time-nuts] Lady Heather's latest bells and whistles
holrum at hotmail.com
Wed Sep 14 16:56:16 EDT 2016
While on the subject of how GPS receivers do or don't report the day that a leapsecond will occur, I fired up a Nortel/Trimble NTWB receiver to see what it said. It has the standard Trimble packet (0x58:0x05) from which one can calculate the day, but Lady Heather was reporting "Leap Pending!", instead of Leap 108 days... That meant that it has a leap second pending flag, but the leap day calculation was over 183 days away. Hmmm, that shouldn't be. Digging deeper, it appears that it is reporting the wn_lsf (GPS week number of the leap) as week 2953 (sometime in 2036). It appears that they are adding the week rollover correction to this field twice.
Switching the receiver to SCPI mode (yes they can speak a dialect of SCPI) the time of the leap is reported correctly. The error appears to be in their TSIP packet output routine.
For receivers that report a leap second pending flag, but don't report the day of the leapsecond, I just added a feature (that is sure to give TVB apoplectic fits). I ASSume the day of leapsecond as being June 30 or Dec 31 and show the number of days until leap based upon that (but include a '?' after the value). Yes, we all know a leapsecond can be added at the end of any month, but these seem to be the days preferred by BIPM lately (March and September being the second most likely months). We'll see how long that ASSumption holds up...
More information about the time-nuts