Firstly, I am aware there are much better ways and materials to make this work well. However, I have a fully 3D printed 10" FPV quad that I am experimenting with and for the most part, it has gone surprisingly well considering the shortcomings of the materials.
Issue: 3 times now, and completely at random, I will experience a hard right yaw, and immediate loss in altitude resulting in a crash. In the goggles, it appears like an unrecoverable flat spin to the right. All three times I’ve seen it have been to the right.
I’m unable to pinpoint anything that I’m seeing in the logs as the root cause of it, and the ESC and motors are fine. I initially thought maybe a desync, but haven’t been able to confirm anything. I require some expert log analysis to see if anything sticks out at the moment of failure.
I have two logs of this occurring. The big one (34mb) is an extensive test flight ending in me disarming prior to impact with a patch of trees. The second is low altitude in Optical Flow EKF Source 2 in the backyard where the same exact issue happened.
I’m open to any feedback based on the logs. Again, I know the logs are going to be less than ideal in many areas due to the materials and vibrations. This is an experimental platform but the random yaw or loss of front right motor issue, I cannot figure out.
I afraid your frame is too flexible, your vibrations peaking to 70 which is too high.
Also you need to start with the proper filtering, you are using multi source but your attenuation/bandwidth is too high, so it can give unacceptable delay. You need to enable batch or raw imu logging to log gyro data.
Also there is a problem with YAW imbalance, maybe your props not in level.
Than in the end your ESC #3 stopped providig RPM data, which can be caused by ESC failure or desync.
@mike_tk thank you for taking the time to review the logs and offer advice. Super helpful, I’ll work on narrowing bandwidth filters to decrease delay. Unfortunately the flex and vibes are something I’ll have to deal with for this experiment. Like I said, it’s definitely a “less than ideal” build with much better ways of achieving success available. I just wanted to first make sure that there’s nothing going on in the software side (EKF confusion, runaway motors, etc.) causing the uncommanded hard right yaw. It’s a problem that has very recently gone up. The quad, despite all its less than ideal foundation, actually flew and performed quite well considering, for several weeks of intense testing.
Since I’ve posted, I’ve had the motor stop 5 additional times. I’m hovering at roughly 1 meter altitude over grass so it spirals in softly when it happens. Each time it’s the front right motor which causes the right yaw and drop. That motor appears physically fine, spins freely, and is able to arm and fly perfectly each time until it randomly shuts down. Only thing it’s there is a little bit of bearing noise in it, and you can feel a very slight grind for a few mm each rotation by hand.
I’ve tried updating to AM32 2.19 and adjusting timing from 7 all the way to 22.5 degrees incrementally but the problem persists. The ESC isn’t failing because I can immediately re arm and takeoff again. Although I have tempurature protection disabled, there’s no way it’s ESC heat because I’ve had it occur on the first flight within 20 seconds.
What I’ve currently done for todays testing is changed the front right motor to a new one, used auto timing for the ESC and turned off “Stuck Rotor Protection” in case the bearing grind was interpreted as a stuck rotor causing shutdown. It survived about 45 minutes of maneuver and hover last night and I will test extensively today.
If you don’t mind looking again at my params, there’s nothing in my DSHOT setting, BLH, or any ESC settings in Ardu that would be causing a desync issue, I hope?
Thank you again for taking the time. I really appreciate it!
We have a couple of 10’’ 3D printed quads. None has vibration levels of 70.
You did something wrong in your 3D model. @Quadzilla also builds a lot of 3D printed frames, and he also does not report such high vibes.
@amilcarlucas at the moment, I have long unsupported arms. They will soon get a bracing solution or rib geometry to aid in stiffening. Early version at the moment.
The log with vibes to 70 m/s/s was carrying a rather heavy repeater payload. Here are the vibes on the airframe with no payload. I clearly need to fix induced vibes resulting from the payload attachment, but do these look acceptable or a least manageable for the airframe itself? This is from the desync issue happening in at roughly 1m AGL in my yard with no payload.
