Mavic 3 Raptor
Second Officer
Flight distance : 3014675 ft
United States
Offline
|
Suren Posted at 3-26 18:14
That is truely the real mystery here. How come no one actually complained about long gps wait times prior to the December FW, not even me. After this update the complaints started pouring in. I even rolled back to 0300 yesterday and the lock time was 30 seconds to less with 8 sats only, what changed is a mystery.
It actually might not be as much of a mystery if the root cause posted by one of the other forum members (I think it was AerialZup2, but I'm getting old, so might be remembering that incorrectly :-) ) is correct on what has happened by DJI selecting the automotive chip.
Basically, because of all the added features that utilize the memory and GPS system that were added in Dec and Jan updates are creating a bottleneck in the memory for satellite acquisition data.
I have kept quiet about the selection of the automotive chip over the professional chip, but I will actually defend DJI on this one a bit. There is a real shortage of chips on the market...just look at the mess with the automobile industry! It sounds like the automotive version of these GPS chips were easier to 'come by' and could actually save the production line. And, I know from experience, automotive chips go through a LOT LOT LOT more robust testing in manufacturing because automotive industries literally expect ZERO defects. Now, if in the specs these two chips looked technically 'identical' as far as application use of them (and I don't know, I'm speculating) then it would make sense to switch to these chips.
However, it does appear that there is a difference in implementation for these chips and (again speculating) perhaps the 'rush to market' failed to test these enough prior to release to catch the issue.
I wouldn't think there is a hardware issue with these GPS chips themselves. Rather, it is how the logic architecture around the chip and the software (firmware) using the function are implemented.
I'm hoping that the logic architecture (this includes external memory usage, etc) is good enough that as DJI becomes more familiar with proper usage of this chip that, perhaps, firmware tweaks will improve the cold boot time of satellite acquisition...I don't know if they can store the ephemeris data in the internal 'non volatile' memory (i.e. won't erase when power is removed) or even our externa SD memory on a phone or RC Pro controller, but bottom line is they need to focus on the cold boot specifically to help relax the community. Subsequent boots are very very fast once you've already go that ephemeris data downloaded and go through the 'pain' of the first cold boot. If they can work with the chip manufacturer to really tweak their firmware, we could see even more improvements down the road.
I'm paraphrasing the details given in that post, but it is a good read if you search for it. |
|