Mini 3 Pro Yeeted Itself Into The Water Today

[whitnasty1]

So I purchased a Mini 3 Pro 5 days ago and have flown it a total of 11 times until it's last flight today... During 5 of the 11 flights, I am looking at my Flight Logs on my RC Controller, there was an uncontrolled, extremely rapid ascent, over the max altitude that I set for the drone. Each time this happened, I was worried, but chalked it up to me being new to the drone world or just not knowing what I was doing... After a period of 5-8 seconds, the ascent would stop and I was able to re-gain control of the Z-axis movements of the aircraft.

That changed today... I live at the beach and was flying over an inlet, to get some pics... I had VLOS the entire time, and was at max signal strength on my RC Controller at all times. Battery was fully charged. On the way out the inlet, I can see where one of these rapid, uncontrolled ascents happened. It shot from 40m to 140m in a matter of seconds, above the set flight limit. But again, after 5-8 seconds, I was able to regain control and descend. So I turned around and started heading back towards me since I was over water and the rapid ascent happened. I remember thinking, well as long as it doesn't do a rapid descent, I should be okay...

Everything seemed to being pretty normal, so I flew the drone back and forth a few times about 50' off shore from me where it was easily observable. When I passed back the starting point, another rapid ascent began happening, but topped out around 50m according to the flight log on the RC controller. As soon as the ascent happened, it began an uncontrolled descent... Uh oh... Once I realized this thing was going down and not stopping, right in front of me, I quickly pressed the RTH button at approximately 20m above the surface of the water. I didn't see it at the time, but looking at the log, it clearly shows after pressing the RTH a notification "Ascending. To RTH" or something along those lines, however while that is being shown on the Controller the aircraft is still steadily descending... At this point, I can only watch in horror as it just sends itself right into the water in front of me, and I was completely unable to do anything to recover from this. Battery was at 68% when it hit the water, and I received no error messages other than an altitude alert whenever it go stuck on the uncontrolled ascent about 3mins prior to hitting the water. Wind was calm and surface of the water was glassy.

I will upload the log file to here tomorrow, but looking back, it appears this has happened multiple times in the 5 days I have had this thing. Luckily, I paid for the Insurance from DJI. But now looking at the fine print, I guess this considered a "fly-away" because the water is freezing cold right now and I wasn't about to get myself killed trying to retrieve a drone that was probably already at the bottom of the inlet on the way out into the ocean.... I haven't spoken to anyone at DJI yet because their customer service is closed right now, but I will be doing that first thing tomorrow...

The replacement fee for the Mini 3 Pro is $235, and I will be damned if I am paying that after watching that drone yeet itself into the ocean today. Clearly something happened with the IMU or something and it just did what it was going to do. If I was being an idiot and just tried to fly as far away as I could and I lost it, I could understand having to pay $235 for a replacement. But after watching the DJI Mini 3 Pro just send it into the water with everything it had, I'm not paying that for something that was clearly defective... And I am an electronics manufacturing engineer, and I make PCB's everyday, so I understand stuff happens sometime. But my confidence in the Mini 3 Pro is now gone and even if I do get a replacement, I'm not sure I'll ever feel comfortable flying one again after what I witnessed today. Just watching it and not being able to do anything about it was the worst part. You could just see it coming....

I'll update with the flight logs once I get them off the RC controller tomorrow and see what DJI has to say....

Anyone that has been around for a while seen or heard of this kind of problem with the Mini 3 Pro?

Thanks!

 

[Yorkshire_Pud]

Your description makes it sound more like a fault than a flyaway i.e. a warranty claim.
The ,txt log may tell the story but if you have not already synced logs with DJI then DO NOT sync until you have saved the ,txt logs and hopefully the DAT logs, from the controller or phone, to external storage e.g. your computer.
Syncing may delete the DAT and DJI might ask you for it/them.
No one, apart from DJI, can read the DAT but it is handy to have a copy of it ...... just in case.

The only possible fly in the ointment that can imagine is that you flew 5 other flights where the uncommanded ascent occurred without start to question what was going on.
But it will all depend on what the logs say. It might be an idea to upload the 5 other logs as well as the crash log,

 

[whitnasty1]

I'll upload all of the files that I see that issue, all of my microsd's have stuff
I don't want to delete on them, so I'll go grab a new one in the morning and dump all the logs onto a drive and upload... I haven't sync'd anything yet, so I can still see all of them... I was definitely questioning it after the first few times, but since I'm not super experienced with them, I just began to think that was something that happened in certain circumstances. It was definitely the most severe today and I kind of wrote it off since I never saw an uncontrolled descent, so I kind of shrugged it off... That proved to be a mistake over water today. I'm not sure if the IMU's and sensors they are using are MEMS sensors, but after building and testing thousands of boards over the years, I've definitely seen MEMS sensors stick before. Either going high or low, and that's kind of what it feels like was going on here...but I'm not sure how their circuits work exactly, so I'm just guessing here. I'll upload them all here tomorrow before I reach out to DJI, so I can know what's up going into this. But I'm certain something will come back abnormal based on my observations, it has to.

Thanks!

 

[Yorkshire_Pud]

Two questions, was the drone facing into the sun when the climbs happened and was obstacle avoidance on or off?

 

[whitnasty1]

The rapid ascent happened facing the sun, but the rapid descent occurred facing away from the sun.
I was filming in 4k/30, so from what I understand, obstacle avoidance would have been defaulted off at this point? I tested the obstacle avoidance in my back yard yesterday with a tiki torch and it seemed to work well at avoiding it.

 

[Meta4]

Anyone that has been around for a while seen or heard of this kind of problem with the Mini 3 Pro?

What you described doesn't sound normal.
Lets see the .txt file and see if it explains what was happening.

I'll upload all of the files that I see that issue, all of my microsd's have stuff

The file that's of most interest is the .txt file that is in your controller.
Don't worry about other stuff for now.

 

[whitnasty1]

Here is the log from the flight yesterday... Definitely seems like something was wrong with this unit.

DJI Flight Log Viewer - PhantomHelp.com

 

[whitnasty1]

Here is another log from my very first flight where it did something similar.... Forward motion basically dropped to 0mph and it went straight up... Upon landing, you can see the VPS is reporting accurate data to the ground, however the IMU is reading 65 feet BELOW ground.

DJI Flight Log Viewer - PhantomHelp.com

 

[slup]

Here is the log from the flight yesterday...

Nope... this isn't normal at all. Don't sync your logs to the DJI cloud service (mobile device .txt & .dat) before you've saved them to your computer, then contact DJI.

The chart below is depicting the uncommanded ascent just before that bridge you flew over before you turned around.

The blue dashed is your throttle command (value 1024 is neutral stick)
The green is the vertical speed (positive = descent & negative = ascent)
The red is the height above HP derived from the barometric sensor
The purple is the VPS height (here just flatlining on the last readable height)

Have placed the chart marker at 303,4sec, just where you've released the throttle stick... the vertical speed is there moving from negative up towards 0mph. After this nothing makes any sense... the height increases, you command descent & the vertical speed goes moderate on the positive side but not near to the speed it should considering the level of throttle command. Looking at the green & blue the drone should descend & the red should decrease but it doesn't.

Purely out from the log data (not considering what you saw happening) I say something is wrong either with the IMU or the barometric sensor. The .txt log doesn't contain enough data to say exactly what is failing... a readable DAT log is needed for that.

Here is another log...

It's the same here... a lot of very clear uncommanded vertical movements.

Here I've placed the chart marker where you don't touch the throttle at all but the vertical speed increases to the positive side (descent), but the red height increases anyway... after this it just becomes worse.

 

[whitnasty1]

That's awesome man, thanks very much for taking the time to check these logs out. Makes me feel a lot better before contacting DJI. Hopefully I don't have to pay some ridiculous fly-away fee to get another one...

I have copied all of the .txt files off of my RC controller to my Computer.

Not sure about the .dat files however, trying to locate those now.

 

[slup]

...Not sure about the .dat files however, trying to locate those now.

They should be in the same place as the .txt logs but in a subfolder named MCDatFlightRecords

The .dat logs for the flights you've posted, the correct ones ends with FLY018.DAT for the first log you put up & FLY010.DAT for the second.

If they are gone, your logs have been synced to the DJI cloud service (shouldn't be any catastrophe, as then DJI already have them).

 

[whitnasty1]

Awesome, found the .DAT files... The .DAT files for the first FLY010 is not showing up, but the most recent one from the crash is showing, and I have copied it to my computer.

Is there anyway we can view the .dat files or is that just for DJI to view?

 

[Yorkshire_Pud]

DJI only unfortunately.

 

[whitnasty1]

That's what I figured. Thanks for all of the help!

I will report back after I give DJI a call this afternoon. Hopefully have all of my ducks in a row, so it isn't a major problem with them.

 

[slup]

...Is there anyway we can view the .dat files or is that just for DJI to view?

DJI only unfortunately.

It's actually readable... don't give much more than it more or less confirms that the ascent was real.

When the uncommanded ascent started, a message is written in the dat log event stream about "[L-FDI]NS(0) GPS(0): fault on , height_drift". And then this comes when the craft start to listen to the stick command again ... "[L-FDI]NS(0) GPS(0): fault off, height_drift".

I believe that the ascent is real as the motor rpm increases when the height increases, so does the command percent from the flight controller & the current draw from the motors... all these relaxes and decrease when the craft start to respond & descends.

DJI will for sure be able to read out much more from this... but it seems like the barometric sensor had correct values & the ascent really happened, which should mean that the FC or IMU stood for the error.

 

[whitnasty1]

Thanks for checking that out!

The ascent was definitely real, as it was easily viewed from the camera. It was moving up at an extremely high rate of speed. I agree the barometric pressure sensor readings appear to be correct, I think that part was working.

The IMU is beginning to seem like that culprit, in my opinion.

 

[JimSteadman]

You have done a great job of documenting the problem and I am confident that DJI will replace your unit. Please let us all know how they deal with you. I have found them to be reasonable and prompt to resolve issues.

 

[whitnasty1]

You have done a great job of documenting the problem and I am confident that DJI will replace your unit. Please let us all know how they deal with you. I have found them to be reasonable and prompt to resolve issues.

Just got off the phone with them for the first time.... Explained everything, but I'm sure it was just a customer service rep that had no clue about what I was saying lol... He instructed me to fill out a Fly Away Service Request on the website so I am about to begin that process now.

He stated to not use the DJI Care Refresh because it would try to charge me the $235 fee for the unit "flying away", so as long as I can avoid that and just get a replacement, I'll be happy.... Even though that footage would have been sweet to have..

Thanks!

 

[Yorkshire_Pud]

@slup what version of CsvView are you using to read the .txt?
When trying to read the .txt I'm getting "internal error" messages or gibberish from 4.2.5, 4.2.6 & 4.2.7
Are you using the PH csv? If not what are you giving CsvView to process?

Oddly 4.2.7 reads the DAT but do you think the motor speeds are out by perhaps X10?

 

[slup]

@slup what version of CsvView are you using to read the .txt?
When trying to read the .txt I'm getting "internal error" messages from 4.2.5, 4.2.6 & 4.2.7
Are you using the PH csv? If not what are you giving CsvView to process?

I'm using v4.2.7 & in this incident the Airdata csv as the PH csv didn't work. Have already notified the creator of CsvView... I suspect that it's due to some kind of incompatibility when PH's msinger (member here) changed certain thing's regarding signal naming recently.