[Space] Fwd: KML from G1 data

Erik Ebert eebert at gmail.com
Fri Jun 11 20:49:41 UTC 2010

More information from Ken, our club's GPS guru.

I suspect that the weird stairstep ascent and descent you can see in
the Google Earth plot of the data is caused by the filtering issue
that Ken describes for "high altitude and/or high dynamic flights".
That's a typical symptom.

My guess is that what is going on is when the GPS is in 'ground mode'
and it sees a large change in altitude it assumes it must be a
multipath artifact and tries to smooth it out.  But when the altitude
change exceeds some threshold or goes on for long enough it decides
the altitude change must have been real and it resets the baseline to
the new value.

  -- Erik

---------- Forwarded message ----------
From: Ken Biba <ken at bibafamily.com>
Date: Thu, Jun 10, 2010 at 10:09 AM
Subject: Re: [Space] KML from G1 data
To: Erik Ebert <eebert at gmail.com>

Love to help.

There are really two problems with GPS chipsets:

1.  Some have a legacy limitation on speed and altitude - and I
suspect the Garmin had that problem.   There are legacy restrictions
on GPS from the old days (now expired) so the these devices would not
work to build missle guidance systems.    Now, since most of these
devices are built in China anyway - that is a pretty silly
restriction.   And further, they were often incompetently implemented
... the restrictions were basically preventing a device from reporting
faster than the speed of sound AND above 60K'.     Sadly .. some folks
implemented OR rather than AND.   Confusion reigns.

2.  A second problem, and somewhat distinct - is the filtering used in
the GPS microprocessor.    All of these devices use some kind of
filtering to improve their performance in changing environments.
Most are optimized for the GPS reception environment in urban canyons
on the ground (car and foot navigation) and make assumptions when
filtering consistent with that model.  Some - have multiple filters
supporting different movement models - the uBlox and the Trimble are
in the latter camp.

I think Garmin no longer makes its own chipsets and uses SiRFStar
chipsets ... which would explain all of the above behaviour.

The only chipsets that really should be used on high altitude and/or
high dynamic flights are uBlox or Trimble Lassen.     Trimble is
chipset used in Big Red Bee GPS.   Both need to be correctly
programmed (in non NMEA mode) to use the right movement filter.
Actually - that was the problem I think you had - BigRedBee had a
manufacturing glitch and forgot to properly program the GPS chip.

Parrot is good ... so is GPSFlight ... frankly for their application,
I have 1W GPSFlight devices left over from the 100K project that
should work well.


More information about the Space mailing list