firts step is finding the right brushless motors. then tearing one of these open so the electronics can be analyzed and lower rated componnts used. those things are nema 23 and way more powerful than any nema 23 stepper. definitely flies in the face of those arguments about not enough torque. Come on 245 people paypal me a dollar and i will do a tear down and shater the videos and pics. We can crack this as a team. Thats why open development has expanded 3d so far.
Cheapest servos use brushed DC motors. While I do agree a brushless motor is preferred (longer life, quieter), I do not see that happening cost-wise. It is the electronics drive what gets more complicated if you go brushless.
If you are interested, a number of us have already wandered down this path - @Steve_Graber is actually using these SDSK’s on his Gigante delta design. I’m using Gecko G320Xs on one of my CoreXY designs. Both of us are just feeding step and dir from semi-traditional hardware and software - in my case from an Azteeg X5 running Smoothieware.
Cost aside, you are still missing a proper closed feedback loop, and the firmware to support it. With the build in shaft encoders, you know if you motor missed a step, and correct for it - but you don’t know if you lost a step anywhere else in your motion system.
The other part is proper acceleration and jerk calculation in this motion control system. @Alden_Hart and the tinyG folks are making progress, but when you look at some of the industrial (and patented!) solutions you get a good idea of where we could be.
Pretty much all of this means a move away from 8-Bit AVRs.
I guess my point is that it’s not just hardware we need - it’s that hardware integrated with software as well.