An update on my fiddlings:
I tried the suggestion of removing the set screws, blowing them out, then replacing them on the Y axis pulleys... and managed to scare myself half to death. After reassembly the motor would just give off an angry buzzing noise any time I tried to make it move in the Y plane.
Two hours of loosening, fiddling, tightening, buzzing, loosening, fiddling, tightening, buzzing, etc... later I finally fixed an entirely different problem.
When my Solidoodle first arrived it would hit a point in the +Y direction where it would give that same angry buzzing and not move. I regreased the rods, tightened the belts, and played with it until it (mostly) stopped. It would still do that every once in a while, but so rarely that I decided I'd just forge ahead.
Now, though, I think I've figured out what was causing that and (maybe) fixed the root cause of the shifting Y axis.
It seems that the motor and the first pulley in the Y stage weren't aligned. They were really, really, really close. But off just enough that as the motor moved it would slowly move the belt along the pulley itself. Eventually it would hit a point where that belt would reach the end of the pulley and start to bind on the motor, making it unable to turn and produce an angry buzzing sound instead.
Why would it only do this sometimes, and only at the very back of the Y stage? I'm not really sure, I don't know enough about gearing and alignment in situations like this to say for sure, but that misalignment seems to have been the root cause (and after all my fiddling the alignment is now right, so the root cause is gone).
My first attempt to run the same print resulted in a very clearly jumped tooth on the Y axis, so I checked the belts and now I'm trying again. It'll be a bit before I know for sure if it's fixed, but I'll check back in at the end and we will see.
On the other suggestions:
I've always felt iffy about the included PVC spool holder, and especially about threading it through that rear hole. I plan to print this: (apparently I can't post links, but it's thing:36931 on thingiverse) when I get a chance because that looks like a much better solution to me. I don't know if the filament tension was causing this problem, but I've felt sure it has caused some intermittent layer alignment issues I've seen in the past.
For the Slic3r code, I have most of that, but I don't issue G91 at all, I just set to absolute system then move the platform down and find home. I'm game for trying incremental then absolute like that, but what is the effect of that, and what is it intended to fix?
