DJI Air 3 suddenly started to drift with the wind and did not obey the controls

[jarimy]

My guess is that something went wrong when the software ended the pano mode which limited the max tilt angle... which in turn caused a blow away with a unresponsive stick feel. This glitch ended when the RTH was manually initiated

This was a pretty in-depth analysis! If I understand correctly, it's also possible that there's a software bug in the firmware that manifests itself under certain circumstances?

 

[slup]

...If I understand correctly, it's also possible that there's a software bug in the firmware that manifests itself under certain circumstances?

Yeah, that's my guess. The drone was blowing away, but not due to a too strong wind, instead the drone was limited in tilt which made it unable to fight the wind & hold position.

If comparing the brownish area (where you had problems) with the blue area where you had initiated RTH, the drone easily reached speeds there up to 17-19mph into the wind on the way back home by utilizing tilt angels between 25-30 degrees... during the incident phase the tilt was in the ballpark of only 10 degrees & the drone drifted away in the wind direction with approx. 5mph.

A SW glitch during the deactivation of the pano mode could be the reason for the limited tilt...

 

[msinger]

During period 1 the drone slowly ramp up the speed to approx. 5mph uncommanded, here the drone tilts correctly into the wind with intention to hold position... but reaches only a tad above 4 degrees (of maximum 36 degrees) which isn't sufficient to prevent wind drift.

The drone started to "drift" about 1 second before it switched from Pano to Sport mode.

Prior to that drifting, the drone was at about an 8-11 degree angle (pitch) since it wasn't facing the wind at that point. After the mode switch, the pitch and wind remained fairly consistent.

 

[slup]

The drone...

Meaning..?

 

[msinger]

Meaning..?

You mentioned that "the drone was limited in tilt, making it unable to fight the wind and hold position." However, that doesn't seem to be the case here, as the drone maintained its position just fine while tilted at the same angle before the flight mode change.

Do you have any thoughts on this? Or could you elaborate on your tilting theory in more detail?

 

[slup]

...could you elaborate on your tilting theory in more detail?

Exactly where the drift starts, the tilt angel (combined pitch & roll) change from an earlier average of 14 degrees while holding position, to a rather constant tilt just a tad below 10 degrees. This seems to generate a drift reaching 5mph during those 7sec it took before the OP started to command the drone.

The tilt change could correlate to that drift speed & that it happens just in the end of the Pano phase could mean that the tilt change was SW related.

 

[msinger]

The tilt change could correlate to that drift speed

It's an interesting observation, but I can't determine whether it's the cause or merely a coincidence. I suspect the drift was caused by a change in wind speed without the proper counter adjustment in tilt, but without knowing the exact wind speed at that moment in the flight, there's no way to confirm.

 

[slup]

It's an interesting observation, but I can't determine whether it's the cause or merely a coincidence. I suspect the drift was caused by a change in wind speed without the proper counter adjustment in tilt, but without knowing the exact wind speed at that moment in the flight, there's no way to confirm.

If comparing the difference in IMU & GPS velocity during that whole Sport mode period between the end of the pano & RTH initiation, we see that the differences are nearly zero... this indicate that the FC very well knew that the drone was drifting uncommanded. This should in normal cases, with a good quality GPS lock, mean that the FC should command a steeper tilt into the wind direction to hold position... the tilt would continue to get steeper & steeper to hold position until the drone reaches it's max allowed tilt, and first after that the drone would start to blow away if the wind was strong enough.

What tells me that something fishy is going on is mainly 4 indicators.

1. it happens just in the end of an automated function that " controls the flight"... the pano mode.
2. the tilt change just when the drift starts... a tilt that is pretty consistent both before & after.
3. the drone seems to get blown away directly in the wind direction, this without not even trying to utilize the full tilt angel.
4. as soon as another automated mode is initiated (RTH) the drone snap's out of the seemingly restricted tilt angel & start to use angels up to 30 degrees, both to stop the drift & to reach a speed home up close to 19mph.

This incident wasn't at all pilot related & the wind speeds wasn't at all up on levels that would risk a blow away for an Air 3 during the whole flight.

The data available in the .txt log isn't enough to reveal the true cause unfortunately... we can only guess.

 

[msinger]

This incident wasn't at all pilot related & the wind speeds wasn't at all up on levels that would risk a blow away

Nope on both counts. However, an increase in wind speed likely caused the drone to drift for a while.

Regardless, it seems DJI has a firmware bug that needs fixing.

 

[Felix le Chat]

Nope on both counts. However, an increase in wind speed likely caused the drone to drift for a while.

Regardless, it seems DJI has a firmware bug that needs fixing.

Seems like 'firmware bugs' are becoming a little too frequent of late... a bit like DJI dropping the ball to AUTEL levels... the picture quality on the Air 3S being a case in point.

 

[Zeprider]

Mechanical and firmware aside. I use Airdata UAV to check flying conditions; I'm sure many here already do. Follow the app's advice.

My personal recommendation for larger DJI drones (M3Pro), avoid flying in winds exceeding 25mph. Smaller drones limit flying in winds above 23mph. These figures are according to Wind Resistance specs. Mother nature's a ***** and she will steal your equipment.

 

[MS Coast]

Mechanical and firmware aside. I use Airdata UAV to check flying conditions; I'm sure many here already do. Follow the app's advice.

My personal recommendation for larger DJI drones (M3Pro), avoid flying in winds exceeding 25mph. Smaller drones limit flying in winds above 23mph. These figures are according to Wind Resistance specs. Mother nature's a ***** and she will steal your equipment.

Given the variation of wind speed with altitude, land forms, buildings, and local airmasses, and considering the difference in local speed from the forecast station, I really doubt that a 10% difference in the forecast speed has much significance.

What do the wind resistance specs mean, anyway? I'm not aware that DJI offers an explanation.

 

[Zeprider]

Given the variation of wind speed with altitude, land forms, buildings, and local airmasses, and considering the difference in local speed from the forecast station, I really doubt that a 10% difference in the forecast speed has much significance.

What do the wind resistance specs mean, anyway? I'm not aware that DJI offers an explanation.

Get the local stations reading, then use your judgement to fly. Bring an anemometer, they are truly cheap. Now I can only presume that the wind resistance variable is the point at which a drone will begin struggling to resist the head winds. The drone is overpowered and is stopped or carried away. Perhaps someone flew multiple DJIs to determine when they are losing the battle. Oh, and...

DJI Drones Compared

 

[slup]

...Perhaps someone flew multiple DJIs to determine when they are losing the battle.

This isn't a mystery at all...

A Mavic 3 Pro have a max speed of 21m/s according to spec... a direct headwind of 21m/s will then stop it, & faster winds will push it backwards. DJI states 12m/s as "Max wind speed resistance" ?

A Mini 4 Pro have a max speed of 16m/s according to spec... a direct headwind of 16m/s will then stop it, & faster winds will push it backwards. DJI states 10,7m/s as "Max wind speed resistance" ?

 

[Meta4]

I can only presume that the wind resistance variable is the point at which a drone will begin struggling to resist the head winds.

This is as vague and meaningless as DJI just giving a number with no definition or explanation.
How would you define the point at which the drone begins to struggle to resist a headwind?

The drone is overpowered and is stopped or carried away.

DJI's mysterious and undefined "max wind speed resistance" is well below the wind speed at which your drone is overpowered and stopped or carried away.

 

[Meta4]

A Mavic 3 Pro have a max speed of 21m/s according to spec... a direct headwind of 21m/s will then stop it, & faster winds will push it backwards.

It's not that simple.
The Max Speed in still air that's shown in the specs is misleading.
It's not the maximum speed the drone could achieve, it's the max speed that DJI allows the drone to achieve.
Look at the flight data for a Mavic 3 (or any other DJI drone) at max stick in still air, the drone is taking it easy at much less than max tilt angle.
You can still achieve the same max speed flying against a significant headwind.
The drones are throttled back and still have a good reserve when flying at max speed in still air.

 

[slup]

It's not that simple... it's the max speed that DJI allows the drone to achieve ...in still air, the drone is taking it easy at much less than max tilt angle.

I think it's that simple in practise...

DJI photo drones have always been limited in tilt+speed by the firmware. The drone will be able to make headway at max speed (here meaning ground speed) all the way up to when the (fw limited) tilt angle is fully utilized, here the ground speed will start to be affected (and yeah... this can maybe be called "a reserve"... or a delayed affect on ground speed).

When the drone doesn't have a ground speed anymore, the full (fw limited) tilt is usually utilized... the air speed here is then very likely equal to the specified max speed in still air, which in turn then is equal to the speed of the headwind.

In order to calculate all this in more detail the exact wind speed is needed... & that data we rarely have, so the specified max speed in still air is the only data we have available.

*Below added*

Another way of looking at it & relate to the specified max speed in still air...

A Mavic 3 Pro have a stated max speed of 21m/s... that's approx. 47mph or Beaufort 9-10. Sustained windspeeds on that level are classified as a storm. It's quite logical that a Mavic 3 Pro is stopped up on headwinds like that (& in reality have started to drift backwards in gusts stronger than that)... & at the same time it's really impressive. Wouldn't count on that a Mavic 3 Pro have further capacity to fight even stronger winds.

 

[DaHawaiian]

I had a similar experience with my Mavic 3. It took me a while to figure out it wasn’t the drone, but the wind at 300’ was obviously much stronger than where I was standing. Good thing RTH had 25%. I needed all of it. I could hear the motors fighting, loosing. After panicking, I tried to go straight down to confirm my wind theory, it obeyed that command. I then guided it home — it was straining like it was being tortured. Out of nowhere, the wind picked up considerably. It was running out of battery, obviously not going to make it back. I found a decent LZ not far away, guided it down to a hard but OK landing.
At that point, the wind was blowing so hard I could barely walk to get it, a thunderstorm cell appeared out of nowhere and dumped heavy rain. That’s not uncommon for desert weather, but it caught me off guard. Scary. I now read much more detailed weather forecasts, and I’m more conscious of wind and condition changes.

 

[jarimy]

At that point, the wind was blowing so hard I could barely walk to get it, a thunderstorm cell appeared out of nowhere and dumped heavy rain