Possible cause of GNSS signal fluctuation

[Jetrell]

I have been looking into finding the cause of this EPH spike for a while.

It is noticeable in the log of every model I own using INAV firmware. Both planes and copters, with a range of GNSS modules. M8, M9 and M10's.
However there are flights when it doesn't appear on some of those models. But it mostly does.

Yesterday I ran some ground tests by placing 2 modules and connected flight controllers at the same location. Both connected to different PC's to monitor the GPS status tab.. One was a Matek M10Q and the other a MicroAir MG-903 [M9]. There was no other peripheral hardware connected to these FC's.

I ran them through multiple changes in settings and INAV firmware changes between 6.0, 7.0, 8.0 and 9.0.
I didn't make any alteration with U-center.
These setting changes included -

• Always using GPS and Galileo. But also using Glonass and Beidou when the firmware allowed for it.
• Adjusting gps_ublox_nav_hz
• Enable/disable gps_auto_config
• Enable/disable gps_auto_baud
• Altering gps_auto_baud_max_supported
• Altering port baud rate

The main things that came from this test were.

• INAV 6.0 appeared to provide more stable position accuracy.

• While using gps_auto_baud in INAV 8.0 and 9.0. Based on the port baud rate. If it was disabled when the FC was still powered. The signal and Sat count improved noticeably. But if the FC was power cycled, communication between the two could no long be established.

• The precise frequency gps_ublox_nav_hz was set too. Made a large difference to the satellite fix, and therefor the HDOP and EPH accuracy, with both modules when using INAV 8.0 and 9.0.
  When using 4 constellations for both modules. And having gps_ublox_nav_hz set to the 10Hz default.
  The M9 would only see an average between 15 to 18 Sats.
  And the M10 would only see an average between 13 to 17 Sats. With both experiencing wild fluctuations in HDOP and EPH.
  But when I found the sweet spot of 6Hz and below for the M10. And 9Hz and below for the M9. The signal improved considerably.
  The M9 seen the average Sat count increase by 8 to 10, and the HDOP drop from a wild ranging 2 to 5, down to a stable 1.3.
  And the M10 also experienced similar results, but not quite as distinctive.
  This was also tested by turning off one constellation at a time. Either Beidou or Glonass to check the result. And the stability remained constant. With just a little lower fix count and EPH position accuracy.

However in INAV 7.0. The gps_ublox_nav_hz would default to the lower 5Hz, even when set to 10Hz.
While INAV 6.0 didn't have gps_ublox_nav_hz. It just set the default based on UBLOX or UBLOX7. either 5 or 10Hz. But also only supported GPS and Galileo. Which likely made the update more stable at 10Hz.