Better think about a way to keep the table from moving on those wheels. You will be amazed at the reactionary forces involved in slinging the gantry around at 100 + IPM and the Z back and forth (Newtons Third Law). You will find yourself chasing that table across the floor as it cuts. It will be a Self-Propelled CNC Plasma Cutter
I can't tell but I don't see any belt (speed) reduction. If not you can get astounding speeds (in the range of 1800 IPM) but you pay for that with having less TORQUE and resolution and less torque means less acceleration. While you can run the feedrate lower you don't gain any torque or resolution from that. Direct coupling of the motors to the spur (pinion) gears follows this: You get a speed gain of PI times the pinion diameter and a torque loss of the raw motor torque divided by the pinion radius. 620 motors at normal cut speeds will have about 400 oz-in or rotary torque. In any transmission you always trade speed for torque and resolution.
Here is the deal with acceleration. The ability of the mechanics to make rapid stops and starts (deceleration and acceleration) defines how close you can adhere to a toolpath (variance). Poor acceleration forces compromise and rounded corners and angles. The problem is amplified by the normal level of feedrates. The acceleration from 0 to 200 IPM is longer than from 0 to 50 (like on a router) so the toolpath is more effected at higher cutting speeds. Since acceleration = Force/mass changing either part of the equation creates a proportional change in the number across the equal sign.
One of the reasons I discourage the builds of mixed use (router/plasma) machines besides the obvious one of fire and wood chips, is that you have to compromise optimal performance unless you are willing to grossly over design. Sort of designing a Sports Dump Truck.
TOMcaudle
www.CandCNC.com