I’ve tried various heights for the center of mass offset, -80cm (i.e. 80cm below the vehicle body) was what I found to be good compromise between chassis stability and “feeling good to drive”. at -40cm it was very unstable and still regularly toppled over. The car is sort of a 1920s-style racing car with bicycle wheels.
Is your mesh volume centre a large distance from the coordinate system origin for that mesh?
Not terribly. I’m still trying to understand your line of reasoning here; why would moving the center of mass below ground invert my physics?