Piston tolerances

I have what amounts to a cylinder and shaft which needs a tight fit but to allow smooth motion as well. Technically that isn't quite right - the 'piston' is actually curved, but close enough. The last two prototypes I did of this both make the piston be .13 mm smaller than exact in order to allow it to move (the pieces aren't touching when they're printed so they don't fuse). The weird thing is, one prototype is very loose, and the other requires significant force to move, even though that part is exactly the same in both of them. (Yes, they're both properly worn in).

Does anybody know which of these runs is more typical, and if there's anything I can do to get consistent results in the future?

Please, stop posting the same message over and over (6 copies so far, 01:38, 01:09, 00:09, 22:54, 21:51 & 21:43 UTC). The servers have issues, but sometimes the posts arrive, you just get an error saying something (else) failed. Check the index before trying again.

About accurancy, SW ships models without meassuring them. It's more art oriented than engineering, while what you describe sounds like something that is normally solved by tight specs (WSF is +-0.15 mm, and surfaces vary with orientation), multiple parts in different materials with different properties (piston rings, oil) working together, and even mix & match the parts that work or tune until they fit.

You could print everything bigger than needed, and then file down to the right size. Your best bet otherwise is design the parts so they have some kind of lips or other flexible shapes to do the matching (hard to guess without better description of what you have and what it should do). Or replace with non 3d printed parts, like people using metal thread insets instead of modelling the threads.

Sorry for the repeat posts. I got a 504 Gateway Time-out every time, so I just assumed it hadn't posted. Unfortunately there doesn't appear to be a way for the original author to delete a whole post.

The shape I'm using is, uh, kind of hard to make with any technique other than 3d printing. It's for a puzzle, and is intentionally something with exotic properties.

I'll try increasing the gap and cross my fingers that that works.

Yes, 504 is the typical error in forums. Other things around SW are images or CSS that fail to load, or models that vanish for some minutes.

Increase the gap and then add some kind of pattern that wears down with use until it slides smoothly? Imagine some bumps that are as big as original size, half spheres, the contact would be just their top zone, and with use they will wear down to a river head shape matching the real size. Or try springs... like in banana plugs.

What material did you select?

Sorry, forgot to mention that it's WSF.

Unfortunately it needs to be fairly robust once it's worn down, because people are literally banging on it. It is, after all, a puzzle.

Banana plug shape for the cylinder or leaf springs protuding from the shaft walls, maybe even some kind of torsion force to the cylinder as it slides into the shaft. All those would wear down, but also make use of material flexibility to keep working. Hard to tell without clues about dimensions or shapes, banana method would require some mm to build a bunch of separate wires, simplest "banana" would be a prism with slit from side to side, so say 1mm for wall in each side and 1mm hole, 3mm.

The bumps trick works for items sintered assembled, with a "big" separation all over (to avoid fusing) and them some tight spots (at fuse limit) caused by the bumps.

The shapes are a little hard to describe, but the 'piston' is basically toroidal, and the hole it goes through is flared on the sides to accomodate it exactly. The two things move in a full two dimensions as the object is manipulated, in/out and clockwise/counterclockwise. There's no way to keep it from simply wearing down as much as it possibly can from manipulation.