I have a couple of BerryIMUs v 2.3. Using them for robotics. Built calibration in so that it calibrates before each run. This makes it more accurate, but I had to reduce the level of Kalman filtering since there was too much lag to be useful. But this surfaces a ton of noise resulting in swings of 5 degrees or more, which also reduces its usefulness. Is there any way to filter out the noise without introducing so much lag? Or is v3 any better when it comes to noise? Thanks.