Yaw Drift re-examined

[JDNC]

After reading other threads that suggest leveling the IMU inside the bird due to DJI's bad installation I decided to give it a try. After receiving 3 quality circular levels as I wanted very accurate levels I preceded to go into my bird. My bird had been drifting left about 6-10 ft when yawning to left and also yawning right it appeared to be drifting but not as bad. Also small winds influencing this somewhat.

Anyway I first performed an advanced IMU calibration using a 13"X13" 1/8" thick piece of glass across the motor shafts first as I didn't really want go inside the bird. This brought the drifting inside of 4 ft I would say and I was thinking I would leave well enough alone but I just couldn't. My first crash was because of this yaw drift and my inexperience in flying the thing!

Today I went inside the bird and after leveling I have reservations on performing an advanced calibration by leveling the unit inside. As I was waiting on a battery to charge I noticed something maybe no one has noticed and would really like input from you guys. Some have said that DJI has done a poor job at installing the IMU and that due to it not being level is what's causing this yaw drift. I agree that it's caused by the unit not being level, but I think the reason is that the circuit board underneath is swaying in the center due to weight of components.

If I calibrate by the IMU unit will this cause the two left motors to increased RPM all the time? I would rather deal with the yaw drift than put more load or stress on two motors, not to mention the bird not flying level. Am I looking at all this wrong?

 

[JDNC]

Well, I bit the bullet and done an advanced calibration. I can always do another with the glass on the motor shafts later. I sure hope it doesn't tip on take off now!

After looking again its hard to determine if the board is low in the center or not. If it isn't I can't tell if the mount tape is compressed on the edge near the center or not. One thing for sure, the IMU is not level with the motors!

 

[RichWest]

Just throwing this out there... If I was to level the unit I would remove the variable of the legs. Once your platform is confirmed level, I would support the craft at the point of the motors (underneath). I believe this is the relevant reference plane for what you are trying to accomplish, the motors/propellers.

Further, I believe the IMU will adjust to what ever you consider level when calibrating. So additional hardware, like antennas or tracking modules will change the balance point reference. Just a variety of factors that can influence your outcome.

I have one bent arm and the propeller's orientation is noticeable off by 1-2 degrees. It has not been a factor to any "balance" issues, it yaws well and stays level.

 

[BioTeq]

Unless the box is actually crooked after a crash trying to level it is just a waste of time. That's what the calibration is for.
Calibrate the bird on a flat surface and it will be enough, all offsets will be incorporated into the position calculations.
Ultimately, if external disturbances are eliminated, it will be the poor sensor quality that's responsible for the drift.

 

[vico36]

Hey, I'm also experiencing yaw drift and it's driving me mad!
When you say "way off level" in your first image, only one of them is way off, is the imu unit right below this level? it's my undestanding that the imu unit is somwhere inside the naza module.
Let us know the results, I really want to fix this yaw drift too :/

 

[JDNC]

Hey, I'm also experiencing yaw drift and it's driving me mad!
When you say "way off level" in your first image, only one of them is way off, is the imu unit right below this level? it's my undestanding that the imu unit is somwhere inside the naza module.
Let us know the results, I really want to fix this yaw drift too :/

I posted the picture to show just how far off the motors are off level when the IMU unit is level. The larger circular level is sitting on glass which was placed across the motor shafts.

I later performed an advanced IMU calibration with the setup in the picture (IMU level). I just got in from attempting to fly and got the results I thought I would. The bird tried to tip over during lift off. I mean pretty bad, so stopped and come back in. I'm going to place the glass across the motors and level as I had done prior and re calibrate. Then if it's flying ok as it did prior, I'm going to leave it the hell alone and fly it.

It was a learning experience for me but most just a waste of time. I will say, I think someone's idea of using a sheet of glass across the motors is probably best. At least it reduced my yaw drift from several feet to only about 3 ft at most.

 

[vico36]

Ah ok I didnt realize there was glass under the bigger level in the first pic. Makes sense now.
Ill probably give the glass on the 4 motors a try.

 

[JDNC]

Just throwing this out there... If I was to level the unit I would remove the variable of the legs. Once your platform is confirmed level, I would support the craft at the point of the motors (underneath). I believe this is the relevant reference plane for what you are trying to accomplish, the motors/propellers.

Further, I believe the IMU will adjust to what ever you consider level when calibrating. So additional hardware, like antennas or tracking modules will change the balance point reference. Just a variety of factors that can influence your outcome.

I have one bent arm and the propeller's orientation is noticeable off by 1-2 degrees. It has not been a factor to any "balance" issues, it yaws well and stays level.

That is what I'm trying to do with the glass and it appears that you're right that the motor and props should be the platform standard base line.

To get the NAZA unit level it to approx 3/16" under the left side legs to level. As you can see the larger circular level was thrown completely off. I just had to know so I advanced calibrated at this point. As I thought my bird almost tipped over attempting to take off so I stopped. Never got off the ground. I might add the take off platform was level.

I then calibrated again using the top of the motor shafts as a platform baseline using the glass. I just got back in from flying and I'm back in the ballpark but not as good as I've had it. I guess it's pretty sensitive to error. It will turn quickly within probably a foot circle, but when turning slowly it will drift 4-8 ft. I hovered in ATTI to see how much it would drift in a small breeze and it was very little.

There is no doubt (in my mind anyway) that the problem is the naza unit not being level with the motor/prop platform.
Do I want to remove the naza and level it? I don't think so, I'll just deal with it.

Thanks you guys for the input and anymore if you have it.

 

[danillll]

Very interesting discussion and findings. So the main thread, where people suggested to level the IMU or use stack of papers, never mentioned off balanced motors and the risk of tipping on take off, but it makes sense.
So I guess your best bet is to lift the IMU until it gets leveled with the motors/glass and recalibrate.

BTW where did u get the circular levels from?

Looking forward to hear about your findings, good luck!

 

[danillll]

..... One more thing, make sure to have the battery installed when you level, at least on mine, once the battery is installed it pushes the board a little bit.

 

[JDNC]

Very interesting discussion and findings. So the main thread, where people suggested to level the IMU or use stack of papers, never mentioned off balanced motors and the risk of tipping on take off, but it makes sense.
So I guess your best bet is to lift the IMU until it gets leveled with the motors/glass and recalibrate.

BTW where did u get the circular levels from?

Looking forward to hear about your findings, good luck!

I really don't know why I didn't get the results others have. But in my mind if you lean the bird by a great amount and the IMU thinks that is level, then when the bird IS LEVEL, the IMU is going to try to adjust it to what it thinks is level. If that makes any sense at all. When I attempted to take off, by bird was violently trying to tip so I just shut it down.

As for the levels I would recommend https://www.geier-bluhm.com/index.php. They are the best in my book. I use them for leveling parts prior to machining. I have several and they agree with each other. I can't say that for others. My $.02.

 

[danillll]

Thanks for the link JDNC, did you end up lifting the IMU to level it?

(BTW, is it JD North Carolina? I am in the triangle area)

 

[JDNC]

No, I didn't lift it, that's a future project. For now I just leveled the motors and calibrated and will deal with the drift.