garageFab.cc

Over the last couple of weeks I tested the new design of the sliding element. It seems to work well and not have much torsional play. The Teflon angle is cut into 1/2" lengths which are placed in 3 inside corners of the tube, completely enveloping the outer side of the rail, with overhanging pieces pushed towards the rail by small 6-32 screws. This allows for some limited adjustment of friction.

Otherwise, the sliding element is similar to the design A with ball bearings - i.e. there is 1 1/2" mounting section with 3-hole pattern on each side. The Sketchup file is here, make instructions are WIP.

Now that one axis belt drive looks more or less satisfactory, I'm about to build and test XY belt drive - essentially, a plotter (model of 2 axis plotter contraption made in Sketchup). I have a couple of 24" angles with -1/32" total error which I thought I'd use. However, being 1/32" shorter, they turned out to be useless as a belt drive rail - the belt slops way too much.

This means that I need the long angle without any compounded error. I went back to the drilling contraption, ran some more tests and made sure that X axis is not skipping/missing any steps anywhere. The conclusion was that the error is related either to the leadscrew, or to the flexing rail (32" long), or to combination of both - and it doesn't look like something straightforward to fix. Since the error is fairly stable at ~1/32", I implemented a simple cheat to correct for it - when X axis advances to the next hole, it moves 1 inch plus a tiny bit (1/32" over 24 inches works out to be 8 more steps after each inch). This removes the compounded error and makes (almost) perfect perforated angle.

When I was running tests on drilling contraption, I also noticed the increase of backlash (from ~45 steps to ~75). In all likelihood, this is due to Delrin nut wear. This got me thinking about anti-backlash nut designs and googling what's been done on this problem. One idea (see pic) is based on split nut solution I saw on figNoggle.