the main thing is that pitch flips look strange to me in the log because roll values seem to flip too. But if one looks at the RCin only 1(throttle) and 3(pitch) were used to pitch flip.
And for roll flips it’s like I would expect. 1(throttle) and 2(roll) sticks were moved and the roll values do what is expected. The pitch values stay close to neutral which is also ok.
But what happens when pitch flipping???
I couldn’t reproduce the same by accident.
Here’s a couple of pics from a log including some rolls and a double roll combined with a pitch flip.
Quad 450mm in Acro mode, Pixfalcon FC (like a Pixhawk-mini)
Roll input and output
Pitch input and output
Combined pitch and roll 
looking at your diagrams which show pitch and roll values I assume:
That’s the point!
I’m running 3.6.9
Which firmware are you using?
At about 32500 it did do two rolls and a flip all in one sequence. It was commanded, you can see the roll input stays “rolled” through that whole sequence.
There’s even some of the rolls that included complete yaws too but I didn’t picture those details.
Summary: I couldn’t find anything in the log that looked uncommanded or unresponsive.
AC 4.0.2
I interpreted your inputs in the following way:
RCIN.C1: roll
RCIN.C2: pitch
correct?
Yes:
RCMAP_PITCH,2
RCMAP_ROLL,1
Here’s the whole log, check it out and let me know what you think. I’m more than happy to be corrected on anything I’ve got wrong or misunderstood too.
I thought you could download from that link without signing in to dropbox?? Maybe not…
If it’s requesting sign-in or asking you to create an account I think you can skip it.
EDIT: Let me know if you want the log hosted somewhere else.
RCIN.C3: throttle
RCIN.C4: yaw
so looking at the “double roll combined with a pitch flip” it seems to be some roll moving together with pitch according to the lines drawn in the log.
But what did you steer?
The throttle(3) went to idle and pitch(2) went below 1000.
And roll(1)? To me it’s clearly around 1500 for the whole maneuver.
That’s my point: The log is not really showing what the copter does when pitch flipping.
If you have a look at real rolls (the ones inititiated by the pilot) the roll curves look different. Roll and DesRoll are in phase.
hope it is clear what I meant 
