Well. I tried anti-gravity and lowing P rate but still having that hard jerk at the top of my split-s just as it nears or gets to the upside down position. Maybe I'll post on YouTube to show what I'm taking about.
Anti-Gravity momentarily boosts the " I-Term" of your PIDs to prevent unwanted interactive results
in attitudes when making abrupt throttle changes. The most commonly affected axis are pitch and yaw
though which leads me to suspect that, while it would be helpful overall to have antigrav tweaked, that
it isn't going to specifically address what you're experiencing. I would recommend against scrapping your
PIDs entirely though, as they're presumably fine other than needing some minor adjusting.
Aside from the established functions of each term in PIDs, it may be helpful to consider some of the
peculiarities of each term. While the P term is faster acting than I, it has a shorter "memory".
While the I term is slower in reacting, it has a "longer memory". This having been said, my advise
would be to experiment with the P and I terms for roll only with an INCREASE of the P term primarily
(you said you lowered it ?), accompanied with minor increases in the I term and leaving the antigrav
settings where they are (or, better yet, where they were).
Addendum: You say that a straight roll doesn't exhibit this problem ?
If it's happening only during a split-s or Immelmann maneuver then is it upon pitch command that it occurs ?
If so, then you may need to also address the P and I terms for pitch as well. Making small adjustments to each term for each axis until it's fixed is the key, because overdoing it will cause the quad be feel too "stiff".
It's kind of a game of "hunt and peck" resolving such problems, but if you're performing acrobatics like these I'm hoping you'll also have the "feel" for the improvements as these adjustments are being made.
