@mcdanlj Huh. Are the printed parts maybe just very slightly not-straight, and the longer bearings have leverage to self-align better?
I would try putting a very slight taper to the hole, so the very center is tighter than the ends of each clamp, so the exact alignment can “float” slightly.
In fact, as a check, might go all the way to a 1mm inset “waist”, just to see what happens.
… and if that works, will have to update my all-plastic design. 
@Preston_Bannister I’m fairly confident that they have some slight trapezoidal error from stiction—I’m still running on the accidental test of printed plastic bearings. A 1mm waist is indeed exactly what I was thinking about for an alternative if running with the 30mm brass/graphite bushings doesn’t solve the problem. But first, I’m planning to see how easily they run with the graphite/brass bushings, and probably print new sets with better bearings to compare. By then the drylin bearings should have arrived and I can test them as well. I can also compare to putting the drylin bushings into pillow blocks.
Yup. About 1° trapezoidal error. I think the printer has a design flaw making it tricky to keep square. I don’t think stiction is the main culprit here.
Five iterations that I’m not using. First iteration trapped bushings between end caps. Second changed to clips. Third moved clips to end. Fourth fiddled with clearances. Fifth had screw-on brackets too small and too much clearance inside. Sixth iteration installed but a bit late to test printing with tonight.missing/deleted image from Google+
@mcdanlj Heh. Iteration. At the end of the day, that might be biggest deal about 3D printing. You would not go through as many iterations, if you had to fabricate by hand. That means you get to better designs, more quickly, and likely your design skills improve. Then that better-design is easily replicated by anyone sufficiently skilled in using a 3D printer.
That snowball effect should be the main story of 3D printing.
Truth. I can spend as much time with lathe or mill making a part as I do in openscad designing a part to print… Though with substantial differences in precision!
@mcdanlj Heh. Read your comment two ways, though expect you meant the first.
For a simple part, with a single or few critical dimensions, pretty sure manual fabrication with a lathe or mill is a lot more precise.
For a more complex part, with a chained series of interrelated critical dimensions, manual fabrication would be a nightmare. For a printed part, chained tolerances do not cost near so much. If we can design for looser tolerances, we can design more complex things than practical with manual fabrication.
And replication is almost free.
As a profession, we have several decades of designs meant for old forms of fabrication. Calls for some refactoring of how we design.
Makes for an entertaining topic.
OK, printing the silly plastic bushing, to see if they work. No immediate plans to put on my i3 clone (the existing bearings seem OK). Found a (mostly) smooth rod with which to test.
Was watching a video, and at this point:
… he shows a bearing design not suited for 3D printing. Was more(!!) than a bit annoyed. The design is not suited for a 3D printer. We should use designs suited to the materials and method of manufacture.
BTW, my X5S is sitting on a desk I build in 1981, as a poor college student, with a handsaw, electric drill, and limited skill. Bit battered, nothing fancy, but solid strong and true. Designed something I could fabricate easily, and it worked.
Let you know in a few hours…
Sorry, just saw that google had “helpfully” marked your comment as spam without notifying me; I just fixed that. Now I know what set off your separate thread!
My printers sit on a desk that I designed and built shortly after I graduated from college, with a saw, electric drill, and limited skill.
I’ve been thinking about my trapezoidal error problem. I think that it is more than the printer being out of square. I think that I need to lower X acceleration and jerk as well as Y acceleration and jerk. If that is the real source of apparent trapezoidal error, I should print the clips larger rather than smaller to reduce the impact.