Forced land at 50%. Why?

[Krag]

So we lost a Mavic today and I would like to figure out exactly why. Totally our fault. We were caught unprepared by the autoland and didn't cancel it before the drone dropped below the trees. But the reason why it went into auto land is what I want to know. This is a complicated one so bear with me.

A friend and I were out doing distance runs with virtually identical Mavics. The first run we used a recent DJI Go 4 and were also using the DJI goggles. The flight is about 2 miles up the side of a mountain with a significant height climb. On my trip out I get a warning at around 55% battery that says, effectively, "you only have enough battery to return home, starting RTH unless canceled". So using the Goggles I cancel the warning. I know I am going "uphill" and will have plenty of battery to get back. I finish the flight with 40% battery, no problem.

Now my friends turn. Everything identical except he is using Litchi. We are still using the Goggles (one person observing). Now on the way out we get the warning again about RTH unless canceled. We cancel in the goggles again. Everything seems fine but 10 seconds later at exactly 50% battery there is a warning in the goggles that says something about battery and starts auto landing. Due to where it was we only had a few seconds to react and were unable to stop it from dropping beyond line of sight and it presumably landed right where it was. Not recoverable until at least the snow melts.

We have the Litchi .csv logs in the phone and the DJI .txt logs from the goggles (thank god for that btw). I can't share them yet because they would pinpoint the location of my friends drone and he wants to try to get it.

Battery problems seem obvious but it was a good freshly charged battery and looking at the logs shows NO battery problems. When we lost contact it said 48% battery and all cells had steady 3.7 voltage on all cells and no spikes.

As far as the messages go Litchi doesn't seem to record them, at least in the csv. The goggles logs also don't seem to show verbatim messages. The message in this log from AirData is "50% Battery at maximum distance" just before it started its autoland.

Flight time    Altitude    Home Distance    Type    Notification
A    00m 00s    0.0 ft    0 ft    Mode    Mode changed to Motors_Started
B    00m 00s    0.0 ft    1 ft    Tip    Setting new Return-To-Home altitude to 30m (98 ft). 
     00m 00s    0.0 ft    1 ft 

100% Battery
C    00m 01s    0.0 ft    1 ft    Mode    Mode changed to Assisted_Takeoff
D    00m 01s    0.0 ft    1 ft    Mode    Mode changed to GPS_Atti
     01m 36s    x167.6 ft    1,526 ft 

90% Battery
E    01m 56s    610.9 ft    ,155 ft    Data Loss    Downlink data connection lost for 1 seconds
F    02m 00s    673.5 ft    ,248 ft    Data Loss    Downlink data connection lost for 1.1 seconds
G    02m 00s    673.5 ft    ,248 ft    Mode    Mode changed to WiFi_Reconnect
H    02m 01s    676.2 ft    ,269 ft    Mode    Mode changed to GPS_Atti
     02m 42s    536.4 ft    ,464 ft 

80% Battery
     04m 35s    463.2 ft    ,721 ft 

70% Battery
     06m 55s    947.8 ft    ,266 ft 

60% Battery
     08m 52s    138.1 ft    ,293 ft 

50% Battery
     08m 54s    138.1 ft    ,296 ft 

50% Battery at maximum distance
I    09m 04s    138.7 ft    ,276 ft    Mode    Mode changed to AutoLanding
J    09m 16s    063.3 ft    ,272 ft    Mode    Mode changed to WiFi_Reconnect
K    09m 17s    049.5 ft   ,273 ft    Mode    Mode changed to AutoLanding
L    09m 18s    033.1 ft    ,273 ft    Mode    Mode changed to WiFi_Reconnect

So what happened? What combination of factors resulted in it deciding to land where it was. I think in part it was the low battery behavior that should have caused it to go into RTH somehow made it decide that because it didn't have enough battery to get home its best course was to land now. Or was it because the RTH was canceled in the goggles but not in litchi? We are not totally sure we saw even saw the RTH warning on the litchi screen. It may have only been in the goggles.

As I said. Complicated. Anyone hear of anything like this?

 

[vashon100]

48% and 3.7v doesn't sound consistent. Isn't 3.7v somewhere around 30% (or less)?

 

[Krag]

Is it? I thought it was about right for 50%. They started at about 4.2v and dropped steady. AirData says no sudden spikes. And it is consistent with my successful flight where voltages were also 3.7 at 50%.

Last 5 cols are Total Volts, Cell 1v, 2v, 3v and voltage change.

9m 16.2s P-GPS 20satellites ,138.0ft 5.1mph ,283.9ft 49% 11.152V 3.724V 3.726V 3.702V 0.024V
9m 17.7s Auto Landing 20satellites ,138.0ft 6.1mph ,274.3ft 49% 11.152V 3.724V 3.726V 3.702V 0.024V

 

[RayOZ]

Not sure if the mavic knows it's going up the mountain. Surprised it let you fly past 1640ft from launch altitude. I wonder if it calculated it needed 40% battery to get down from 4100 ft to 0 ft.

 

[Krag]

Yes, it does a calculation of that sort. Hence the warning about RTH, that was then canceled. I can understand it wanting to RTH. The landing part I don't understand yet.

I don't think its really the intended behavior that "oh I can't get home, guess here is as good as anywhere to land".

 

[Keule]

Without having the flight logs, it's just guessing or looking into crystal ball.
The variables of interest are all battery and SMART_BATTxxx related variables.

 

[Kilrah]

Yup, not much can be said without logs, ideally both his and yours for comparison.

The smart battery system is pretty complex and takes a bunch of things into account. If it saw an abnormally faster drop before that time it can take that decision, so if say his battery is a bilt older/worn it could explain the different behavior.

 

[RayOZ]

And since you mentioned snow, I guess the cold does affect battery too.

 

[Akoztha]

I lost mine the other way around. Flying off a cliff which is 1200 meters above sea level.
My drone started to go down (to level up with the 500 meters limit).
The Mavic will "even up" with altitude on those scenarios
Maybe that's how yours got lost.

 

[Krag]

Surprisingly the Mavic has already been recovered. Once it is down off the mountain and dried out I'll take a look at the logs and see if I find out anything.

 

[Kilrah]

My drone started to go down (to level up with the 500 meters limit).

That is not correct. 500m are above takeoff point.

 

[Grizzo]

When the drone is auto landing dont you still have control to move where ever you want? You still could have flown back if you had battery power?
Im guessing for whatever reason, the battery went to zero.

 

[Krag]

We might have been able to stop the autoland but we were surprised and didn't even try until it was too late.

 

[Krag]

Autoland made me crash once as well. But it didn't happen with 50% battery. Since I fly a lot doing range tests with external batteries, I changed the flight parameters so that it will not descend at all for autoland.

Ah! What parameters specifically?

 

[Krag]

Alright! Now we are getting somewhere. Half of the mystery is solved. I just check the drone that made it and it had the battery parameter changes and the other one did not. That's probably why one didn't made it and the other didn't. I am still not sure why it would try to land at all though. Hopefully the logs will solve that part.

 

[Divebomber]

And since you mentioned snow, I guess the cold does affect battery too.

I work with back up batteries in data centers for a living. Batteries are definitely affected by temperature. You also have a situation you may not have considered. For each 1000' feet altitude gain, the temperature in still air drops between 2 and 3 degrees F. Additionally, if you were next to a mountain with snow on the slopes above you, but you were in warmer air, you were probably dealing with "downslope wind", meaning colder air from the snowy area was blowing down slope to warmer air. Colder air is heavier than warmer air, and will sink down slope, causing a breeze or wind, depending upon atmospheric conditions. What this phenomena does is pulls the even colder air from the higher altitudes down to lower altitudes. You didn't mention how cold it was, but you did mention snow, so I assume it must have been near freezing. Batteries lose power quickly depending upon how cold they are. If you flew high enough to get the battery cold enough, it could have easily lost power very quickly to the point that the drone performed a critically low battery power landing. Logs will tell the tale, maybe, depending upon how quickly the battery lost power.

 

[NickBuckley]

I work with back up batteries in data centers for a living. Batteries are definitely affected by temperature. You also have a situation you may not have considered. For each 1000' feet altitude gain, the temperature in still air drops between 2 and 3 degrees F. Additionally, if you were next to a mountain with snow on the slopes above you, but you were in warmer air, you were probably dealing with "downslope wind", meaning colder air from the snowy area was blowing down slope to warmer air. Colder air is heavier than warmer air, and will sink down slope, causing a breeze or wind, depending upon atmospheric conditions. What this phenomena does is pulls the even colder air from the higher altitudes down to lower altitudes. You didn't mention how cold it was, but you did mention snow, so I assume it must have been near freezing. Batteries lose power quickly depending upon how cold they are. If you flew high enough to get the battery cold enough, it could have easily lost power very quickly to the point that the drone performed a critically low battery power landing. Logs will tell the tale, maybe, depending upon how quickly the battery lost power.

I would definitely support this as a theory - really interested in what the logs tell the OP

 

[Krag]

I don't think temp was a factor. The battery was kept at room temp (home and then car) from the time it was charged until it was used. It was mid 50's at the bottom and mid 40's at the top.

I got the logs late last night and haven't had time to analyze yet. I did see a "smart battery error" but I am not sure what that means yet. Whatever the problem was it could not have been too bad because the Mavic then spent 15 minutes hovering and fighting wind until it shut down because of low battery. We got a nice time lapse out of the deal.