I'm working with breakable collections of objects and need to use joints rather than parenting to link the motion of one object to another.

I had noticed that when a low mass object is connected to a high mass object (ie where there the mass ratio is say 30:1) then a fixed joint originating from the low mass object (and with the high mass object as the connected body) isn't really fixed at all and behaves more like a spring. I then tried the same set up with a hinge joint with limits switched on and set to zero to simulate a fixed joint and found the result much more solid.

So my initial assumptions were either that the implementation of the fixed joint was broken for large mass disparities or that the hinge joint perhaps performed additional iterations of the physics solver and thus was more solid.

Before I launched into timing tests of the two joints though to see if I could get away with using locked hinges everywhere I tried reversing things and made the high mass object hold the joint component with the low mass object as the connected body. And in this case the situation reversed, the fixed joint was solid and the "locked" hinge was wobbly.

So I now think it's just down to the coded implementation of these joints. At least I'm hoping it is because then at least I can ensure that if I use a fixed joint I always put the joint component in the high mass object. If it's down to something more random like the force evaluation order of individual objects I can't see that I'll be able to get consistent behaviour.

Does anyone else have experience or insight on linking objects of varied masses? Thank you!