[time-nuts] GPS jumps of -13.7 us?
martin.burnicki at burnicki.net
Tue Jan 26 17:46:43 EST 2016
Paul Boven wrote:
> Hi everyone,
> Has anyone else seen GPS time jump by -13.7 usec today?
> I just heard from several geographically quite distributed radio
> observatories that they have seen their GPS receiver(s) jump compared to
> their in-house standards.
We were able to track this down today at Meinberg.
The problem was that some satellites were sending invalid UTC correction
parameters. The UTC correction parameter set not only contains current
leap second information but also coefficients (A0, A1, WNt, tot) for a
polynomial used to compensate the fractional difference between GPS time
Normally A0 (the constant offset) is very small, close to 0. WNt and tot
give the week number (truncated to 8 bits) and second-of-week for which
A0 is valid, i.e. the reference time for the time correction parameters.
WNt is currently about 89.
Today the faulty satellites sent about 13.7 microseconds for A0, and WNt
as well as tot were both 0. So when the GPS receiver updated its UTC
correction parameters from a faulty satellite the UTC correction jumped
from close to 0 to about 13.7 microseconds, which let the UTC time step,
and when the GPS receiver received the UTC parameter set from a healthy
satellite the UTC time stepped back.
We have recorded a few of the faulty subframe words. If someone is
interested I can provide more detailed information. However, I'm
currently out of the office and don't have the information here right now.
More information about the time-nuts