Simply reflashing the smoothieboard with the latest firmware caused the PWM to start working correctly. Quite an anticlimactic result considering i picked up a new Rigol scope to troubleshoot the issue!
As seen, a 0-100% duty cycle reflected correctly on both the scope and the observed intensity of the laser.
Here’s a quick video of the kit cutting up some dense 1/2″ open cell foam.
I am still trial-and-erroring the focal depth of the laser. Right now it takes 5-6 passes to cut all the way through the foam. It starts around 12mm off the bed and moves down 1mm per layer. This is pretty dense foam though.
For basic shapes like circles for these marbles I’m simply creating a cylinder in Google Sketchup, exporting it to cura, generating the gcode, and then manually editing the gcode in Notepad++.