[allrounder799]

Hello all,

I have been tasked with building a 1 meter X 1 meter Cartesian bot. I already have some experience in building 3D Printers, so I had assumed that I can just use 1 meter lead screws and smooth rods and be done with it. But I was so wrong. Soon I discovered that lead screws are not fast enough and smooth rods(10 mm) might sag.
I have already spent few hours reading many posts on the forum and learned a lot regarding ball screws, R&P, linear guides etc. I am confused between what to choose. I will list out the specs of my future machine:
X-Y axes length - 1 meter (Z axis not required for now)
Payload - 1 to 5 kgs(2 to 11 pounds)
Accuracy/Resolution desired - 30-50 micron(0.001' inch) [I arrived at this figure by taking average of accuracies offered by a Rolled C7 Ball Screw and a industrial Cartesian robot)
Speed - 1000-1500 ipm [ This is the speeds what a Industrial Cartesian robot offers]

There will be no forces acting on as it does nothing heavy and will be used for glue dispensing.
I am torn between ball screw and Rack & pinion. Does the price of a Ball Screw justify in a machine with no heavy loads? I am leaning towards R&P but still haven't found anywhere in the forum what is the actual accuracy offered by it. And then I also discovered that you can't drive the pinion directly and need a 3:1 reduction.

Second doubt is of linear motion. Will 16mm linear rods with bearing blocks suffice or linear guides are necessary?

For driving both the axis I will be using NEMA23 motors having 19 kg-cm torque.

[RCaffin]

1 m x 1 m?
Try 50 mm ground hardened steel rods with linear bearings, or 75 x 75 mm box frame with linear guide rails glued and bolted on top.
Accuracy for a gluing machine: be reasonable!! 0.5 mm will probably do!!!
Either 20 mm dia ball screws, or very possibly open toothed belt drive, using GT2 belts 10 - 16 mm wide.
Rack and pinion over 1 m - maybe 1 mm accuracy. OK, maybe 0.5 mm. NOT good enough. (Others WILL disagree.)

Cost of materials: $N
Cost of labour: $10N
Cost of time spent trying to do it on the cheap (and failing): $100N

Cheers
Roger