Tons of errors, disconnections, crazy fly away 1,000 feet, able to get back for scary/ugly landing - 2nd event

[StinkiePants]

2nd time this happened
I started my Mavic Pro from my wood deck as I usually do, waited for it to go into GPS mode, went up to look around, it was too dark to see anything really, so I RTH'd. But it started to come down, and headed in wrong direction, so I switched to S mode. It was erratic, but I was able to get height to miss the tall trees (100-120' trees). Off she went. Drifting away. Disconnections, errors (yaw, GPS, compass , system)... It was uncontrollable. It stayed connected for a few minutes, but eventually reconnected. It was very scary, I was sure it was going to go crash somewhere. Eventually I maneuvered it back over my yard. It was going toward the trees and pool and wherever. I got it low to the ground, and it started to go fast backwards. I tried killing it (sticks down and in), but a second or two later, it ran into the ground or tree and stopped. Doesn't appear to have any damage. Following is data file. Since 2nd time, seems unit is either defective or samething is cusing the anomaly to happen when I am hovering above yard. It's kind of rural area with, large yards, mostly wooded area. Any insight would be greatly appreciated! One more note: since receiving this drone, it turns or drifts while hovering.

DJI_ASSISTANT_EXPORT_FILE_2020-07-12_22-06-09.DAT

Drone: Mavic Pro (A few week old Refresh replacement from DJI)

Was unit in a crash? Uneventful crash landing.

What have you tried so far?: I previously recalibrated compass and IMU. I have done nothing since.

What device are you using? iPad Pro

What firmware are you running ( aircraft, remote controller)? Newest

What Go app version are you using? 4

Any modification? None

Did you change anything or install any apps? No

Do you have a video or pictures of the problem? No

This is the second time this happened. (First time was July 1st Mavic drifting, then system errors, drifted away and luckily recovered)

 

[sar104]

You still have serious compass issues. What did you do to try to fix it after I pointed those out in the previous thread? In this case, compass 0 was active.

 

[StinkiePants]

You still have serious compass issues. What did you do to try to fix it after I pointed those out in the previous thread? In this case, compass 0 was active.

I recalibrate compass, recalibrated IMU, I hovered and flew around yard a little burning through 3 full batteries. I researched how to demagnetize, but didn't actually do it.

 

[sar104]

I recalibrate compass, recalibrated IMU, I hovered and flew around yard a little burning through 3 full batteries. I researched how to demagnetize, but didn't actually do it.

Well now it's time to do it.

 

[StinkiePants]

Well now it's time to do it.

I ordered one - hopefully it arrives this week.

 

[StinkiePants]

It continues. I demagnetized my Mavic, then flew gently around the yard, I spun it around manually a few times in both directions also, and went through a couple of batteries. The data file is the most recent of 4 flights.

1. When I start to hover, it continues turning right or left and will go 15-20 degrees it seems before stopping. One hover, it just kept going left very slowly.
2. When I RTH’ed (2 times), it just started straight down, rather than returning to the actual home. I’m using RTH, because the last fly-away happened as soon as I hit RTH. I want to identify if coincidence or not.
3. On my 3rd flight, I tried a waypoint using Litchi and I paused after a couple minutes and it started hovering crazy. Looking in the same direction, but started orbiting in bigger and bigger loops. I landed safely without a struggle.
4. 4th flight, went up and hovered and it started doing bigger and bigger circles, then it was going to fly away sideways. I have a video of this. I was able to gain control without much effort and able to land safely, although everything was moving way too responsively.

I noticed on one flight I received a strong wireless interference message, which seems odd to me, in this fairly large and private area.

Any thoughts on this? Is anyone able to look at the dat file to identify the issue, please?

data file (I hope): DJI_ASSISTANT_EXPORT_FILE_2020-07-27_14-56-16.DAT

 

[StinkiePants]

Well now it's time to do it.

I demagnetized, and experienced continued erratic flights. data in separate post.

 

[Alex Baxter]

Steel screws in your wooden decking?

 

[StinkiePants]

Steel screws in your wooden decking?

Maybe.

 

[RichR]

I'm not sure if you've already done this - but could you go to somewhere that's definitely away from any magnetic influence (perhaps a park or field) and do a test flight there? Keep it low, slow and fairly close so you can land if it starts to misbehave. Then have a look a the log files from that location to see if they show any difference to when you take off from your deck?

 

[wippo76]

Normally if the compass is acting wrong before take off
Your app will not let you fly but you get an error message
But all the problems and behaviour is definitely a compass problem

We call it the toilet boil effect

 

[sar104]

Normally if the compass is acting wrong before take off
Your app will not let you fly but you get an error message
But all the problems and behaviour is definitely a compass problem

We call it the toilet boil effect

That's not entirely correct. The aircraft will produce an error if the measured magnetic field is outside a reasonable range, but if it is just distorted enough to incorrectly initialize the IMU then it will not detect that or produce an error.

And the "toilet bowl" flight path is only one specific case of an IMU error where the magnitude of the error is approximately 90°. Other errors produce different results.

 

[StinkiePants]

I sent it to DJI. We'll see what they say.