Probably this frame is stiffer than the other and this corroborates your experience that stiffening the frame improves things.
I changed frame only because the new one has more space to mount two stack, one for ESC + FC and one for split cam, but I have to give the old one another try to confirm that the new frame is the cause of the leaning.
With larger copters I have had similar experience to yours. I have one 5 inch and one 6 inch (not so large) both with Pixracer and no problems with leaning.
For the low frequency noise on smaller copter I have this hypothesis (to be confirmed).
Small copters with high KV motors have high frequency vibration, it is possible that the harmonics goes well beyond the Nyquist frequency and re-enter as low frequency aliases.
[Edit]
Another cause of noise from sensors is a not so good filtering of power supply.
A good test could be to set GYRO_RATE = 1 on a copter with a Kakute F7 Mini V1, that board has an ICM20689 that should go to 2KHz both for gyro and accel, and see if the low freq noise decrease.
Has anybody gotten things to work on prop sizes < 3"? I’m having terrible accel noise when I go down to 2.5" with an EMAX RS1106 II. Would INS_ACCEL_FILTER fix this?
Hi, I use a HAKRC mini F4, which although not documented, also works with arducopter (4.0.5). The problem is that it has no flash memory for the log. In fact I have a tremor in hovering that I can’t remove, but for the rest it works well flying about 11 minutes at 2s 18650. For the engines I use the kingkong 7500Kv but I recommend the 4500Kv fullspeed which are cheap and consume little.
Can you please tell more about the temperature issue with Kakute mini? Does it affect the IMU?
I have an issue with the large kakute F7, when its integrated 3.3V regulator heats, it influences baro which is just under the regulator on the other side if the board. When cold air from the props cools the regulator, baro altitude changes.