Note: This article does not provide the official answer, nor does it provide a solution (yet). This article is meant to get to the core of the problem, and to urge DJI to fix it or at least find a way to make customers whole. The DJI Mavic 3 is an expensive drone and it should work as advertised. Simple as that. Please share if you agree.
The DJI Mavic 3 was rushed to market
The DJI Mavic 3 was rush-launched so that DJI’s latest and greatest drone would hit the shelves before the fourth-quarter holiday shopping season.
At the time of the launch, DJI explained that many of the promised features of the Mavic 3 would become available with a scheduled Firmware Update for January 2022.
The DJI Mavic 3 GPS problem
However, after the December firmware update, many drone pilots complained about the time it would take the Mavic 3 to acquire sufficient satellite signals to take off, fly at a sufficient altitude and safely return home if and when needed.
The DJI Mavic 3 GPS problem caused many drone operators to have to wait several minutes before they could fly their drone.
Interestingly enough, DJI promised us that the DJI Mavic 3 would lock on faster than ever.
“Mavic 3 comes with a powerful positioning algorithm that improves hovering precision with signals from GPS, GLONASS, and BeiDou satellites. This enables Mavic 3 to lock onto multiple satellite signals faster than ever,” it said in the original press release. “The increased positioning precision also makes Mavic 3 less likely to drift in the air and more stable when shooting long exposures and timelapses.”
The DJI Mavic 3 is the drone maker’s flagship drone for both commercial and recreational drone pilots. It is expensive and offers almost all the features one could ask for in a drone.
DJI promised “Imaging Above Everything” with the launch of the Mavic 3, so it should not come as a surprise that customers are upset when the flagship drone fails to deliver on all of its promises.
DJI acknowledged the Mavic 3 GPS issue
Earlier this month, DroneXL spoke with DJI about the Mavic 3 GPS issue and the drone maker responded by email with an official statement acknowledging the problem and saying it would be fixed soon.
“We have identified the issue behind the slow satellite acquisition and will address it in a firmware update coming soon,” DJI said in an email to DroneXL.
On March 23rd, DJI issued the third firmware update which was supposed to fix the GPS problem of the M3 for once and for all. However, while it seemed to have worked for some drone pilots, it clearly hasn’t for many others if you follow the various forums.
DJI Forum: The Cause of the Mavic 3 GPS Issues- Revealed
On the official DJI Forum, AerialZup2 posted a very lengthy and detailed possible explanation as to what may be the cause of the Mavic 3 GPS issues.
I do not if AerialZup2 is indeed a troll. I also don’t know if his explanation is 100% accurate. It seems plausible but I am not a GPS expert. As I stated before, this article is meant to get to a solution so that we all can enjoy the DJI Mavic 3 as it is supposed to work.
With that, here is the complete original text of “The Cause of the Mavic 3 GPS Issues- Revealed” as it was shown on the DJI Forum, sourced from Google’s cache as the internet never forgets anything.
You can choose to believe this or not, but I’m going to tell you the cause of the GPS issue, why it affects some craft and not others, and what is going on. If you don’t believe this, that’s fine. But this has been communicated to me by some very credible people within DJI (and please don’t ask me who – I can’t give them up).
I am telling you all this because how you respond to it is up to you. You can blast me, you can reach out to DJI, or you can demand some type of action – or you can do nothing:
When the Mavic 3 was manufactured, they decided to use their tried-and-true uBlox for their GPS chip. They wanted to use the same UBX-M8030 series – the same series that they used for years in the Mavic 2, Air 2, Air2S. The same chip that gets great performance in the Air2S as a matter of fact.
There are 3 “versions” of the uBlox chip:
“standard grade”,”professional grade”, and “automotive grade”. (picture) In the prior drones, DJI used the “professional” grade UBX-M8030 (KT) sku.
As manufacturing commenced, there was a shortage in this chip, so DJI decided to use the “automotive” grade UBX-M8030 (KA) sku.
On paper, the chips have identical specs in terms of GPS. They both have the capability of using 4 separate satellite systems (GPS, Galileo, Glonass, and Beidu) – but only 3 can be used simultaneously as dictated by the DJI craft firmware. The only “paper” difference is that the “automotive” grade has some more thermal shielding which per the spec sheet, gives a wider temperature operating range. If anything, DJI felt this was a net positive, and an “upgrade” if anything.
The Problem, And How it Revealed Itself – and Why it Came with the December Firmware:
The problem with the UBX-M8030KT is that its onboard memory is different to the other chip variant. Some clever individuals in fact, in another forum, did a tear down on this chip in the Mavic 3 and noticed some slight differences in the physical memory configuration on the chip (although the dimensions and connectors on the chip itself remained the same) It has a much more limited memory for “ephemeris” data.
What is “ephemeris”?
GPS satellites transmit information about their location (current and predicted), timing and “health” via what is known as ephemeris data. This data is used by the GPS receivers to estimate location relative to the satellites and thus position on earth. The Ephemeris Data can also be used to predict future satellite conditions (for a given place and time).
This information is stored in the memory module of the GPS chip for a distinct period of time. It is why a “cold start” “warm start” and a “hot start” have vastly different performances of satellite acquisition. The way the memory module stores the ephemeris data and accesses it also is different between the 2 SKUs of this uBlox chip.
Prior to December, the GPS module performed relatively well. There were no complaints of GPS performance, as expected. However, the December update provided some new “features” to the product, namely “FocusTrack, Hyperlapse, MasterShots”, and a new way that RTH is implemented. These “features” use GPS data within the chip, and take up some of the memory module on the UBX-M8030 (KA) chip itself. Because of that, with these features, there was limited memory for the chip to download and store the ephemeris data. Every time the drone had a cold start, it had to go through the process of downloading the ephemeris data, slowly, on a limited bandwidth, to a small pipeline. In fact, the January 2022 firmware included further features, and some users reported that the GPS “issue” actually worsened. (There may have been a good reason why AirSense is somewhat “broken”. It relies on GPS data as well)
This is why DJI initially advised to be “connected” to a data network. Theoretically, the ephemeris data could be downloaded through a network faster than directly through “the air”. But unfortunately that’s not how it panned out.
Upon discovering this problem, the primary way DJI felt it could be dealt with was “shrinking” the allotment of memory taken up by some of the very features which were causing the GPS problem to being with. Thus DJI had to work very hard to limit the memory allocations that its “features” were using in the limited memory it had to work with. This is not “ideal” as it is more of a “software” bandaid that doesn’t address the core issue. But it is why some have seen improvement.
Why are others not seeing improvement while some are?
Ephemeris data contain information on week number, satellite accuracy and health, age of data, satellite clock correction coefficients, orbital parameters, amongst other data. Location, position, and a number of other factors play a role in that data download. It can very well be location-dependent.
Thus the issue is “hardware” and “software” related, in a sense.
I’m sure this isn’t “exact” but it gives a good picture as to what is going on. It isn’t clear if this issue can truly be “solved”, and I hope it provides some clarity on why it took so long for DJI to attempt to address this issue.
I’m sure it is an even more complicated decision about what to do about this now.
Can a manufacturing line be changed to use the appropriate uBlox chip that was intended?
Can further “software modifications” be optimized to improve the craft without further feature sacrifice?
Should there be a “silent recall” with a swap out?
I’m sure you can all imagine a number of questions and possibilities here. But one way or another, I would hope most people agree that the issue needs to be owned up. Given the above, I would assume you can understand why DJI has been less-than-transparent about what is going on.
Again, take this or leave it. But this is what is going on.
Mad’s Tech counter perspective
I’m not a GPS chipset expert so it is hard for me to say if this is correct or not. however, here’s Mad’s Tech’s response to the above post. Spoiler alert, Mad’s Tech, who knows a thing or two about tech, says AerialZup2 is completely wrong in the comments on OriginalDobo’s YouTube channel.
Here we go:
Ken, that article on DJI forum is not factually correct, people are so desperate for info they are starting to make stuff up. The Chipset they are using is the same it’s just the automotive version that has higher compliance requirements for car use.
The UBX-M8030 no matter the version has the same GPS performance. The automotive industry version simply costs more for those applications usually with higher temperature requirements and it’s designed to comply with Electronics Council’s various grade specifications including AEC-Q100 Grade 1.
I put a big post on there and he removed it with the thread. 99% of what he is saying is totally wrong. There is a lot of hype and fake info around this right now it’s getting silly. Yes DJI have an issue here but it’s not a specifically related to that automotive chipset.
This was my full response to him.
The M8030 models have the similar internal layout of CPU and SRAM including the amount. The differences between the automotive and professional grade is electrical characteristics for heat, EMI and some other things mostly around reliability in harsh environments such as vibrations and temperatures. Devices labels as automate grade will…
Automotive variants are more expensive due to the Additional checks but have no features drawbacks like less rom or ram. It’s a the same Ublox core.
DJI do not use GPS memory on board the Ublox chip to store data for smart features. any GPS data needed will be stored in lookup tables in the companion SOCs running the smart shots in ram on DJI side of the Mavic 3. The Ublox chipset is fairly simple in the sense it handles everything and that’s passed out over Uart, USB, 12C or SPI. The Ram and ROM are onboard the digital block in the modules and not externally accessible for use other than writing firmware.
DJI choosing the Automotive variant changes nothing major over the non automotive other than costing them more. There are many ways DJI may have had to change the implementation of the GPS such as not using external ram for storing additional data or changing the Ublox configuration however there is some things people need to understand.
There is a difference between Ublox 3D Lock and DJI GPS Lock. A GPS can be configured to lock at different accuracy levels however that has zero impact on what DJI are telling their flight stack to accept as a valid lock. For instance the Ublox could lock at a HDOP that’s higher than DJI want and they will simply wait for it to drop to the relevant level.
That image being posted around of the Mavic 3 GPS module with the unpopulated area is very likely a second compass or controller for something like beacon, it is not SQI flash as it’s the wrong pin and it’s out side the shielding. The M8030 off the top of my head has 2Mb of onboard flash in all models. While it does not require external SQ flash it is still recommended for certain applications. For instance when using Glonass it’s advised to have SQ flash for additional configuration data. That would be located under the shielding can and not off to the side of the PCB. Remember DJI often make multiple variants and even design it with features they never use due to costing and marketing. This has been the case for many years.
Right now there is no evidence to say they are not using SQ flash, it is recommended for additional configuration data with the likes of Glonass, if they have it would also means they can’t properly use Ublox AssistNow but I don’t see DJI doing that tbh, we just need to see below the can to confirm but I’m expecting it to be there.
There is too much hearsay on this and there needs to be far more proper testing. This may still be a result of the shift to BeiDou but we need proper data, we need to see results from ublox u-center at the same time and location to see what’s going on. We need proper data and not just assumptions.
People also need to stop looking at this as sat count, what’s more important here is HDOP and how DJI are interpreting the Data the GPS provides. Everyone is looking at the module but that may have zero cause on what’s going on here. The DJI Lock is all on their side, the GPS is simply streaming its data out, what DJI do with that is another matter.
A final note on DJI themselves, as some know I have many years dealing with them in many different ways, they are a fantastic company however often what is publicly said or share is not based in full reality or fact. Often the people who really can sort issues or need to know about them don’t know and are in siloes. And do not assume for one minute that posts on social media, Facebook, and this group even get to the right people either. Most of the support staff responding on there and other places are low-level customer services, the odd PM does post but it’s like walking into the Genius Bar at Apple Store and expecting them to give you details on Apple’s M1 potential SSD architecture issues. Yes, the engineers used to swing buy here DJI Forums from time to time and probably still do now and then but DJI has grown well beyond that now for the most part.
It’s very simple overall, DJI have years of designing GPS, but that’s not to say something has not bitten them here, don’t expect public acknowledgments of issues, they never been their way on thinks like this. They will quietly fix it or they won’t and they will move on. History is littered with some DJI products with unresolved issues, I own half of them.
For those not happy with the current gps wait as long as you’re happy too, if you’re not happy return it or move onto something else. Posting here every day for answers won’t get them or change anything, Provide them losses through returns, data and info from logs but understand that may never get to the people who really need it.
Here’s the video to which Mad’s Tech responded
DJI Support blames faulty USAF satellites for Mavic 3 GPS headaches
One of our readers reached out to DJI support for an answer and they got a response. Now, before you go ahead and tell me that DJI support doesn’t know anything, the DJI Mavic 3 GPS problem was brought to the attention of DJI’s “Senior Engineering Department.” That should mean something, right?
Spoiler alert, DJI Support partly blames faulty United States Air Force satellites for the GPS issues of the Mavic 3…
Glad to hear back from you and thanks for the information. The video you sent has been received by us.
The following analysis is what we have found from the flight controller data you provided earlier.
1. The carrier-to-noise ratio of the satellites (GPS, Galileo, Beidou) captured is well behaved, which demonstrates that your drone was working in a good condition.
2. The satellites mainly involved in positioning from the flight controller data you provided are GPS02, GPS05, GPS11, GPS12, GPS19, GPS20, Galileo01, Galileo07, Beidou19, Beidou21 and Beidou22.
In addition, we have also found the following things.
a. Several satellites that should be captured by design have not been captured. It is suspected that there is occlusion in your surrounding environment. The occlusion seems to appear in the northeast to the northwest.
b. Several satellites should have been used after being captured, but it hasn’t.
c. The health of GPS11 and GPS20 satellites is not ideal.
With above said, we would like to confirm/share the items listed below with you.
i. Whether there are possible obstructions. If it is convenient, you can provide photos (northeast to the northwest) to help us to confirm;
ii. We confirm that discovery b can be optimized in algorithm, we are working on it, and it will be optimized in the firmware update as soon as possible;
iii. As for information mentioned in discovery c, we need to wait for USAF to repair their satellites.
You mentioned earlier that you have an Air2s, we recommend you try the DJI Mavic 3 and DJI RC-N1 together, the AGPS (Assisted GPS) function will work when the phone is connected to the Internet. It should speed up the searching process. In order to further analyze, we also sincerely recommend that you send us the flight controller data after testing with the RC-N1.
The link is the same as before: https://pan-sec.djicorp….
Once again, apologies for the inconvenience caused so far. Looking forward to your reply.
So what is the solution to the Mavic 3 GPS problem?
Well, if you haven’t updated our DJI Mavic 3 with any of the firmware updates, you might be in luck. You might not at all experience any GPS issues with your new and expensive DJI drone.
However, in that case, if you were to update your Mavic 3 to the latest firmware update, be warned, there is no roll-back.
Or, you might just be lucky and not have had any issues at all. That could be.
However, for all the unhappy DJI customers on the forums, Twitter, and Facebook, including the folks who have reached out to DroneXL, the DJI Mavic 3 GPS problems are real and need to be addressed.
We have reached out to DJI for comment on this article and asked them if and how this issue will be addressed. Hopefully, this article will help to bring a solution sooner or bring about some other way for DJI customers to be made whole.
Having to upgrade to the next Mavic version to fix the GPS issue, does not seem like it would satisfy most drone pilots.
We will update this post as we get more information from either our readers or DJI itself.
Share your DJI Mavic 3 experience
In the meantime, please be sure to share your DJI Mavic 3 experience in the comments below. We are very interested in hearing from you. Maybe by comparing notes we can get to the bottom of this or at least get an answer from DJI.
Please share this post with other DJI Mavic 3 owners. Thank you!
Get your Part 107 Certificate
Pass the test and take to the skies with the Pilot Institute. We have helped thousands of people become airplane and commercial drone pilots. Our courses are designed by industry experts to help you pass FAA tests and achieve your dreams.
FTC: DroneXL.co uses affiliate links that generate income.* We do not sell, share, rent out or spam your email, ever. Our email goes out on weekdays around 5:30 p.m.