(First, I should note that we have a Mavic 3M RTK, but the RTK unit [D-RTK 2 Mobile Station] is the same as that used for the Phantom 4 multispectral, and there's no topic or subtopic type for the Mavic 3M. And also the folks following this section, like @patiam, know a lot about this stuff...)

We're having problems with the Pilot app showing RTK not connected when it would seem to be in range, with the documentation saying "OcuSync Communication Distance" [I'm assuming that means range from the base station to the rover in the drone] goes 12 km or 6 km, depending on whatever NCC/FCC or SRRC/MIC/KCC/CE represent in this case, and we're experiencing this lack of connection when at less than 2 km.  We've even manually launched the drone to plenty of height to ensure a line of sight connection.  

We may try setting up our base station closer, however I'm not sure if that's the problem.  We've had success for some flights, that were a bit closer, but haven't really systematically tested it.

Also the notification display at the top of the main display shows RTK with a number of satellites, not in red; it's just when we start the launch process where we see something about the RTK not being connected.

Hey Jerry-

All DJI RTK gear that I know of is notorious for showing obfuscating and misleading error messages when it is not happy (for example "weak RTK signal" when using NTRIP can sometimes happen despite having a super robust internet connection but your baseline to the NTRIP caster is too long, which can indeed be a problem but is not related to any sort of "weak signal"). Likewise, the one you mention may or may not have anything to do with the actual problem. You can get the same error on a P4R by trying to switch between RTK sources (D- RTK 2 & "Custom RTK Network") without restarting the aircraft (you have tried restarting to see if it helps, no?).
You're on the right track, systematic testing is often the only way to figure out what the real problem is, given the dismal documentation and minimal support available. Sorry I don't have one of these newer birds in my squadron, so I can't offer personal experience on it. Hopefully others will have some insight.

BTW the M3M's brethren (M3E & M3T) have forums in the Mavic section:
https://forum.dji.com/forum.php?mod=forumdisplay&fid=129&filter=typeid&typeid=556

Thanks, Pat, very useful.

I've done some more testing:
(1) at Pacifica State Beach, with base station by San Pedro Creek mouth and drone taken to various distances.  Occasionally didn't work (e.g. by Taco Bell), but ultimately worked the length of the beach ~1200 m.  At pump outlet between Taco Bell and San Pedro Creek (168 m from the base station), the first test connected, but the second one did not until I picked up the drone to get it higher.  
(2) Testing at home, with base station on back deck and drone taken to the other side of the house (to block connection). After connecting, I could walk down the street, clearly blocked from the base station by houses, yet it continued to show connection in the Preflight dialog; however after powering off and back on, was not connected, though the flight display might show it working.  Possible diagnosis:  the drone had a good GNSS connection, but the Preflight Check indicated that it was not talking to the base station.

I'm thinking the most likely solution is to set up the base station at a ground control point within the flight plan.  It adds the need to lug that equipment out to the site (fortunately I always have multiple assistants on the project), but would provide the advantage of better accuracy anyway, and also the ability to change batteries in the base station if necessary.  

But I'm wondering if the system was actually working anyway (with the RTK in the main display as white, not red), and if we can diagnose this from the log files.  The *PPKNAV.nav, *PPKOBS.obs and *Timestamp.MRK files seem to have data (readable by a text editor) that might be useful, and I'm not seeing any obvious differences between log files from flights showing RTK connected in the preflight checklist and those showing it disconnected.

JerryDavis Posted at 6-6 09:48
Thanks, Pat, very useful.

I've done some more testing:

You can tell pretty quickly by looking at the MRK file with notepad. The last columns are those of the precision of the coordinates.

Ah, perfect, thanks!  I'm seeing numbers like 0.011650, 0.009813, 0.022294        in the MRK file for the flight that did not show RTK connected in the preflight checklist, and similar numbers like 0.011432, 0.008916, 0.024303 in a flight that showed connection.  All of those are about 1 cm horizontally and 2 cm vertically, so it looks like it worked after all!

We may go ahead and set up the base station at a nearby GCP, but it's good to know that we might not have to.

JerryDavis Posted at 6-6 11:00
Ah, perfect, thanks!  I'm seeing numbers like 0.011650, 0.009813, 0.022294        in the MRK file for the flight that did not show RTK connected in the preflight checklist, and similar numbers like 0.011432, 0.008916, 0.024303 in a flight that showed connection.  All of those are about 1 cm horizontally and 2 cm vertically, so it looks like it worked after all!

We may go ahead and set up the base station at a nearby GCP, but it's good to know that we might not have to.

On M300 there is a function that puzzles me, it is the maintain of the RTK. If the connection to the NTRIP is broken, it maintains a level of precision which decreases and which is interrupted after 10 minutes. For photogrammetry it's good I think, for LiDAR much less. But in the case you're describing, it's possible that's what's happening with your M3M.

2km is the practical range of DJI devices in CE mode. I have never tried FCC. 12km and 6km is theoretical, certainly achievable in optimal conditions.

Now I do almost everything in PPK, unless I'm close (less than 20km) to a national CORS NTRIP station. This saves me a lot of problems that I encountered before with the RTK.

I can probably shed some light on this puzzle, based on my experience with using a GNSS receiver (Eos Arrow Gold) that connects to the California Real-Time Network (CRTN), which provides RTK via an internet connection.  Basically, the correction continues to work even after you're disconnected from the RTK (this happens to us where we have spotty cell coverage), but the accuracy degrades as the time since connection increases.  I'd like to figure out a PPK solution for us, due to this problem.

JerryDavis Posted at 6-6 11:34
I can probably shed some light on this puzzle, based on my experience with using a GNSS receiver (Eos Arrow Gold) that connects to the California Real-Time Network (CRTN), which provides RTK via an internet connection.  Basically, the correction continues to work even after you're disconnected from the RTK (this happens to us where we have spotty cell coverage), but the accuracy degrades as the time since connection increases.  I'd like to figure out a PPK solution for us, due to this problem.

For PPK, my best recommandation is :

https://www.sparkfun.com/products/17370
+
https://www.boschtools.com/us/en ... pods-bt160-34307-p/
+
3D files provided in post nr1 here : https://forum.dji.com/forum.php?mod=viewthread&tid=285647

With this software (free) https://docs.emlid.com/emlid-studio/

I use this setup daily for LiDAR, photogrametry and GPR, I am completely satisfied. No more costly system like Trimble, Leica... which are very good for the use of their dedicated software. Not for our use with DJI drones in a university environment..