Axis Oscillation

I have retrofitted a Bostomatic 312 VMC with LinuxCNC using a Mesa 7i80HD FPGA card and 7i40HV h-bridges. The motion axes work well, but the ATC is giving me a lot of trouble. The ATC is an umbrella style, it is driven 1:1 by a rather oversized motor with 1024ppr encoder and has ~20lbs of tool holders. Part of the problem is that the axis has almost no friction and a pile of inertia. The main issue appears to be the motor does not have a linear response to the PWM output. The 7i40 is configured for slow decay which should give the most linear response, but there seems to be a large dead zone where it doesn't produce any motion even though it is outputting the correct voltage. I believe that this is due to the brushed DC motors being slightly oversized for the 7i40 cards. Not sure because I have been unable to find any information about these motors as they appear to have been custom made for Boston Digital. The workaround I've been trying is using lincurv to scale out of the dead zone. It seems to fix the linearity problem when looking at it in halscope (using limit2 to make a ramp and comparing it to encoder velocity), but the axis is either buzzing from too much D gain or oscillating when I use the PID component.

Configs here: github.com/stevenBowtie/BosuMilll
The manualMotor branch has the HAL file I was using to test the linearity of the motor response.

1. Has anyone else had a similar issue with 7i40 cards?
2. I've programmed the ATC as the B axis because it was very straightforward to program it as a rotary axis, is there a more sane way to go about this?

Thanks!
-Steve

Are the drives current limiting? (red LEDs) If so they will be pretty much impossible to tune.
Reducing the acceleration may avoid current limiting