Quadcopter yaws wildly for a few seconds on takeoff

ArduCopter Copter 4.3
1

Hello All,
I am reaching out for any assistance in an issue I have with this copter I have( see pictured below).

Drone

On takeoff it will yaw left and right more or less out of control ( I think if i correct it, it will rotate the other way). In a second or two, I get the ‘Yaw re-aligned’ message and EKF3 lane switch 1 message and then it flies normally. Although whenever it is hovering with no stick input, there is a undulating movement, perhaps like the described “toiletbowling”.

I have had this craft flying well, earlier this year, in a slightly different configuration, with the Li-Ion batteries undernealth and FC + ESC on top, flying in a H configuration.

I have tried ( and can provide other flight logs) a H and its current X configuration. It is effectively a H-frame flying sideways, which I have read should be setup as an X-frame.
Either way they have both done the same things. Most recently I have also just tried a compass-motor calibration (see attached log) to no avail and the Mag interference in the log file looks worse on this current one.
I have tried a few other small things like the battery wire reconfigure and twisting and remounting and changing the direction of the FC and softening the mounting as the vibes are a little higher than I would like.

The FC is mounted upside down and 270 degrees off from the direction of flight. The hud moves correctly in the GCS and I have done multiple Accel, and healthy Compass calibrations.

Flight characteristics have been the basically the same no matter the changes.

https://drive.google.com/file/d/1MU2ijJSNMDIAEnllUxwzh_CBxQOmiPQr/view?usp=share_link

Thank you for any and all advice
JP

2

Just for a start you could set these compass values. Yours were already extremely close but this might make a small difference.
COMPASS_OFS_X -63
COMPASS_OFS_Y -36
COMPASS_OFS_Z -47
COMPASS_DIA_X 1.028
COMPASS_DIA_Y 1.098
COMPASS_DIA_Z 0.830
COMPASS_ODI_X 0.044
COMPASS_ODI_Y -0.013
COMPASS_ODI_Z 0.017
COMPASS_MOT_X 4.364
COMPASS_MOT_Y -0.754
COMPASS_MOT_Z -7.775
COMPASS_SCALE 1.00
COMPASS_MOTCT 2
courtesy of Magfit

image

1 Like
3

I would use Magfit to calibrate the compass and generate the compass motor coefficients. Use the tool in MAVExplorer.
Magfit
Is that one of those Radiolink GPS/compass modules?

4

@xfacta @dkemxr
Thanks for these parameters and suggestions. I will go try them out.
I haven’t come across Magfit yet, I just used for the first the one in Mission Planner. I also tried in QGC but it only froze on me.

And yes it is an older RadioLink SE100 m8 GPS with whatever compass it has.

5

So far nothing I have tried has really made a noticeable difference. I have run Magfit on a few flight logs and have got various parameter changes back.
I also have tried a compass from a 3dr Iris+ gps/compass installed diagonally opposite the other one, on a lower plate under the main frame with the same results.

In Magfit, the below log gave me results in the third panel, Yaw Change, of less than 10 degrees of errors, which by the sound of it, isn’t all that bad.

https://drive.google.com/file/d/135E-d_EpiLh8KMOaD3nZVlOuisGHlRmm/view?usp=sharing

The suggested changes gave me this short log.
https://drive.google.com/file/d/18sdHVPVnT1BsFD9SirIczyhdvaMcFdbX/view?usp=sharing
With no real yaw control input, the quad made short left and right yaws continuously. It also has given me, on occasions, AHRS error warnings from 26 - 86 degrees. One such warning was on this log I think.

Could it be something other than compass related? Could it be a problem with having the AP installed upside down and rotated? This has been built such a way for for convenience, easy battery changes and stuff. But with a little work, it could all be reconfigured, FC and esc up top again, and batteries on a plate underneath. However depending on what might be happening, this may not fix it.
Cheers
JP

6

I cant see anything in parameters to change related to the yaw issue.
Where is the compass located? In the GNSS unit on the stand?
If that’s the case then it’s probably a very bad implementation of the compass, since it should be far enough away from interference, and shouldnt be having these problems.

You could try a small sheet of mu metal to shield the batteries from the rest of the system.
Also mark the batteries across the top of them so they are always fitted in the same order, orientation and rotation. LiIon batteries have a much greater magnetic field than LiPos.

7

Or replace the GPS/Compass module. Radiolink is sketchy hardware at best IMO.
Li-Ion can be a problem on small craft but I flew one a lot smaller with one for awhile w/o too much trouble. Switched to Lipo though.

8

Ok, Thanks for your respones guys.
I might try another GPS/compass module installed on the mast. I have the mini Matek M8n one, so thats a bit newer. It just doesn’t quite get as good or as fast a gps fix. Probably because it’s tiny.
I always use the same three batteries, and because of the way it is wired, they always go the same orientation. However, good idea about the same order. Never thought about that.

LiIon definitely comes with its challeges.

I’ll try a few different hardware / configuration changes and let you all know what worked out.

9

That’s what I use on all small craft. True it can take a bit longer but in use I have had good luck with it. The M10 module has a bit bigger antenna but I haven’t tried it yet.

BTW-I tried some Mu metal foil on a 2S 4000mah Li-Ion pack I had on a sub 250g (flew forever) but I didn’t see much if any improvement. Worth trying though if you have some which I do from some other projects.

10

Well, I’m a bit stumped here.
I have made a few physical changes, including the current layout pictured below, with the Matek mini m8 installed above the f/c, and also with the gps/compass installed back on the original mast.

Quad_2
Quad_21920×1536 237 KB

All have resulted in basically the same way, with instant yawing on takeoff, until it does a lane switch. Sometimes left and right quarter turns, and in the short log below, a constant right rotation until I set it back down again quickly. Once it does a yaw alignment and a lane switch it goes back to flying relatively ok, albeit untuned at this stage.
https://drive.google.com/file/d/1REUFvL-rS2Qp9Li_MpjuoxHyQZq-9RmL/view?usp=sharing

There’s not a lot to see it this short flight as far as I can tell. I did notice though, that the innovations in EKF3 - IYAW appear fairly wild.

Screenshot (3)
Screenshot (3)1153×455 14.8 KB

One thing to note, is that wit the changes to the weight distribution, the initial tuning is very sluggish. Hopefully tomorrow I might get a chance to do better flight a set at least a basic tune. Not sure if this will change anything though, because I’ve never had anything give me such persistent yaw issues before.

I have a piece of coated copper foil leftover from an Iris+, that I have used to cover the batteries in the earlier setup, and in the current one, I have it under the gps/compass as pictured above. I’m not sure if this actually emf shielding like Mu metal, or something to enhance the gps signal.

I have noticed, that the props are not quite dead flat across the length of them (8inches) with the worst one being maybe 3-4mm out. I have heard that some quads can be quite sensitive this, but I’m not sure what is consindered bad. It is obviously a home built quad but is built flat and square to a pretty tight tolerance and with no flex at this weight. It does have a fairly high thrust at about 3.5:1 or so. Again, its weird because I have had this quad flying well and tuned up moderately well about a year ago when it looked like this below.
Screenshot (5)

It was in a H configuration with a slightly different weight, carrying a camera system I was experiment with. 2s Liion at first but I changed to 3s becuase the esc had voltage issues at 6.3 volts.
However, all components are the same except for the esc.

Sorry for the long post.
Thanks again if anyone has any suggestions.
( I havent yet run Magfit over this new setup, hopefully I might get a better flight log in tomorrow!)