[time-nuts] GPS week rollover

[Steve Olney]

Hi Jeff,

On 6/04/2019 3:30 pm, Jeff Zambory wrote:
> Yes Steve, you are correct.
>
> https://www.gps.gov/
>
> Scroll down a bit and you can find a count down to the roll over. And it states the time when it will roll over. Just what you have said.

Thanks !!!

I'll watch that and watch my receivers.

It seems, from posts here, that different receivers exhibit different 
WNRO behaviour - with effects noted *before* the RO epoch.

I am curious about this and set my Garmin 16x to output as per a 
snapshot below...

$PGRMF,1023,590391,060419,195933,18

...which is GPS week, GPS seconds, UTC date, UTC time and leap second count

As GPS seconds wraps around at 604799 (and GPS is incremented at that 
moment) then there is 604799 + 1 - 590391 = 14409 seconds to go from 
19:59:33.

14408 seconds = 4 h 9 seconds.   19:59:33 + 4:00:09 = 23:59:42.

This agrees with the www.gps.gov data.

So why are people reporting effects already ? My *guess* as to why some 
receivers have reacted before the RO epoch is that attempts to address 
the issue have been different depending on the vintage of the GPS receivers.

  * Receivers manufactured just a few years before the first rollover
    data would need to cross the 1023 - 0 week boundary a short-ish time
    after manufacture.   So some kind of offset would need to be added
    to the week number.  These units which have not had firmware updates
    would presumably react early to the second rollover depending on the
    date of manufacture.
  * Receivers manufactured after the first rollover would presumably
    react late to the second RO by the same logic.
  * Receivers of any manufacture date which has had its firmware updated
    regularly would presumably not be affected as the updates would keep
    pushing the internal wrap around epoch further into the future.
  * Old receivers (like my GPS 35 with no updates) already have gone
    through one rollover (currently showing 21 August 1999) and will
    presumably go back to 1980 dates with this rollover.

So - my bottom-line note to myself about this is that just because a 
receiver looks fine tomorrow doesn't mean it won't fall over next week, 
or next month, or next year, etc.

Cheers

Steve

P.S. for some reason time-nut posts arrive in my inbox uncorrelated to 
their time stamps.  So I find myself replying to later posts before 
earlier ones.  Gets a bit confusing for this Senior Citizen... LOL...