Sunlight readable PNAs sold with faulty GPS for flight

Dear all,

On the internet there are plenty of announces concerning 5 inches PNAs for
flight, expecially on Alibaba.
People should be warned that these units will not currently report a correct
position while turning in the air with an aircraft.
It is a much worst problem than the HP31x, because the error can be even in
the range of 300m, 100m altitude, and last for up to 30 seconds.

While this is not a problem for all pilots that are connecting such units to
an external gps (like Flarm, LX etc.), those that want to use it "as is",
with the internal gps receiver, will most likely not get a wind calculation
at all, or incorrect calculation. They will read incorrect ground speed
sometimes, even while going straight (up to 30seconds after exit
thermalling).

Paragliding and Hang Gliding pilots will most likely not suffer of this
problem, because their speed while turning is slow enough to be kept valid
by the GPS filter.

Two serious companies, one in the UK, warned their customers and offered
money back for those that wanted to change product.
The principal distributor, on Alibaba, is still selling these units without
warning about the problem. The distributor knows the problem very well of
course.

Technically, the problem is that the GPS all of a sudden does not believe
the aircraft is turning with such a bank angle and such a speed, and invents
a route going straight.
At a distance of about 280m, sometimes, the flight direction is corrected,
but not adjusted to the real position.
The NMEA stream is reporting a real valid position fix, not even with "dead
reckoning" status.

The problem seem to be related to the Kalman filter working with an
inappropriate track smoothing, or " car mode ", whatever it is, inside the
AtlasV GSD4t chipset.

There seem to be no chances to change the GSD4t (aka SirfIV) behaviour using
the native "OSP" binary protocol: there is no such protocol available in
this chip version. The OSP is available in GSD4e , not in the "t" version
which is "hosted" by the AtlasV.

The same problem appeared two years ago on the Hewlett-Packard HP31x serie
(although much more light, with an error of a few dozen meters, and for a
few seconds, not for 280m for 30 seconds).

I suspect the same problem can exist also on many other devices using Atlas
(centrality titan) chipsets.
If your PNA+software cannot calculate the wind, for example, this can be the
reason.

I am able to detect if a GPS has this problem by looking at an IGC log with
1 second interval between fixes (1 second, not 4 as most loggers do).

In straight flight, and after at least 20 seconds have passed since last
tight turn at high speed, the problem is not appearing.


That's all about it.
paolo