Strange Flight

[slup]

Pretty sure a compass problem will cause yaw errors when the drone pitches. The drone yawed abnormally every time it stopped, and the yaw was directional. Right headed west, left headed east. May also be an uncalibrated gyro in the IMU, but that would typically be unidirectional, not based on compass.

As said already many times before ... the compass doesn't handle the flight, it's the IMU. A compass in need of calibration will at most cause a slight crabbing in flight, not a sudden turn of nearly 90 degrees with erratic IMU heading speed readings & big velocity deviations between IMU & GPS... and all during just a short time span.

You are just stuck on the 2 short compass errors a couple feet from ground during landing & speculating wildly without any support from the log ... You're equal wrong here as you were when you speculated about loss if signal & signal interference before.

 

[eEridani]

All I am saying is the aircraft goes into a tizzy-fit any time it pitches from throttle changes.

Also, I am pretty sure the aircraft can't physically rotate 133 degrees in 0.5 seconds. An uncalibrated compass can mistake pitch changes as yaw. As can a gyro ... but wouldn't it take two or more failing gyros?




By the way, the elevator control data is lagging IMU data - which we know isn't reality - not sure why, is there a problem in CSVview or the way a Mini logs data?

 

[slup]

All I am saying is the aircraft goes into a tizzy-fit any time it pitches from throttle changes.

No ... that's not all you're saying, you're, without any log support at all, saying "-Pretty sure a compass problem will cause yaw errors when the drone pitches". You don't have a clue, you are just speculating. And that isn't helping the OP at all ... it just sending him out on a wild goose chase ...

The only time a compass can be effected by a pitch movement is when some part of the drone itself have been magnetized & by that the combined magnetic field strength of earth & the drone will change either by flying direction or drone tilt. The result in cases like that, is that the mag. field modulus will change & affect the compass differently ... this will in turn slowly feed in corrections to the IMU causing the drone to drift off course, not suddenly turn around.

That the Mini show a slight yaw instability during braking in other places along the flight path isn't anything new at all ... it's equally famous for that as for height & tilt instability due to the flimsy props... you don't need to reinvent how thing's work in order to come up with plausible explanations for these smaller deviations ... it's enough to go back and look through earlier cases regarding uncommanded descents with a Mini involved.

Also, I am pretty sure the aircraft can't physically rotate 133 degrees in 0.5 seconds.

You don't have a clue what angular rotational rate the Mini is capable of if the IMU fails ... my own Mavic Air have an controllable angular rate of 250degrees /sec ... & that is with a working IMU. My 3" quads which have weaker motors than the Mini is capable of more than 1000degrees /sec. ... so you're guessing & confusing here also.

An uncalibrated compass can mistake pitch changes as yaw. As can a gyro ... but wouldn't it take two or more failing gyros?

Can you show this thesis of yours in the log from this incident ... or in any log? Do the OP a favor ... keep it to what you can find support for through the available log, nothing more nothing less.

 

[BudWalker]

@slup @eEridani I managed to extract the .DAT from the fc_log.log file. The trick is to tell DatCon that you don't want Smart Time Axis Processing.

Here is a link to the .csv created by DatCon. You can submit this directly to CsvView
fc_log.csv

 

[SJGMoney]

Pretty sure a compass problem will cause yaw errors when the drone pitches. The drone yawed abnormally every time it stopped, and the yaw was directional. Right headed west, left headed east. May also be an uncalibrated gyro in the IMU, but that would typically be unidirectional, not based on compass.

Also, the crazy event 90 secs in was with right stick full forward, no stopping or slowing until the unexpected 90 degree turn right

 

[BudWalker]

So I took my Mavic Mini out for a lunch flight and was about 1:30 into the flight when my bird rotated 90 degrees and seemed to yaw to the right totally out of the blue. Hopefully you can see this U-turn in the picture attached, only stick inputs were forward thrust yet it rotated at least 90 degrees and seemed to change position strangely. Unfortunately I wasn't recording video at the time, but watching thru my phone was enough to get the blood pumping fast. Attached is the log, wondering if you guys see anything strange. Flew back and then continued to fly in the same general area for another 15 minutes or so with no issues so doubt it was a mechanical issue. Bumped by a bird, sudden gust of wind? In the picture attached you can see it looks like I was trying to draw a heart or something, that's when it changed course all of a sudden.

No crash or disaster but I appreciate your input.

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

I'm looking at the .DAT and it looks to me like the MM didn't actually rotate at 90 secs. Are you sure that it actually rotated or are you going off what the .txt shows?

Also, what was the heading at launch? I think it might have been -47° instead of -7° as shown by the .txt.

 

[SJGMoney]

I'm looking at the .DAT and it looks to me like the MM didn't actually rotate at 90 secs. Are you sure that it actually rotated or are you going off what the .txt shows?

Also, what was the heading at launch? I think it might have been -47° instead of -7° as shown by the .txt.

I'm going off what I saw with my own two eyes on my phone screen as I was flying lol. It was flying at about 11 oclock to my position and a second later it was facing 3 o'clock and for a minute I thought it was going down as it appeared to be yawing to the right

 

[slup]

....Here is a link to the .csv created by DatCon. You can submit this directly to CsvView
fc_log.csv

That's splendid @BudWalker ... strange that I could see 2 other flights but not that with the incident.

It really looks like the IMU is working really wonky during a period of 15sec where it turns ... it's a true yaw error created there & the deviation between the compass & IMU goes up to 93 degrees ... fully sufficient for a circular flyaway path as the one we see.

Blue bar in the sat picture= compass direction, yellow bar the heading direction of the IMU ... here where the deviation is as largest.



Comparing the gyro with the unwrapped IMUyaw (yaw angle value adding and subtracting without a locked scale in order to compare to the gyro which do the same)

They don't match ... so which is wrong.


Comparing the IMUyaw (on a +-180 degree scale) with the compass (magyaw) + the vio yaw that gives us values just where the incident happens.

The compass & the VIO yaw match ... & adding in the gyro shape from above chart I would say that the gyro match.

This indicate that it's a IMU failure that's going on during approx 15sec ... then all goes back to normal. This also indicate that the Mini never did that abrupt quick turn in reality ... instead it turned CW much slower.

Any conclusions from your side @BudWalker ?

 

[SJGMoney]

Fascinating stuff.....except that bird turned clockwise very quickly, not slowly lol

 

[eEridani]

@BudWalker - can you explain the lag/delay between the RC plots and the IMU plots?

Also - seems this sort of problem isn't new to DJI drones:

Fixing DJI's Compass Problem

UPDATE: It's fixed. Download 3.06, install, recalibrate everything and enjoy. Below is the history of the issue: As of July 14, 2014, there are 139,765 P2 users of all sizes and flavors reporting the same issue. If you are affected, make sure you say so here! Apparently Barack Obama crashed...

phantompilots.com

 

[eEridani]

No ... that's not all you're saying, you're, without any log support at all, saying "-Pretty sure a compass problem will cause yaw errors when the drone pitches". You don't have a clue, you are just speculating. And that isn't helping the OP at all ... it just sending him out on a wild goose chase ...

The only time a compass can be effected by a pitch movement is when some part of the drone itself have been magnetized & by that the combined magnetic field strength of earth & the drone will change either by flying direction or drone tilt. The result in cases like that, is that the mag. field modulus will change & affect the compass differently ... this will in turn slowly feed in corrections to the IMU causing the drone to drift off course, not suddenly turn around.

That the Mini show a slight yaw instability during braking in other places along the flight path isn't anything new at all ... it's equally famous for that as for height & tilt instability due to the flimsy props... you don't need to reinvent how thing's work in order to come up with plausible explanations for these smaller deviations ... it's enough to go back and look through earlier cases regarding uncommanded descents with a Mini involved.


You don't have a clue what angular rotational rate the Mini is capable of if the IMU fails ... my own Mavic Air have an controllable angular rate of 250degrees /sec ... & that is with a working IMU. My 3" quads which have weaker motors than the Mini is capable of more than 1000degrees /sec. ... so you're guessing & confusing here also.


Can you show this thesis of yours in the log from this incident ... or in any log? Do the OP a favor ... keep it to what you can find support for through the available log, nothing more nothing less.

Go read this:

What is the physics of tilt compensation of an electronic compass

This question concerns the physics behind the implementation of electronic compasses to find the orientation of a device. In the robotics community, 3-axis magnetometers are often used for this pu...

physics.stackexchange.com

Tilt Compensated Compass

Tilt Compensated Compass: a.articles { font-size: 110.0%; font-weight: bold; font-style: italic; text-decoration: none; background-color: red; } a.articles:hover { background-color: black; } This instructable explains how to make a tilt compensated compass using an Ar…

www.instructables.com

 

[slup]

Fascinating stuff.....except that bird turned clockwise very quickly, not slowly lol

Yep, but in comparison to what the IMU indicated it was slower

 

[BudWalker]

@BudWalker - can you explain the lag/delay between the RC plots and the IMU plots?

Could you be more specific? An actual plot with corresponding data. And, an description of what you think is incorrect.

 

[eEridani]

@BudWalker - The plot in post 22 shows the issue. But I have since looked at the actual CSV coming out of flightreader and it is skewed. So it's not a plot issue. Either flightreader is doing something odd or the Mini timing data is flawed.

The issue is that the RC elevator command should precede stopping and starting, but it lags what the drone is doing by about a second, making it look like the drone is doing something BEFORE it is commanded. I've always had a problem with anticipatory logic.

A deeper dive -- well, I was chasing a red herring thinking the odd large turn was pitch related. Seems the FC was going nuts all by itself. The initial push forward timing is normal. The spin which happened at marker was clearly abnormal and uncommanded - it precedes the elevator command. I thought it was a plot timing error.

 

[eEridani]

So I took my Mavic Mini out for a lunch flight and was about 1:30 into the flight when my bird rotated 90 degrees and seemed to yaw to the right totally out of the blue. Hopefully you can see this U-turn in the picture attached, only stick inputs were forward thrust yet it rotated at least 90 degrees and seemed to change position strangely. Unfortunately I wasn't recording video at the time, but watching thru my phone was enough to get the blood pumping fast. Attached is the log, wondering if you guys see anything strange. Flew back and then continued to fly in the same general area for another 15 minutes or so with no issues so doubt it was a mechanical issue. Bumped by a bird, sudden gust of wind? In the picture attached you can see it looks like I was trying to draw a heart or something, that's when it changed course all of a sudden.

No crash or disaster but I appreciate your input.

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

Another thing that just occurred to me that may explain some of the anomalies, is temperature.

The Mini is spec'd 0c to 40c.

Operating Temperature Range	0° to 40°C (32° to 104°F)

Weather that day was only 32 on the ground.

 

[SJGMoney]

I've flown it when it was colder. I recalibrated the IMU and have flown it since, including in the same area with virtually same flight path today, no issues.

 

[BudWalker]

I've never seen data quite like this. As others have noted it appears form several data sources, including the pilot's observations, that the MM rotated CW about 90° in about a second. But, the magnetometer data doesn't show this.

magYaw is computed by DatCon and depends only on the 3 magnetometers and the roll and pitch data. I've never seen anything like this.

Another thing to note about this flight is that it was launched from a geomagnetically distorted location. This can be seen by looking at the magMod data as the MM ascends through the 1st meter.

magMod is the X,Y and Z magnetometer modulus and is proportional to the geomagnetic field strength. Without distortions it should be constant. Here, while sitting on the ground magMod is 4800 - way above the normal 1500. As the MM ascends through 0.3 meters the magMod drops to 495. Then it rises to a more normal as the MM continues to ascend through 1.0 m.

I suspect that the .DAT extracted from the fc_log.log is identical to that of the on board .DAT. AFAIK, this is the only time this type of .DAT exists for a MM incident that involves geomagnetic distorted launch site.

 

[BudWalker]

Pretty sure a compass problem will cause yaw errors when the drone pitches. The drone yawed abnormally every time it stopped, and the yaw was directional. Right headed west, left headed east. May also be an uncalibrated gyro in the IMU, but that would typically be unidirectional, not based on compass.

That's because negative rudder is (inadvertently?) being applied when the pitch is reduced from full down to neutral.

 

[SJGMoney]

Interesting observation about the geomagnetically distortion, I referenced the material that is on floor of this parking garage I launch from being some weird stuff. Yesterday I calibrated the compass before launch and hand launched and didn't have any issues at all. Of course I also re-calibrated the IMU the day before so I don't know which one helped the most.

 

[BudWalker]

I extracted more record types. For those not wanting to slog through getting this.
fc_log.all.csv