[time-nuts] Warped back to 1993

Martin Burnicki martin.burnicki at burnicki.net
Sun Aug 11 11:11:02 UTC 2013


bg wrote:
> As has been discussed here before.
>
> The GPS firmware programmer finalizing his stuff at GPS week X
> (approaching 1024) will make sure his receiver will survive until X
> weeks into the next era. What X is is hard for the end user to know.
> 1024 weeks is long time beyond normal warranties.

Right, this is one possible reason.

> I have seen era predictions based on # of leap seconds. But that
> particular implementation did not fare well with long leapsecondless
> periods we had.

I remember a discussion about this way to guess the epoch. Depending on 
the implementation and the intervals between leap seconds about 15 years 
ago, and looking at the large intervals between the last leap seconds, 
this may also fail.

A much simpler way to handle this is to provide a way where the user can 
enter the current date, and the firmware determines the correct epoch 
and week number.

This also had to be done only once if a receiver was powered up from 
scratch. If the receiver knows it has already been operating at a date 
after one or more rollovers it can easily get the right epoch.

Martin




More information about the Time-nuts_lists.febo.com mailing list