So I definitely need to figure out how to improve the accuracy situation, but before I experiment further, it’s time to get some sort of hold-down clamp system going, because clamping for the gears above took more time than routing them.
First photo: I used the inkscape-gcodetools pipeline to cut a series of 1/4″-wide slots in my table, on 200mm centers. (Feed rate was measured in farthings per fortnight, of course.) Not all the slots are the right size. First I forgot to convert object-to-path in Inkscape, so I wasn’t subtracting off the bit diameter. Then I was battling with too-narrow slots due to machine slop; reversing the cut path to be clockwise put the error in the other direction, so the slot’s a little wider than designed, but quite usable.
Second photo: I bought some 1/4″x20 x 3″ carriage bolts and ground the heads down to 1/4″ wide. I can drop one into the slot, turn it 90 degrees, and the square base of the bolt locks into the slot; then a wingnut from the top can clamp down on some holding piece.
Third photo: I want some pieces designed to clamp along edges. Here’s my first attempt. It needs a rabbet cut on each end, but I haven’t quite mastered the depth controls in gcodetools yet.
I made a fair amount of sawdust; it’s probably time to move this contraption to the shop where there’s a shop vac and air cleaning equipment. Cough cough. I also made a burning smell — I suppose I should increase the feedrate, but I have it pretty slow (250mm/min) because I was trying to reduce lateral forces that would contribute to path error. Maybe I should increase the feed rate and the number of passes?
Speaking of rigidity, I’ve noticed that I can spin maybe half the bearings in the system with a finger. Is this an indication that I should be tightening the “clamping bolt” (that goes through the middle bearing and squeezes the sides of the part together) in eaach assembly more?