Tailsitter First hover flight FBWA a Disaster

Vertical flight was a success, but FBWA was a disaster. In the video you will see the plane rotate into horizontal mode, then spin toward ground. I was lucky to be able to save the day by switching to Loiter again, the plane recovered. Later in the video I tried it again but maybe I have the COG too far back. I will rebalance and try again. There were also some strange results I don’t understand, the plane seemed to go into RTL without be commanded. This will need further investigation.
Here is the link Tailsitter First hover flight FBWA a Disaster - YouTube

Hover looks good.
For CG location, this calculator gave very good results to me.
5% static margin is a good value to start with a tailsitter. Less static margin will improve hover modes but need careful tests in fixed wing mode.
But this spin could be induced by motors differential thrust, To check if left and right motors have same thrust you can look at pwm values in hover mode. They should be equal ± 20. An other possibility is to get higher speed before transition.

After studying my videos in very slo- motion, I can only account for probably Three reasons for this constant crashing when switching from Qhov, or Q ltr to FBWA.

  1. can it be the motors out of balance, I checked and at very low speed in manual left motor had started a tiny amount before right motor. I just tweaked min pulse width to get them equal.
  2. On the transition from Q hover to FBWA, the rotation begins, the plane begins to pick up some speed, but every time, it goes into a left hand flat spin, followed by a spiral dive that could be caused by the right motor being more powerful, but as I wrote above the motors appear balanced.
    Or is it just the plane stalling???
  3. this plane weighs nearly 4 lbs 1760 gms, so it is obvious the forward speed needed is going to be a light higher than say a 2 lb 970 gms model. I don’t see the horizontal speed increase enough to support this weight and as such consider this the primary cause.
    So lets assume a model weighs 1 lbs and as it transitions from vertical to FBWA, it begins to accelerate, rotation continues and as the plane reaches +30 degrees above level flight, it has enough speed to support level flight.
    Now, lets move up to a 4 lbs model and again we transition to level flight. What settings can I adjust to make the forward speed increase to a point where level flight is sustained. Can it be any model after the transition dives slightly to attain speed, then levels out?, I don’t know.
    I need to be able to some how increase this flying speed so at the transition end my model has enough speed to sustain flight.

This is not a guaranty nor an indication that motors produce the same thrust for the same input. A good indication is to check pwm hover value for left and right motor. I would check ESCs internal setting too.

You can set parameters for transition angle (Q_TAILSIT_ANGLE), pitch rate (Q_TAILSIT_RAT_FW) and the amount of thrust depends on Q_M_THST_HOVER

A continuing saga of my effort to get a tailsitter flying. After my last crash, I suggested the motors might be out of balance. In fact, this was the case and I discovered one esc did not produce power in line with the other, even after calibrating. I changed the motor and speed controller and tested in the workshop and motor power output matched.
Moving along to 2022, Friday 16th Jan was the first day for another test. The intention was to take off in Q-Loiter,climb to a safe height and switch to FBWA. This was achieved and the plane flew in fbwa, all be it, a bit out on the pids. After an input to Aileron, there was a wobble that settled. This was probably pids too sensitive. But it was flying, Great. Now it was time for QRTL. The plane returned, but was a little too much overhead. It did a transition to somewhat vertical. The next few seconds were a bit of a blur, I tried to move the plane away a bit for safety, but it appeared to fall over into inverted position. It then began flying away from me doing some left, then right rolls. Height appeared to be controlled, but I did not have any control of the plane. it flew a long way out still rolling with no response to tx stick input. At this point I had no choice but to disarm the plane and it crashed into the ground. This time it was a total bin job.
For now, I am wondering if I can ever get this type of plane to be reliable. The pixhawk board is intact and I have the log for any dev to give me an opinion.
On thinking back, this is the second time the plane appeared to change into a state where I have no control, There is a video that might give a clue as well
Last Tailsitter Flight - YouTube
Here is the link to the log file https://drive.google.com/file/d/1mqMVY9AFC-Qr7vsa7lK6dFf2gsvDsfhs/view?usp=sharing