Did some more testing this morning on 3.05b2. It does indeed appear to remember the corrections after a power cycle, so whatever "learning" the MC memory gained, it is retained between battery changes. Hopefully the weather will permit me to fly again later today or tomorrow and we'll see if this holds for a longer power-off period. Apparently, the corrections are stored in non-volatile memory. Hurray!
The more flight time I get, the better the correction appears. The aircraft is flying almost perfectly straight now. I did notice that making runs along different directions of the compass rose seems to show some deviation, but it damps out quickly and does not reappear on subsequent runs, so the learning seems to be sensitive to the initial magnetic heading taken. Makes sense.
Judging by the consistency of the reports from the enlarged beta test group, I believe that DJI should now feel confident in officially releasing the new firmware. I hope they do so soon so that all Phantom owners in high declination areas can enjoy the benefits of the new software.
The more flight time I get, the better the correction appears. The aircraft is flying almost perfectly straight now. I did notice that making runs along different directions of the compass rose seems to show some deviation, but it damps out quickly and does not reappear on subsequent runs, so the learning seems to be sensitive to the initial magnetic heading taken. Makes sense.
Judging by the consistency of the reports from the enlarged beta test group, I believe that DJI should now feel confident in officially releasing the new firmware. I hope they do so soon so that all Phantom owners in high declination areas can enjoy the benefits of the new software.