Re: RINEX update

[Gary E. Miller]

Yo John!

On Wed, 29 Apr 2020 11:28:44 -0400
John Ackermann N8UR <address@hidden> wrote:

> Last night I saved a couple of hours of raw observations from my M8P
> using gpspipe.  I thought using the single-frequency receiver would
> reduce the number of variables.

Yes, good tests.  But we only need a few minutes of data to see the
differences.

> I sent both .obs files to NRCan.  It processed the version from
> convbin, and rejected the version from gpsrinex.

A good A/B baseline.  Any obvious differences between the two files?

> If the list accepts files this big, attached are a zip file containing
> the raw .ubx log and the two RINEX files, and another zip with the
> NRCan results for the convbin file.  If they don't come through, I'll
> share them via Dropbox.

I see the obs and ubx.  I can look at that.

> I tried comparing the two RINEX files by eye, but the differences
> between version 2.11 used by convbin and 3.03 used by gpsd are such
> that it's hard for me to spot much except:

Yup, welcome to my world.  Sort of like reading the matrix raw.  The
data is the same, just shuffled around a bit.

> 1.  The convbin file includes only C1 and L1 data (not D1), and only
> specifies those two fields in the header, while the gpsd file
> specifies six fields in the header and includes C1 L1 and D1 in the
> data with the other fields left blank (which I believe is OK).

AFAIK, NRCan only uses the C1 and L1 data, not the D1, signal strengh and
LLI that gpsrinex includes.

Easy to make gpsdrinex not include those fields for testing.

> 2.  The convbin file starts at 23 22 29.9940000, while the gpsd file
> starts at 23 25 00.9940000 -- 2 1/2 minutes later.

Yes, gpsrinex throws away data that it knows PPP can not use.

> 3.  Both files wrote at 30 second intervals, but the convbin sequence
> is at 29 and 59 seconds, so apparently the timetags do *not* have to
> be at the 00 and 30 integer second points, at least for NRCan.

I disagree.  I think your data may show the opposite.

The convbin 29.9940000 is pretty much 30.
The gpsrines 00.9940000 is pretty much 31.

Is the PPP needs 30, then gpsrinex is not providing that.

I will look into that.

> 3.  Looking at contemporaneous data stanzas, the observations do not
> seem to line up at all.

Nor would I eexpect them to.  They are for different timestamps.

See #3 above.

> Although the SVs seem to be listed in
> different order,

gpsrinex sorts them to make reading the file easier.

> 4.  I haven't fully figured out the rules for the LLI bit, but where
> shown it doesn't seem to match up between the two files.

And, if you read the u-blox doc, setting the LLI is mostly guessing.

Not to worry, most PPP ignore it.  The few that use it do so to throw
away the dodgy samples.  When carrier tracking slips, that means the
receiver is slightly confused, its kalman filters abused, so those are
good measurements to ignore, if others are good.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        address@hidden  Tel:+1 541 382 8588

            Veritas liberabit vos. -- Quid est veritas?
    "If you can't measure it, you can't improve it." - Lord Kelvin

pgpt9wT2Tmqku.pgp
Description: OpenPGP digital signature