Testing the firmware V01.00.648 (Beta Firmware) looks like the issue has been improved or resolved.
I was able to adquiere GPS signal and homepoint in about 40 seconds when with the firmware 642 was 1 minute and 20 secods and the releases firmware .600 (current firmware) I always pass over 2 minutes.

I am ataching a video testing.

Also APAS has improvements on the firmware 642 but some issues (but still good) on firmare 648. Now i have good hopes for the next official firmware.
Thanks DJI !!!!!
Thanks to Hans Groover for provide me the firmware !!

Nice video!  I actually had it come up on my Youtube suggestions first.  Thank you for showing the improvement in the zoom function too!

Does Airsense work now?

Does firmware 648 support RAW and manual control for 7X zoom camera?

Airsens works now but not all the time. I have screen recording working. I have only 6 flight with that firmware.

jerryway Posted at 5-15 18:34
Does firmware 648 support RAW and manual control for 7X zoom camera?

Not with the firmware 648. Maybe once they release the new fly app. I am just guessing

ZeuS-FL Posted at 5-15 18:38
Not with the firmware 648. Maybe once they release the new fly app. I am just guessing

Thanks. Also can you clarify what do you mean by "some issues" on firmare 648? What are these issues exactly?

When are we expecting this firmware update?

Hello there ZeuS-FL. Good day and thank you for sharing these information and video with us. Have a safe and a happy flying always.

Good news.   Hopefully others will see faster times (<40s) when DJI's officially releases of V01.00.648 (or ++).

I doubt it is fully fixed because for me I am waiting 90 seconds on a cold boot with RC Pro, 4G Network to RC Pro and on 0648 as well. Dji still has work to do before sending out this FW. It needs to work for all. I must say it is an improvement from previously being 6 to 11 minutes but I want it at 30 seconds or less

I hope so, when .600 arrived situation seems bright but now with .600 I have half flights with no home point as 5-7 minutes of waiting are far from acceptable

That looks promising, thanks for sharing.

Suren Posted at 5-15 22:10
I doubt it is fully fixed because for me I am waiting 90 seconds on a cold boot with RC Pro, 4G Network to RC Pro and on 0648 as well. Dji still has work to do before sending out this FW. It needs to work for all. I must say it is an improvement from previously being 6 to 11 minutes but I want it at 30 seconds or less

I must say it is an improvement from previously being 6 to 11 minutes but I want it at 30 seconds or less
You mean that you want what they advertised at release time?
>>  This enables Mavic 3 to lock onto multiple satellite signals faster than ever.
It looks like that's never going to happen.

DJI Stephen Posted at 5-15 20:06
Hello there ZeuS-FL. Good day and thank you for sharing these information and video with us. Have a safe and a happy flying always.

Thanks DJI Stephen.

good tests, but a sample size of 1 cant prove much of anything. GPS will differ from day to day, and sometimes hour to hour, but your trend at least suggests its possibly better rather than possibly worse. Regarding APAS, I doubt you should be seeing much changes build to build. APAS is most certainly algorithm based, so how it responds to an environment (like you walking around under tree branches and stuff) will vary from run to run, even on the same firmware. Just a few feet left or right or up or down from the previous run will cause a different calculation to be made and a different behavior. Testing APAS changes would be difficult, without some very tight controls on the testing environment I would imagine.

Tornado12 Posted at 5-16 07:02
good tests, but a sample size of 1 cant prove much of anything. GPS will differ from day to day, and sometimes hour to hour, but your trend at least suggests its possibly better rather than possibly worse. Regarding APAS, I doubt you should be seeing much changes build to build. APAS is most certainly algorithm based, so how it responds to an environment (like you walking around under tree branches and stuff) will vary from run to run, even on the same firmware. Just a few feet left or right or up or down from the previous run will cause a different calculation to be made and a different behavior. Testing APAS changes would be difficult, without some very tight controls on the testing environment I would imagine.

I agree. Sometimes acts very smart, sometimes dumb. Same firmware.
There are so many variants.

I will repeat the GPS test in a clean environment. open space at the lake. Loading both firmware. .600 and .648

Still is not presice since I need to wait at least 24 hours for a cold boot but still can be useful.

jerryway Posted at 5-15 18:45
Thanks. Also can you clarify what do you mean by "some issues" on firmare 648? What are these issues exactly?

When I say "some issues" I meant there are different ways the drone is great. Seems to be very smart and sometimes dumb.

I need to keep testing to have a conclusion.

ZeuS-FL Posted at 5-16 07:08
I agree. Sometimes acts very smart, sometimes dumb. Same firmware.
There are so many variants.

I think looking at tracking footage it resembles very much a skydio which is also very unpredictable, I think there are limits to AT and realizing them will get the best results. The testing is fine but the scenario of following someone in and out of the bushes could be done much easier and better with a handheld camera on a gimbal. Stretching the limits as you were doing shows how good it can work but the risk will always be with the camera man pilot etc.

What may be part of problem is DJI trying to get a Satellite system "Lock" using multiple GPS systems (countries).

While back, was discussed issue of mixing multiple GPS systems' satellites.  For a GPS system to work, all satellites have to have a clock signal which is synced and compensated for slowing of time due gravitational field of earth vs. satellites in orbit.
Problem is clocks between different GPS Satellite systems (countries) are not synced.  Leading to GPS receiver having to either compensate for clock (time) difference between different GPS systems clocks or use four or more GPS satellites of same system.  Examples, the GPS receiver will take a long time to get a GPS fix using 3 GPS satellites of GLONASS and 3 satellites of NAVSTAR and 3 satellites of Galleio vs. using 5 GLONASS and 4 NAVSTAR or using 9 NAVSTAR.


"navigational signals transmitted by GPS satellites encode a variety of  information including satellite positions, the state of the internal  clocks, and the health of the network"

"first subframe of each frame encodes the week number and the time within the week, as well as the data about the health of the satellite. The second and the third subframes contain the ephemeris – the precise orbit for the satellite. The fourth and fifth subframes contain the almanac,  which contains coarse orbit and status information for up to 32  satellites in the constellation as well as data related to error  correction. Thus, to obtain an accurate satellite location from this  transmitted message, the receiver must demodulate the message from each  satellite it includes in its solution for 18 to 30 seconds"

Labroides Posted at 5-16 02:32
I must say it is an improvement from previously being 6 to 11 minutes but I want it at 30 seconds or less
You mean that you want what they advertised at release time?
>>  This enables Mavic 3 to lock onto multiple satellite signals faster than ever.

100%, I want that and not asking for much. Just to gives us what was promised as features nothing extra.

Correct. We are not asking to break a record on acquisition speed. I agree that 20-30 seconds is something acceptable.

HedgeTrimmer Posted at 5-16 09:29
What may be part of problem is DJI trying to get a Satellite system "Lock" using multiple GPS systems (countries).

While back, was discussed issue of mixing multiple GPS systems' satellites.  For a GPS system to work, all satellites have to have a clock signal which is synced and compensated for slowing of time due gravitational field of earth vs. satellites in orbit.

Using multiple constellations is extremely common in most of todays GPS systems - Including all of DJI's drones. Without getting into a long technical post This should not present a problem at all. The data which is downloaded from the gps constellations is akin to small text files. Very very small bits of data.

What I have speculated on with Mavic 3 is the possibility that DJI attempted to increase GPS accuracy in the Mavic 3 over previous drones. If they set parameters to increase accuracy that could cause delay in acquisition. That is just speculation though. There is a lot that can go into this acquisition, and global location plays a big role. This is why DJI's official response on the continued GPS issues is interesting. I don't have it in front of me but they essentially acknowledged existing problems in the northern hemisphere, which is where the vast majority of the worlds population lives. Back before they released any fix the only hint they dropped to the cause was that it was "In the algorithm". This lead me to speculate that they are doing something different with Mavic 3. Whether that is increased accuracy in this drone, or something else, we cant know.

Using miltiple constellations though should NOT be the problem here. The only thing that could have been a possibility was the change to beidou, but the mavic mini 3 pro also uses beidou and so far appears to not suffer from the longer acquisitions.

Sometimes the road to hell is paved with good intentions.
Probably DJI tried with the Mavic 3 to make a better drone (in their intentions) but the cost of that causes more problems.

Tornado12 Posted at 5-16 10:14
Using multiple constellations is extremely common in most of todays GPS systems - Including all of DJI's drones. Without getting into a long technical post This should not present a problem at all. The data which is downloaded from the gps constellations is akin to small text files. Very very small bits of data.

What I have speculated on with Mavic 3 is the possibility that DJI attempted to increase GPS accuracy in the Mavic 3 over previous drones. If they set parameters to increase accuracy that could cause delay in acquisition. That is just speculation though. There is a lot that can go into this acquisition, and global location plays a big role. This is why DJI's official response on the continued GPS issues is interesting. I don't have it in front of me but they essentially acknowledged existing problems in the northern hemisphere, which is where the vast majority of the worlds population lives. Back before they released any fix the only hint they dropped to the cause was that it was "In the algorithm". This lead me to speculate that they are doing something different with Mavic 3. Whether that is increased accuracy in this drone, or something else, we cant know.

Mavic Pro uses (GPS / GLONASS)  
Mavic-2 uses (GPS+GLONASS)
Mavic-3 uses (GPS + Galileo + BeiDou)

I believe DJI specs mean:
Mavic Pro could use either USA GPS or Russian GPS, but not both at same time
Mavic-2 added ability to use USA GPS and Russian GPS, at same time
Mavic-3 added ability to use Galileo GPS and BeiDou GPS, at same time

Thus Mavic-3 firmware for GPS would be different from Mavic-2.

HedgeTrimmer Posted at 5-16 09:29
What may be part of problem is DJI trying to get a Satellite system "Lock" using multiple GPS systems (countries).

While back, was discussed issue of mixing multiple GPS systems' satellites.  For a GPS system to work, all satellites have to have a clock signal which is synced and compensated for slowing of time due gravitational field of earth vs. satellites in orbit.

What may be part of problem is DJI trying to get a Satellite system "Lock" using multiple GPS systems (countries).
But this has not been a problem for any other DJI drones.
The Air 2s for example uses GPS+GLONASS+GALILEO without any problem.
The new Mini 3 uses the same constellations as the Mavic 3.
And it wasn't a problem for the Mavic 3 before the December firmware update.
Ublox who make the GPS chip that's used seem to have that sort of thing worked out.
But something in DJI's implementation post December is the issue.

Tornado12 Posted at 5-16 10:14
Using multiple constellations is extremely common in most of todays GPS systems - Including all of DJI's drones. Without getting into a long technical post This should not present a problem at all. The data which is downloaded from the gps constellations is akin to small text files. Very very small bits of data.

What I have speculated on with Mavic 3 is the possibility that DJI attempted to increase GPS accuracy in the Mavic 3 over previous drones. If they set parameters to increase accuracy that could cause delay in acquisition. That is just speculation though. There is a lot that can go into this acquisition, and global location plays a big role. This is why DJI's official response on the continued GPS issues is interesting. I don't have it in front of me but they essentially acknowledged existing problems in the northern hemisphere, which is where the vast majority of the worlds population lives. Back before they released any fix the only hint they dropped to the cause was that it was "In the algorithm". This lead me to speculate that they are doing something different with Mavic 3. Whether that is increased accuracy in this drone, or something else, we cant know.

This is why DJI's official response on the continued GPS issues is interesting. I don't have it in front of me but they essentially acknowledged existing problems in the northern hemisphere, which is where the vast majority of the worlds population lives.
What they said was that they had identified the problem  and that it was particularly affecting users in high latitudes.
But users in any latitudes were observing the same issue.
DJI's announcement was damage control spin and just something to calm the users and buy time.
They had already been working on a fix for a number of months and said it would take another 2 months to fix.
It seems unlikely that they could forecast that a fix would take another 2 months, when they hadn't been able to fix it in the preceding 4 months.
Anyway the two months is nearly up and we should soon see what progress has been made.

As for suggestions of doing something to increase accuracy (although this was never needed), those suggestions only came from forum sugestions.
If they were true, it would have been a very simple matter to simply go back to the old tried and true system that DJI's used for millions of drones previously.

Net net, the GPS issue is a Hardware/Design issue. It's not a Firmware issue. DJI has been writing GPS Firmware for years - they know what they're doing. They are desperately trying to fix this Hardware/Design issue via firmware. Instead of falling on their sword and issuing a recall (like Apple has done) via an engineering change order, they are taking the cheap route. I paid 3K for 6 minute GPS lock. Shame on DJI...

As long as the GPS is improved and they added some promising features, I'll take what they have.

Labroides Posted at 5-16 12:07
What may be part of problem is DJI trying to get a Satellite system "Lock" using multiple GPS systems (countries).
But this has not been a problem for any other DJI drones.
The Air 2s for example uses GPS+GLONASS+GALILEO without any problem.

Several things..

My understanding was Mavic 3 was experiencing slow GPS problems from release day.  But it wasn't until enough people had M3's in use and sufficient people complained did it become a "problem".  

In regard to Air 2s vs M3, DJI did add BeiDou GPS into Mix with M3.  Perhaps it is one constellation to much for their satellite system picking and locking algorithm (code)?

As for Mavic 3 and Mini 3, are two types of drones using same GPS module (or chipset)?  

Labroides Posted at 5-16 12:19
This is why DJI's official response on the continued GPS issues is interesting. I don't have it in front of me but they essentially acknowledged existing problems in the northern hemisphere, which is where the vast majority of the worlds population lives.
What they said was that they had identified the problem  and that it was particularly affecting users in high latitudes.
But users in any latitudes were observing the same issue.

"it would have been a very simple matter to simply go back to the old tried and true system that DJI's used for millions of drones previously."

Which was one of my counter points to slow GPS being simply a mater of firmware.  If it were, DJI would have backed off to previous used block of firmware code for GPS.
Which is why I think DJI got a bad batch of GPS modules, which are senstive to firmware being run.  Pushing GPS module to also provide BeiDou GPS data stream is to much for bad batch(s).   


Whereas Mini-3 either uses a different GPS module (chipset) or Mini-3 being produced later didn't get any of bad batch.  Also fitting with reports of Mavic-3 drones built after March not having slow GPS problem.


I don't know why DJI can't be more forthcoming with these kinds of problems?  Everybody in industry knows DJI has a slow GPS problem with M3.  Good chance DJI's competitors have already been informed as to cause by spying or leaks.

HedgeTrimmer Posted at 5-16 18:16
Several things..

My understanding was Mavic 3 was experiencing slow GPS problems from release day.  But it wasn't until enough people had M3's in use and sufficient people complained did it become a "problem".  

My understanding was Mavic 3 was experiencing slow GPS problems from  release day.  But it wasn't until enough people had M3's in use and  sufficient people complained did it become a "problem".
No reviewers or early buyers mentioned the slow GPS acquisition until after the first firmware update.
The one I bought would pick up 20+ sats quickly indoors, until that update crippled it.

In regard to Air 2s vs M3, DJI did add BeiDou GPS into Mix with  M3.  Perhaps it is one constellation to much for their satellite system  picking and locking algorithm (code)?
The Air 2S uses 3 constellations, but Beidou is not one of them.
The Mini 3 uses the same three as the Mavic 3.
There's no reason to suspect that the Ublox GPS chip has any problem dealing with 3 constellations.

As for Mavic 3 and Mini 3, are two types of drones using same GPS module (or chipset)?  
I don't know what's inside the Mini 3, but the Mavic 3 uses the same Ublox GPS chip that DJI have used for several different previous drone models.

The issue is complicated.
It's not as simple as bad hardware or the constellations being used.
That DJI still haven't been able to fix it yet, 6 months after release should be a big hint about how difficult it is to fix.

Im definitely exited for Dji’s next firmware update on the Mavic 3, I know for my self it’s all been about location, sometimes when I set up my Mavic 3 it’ll set the home point in about 30-40 seconds but then other times it go past 4 minutes but on those times it was cloudy but again can’t wait for the new firmware update.

Fred Fred Fred Posted at 5-16 20:36
Im definitely exited for Dji’s next firmware update on the Mavic 3, I know for my self it’s all been about location, sometimes when I set up my Mavic 3 it’ll set the home point in about 30-40 seconds but then other times it go past 4 minutes but on those times it was cloudy but again can’t wait for the new firmware update.

other times it go past 4 minutes but on those times it was cloudy.
GPS is like the mailman, neither snow nor rain nor heat nor darkness of night prevents the GPS signal getting through.
Clouds make no difference at all to GPS reception.

Labroides Posted at 5-16 18:47
My understanding was Mavic 3 was experiencing slow GPS problems from  release day.  But it wasn't until enough people had M3's in use and  sufficient people complained did it become a "problem".
No reviewers or early buyers mentioned the slow GPS acquisition until after the first firmware update.
The one I bought would pick up 20+ sats quickly indoors, until that update crippled it.

Or (and I hope this is NOT the case); DJI pulled needed resources from working on Mavic 3 slow GPS satellite problem to work on getting Mini 3 out the door.  Would not be first time a company's product got pushed from front burner to back warming plate to make room for what's new cooking at a company.

Really having trouble seeing how M3's slow GPS problem could be so  Firmware complicated.  Gven DJI's lengthy experience with GPS and Mini 3 not  having a problem, and as you indicate M3 didn't originally have a problem.  


If slow GPS truly is a Firmware only problem, one should be able to diff (difference) the previous working GPS firmware code against now broken GPS firmware code to see what changed.  One errant exception, I know about from Admin experience, is changes to revision of compilers, assembler, or libraries being used by a user; whilst said user is making changes to their sofware too.  In such a case, you have to diff the generated machine code, and look at what instructions and possibly static data got changed.

I've read through the discussion here between labroides and hedge and the biggest take away I have are that you both full of bold presumptions. "If it was just firmware then DJI would have just....." how do you know what DJI would have done though? This has always been a pet peeve of mine standing around the morning coffee pot when people start talking about politics, and immediately start making blanket statements about how the whole country could be fixed if this one would do this and that one would do that. The problem with this is that its often extremely naïve, uninformed, and completely ignores the complexity of the situation. I see this happening a lot in this discussion. Its so easy to sit here and be an armchair engineer, when most on here know little to nothing about electrical engineering, or DJI's design process.  You guys are trying to make this a neat little black and white issue that you can tuck neatly into a box, and I think it is clear that the issue isn't that neat. There is obviously some complexity to this issue that everyone here is completely unaware of, a lot of grey area that doesn't fit neatly into our bold presumptions of how and what DJI should and would do.  

Latest reports on beta firmware seem to suggest the issue is improving. While I take these reports with a grain of salt, at least the reports arent going the other direction. If the reports are true though it does show what Ive felt all along, that it is software / configuration.

Tornado12 Posted at 5-17 04:53
I've read through the discussion here between labroides and hedge and the biggest take away I have are that you both full of bold presumptions. "If it was just firmware then DJI would have just....." how do you know what DJI would have done though? This has always been a pet peeve of mine standing around the morning coffee pot when people start talking about politics, and immediately start making blanket statements about how the whole country could be fixed if this one would do this and that one would do that. The problem with this is that its often extremely naïve, uninformed, and completely ignores the complexity of the situation. I see this happening a lot in this discussion. Its so easy to sit here and be an armchair engineer, when most on here know little to nothing about electrical engineering, or DJI's design process.  You guys are trying to make this a neat little black and white issue that you can tuck neatly into a box, and I think it is clear that the issue isn't that neat. There is obviously some complexity to this issue that everyone here is completely unaware of, a lot of grey area that doesn't fit neatly into our bold presumptions of how and what DJI should and would do.  

Latest reports on beta firmware seem to suggest the issue is improving. While I take these reports with a grain of salt, at least the reports arent going the other direction. If the reports are true though it does show what Ive felt all along, that it is software / configuration.

"If it was just firmware then DJI would have just....." how do you know what DJI would have done though?
Are you for real?
DJI have been scrabbling around for 6 months now trying to salvage their failed flagship drone.
It's a no-brainer that they would have tried all the obvious things and all the simple things, and if it was that easy the Mavic 3 would have been flying as it was advertised long ago.

You guys are trying to make this a neat little black and white issue that you can tuck neatly into a box, and I think it is clear that the issue isn't that neat. There is obviously some complexity to this issue that everyone here is completely unaware of, a lot of grey area that doesn't fit neatly into our bold presumptions of how and what DJI should and would do.
Go back and read what I wrote ... you haven't understood it if you think that.

If the reports are true though it does show what Ive felt all along, that it is software / configuration.
It doesn't take any particular genius to come up with that assessment.

Labroides Posted at 5-16 21:06
other times it go past 4 minutes but on those times it was cloudy.
GPS is like the mailman, neither snow nor rain nor heat nor darkness of night prevents the GPS signal getting through.
Clouds make no difference at all to GPS reception.

Well I’m no expert on gps systems but all I know is every time its been cloudy I’ve had to wait longer on my gps to lock  home point or maybe it could be because a guy was eating a sandwich behind on those days which somehow caused an interference I don’t know?

Tornado12 Posted at 5-17 04:53
I've read through the discussion here between labroides and hedge and the biggest take away I have are that you both full of bold presumptions. "If it was just firmware then DJI would have just....." how do you know what DJI would have done though? This has always been a pet peeve of mine standing around the morning coffee pot when people start talking about politics, and immediately start making blanket statements about how the whole country could be fixed if this one would do this and that one would do that. The problem with this is that its often extremely naïve, uninformed, and completely ignores the complexity of the situation. I see this happening a lot in this discussion. Its so easy to sit here and be an armchair engineer, when most on here know little to nothing about electrical engineering, or DJI's design process.  You guys are trying to make this a neat little black and white issue that you can tuck neatly into a box, and I think it is clear that the issue isn't that neat. There is obviously some complexity to this issue that everyone here is completely unaware of, a lot of grey area that doesn't fit neatly into our bold presumptions of how and what DJI should and would do.  

Latest reports on beta firmware seem to suggest the issue is improving. While I take these reports with a grain of salt, at least the reports arent going the other direction. If the reports are true though it does show what Ive felt all along, that it is software / configuration.

Well so far apart from the usual conspiracies there has not been an iota of proof that this has anything to do with hardware. But something went wrong in December that brought about the delay in acquisition and it hasn’t been easy to sort. Nothing to do with manufacturer date we received one made end of March last week, and although it doesn’t take 5 minutes to get GPS lock it isn’t as quick as other dji drones we have and is basically the same as mine Manufacture date 10/21.
So manufacturing process is still pumping out the very same drone .

The only fact that let me think is that with the firmware .400 was working well everything. Now, the only difference from that is that on the January update (implemented at the end of December) they added the new features like master shots, quick modes, and follow me...

Here is the question. Why does DJI not deliver these features from day 1 like the other drones? Do they know something was wrong with the drone while using the features? I know this may be related or may not be to the GPS problem.