I may change to 1610 ballscrews - I'll have to see what sort of torque and rotational speed the X/Y axis stepper motors can deliver once in their closed-loop configuration. If the motors are fast enough, I'll stick with 1605 ballscrews to provide a higher resolution. It has to be said though, this machine is being built for prototyping and small production runs, so breakneck speed is not a priority. There is no point in the machine doing high-speed large production runs as the economics say it is cheaper to farm such jobs out to a CEM.

I'm looking to achieve a resolution of 0.05mm as I believe this will be adequate enough to place 0.5mm pitch QFN and QFP packages and 0.8mm pitch BGA packages.

The idea with the closed-loop control of the X/Y stepper motors is to essentially treat them, in combination with their encoders, as BLDC servos which will increase their maximum operational speed and torque. Of course, stepper motors are not BLDC servo motors and it takes some trickery to make them behave as if they are. Good quality BLDC motors are specifically designed not to cog so that the resolution of a positioning system using them is pretty much determined only by the feedback transducer. Stepper motors, on the other hand, are specifically designed so that they do cog to the extent that no feedback transducer is required at all.

The drive system I'm planning to implement has essentially two modes although the transition between these modes is smooth, not instantaneous. When the torque or speed demand is zero or low then the motor will be multi-stepped in the conventional manner albeit at a peak current much lower than the motor's rating. The aim is for this low current multi-stepping mode to overcome the motor's cogging. During periods of high torque or speed demand, the motor will be driven in a similar way to a BLDC servo, commutated by the encoder and with speed/torque varied by the amplitude of the current waveforms fed through the motor's windings (BLDC servo speed is controlled by varying the amplitude of the voltage waveforms applied across the motor's windings, but with stepper motors it's more appropriate to talk about current).

As well as being used for commutation, the encoder will provide positional feedback into a PID loop to determine torque demand. Controlling the smooth transition between the two drive methods is probably the most complicated part of the implementation and is something I still haven't fully figured out yet.

Matt.