28in Quad Yaws Heavily on Takeoff

I am in the process of test flying a large 28.2 in prop quadcopter using Tmotor U8 85 kv motors, KDE 85a ESCs, Mauch PDB, and a Cube Orange. The frame was designed and built in house. I am having an issue where the drone yaws heavily on takeoff and exponentially gets worse the longer it is airborne, making flight nearly impossible other than a couple seconds of hovering. After testing, it seems to yaw in the direction of the initial transmitter input, however after in the air stick inputs can’t control it.

I am aware frame flex may be an issue, however I’m not sure how to determine this through the logs. Graphing desired vs. actual Pitch, Roll, Yaw reveals, ironically, horrible pitch and roll behavior but the flight controller seems to think it follows yaw fairly accurately. Could this be a pitch/roll tuning issue? And how would that affect the yaw so significantly. Looking at the PWM out values from the motors in the logs, my untrained eye does not see anything noticeably wrong.

I have swapped the external Here 3 GPS for a new one in hopes it was a compass issue. This had no effect. I will post a “flight” video and logs from my most recent flights. Please let me know if you guys have any suggestions! Thank you in advance.

Test flight: https://drive.google.com/file/d/1SB-1lK-txoRpCVWrtfuTBKWzoKBdtUek/view?usp=sharing
Log: https://drive.google.com/file/d/10CPqrYXvY6Pj5OiFma6H2tAxU3briwsm/view?usp=sharing

High currents cause high magnetic fields, have you increased the distance between the GPS and the high current wires and battery?

The GPS is mounted on a mast on top of the drone, next to the FC. The batteries are mounted underneath the drone, separated by two carbon plates. I’m not sure that’s the problem but I can take it into consideration! One thing I have not done is disable the internal compass. I have read this may be more beneficial than having the potential redundancy of internal and external compasses enabled. Would you recommend this?

Yes, please disable the internal compass!

Okay will do! I would think I would get an error if I get too much variance between the two but I’ll try it out. Do you have any other ideas for me to test next time I go flying? It’s a bit of a trek to the flying field so I like to make the trips as productive as possible.

There is no Mag data in that log so hard to say… Checking the Compass box in the Log bitmask would help :slightly_smiling_face:

I’m glad you saw my post Dave, you seem very helpful throughout the forums. It’s very unfortunate I did not have that checked, but I’ll be out for a test flight tomorrow and I’ll post the updated log. Question for you, does the Compass box in the bitmask need to be checked for motor compass calibration to work? I tried performing a compass motor cal yesterday and my interference % line was perfectly linear at 0% which seems very hard to believe.

No, think it’s just for logging. I wouldn’t bother with the conventional method of compass motor cal. Using Magfit thru MAVExplorer is a better method for that and compass calibration:
Magfit via MAVExplorer
Don’t get hung up about the fact it’s a python utility, MAVExplorer makes it very easy to employ. Watch Henry’s video in the Wiki.

1 Like

Sounds great, I’ll give that a try as well. Thank you for the help, I’ll keep this post updated with flights from today.

1 Like

In had an identical problem with a T-motor U8 190 kv motors quad I built and I had the motor order set incorrectly. Cost me a set of props.

Interesting, I know people say it all the time but I’ve quadruple checked the motor order. In my experience, the drone wouldn’t even lift off the ground if motor directions were incorrect. This one will lift off the ground, it just yaws uncontrollably. I’ll check again but I don’t think that’s the issue.

Hey Dave, I went out for another flight and attempted a MavExplorer test. I ran a log through the software and plugged in the recommended compass parameters with no success. One thing I noticed is the original versus corrected graphs look practically identical. I’ll paste it below. I also disabled the internal compass and allowed compass logging for mag data. I’ll paste some logs below as well. If you could take a look and let me know if you see anything, I would be very grateful!
MavExplorer graph: https://drive.google.com/file/d/1GDqHk_L8msZjJ9YM7PatlW3MyXYOquTL/view?usp=sharing
Log 1: https://drive.google.com/file/d/1z9RB7BsNdL8kkara6xGc4Wy9DhqhKibD/view?usp=sharing
Log 2: https://drive.google.com/file/d/1d7hOTWOk64w8oCQHw0R-nf0Jxp8G8Pp4/view?usp=sharing

You need to enter the compass values after the results are presented. But anyway I don’t see any throttle influence on the Compass. Nothing else really to see in these logs.
Update to latest Stable, you are on some old firmware.

It’s very tough to get a good log since the quad won’t stay stable long enough in the air. I entered the compass values after the results and did not see improvement unfortunately. I’ll update the firmware and get back out there tomorrow. Do you see anything out of the ordinary from the RCOUT PWM graphs? I would think maybe these would reveal something with my yaw issue, however I am not super experienced reading these logs.

Neither of these logs would be suitable for running Magfit. It has to be flying under control for some minutes at least. And Magfit typically produces Integer values, what’s in these logs looks like results of standard compass cal.
You did run Motor Test in Mission Planner to confirm the order and direction? You can’t tell that from a log.

Note: It doesn’t look like your current monitor is functioning/calibrated.

Yep, mine got of the ground no problem then started the yaw just like your video. Maybe the 28inch props just delayed everything a bit?? I’m not saying that is your problem but it was mine. :slight_smile:

I just checked my motor numbering again and it looks correct. Motor test also confirms it. I was crossing my fingers that was my issue as well :frowning:

Yep I have checked motor direction and orientation with confirmation from the motor test. I will work on current monitoring. I’m running Mauch electronics so it shouldn’t be too hard with the documentation. I suppose I’ll have to figure out the yawing issue before running a proper Magfit but thank you for bringing my attention to it. Mavexplorer seems like a great tool that I honestly had no idea existed.