I have been investigating 2D fixes in GPS systems and have recently tested a SkyDrop (from SkyBean) and an Oudie 4 (from Naviter). While these instruments use different GPS modules (L80 and PA6H), both of the modules use the MT333x chipset.
The test shows that position reporting is highly inaccurate in situations where the number of satellites in use is low.
Both tests were conducted while driving and deviations from the road are quite evident, even though the NMEA log shows 3D fixes. In the last minute or so the instruments were allowed to acquire additional satellites and the location reported becomes quite accurate.
Both instruments also contain a pressor sensor for measuring barometric height, so it can be seen that the inaccurate position reporting is associated with errors in the height reported. It should be noted that the only time a 2D position is reported when 3 satellites are used in the fix.
As you should be aware, a 3D fix requires at least 4 satellites. Similarly, a 2D fix requires at least 3 satellites. I must stress that a 2D fix may involve more than 3 satellites - I believe the errors being produced are the result of the MT333x reporting 2D fixes as 3D fixes, ie the errors are actually 2D fixes involving more than 3 satellites which have been reported as 3D fixes. This is consistent with the nature of a 2D fix where the height (predicted, but not actually known) is used in calculating the 2D Lat/Long. That height is included in the fix, and an error in the height results in an error in the location. Since the height used in a 2D fix is assumed to be accurate, you can get large errors with low DOP when the height is actually incorrect.
The NMEA logs are available at www.sacra.bix/2dfixes/DEBUG22092018skydrop.txt and www.sacra.bix/2dfixes/NMEAIN02092018oudie.txt
Both instruments also produce IGC files. These are available at www.sacra.bix/2dfixes/igc22092108skydrop.IGC and www.sacra.bix/2dfixes/igc22092108skydrop.IGC
You can use SeeYou (from Naviter) to view these files and it will clearly show the deviation from the road. There are also options that will allow you to compare the heights recorded (IGC files record both barometric and GPS height).
There is also a Garmin NMEA log available which shows 2D fixes using up to 5 satellites at www.sacra.bix/CAPTURE260718garmin.TXT
I would also draw your attention to the PA6H datasheet which states
"Table-6: Mode 2
1 Fix not available
2 2D (＜4 SVs used)
3 3D (≧4 SVs used)"
This appears to be an accurate representation for how the MT333x determines a 2D or 3D fix, however, as I have indicated, 2D fixes may actually involve more than 3 SVs.
Fero from SkyBean can be contacted at: email@example.com
Andrej from Naviter can be contacted at: firstname.lastname@example.org