tailsitter vtol frnot/back transmission failed (no pitch set value is generated)

[xdwgood]

Describe the bug
When I used tailsitter to fly, I found that sometimes(Sometimes it can enter successfully, in the same flight) even if I switch, vtoll can not enter the fw mode

To Reproduce

1. use master branch fly(tailsitter vtol)
2. in altitude mode and try transmission(mc to fw)
3. The drone pitch angle has not changed, there is no tilt acceleration

Expected behavior
Every time a customer uses a stick for transmission, vtol should enter the transmission state correctly

Log Files and Screenshots
https://logs.px4.io/plot_app?log=4df87c34-da2f-46be-b612-ddcc13b135c8

Additional context
I found that fw to mc sometimes fails when I try to send a transfer command using the remote control. I will do some investigations to see where the problem is. I now suspect

numerical calculations under certain extreme conditions? (eg:at the start of transmission, the actual/setpoint angle of pitch are both 0)

here: https://github.com/PX4/Firmware/blob/master/src/modules/vtol_att_control/tailsitter.cpp#L206-L218

Because I compared the two switches, the only difference I found was that the drone angle/setpoint angle was close to 0 before the failed transmission started

FYI: @sfuhrer 😃

[sfuhrer]

@xdwgood Did I understand correct, that the issue is this spike in the transition switch log?

Edit: having had a second look it's now clear what you mean 👍 I'll have a look

[xdwgood]

@sfuhrer
https://github.com/PX4/Firmware/blob/master/src/modules/vtol_att_control/tailsitter.cpp#L216

The calculated tilt sometimes has problems. I checked it using the PX4_ISFINITE(tilt) function and found this problem.

Corresponding code

[xdwgood]

No setpoint value is generated when the value check fails(8)

[xdwgood]

fix here: #15352!