Unexpected Turn

[Wade L.]

I’m hoping one of you can analyze this log and help me figure out what happened here. Not a big deal, but as I was bringing my M3 home to land the other night, it suddenly made a quick, hard turn to the left as I was approaching my LZ in a straight line descent. I thought that maybe I had inadvertently moved my left stick, but when looking at my stick positions in Airdata when the turn occurred, that doesn’t seem to be the case.

Please jump to about 3 minutes, 52.8 seconds into the flight to see when the turn started.


Thank you in advance for your help!

 

[slup]

Sorry ... but this incident will fall into the category "Mavic 3 oddities" unfortunately. In order to maybe find a possible cause either of the 2 DAT logs would have been needed ... but they are encrypted & can't be read by us.

The TXT log mainly confirms your story of an uncommanded CCW rotation of approx 63 degrees at about 232sec into the flight. The winds were moderate & came from a northerly direction.

Have red-circled where it happens ... the purple hand drawn arrow is what the charts further below cover.



Have placed the chart marker where the yaw (purple) starts to go in a CCW direction ... at the time you had a full throttle for descent (dashed greenish) & had just stopped to ease off the forward stick command (dashed yellow)(stick commands have value 1024 as neutral) ... both the heading speed (blue) & the vertical descend speed (green) were still nearly on max... no rudder input was made (dashed black)

(Click on all charts below to make them larger)


Below I have added pitch (red) + roll (green) ... & also the total tilt angle (dashed blue) which is a combination of pitch & roll.

The chart marker is again where the uncommanded yaw movement starts ... where you still have the throttle for max descent, but just make a pause in easing off the forward stick command.

We see that the craft make big tilt changes not motivated by your stick inputs ... or the moderate wind. The pitch angle goes rapidly from 32 degrees to 0 degrees & the roll goes from leaning into the wind with the crafts right side with about 14 degrees ... to easing off to approx. 9 degrees. The dashed blue tilt shows a craft mid the uncommanded turn that is nearly leveled with only a couple degrees tilt, but still with a heading speed of nearly 40mph & a vertical descend speed around 13mph... so mainly like a frisbee in the air.



So ... in the end we see what happens & that it's uncommanded, but we have to little data to say anything about a possible cause.

 

[Wade L.]

Sorry ... but this incident will fall into the category "Mavic 3 oddities" unfortunately. In order to maybe find a possible cause either of the 2 DAT logs would have been needed ... but they are encrypted & can't be read by us.

The TXT log mainly confirms your story of an uncommanded CCW rotation of approx 63 degrees at about 232sec into the flight. The winds were moderate & came from a northerly direction.

Have red-circled where it happens ... the purple hand drawn arrow is what the charts further below cover.

View attachment 142080

Have placed the chart marker where the yaw (purple) starts to go in a CCW direction ... at the time you had a full throttle for descent (dashed greenish) & had just stopped to ease off the forward stick command (dashed yellow)(stick commands have value 1024 as neutral) ... both the heading speed (blue) & the vertical descend speed (green) were still nearly on max... no rudder input was made (dashed black)

(Click on all charts below to make them larger)
View attachment 142081

Below I have added pitch (red) + roll (green) ... & also the total tilt angle (dashed blue) which is a combination of pitch & roll.

The chart marker is again where the uncommanded yaw movement starts ... where you still have the throttle for max descent, but just make a pause in easing off the forward stick command.

We see that the craft make big tilt changes not motivated by your stick inputs ... or the moderate wind. The pitch angle goes rapidly from 32 degrees to 0 degrees & the roll goes from leaning into the wind with the crafts right side with about 14 degrees ... to easing off to approx. 9 degrees. The dashed blue tilt shows a craft mid the uncommanded turn that is nearly leveled with only a couple degrees tilt, but still with a heading speed of nearly 40mph & a vertical descend speed around 13mph... so mainly like a frisbee in the air.

View attachment 142083

So ... in the end we see what happens & that it's uncommanded, but we have to little data to say anything about a possible cause.

Thanks a million for taking a look at it! Very much appreciated! It’s good to have confirmation that the odd behavior wasn’t commanded. It’s such a pity the DAT logs are encrypted.

I really want to love this drone, but DJI has a lot to work to do to make it more stable in flight before I can get there. It's definitely better since the last FW update, but, as this example demonstrates, it‘s still quirky. I suppose I will wait for the next update to see if these unexpected behaviors are resolved. If not, I guess I will sadly be sending her back to DJI.

Thanks again for your help!

 

[MAvic_South_Oz]

Do log this event with DJI, ask for an explanation.
They probably won’t tell you the whole story, if at all, but might find something to indicate your drone should be replaced under warranty.

I think it’s rather disgraceful DJI have made the DAT files unusable outside their own resources, obviously they don’t like others having the ability to dig too deep.

 

[Wade L.]

Do log this event with DJI, ask for an explanation.
They probably won’t tell you the whole story, if at all, but might find something to indicate your drone should be replaced under warranty.

I think it’s rather disgraceful DJI have made the DAT files unusable outside their own resources, obviously they don’t like others having the ability to dig too deep.

I’ll do that! Thanks for the suggestion!

Agree 100% with your comment regarding the DAT files. What are you trying to hide, DJI?

 

[Wade L.]

I contacted DJI about this and provided the log. Their response was to perform IMU and compass calibrations and to make sure the firmware is up to date (it is, which they would have seen had they bothered to look at the log). They stated that if those steps still don’t resolve the issue, I will need to submit a repair request. I’ll perform the calibrations, although I have little faith that they will resolve the issue. I responded and asked them if this was a known issue that would be resolved with the next FW update, as I would rather wait for that than be without my drone for who knows how long while it is being repaired. We‘ll see where this goes...

 

[DougMcC]

A helicopter in rapid descent mode without much forward speed may enter a vortex ring state (VRS) where the blades become inefficient resulting in major pitch and roll changes and increasing loss of altitude. A rapid descent straight down may be the set up for this.
The vortex ring state (VRS) is a dangerous aerodynamic condition that may arise in helicopter flight, when a vortex ring system engulfs the rotorcausing severe loss of lift. Vortex ring state is sometimes referred to as settling with power. The Federal Aviation Administration (FAA) sees these terms as synonymous, whereas Transport Canada sees them as two different phenomena.[2]


Perhaps this flight entered a VRS ? Loss of control would be expected. Remember the loss of the Special Ops MH-60 in the bin Laden raid?

 

[jeff Voigt]

Doug, you sound like a former Army helicopter aviator. I’ll bet you’ve got something there with the VRS thoughts. I’ve often noted the slight instability of my drones as they fly through their own prop wash on final approach. All the best.

 

[Wade L.]

A helicopter in rapid descent mode without much forward speed may enter a vortex ring state (VRS) where the blades become inefficient resulting in major pitch and roll changes and increasing loss of altitude. A rapid descent straight down may be the set up for this.
The vortex ring state (VRS) is a dangerous aerodynamic condition that may arise in helicopter flight, when a vortex ring system engulfs the rotorcausing severe loss of lift. Vortex ring state is sometimes referred to as settling with power. The Federal Aviation Administration (FAA) sees these terms as synonymous, whereas Transport Canada sees them as two different phenomena.[2]


Perhaps this flight entered a VRS ? Loss of control would be expected. Remember the loss of the Special Ops MH-60 in the bin Laden raid?

AH HA! I bet you are absolutely correct. I was returning home (manually, not in automatic RTH mode), and, at the moment the unexpected yaw occurred, I had realized that I was overshooting my approach and eased off quite a bit on the right forward stick but kept my left stick full down. This would have certainly set up conditions ripe for a VRS, and, now that I think about it, the behavior of the drone was indeed what I would have expected in a VRS. It happened about a second or two after I eased off right stick, so I think this is a very likely conclusion.

THANK YOU for this - you’re brilliant! As weird as it sounds, it makes me feel a lot better knowing that this incident was caused by own faulty flying rather than by a defect in the drone itself.

 

[DougMcC]

Doug, you sound like a former Army helicopter aviator. I’ll bet you’ve got something there with the VRS thoughts. I’ve often noted the slight instability of my drones as they fly through their own prop wash on final approach. All the best.

Jeff, I spent two years with the 101st ABN as a Thoracic Surgeon. My best Army friend is a high time Night Stalker (SOAR 160th) aviator. I have 1200 hrs of fixed wing time and spent a fair amount of time with instructor pilots in UH-1 and the occasional Cobras. VRS was an often discussed topic amongst the real helicopter aviators. It seemed to apply here and I have entered briefly at altitude) that state with my MP2. It’s a very real issue. The bin Laden raid was a combination of rapid descent, high density altitude, entering a semi-enclosed space (walls) and rapid heating of the surround air due to exhaust thus loss of lift. Physics rules!
We’re you an Army aviator by chance?

 

[slup]

...without much forward speed may enter a vortex ring state (VRS)

The problem here in this incident & when it comes to VRS is 2 folded ...

-Firstly the horizontal speed was high when this uncommanded yaw occurred ... nearly max in Sport mode (43mph).
-Secondly, VRS just doesn't happen for DJI crafts after the Phantom series anymore ... the Mavic crafts all have design features that prevent it from occurring ... like different angles out from the horizontal plane for the 4 props & props on different vertical heights.

So think it's unlikely that this incident have anything to do with VRS ... would have been a neat explanation though.

 

[Wade L.]

Jeff, I spent two years with the 101st ABN as a Thoracic Surgeon. My best Army friend is a high time Night Stalker (SOAR 160th) aviator. I have 1200 hrs of fixed wing time and spent a fair amount of time with instructor pilots in UH-1 and the occasional Cobras. VRS was an often discussed topic amongst the real helicopter aviators. It seemed to apply here and I have entered briefly at altitude) that state with my MP2. It’s a very real issue. The bin Laden raid was a combination of rapid descent, high density altitude, entering a semi-enclosed space (walls) and rapid heating of the surround air due to exhaust thus loss of lift. Physics rules!
We’re you an Army aviator by chance?

My old man was an Army aviator in the 101st. Started in Hueys and was then selected to be one of eight test pilots for the Blackhawk in the 70’s. When I tell him about this, he’s going to chuckle and say something like, “you should have known that, idiot!”

 

[slup]

...,he’s going to chuckle and say something like, “you should have known that, idiot!”

... and then you can tell him that VRS will never happen for a Mavic with a horizontal speed of 43mph ...

 

[eEridani]

I'd think having four counter rotating propellers would make this single propeller vortex effect a lot less likely. But a test for it is simple: Ascend to a convenient altitude close enough to watch the aircraft, then descend straight down. If any significant unintended yaw is seen then flying through the aircrafts own wake turbulence might be part of the issue.

 

[Wade L.]

... and then you can tell him that VRS will never happen for a Mavic with a horizontal speed of 43mph ...

Haha! I totally get what you’re saying, and it makes sense. But I still have to believe that it was most likely VRS. I know that the logs indicated that I had sufficient forward speed to avoid VRS, but I also know that a second or two before the uncommanded yaw occurred, I did indeed back way off on the forward right stick while keeping the left stick full down, resulting in a very steep descent coinciding with a deceleration of forward motion. This is also confirmed in the log. And while I agree that the design of the Mavic in terms of different prop angles and heights would certainly diminish the chances of VRS, I also believe that there would be conditions that would still result in VRS. While the design of the Mavic may greatly reduce the likelihood of VRS, I doubt very seriously that any design would prevent it in absolutely every case, 100% of the time.

 

[Meta4]

I'd think having four counter rotating propellers would make this single propeller vortex effect a lot less likely. But a test for it is simple: Ascend to a convenient altitude close enough to watch the aircraft, then descend straight down. If any significant unintended yaw is seen then flying through the aircrafts own wake turbulence might be part of the issue.

Haha! I totally get what you’re saying, and it makes sense. But I still have to believe that it was most likely VRS. I know that the logs indicated that I had sufficient forward speed to avoid VRS, but I also know that a second or two before the uncommanded yaw occurred, I did indeed back way off on the forward right stick while keeping the left stick full down, resulting in a very steep descent coinciding with a deceleration of forward motion. This is also confirmed in the log. And while I agree that the design of the Mavic in terms of different prop angles and heights would certainly diminish the chances of VRS, I also believe that there would be conditions that would still result in VRS. While the design of the Mavic may greatly reduce the likelihood of VRS, I doubt very seriously that any design would prevent it in absolutely every case, 100% of the time.

You are both wrong.
VRS was a common thing in early Phantoms but DJI cured that when the Phantom 3 series was released.
They offset the motors from the vertical and limited the descent speed.
That eliminated VRS for DJI drones and it hasn't been seen since.

Plus it would never happen at the speed this drone was flying and if VRS was happening, it would cause the drone to fall, rather than turn.

This incident had nothing to do with VRS.

 

[Meta4]

I contacted DJI about this and provided the log. Their response was to perform IMU and compass calibrations and to make sure the firmware is up to date

That's their standard response to just about anything and indicates you were talking to someone with no idea about the incident.

 

[eEridani]

You are both wrong.
VRS was a common thing in early Phantoms but DJI cured that when the Phantom 3 series was released.
They offset the motors from the vertical and limited the descent speed.
That eliminated VRS for DJI drones and it hasn't been seen since.

Plus it would never happen at the speed this drone was flying and if VRS was happening, it would cause the drone to fall, rather than turn.

This incident had nothing to do with VRS.

How is stating a test for an effect somehow wrong? Jeesh. Grow up.

 

[slup]

...But I still have to believe that it was most likely VRS.

...a second or two before the uncommanded yaw occurred, I did indeed back way off on the forward right stick while keeping the left stick full down, resulting in a very steep descent coinciding with a deceleration of forward motion. This is also confirmed in the log....

Can't change what you believe ... but you don't find support for it out from the log though.

Not much regarding descent or heading speeds had changed when the yawing started ... if you again look at the first chart in post #2 ... (a smaller cutout here below)

You see the purple graph which shows the uncommanded yaw where the chart marker is placed.
At the same time you see that both the blue (heading speed) & green (vertical speed) hardly have changed yet.
And that the red graph shows a constant decrease in height (obviously as the vertical speed doesn't change).

 

[Wade L.]

Can't change what you believe ... but you don't find support for it out from the log though.

Not much regarding descent or heading speeds had changed when the yawing started ... if you again look at the first chart in post #2 ... (a smaller cutout here below)

You see the purple graph which shows the uncommanded yaw where the chart marker is placed.
At the same time you see that both the blue (heading speed) & green (vertical speed) hardly have changed yet.
And that the red graph shows a constant decrease in height (obviously as the vertical speed doesn't change).

View attachment 142127

Understood, and thank you again for the analysis. I’m not one to argue with data. In this case, the data explains what happened but gives no indication as to why. And so the myriad hypotheses and disagreements creep in. At this point, I’ll just chalk this one up as one of those “I’ll never know for sure what happened” things in life and leave the arguing to those with the energy to do so.