I did a hand tune first and was going to see what the numbers were, then auto tune and see what it came up with. My numbers seem really low especially Kd. I did the hand/cradle method using channel 6 knob. I came up with p=.124 i=.4(just changed from zero and have not flown and tuned yet), D=.0100. I hovered it and once a few feet up it seems stable. Down low to the ground it oscillates a bit. I tuned D up until I felt quick oscillations, then backed off some. I never did get P to oscillate and had it up to .14 before lowering it. My motors are nx4008 pancake 620kv, afro 20 amp esc, 4s battery and apc multirotor props 10x4.5 on a qav500.
Prior to this I had different motors and esc on it. I kept losing a motor in flight and just replaced all 4 rctimer motor and esc with the ones on the quad now.
I have a similar QAV500 except my motors are MT2216-12 800 KV on 4S. Props are the same. So with my higher KV I’d expect mine to be a bit more responsive that what you experience and your PIDs might be a bit larger than mine (that’s a guess).
As a reference point for you I initially had decent, respectable flight with Roll PIDs of 0.095/0.100/0.004 and Pitch PIDs of 0.120/0.100/0.004 respectively.
With those in place I then did an Autotune and got Roll PIDs of 0.045/0.045/0.004 and Pitch PIDs of 0.105/0.105/0.011 Autotune set both Stabilize Roll and Pitch at 7.125. It flies just fantastic with those values after Autotune.
Hope that gives you a ballpark to play in.
Thanks for you numbers. I looked at mine and I moved the decimal point around and that was my problem. D of .0500 and .0050 is quite a bit different! I used your numbers to get my QAV in the air, and then used auto tune. Was really pretty fast and took only 3-4 minutes total. It came up with Roll Kp-.0650 KI-.0650 KD-.0050, Pitch KP-.2100, KI-.2100, KD 0150, Stabilize are both at 7.2188. Flies really really well now! I was surprised by how much different the roll and pitch numbers are. Is it normal for KP and KI to be the same number? Or just coincidence?
Glad Autotune worked for you. It’s really a fantastic aid. Your numbers weren’t too far from mine.
QAVs and similar are really a “H-Quad” design with all the mass lined up on the pitch axis. This means the pitch moment of inertia is going to be greater than the roll inertia. So you would expect the PID values for pitch should be greater than for roll as it will take the craft more effort to make a pitch change than a roll one.
I’m not sure why the P and I values sometimes (often?) come out the same in Autotune. I have another quad where that wasn’t the case.
P and I values will always be equal. I term can be higher relative to P on more agile copters. Setting them equal has a reasonable safety margin, about half of what it could be for a 3DR copter.
Is this solved, then?