Maiden - Copter spins on takeoff

So I maidened a quad today and was happy to see it get off the ground but once in the air it started into a spin CCW I beleive. IT ignored any YAW input and I had to land it. It did this each time I got it off the ground so not a successful maiden
It also started tossing all sorts of EKF errors. Indicating that the YAW was off by so many degrees. When I got home and connected it to the PC I can see in the param file it appears to have changed compasses. Thats a new one on me.
Anyway if anyone can have a quick look at the logs and tell me if they see anything that would be helpful and I would be extremely grateful.

Please retry with copter 4.0.4-rc3. Or are you already using that?

Why would I want to try a newer version unless a yaw issue is a know problem in the version I am running.
Currently I am running this version ArduCopter V4.0.3
Also strange observation. I am using a Zubax GNSS V2 on the canbus… So I was told the compass is roll 180 so I set that in mission planner. Do a calibration and when done I reboot it. It shows Yaw 90… I try it againm set to Roll 180 do the calibration. Reboot and it goes back to Yaw 90.
The compass appears to be pointing in the correct direction when I turn the quad around. But it’s Yaw 90…um what the heck. Does this GPS have some function that tells Arducopter what the compass orientation is. That would be new to me for sure.

There are multiple compasses bugfixes on 4.0.4. Did you read the release notes?

bugfixes are good.,
Are you referring to this fix.
a) Compass ids from missing compasses reset after compass cal
This could be the problem in that the Devid for the compass seems to change as if I am moving to a different compass.
I will flash it today and see if that helps.
I also checked the motors to see if they are level. Seems good.

@amilcarlucas That yaw has noting to do with the compass, and recommending the use of a non-stable version of ArduCopter as a troubleshooting tool is down right wrong. A developer should know that.

@rickg32 You need to use something other than Drop Box. Try uploading .bin (binary) files here using the Upload function. “Seems OK” is not good enough. The motors must be exactly square with the frame or you will get a torque yaw. Frames with tubular arms are notorious for this. BTDT

One thing you can look at are the RCOUT/C1-C4 plots. Ideally the plots should overlay each other.

1 Like

Hey @OldGazer. I agree about the Yaw not being compass related. I believe the compass issues perhaps are resolved by the new code. It’s not making much sense to me the way the compass changes.
These are the errors I see.Also the compass orientation changes as does the id…so the new version may help with that.
|09:53:04|PreArm: DCM Yaw inconsistent by 166 deg
|09:53:34|PreArm: DCM Yaw inconsistent by 43 deg
|09:55:00|PreArm: EKF2 Yaw inconsistent by 66 deg
|09:55:04|PreArm: EKF2 Yaw inconsistent by 63 deg
|09:55:13|PreArm: EKF2 Yaw inconsistent by 54 deg

As for the motors being square and level I get your point. I have checked them before and again today but will give them a good going over sometime today. I know that a small out of level can have drastic affects on the yaw.
AS for Dropbox, I have had to use dropbox because I always get a size error when I try to attached a bin file even if I zip the file.


The file I am attaching isn’t bigger then 4500kb.
image

There’s some strange stuff going on with logging of yaw and a bunch of other data too.
Set LOG_BITMASK,141310 and worry about FFT’s and harmonics after this flies better.
Roll and Desired roll are a bit crazy, but strangely pitch is pretty much perfect - although with these “flights” it is very hard to tell anything.
The RCouts are telling:


CW motors are going to near maximum and CCW motors are practically at minimum.
Maybe check prop balances too, but we’ll be able to tell more with subsequent flight.s

Hey thanks @xfacta Shawn
Ya the flights are so short in duration that its hard to tell whats going on.
I can see in the log the variation between the CCW and CW motors.
I have no clue what would drive that. The general thinking is a tilted motor but the CCW motors are definitely being driven harder then the CW which is in line with what I saw. A CCW rotation of the craft.
So I will check the motor level and prop balance…but I need to figure out whats driving the motors like this.

As for harmonics. I am not to concerned with them at this point. I need it to hover first.
I pulled the props and will check them all for balance. I discovered this morning when I got back that I had 6 props and had balanced 4…but I don’t know if I put the balanced 4 on…so will check that.
Will tweak the logging as well.
Going to redo the ESC calibration as well just to be safe.
I am 3d printing a jig to ensure the motors are level as @OldGazer pointed out.
Any insight into these errors
|09:53:04|PreArm: DCM Yaw inconsistent by 166 deg
|09:53:34|PreArm: DCM Yaw inconsistent by 43 deg
|09:55:00|PreArm: EKF2 Yaw inconsistent by 66 deg
|09:55:04|PreArm: EKF2 Yaw inconsistent by 63 deg
|09:55:13|PreArm: EKF2 Yaw inconsistent by 54 deg

Also any thoughts on the compass shifting roll 180 to Yaw 90 or the devid changing after a reboot…this sounds like a bug that one of the devs pointed out.

I was looking some more and see some of the compass things your saying.
I think Yaw inconsistent messages might go away after it’s all set up and calibrated correctly (whatever that is :slight_smile: )

I know Compass 2 (internal) is disabled for use, but it’s still a good comparison. You can see where that MagZ and Mag2Z are going in opposite directions


Roll 180 was set, so that might be wrong, probably redo the compass calibration as per below and lets see a new log and what this graph looks like. COMPASS_AUTO_ROT should save the correct orientation.

  • Get the craft outside up off the ground on a wooden table or similar. Leave it sit powered on for some time to get a good solid 3D fix and temperatures stabilized. Then redo the compass calibration - if you haven’t already done it that way.

  • Test: Power on with ground station connected, NOT facing north - does the HUD indicate North? or does the HUD indicate the true heading (not north). HUD follows true direction.

Compass/Motor calibration might be required too - the two compasses are badly affected by throttle output (motor current).

https://ardupilot.org/copter/docs/common-compass-setup-advanced.html#common-compass-setup-advanced-compassmot-compensation-for-interference-from-the-power-wires-escs-and-motors

COMPASS_AUTO_ROT …wow this is new…that explains the constant shifting of the compass orientation. I will tweak the various items you outlined and then get her outside for the compass work.
If needed I will also arrange to do a compassmot. This never made me comfortable doing it…but when I last was flying this drone I had to do it just to solve a compass issue…so will do it again.

Thanks again. Will let you know how it goes.

COMPASS_AUTO_ROT geeze…missed that.

1 Like

There’s no compass/motor settings saved in that log, so if you did run the procedure on this craft it didnt save.
I find it’s easiest if you get some volunteer (preferably crazy) to hold your craft above their head, with ground station connected (USB lead secured away from props if you use USB) and run the compass/motor calibration that way. Not good with really big craft, and you’d have to follow the documentation for swapping the props around (or if you cant find anyone crazy enough to help).

haha no crazy people I can use…but no tthe compassmot was done on this drone a couple years ago. I have since wiped the fc and upgraded it… I wanted to start with a clean install…so this is why yo don’t see a mot in the params.

But I will do it again. I have a frame that I can bolt the drone to…holds the arms in place and then I don’t need to worry about something odd happening and yes thats with the props reversed.

[quote=“xfacta, post:12, topic:59952, full:true”]

I find it’s easiest if you get some volunteer (preferably crazy) to hold your craft above their head, with ground station connected (USB lead secured away from props if you use USB) and run the compass/motor calibration that way. [/quote]

That idea is stupid and dangerous. The documentation clearly states the safe and proper method for running Compass/Mot is to turn the props over and move them so that Prop #1 goes on Motor #4, Prop #4 goes on Motor#2, and Prop#2 goes on Motor#3 and Prop#3 goes on Motor #1. This way the props are pushing the aircraft into the ground and you don’t put someones life in jeopardy.

Besides that, not running Compass/Mot has zip point nothing to do with this TORQUE YAW.

If you actually look at the RCOUT plots its easy to see that the motors in pair 1/2 are making about the same amount of thrust and so are the motors in pair 3/4. The difference is in the MAGNITUDE of those values. The 1/2 pair is at a significantly LOWER value than pair 3/4. 3/4 is the CW pair which explains the CCW YAW.

I’d like to see pictures of the aircraft and a detailed parts list…

Hi Rick,
I have observed the same with version 4.0.3, setting the orientation of the external compass to “0”, after calibration, it appears at 180 °, which is the actual orientation with respect to the flight controller orientation. Apparently it detects the corresponding orientation and applies it.

Orientation

The AHRS_ORIENTATION must be set correctly for the compass calibration to be successful. In addition,the Accelerometer Calibration should be completed before the Compass Calibration.

The setting of the orientation of external compasses is no longer required. Its determined automatically during calibration now.

So to be clear there are two issues I am up against.
CCW YAW and a compass issue.

So first the YAW…its clear from both @OldGazer and @xfacta that there is an issue with the RCout on the 1 and 2 motors. Its clear in the logs that they running higher then the 3-4 motors. So I will look at the following.
Motor Level. While I have done this twice in the past, It won’t hurt to do it again and rule out the possibility that something twisted.
I will also re calibrate the ESC to ensure there is no issue there.
Apart from that I have no other ideas.

The second issue is a compass problem and your correct is has zip to do with a YAW issue @OldGazer . But if you read the whole post you will see I am faced with a compass issue as well.

@xfacta has laid out a number if items to check.
Also he has called my attention to the COMPASS_AUTO_ROT which I wasn’t aware of. This explains the compass switching from Roll 180 to Yaw 90…this is working as designed apparently. however I do need to follow the calibration process as he outlined and as is documented.

Then I need to do a Compassmot as it’s clear there is a correlation between compass and current on this machine as was last time.This was done in the past on this bird and will need to be done again. I have come to appreciate @xfacta input and I understand the humor he injects into the dialog. So there is no misunderstanding there are no plans to raise a drone with spinning props over anyone’s head, I understand his humor and didn’t take it seriously.
I have a jig that holds the drone in place and ensures the arms are properly supported for the downward thrust that will be caused by the reversed props.

Now back to Torgue Yaw. I still struggle how the problem I am facing is such a form of yaw. While I understand how it would affect the yaw of the machine. The yaw is in the direction of the CCW motors which tells me that the motors are driving the yaw and not a misalignment of the motors.That said I will still give them a going over.

When the question is asked about a parts list…what is it your wanting to know @OldGazer.
18 inch props on Garrt 5010 motors. 6s 16000 mah. Pixracer is the controller…what else is of interest.

Yes the orientation is set right on the machine.
But as @xfacta pointed out the COMPASS_AUTO_ROT would explain the change to the compass orientation setting after a calibration has occurred

I want to see the entire aircraft with the props on.

Its been my experience that 5010 motors don’t like props that big.

While I was thinking about this, it just occurred to me that we have have something backwards here.

Think about it. The quad is yawing CCW. the CW motors are maxed out and the CW motors are almost at minimum. But, to correct for a CCW yaw the FC should have the CW motors spinning faster than the CCW motors.