Troubleshooting VTOL Quad Tailsitter Drone Transition Issues

I am currently working on a VTOL Quad Tailsitter drone with no control surfaces like elevons, rudders, etc. More specifically, it’s going to take off like a multirotor drone and then transition to a fixed-wing UAV. The differential thrust of the four motors will be utilized for controlling the fixed-wing configuration. I’m using the PX4 as the hardware and PX4 1.14 as the firmware.

In the last automated flight prepared with QGC, after the transition began, the drone turned upside down, flew for a few seconds, and then returned to VTOL mode for landing as the transmission time was exceeded.

Below you may find the log files of this flight:
https://review.px4.io/plot_app?log=947a7ff4-3040-4773-8cac-153c452a504a

Can anyone tell me why this happened?

These are some of the parameters I used. If anything else is needed, let me know:

Hi,
I’m also working with a similar platform and have problems with the transition
Have you tried a transition with manual control in stabilized mode?
In my case it was useful to detect that the issue was with the airframe not having enough authority and being quite unstable due to its aerodynamics

Hello Simone,

Thank you for your answer. No, I haven’t tried it with manual control in stabilized mode. I have conducted only pre-planned flights through QGC. I don’t know if you had time to check the log files, as the drone for some reason turned upside down, and I would like to ask if there are any other parameters that I should change in order to make it work without surface controls apart from the ones that I have already modified in the picture I uploaded.

I don’t see anything wrong in the parameters, they are the one I was using (slightly different tuning)
That’s why I was asking if you have flown in manual.
My suggestion is that you start from less autonomous mode to detect any possible problem there. One good thing to spot is how your platform behaves in FW-like attitude. So flying in stabilized, without transitioning, with MPC_MAN_TILT_MAX set as high as possible. This helped me detecting my problem
In your case I guess would be nice to see if the oscillation around roll/yaw are still there.
That said, my knowledge is very limited so I might be tottaly wrong

Where are the aircraft’s centers of gravity and pressure located? Maybe it was doing what it was (unintentially) designed to do?

Hello, the aircraft’s center of gravity is located almost at (slightly in front of) the quarter chord point (c/4) of the mean aerodynamic chord. As can be seen from the logs, when the drone flipped upside down, two of the four motors stopped while the other two caused this 180-degree roll. The airfoil used, is the eppler 423. So, I was wondering if this is a fault of the airframe structure or if a software parameter needs to be tuned.

Some sensor thought that those two motors needed to be turned off. Can you connect to Mission Planner and check that everything is calibrated properly? Can you test what the aircraft does while you hand-hold it and manipulate the controls?

Okay, the next step is to check how the UAV behaves when I hand-hold it and control it manually. My next thought is to add some elevons to make it more stable since it is currently surfaceless. What do you think about this?
If you find any other solution or possible reason for this happening from the logs, please let me know.