[teqc] Ashtech Z12 spurious rollover bug

Lou Estey lou at unavco.org
Wed Nov 9 08:20:37 MST 2016


On 09-Nov-16 07:34 AM, François Meyer wrote:
> Data files follow soon in a separate mail.

Well, this is interesting.  François made two Ashtech R-files available to me,
from Sunday and Monday this week, originally named RBSNCA97.082 and RBSNCA97.083
by the Z-12 receiver.  François renamed them RBSNCA16.311 and RBSNCA16.312.

[2598] teqc +quiet +doy +mds RBSNCA16.31?
2016:310 03:30:30  2016:311 03:30:00   1215338  RBSNCA16.311
1997:082 03:30:30  1997:083 03:30:00   1244650  RBSNCA16.312

These same results were obtained using both the 2016Nov7 version and the 2016APr1
version of teqc.

Now I can do something else: for many formats (esp. the older ones), I can
recompile teqc with one or more preprocessor flags set to see some more details.
Doing this and then rerunning:

[2600] teqc +quiet +doy +mds RBSNCA16.31?
Ashtech_R_session_headers(): GPS week set to 1921
2016:310 03:30:30  2016:311 03:30:00   1215338  RBSNCA16.311
Ashtech_R_session_headers(): GPS week set to 898
1997:082 03:30:30  1997:083 03:30:00   1244650  RBSNCA16.312

An Ashtech R-file has a small binary header where one field is a value for
what is supposed to be the GPS week for the data.  In the above output we are
seeing exactly what that value is.  So it's definitely correct for the first
file which starts on Saturday, Nov 5, GPS week = 1921 and off by 1024 for
the next file which starts on Sunday, Nov 6, GPS week = 1922.

Using the '-week' option with the correct week does work, as reported by
François:

[2603] teqc -week 1921 +quiet +doy +mds RBSNCA16.311
2016:310 03:30:30  2016:311 03:30:00   1215338  RBSNCA16.311

[2605] teqc -week 1922 +quiet +doy +mds RBSNCA16.312
2016:311 03:30:30  2016:312 03:30:00   1244650  RBSNCA16.312

The stored GPS ephemeris messages in RBSNCA16.311 all have the correct GPS
week value, even those that are for Nov 6, i.e. they have week 1922.

The stored GPS ephemeris messages in RBSNCA16.312 mostly have the wrong
GPS week value, 898, except for one message which is a back-tick for PRN 13
to Nov 5:
13 16 11  5 23 59 44.0
and this one message has the correct GPS week of 1921.  If you use '-week 1922',
then the GPS week for those messages with 898 are corrected to 1922 and the one
back-tick message is deleted.

cheers,
--lou


More information about the teqc mailing list