Milling sloppiness update
In pursuit of vicious1’s reported 0.2mm accuracy, I tore the middle assembly all the way down this week. (That involved inserting connectors into cable runs so I could actually thread the middle assembly off the end of the bar with the tubing running through it.) I snugged everything down (until I heard cracking in one case; too tight, Hercules…). Then, because I had it all apart and I had the parts printed anyway, I bolted on Camar0’s extra Z stiffener parst, which more than double the lever formed by the bearings on the horizontal bars. I put it all back together, and ran the test above (400mm/m feedrate, 6mm deep cuts around a 10x30mm rectangle). Things improved a little; now the slop is ~1.7mm (0.8mm per path).
Next experiment: the universal mount feels a little floppy; maybe I should try the original rigid mount for DW660. I printed one out and ran the experiment again, with errors of about 0.8-1.7mm; not obviously any better.
Next experiment: raise the material to operate near the highest Z position, where the tool has the least leverage on the Z bearings. The previous experiments were about 60mm below highest; this one was about 20mm. 1.6-1.9mm, not obviously better. (Actually worse, but N=1, so I can’t say much statistically.)
I tried slowing the feed rate from 400mm/m down to 100mm/min. This seemed to help, with 0.6mm error on one rectangle. I had to cut this experiment short before getting the counterclockwise rectangle cut, because the low feed rate created a lot of burning, and the airflow from the router and the vacuum started an ember burning in the sawdust in the first cut. :v)
The fact that slowing down (or cutting shallowly) helps suggests that the tool tip is creating too much force. Indeed, on some cuts, you can see the router “pop” out as it finishes a circuit and pops through the last bit of wood into the void from the beginning of the cut. This doesn’t sound much like vicious1’s “like butter”comment above. So, ideas:
1. Maybe I’m actually cutting too fast. Is 400mm/min a reasonable feedrate for a 6mm-deep cut with an 1/8” mill in fir plywood?
2. I could slow down the feedrate if I slowed down the spindle. I’m thinking of buying one of these (but with some trepidation about putting my Dewalt router behind a Harbor Freight knob :v). Is there any reason to think this would help? If I slow the spindle, then I’d think maybe the lateral forces would be as high as with a fast feedrate and a fast spindle. So maybe that’s barking up the wrong tree.
3. Maybe I’m using the wrong tool. I bought these. Do they look like the right thing? Or does anyone have a link to a better tool to try for initial playing around?
4. I have a pretty big build area — I built the 25×30″ recommendation from the getting started guide. I can wiggle the X carriage 1.5mm by pushing down on the belt. I can’t tell how much of that’s because the belt is stretchy (it doesn’t seem very stretchy), and how much is due to the loop of zip-tie that secures it to the corner block deforming from oval to a longer oval. In any case, I could imagine this might be an important source of slop. Possible experiments:
4a. Zip tie slop: Replace the zip ties with a printed belt tensioner, so that the only flexible part in the system is the belt itself.
4b. Belt stretch: Create a tensioner end that bolts onto the fixed X and Y rails at some distance (say, halfway) to limit the carriage travel. This would let me simulate having built a smaller addressable area, at least from a belt stretch perspective.
I’d really appreciate any feedback. I’ve gotten through most of the easy things to try; these further experiments are going to be more hassle. So if I’m chasing the wrong squirrel, please let me know!