Garmin 60cx (SSIII) GPGSA message messed up???

Hi gang,

As usual, our group was doing some "nonstandard" testing of GPS
receivers, and we realized that it appeared our Garmin 60cx (newer model
presumably with the SiRFstarIII chipset) has a broken GPGSA message! We had
the receiver hooked up to a constellation simulator, and we selectively
turned off individual satellite signals. The bars on the screen behaved
correctly, and the GPGSV message showed the C/N0 values drop down to zero,
but the GPGSA message said some of the SV's were still being used! Then we
realized the PDOP/HDOP numbers in the message were fantasy as well. Best we
can tell, someone at (Garmin?) who did the parsing of the (presuming) SiRF
binary stream to construct the NMEA got their table indexing wrong...if we
dropped a satellite, some random one of the GPGSA PRNs would drop. Likewise
when we added it back in...sometimes it came back as the same wrong PRN,
sometimes a completely different one. Always one in view, so I surmise that
they're misaligning the "tracking/acquisition channel ID to PRN mapping" to
the "channels being used" list.

We ran a US Globalsat BU-353 (also a SSIII) in parallel, and it had the
correct output. So, I'm thinking it's Garmin's fault. The curious thing is
the bars on the display are correct, though if it marked "solid" a bar with
zero signal it would be hard to tell. Since the chip can still (claim to)
use a signal with C/N0's approaching 17 or so, it's hard to make it just
weak enough to not use.

Has anyone else seen something like this, or do people rarely closely
monitor the GPGSA output? We might have ancient firmware; a call to Garmin
may be going out shortly.

Fun fun fun in the lab...

Dr. Marty
(with much help from his assistant, "Crusher")

Garmin having mistakes in their NMEA output isn't that rare. I've seen
it in some models/firmware versions going back to at least the G45.

It can be a simple mistake like with the G168 sounder which always
reported zero satellites in use in the GGA sentence - later corrected
with a firmware update - to more subtle disagreements between
sentences reporting the (supposed) same value.

Marty Ryba wrote: