Levinslaw / Spicy-CR-10

Quest for the limits of a simple ultrapolymer 3D printer
https://levins-law.github.io/Spicy-CR-10/
0 stars 0 forks source link

Wobblex kinematic couplings for Z-axis lead screws #4

Open levins-law opened 3 days ago

levins-law commented 3 days ago

Problem

There are artifacts/flaws in finished prints of a periodic nature consistent with lead screw pitch.

Click me for details of the problem

The periodicity suggests the lead screws and nuts are not co-axial at all times. Even with dual Z-axis lead screws and independent stepper motors on each lead screw there are obvious artifacts.. MirageC9 (Hevort) developed [isolator couplings called Wobblex](https://github.com/MirageC79/Interfaces-for-WobbleX-integration/blob/main/README.md) that decouple the lead screw and nut from the X and Y axes. This coupling doesn't work well if the load (gantry) isn't heavy enough on both sides to maintain adequate preload on the coupling. > - #2 has been undertaken with the aim of increasing mass and rigidity of the X-gantry. This was done in part to make the gantry weight symmetrical for each lead screw (excluding the toolhead, which moves)

Solution 1 - Wobblex WS8 couplers from Mellow

image

Implementing the Wobblex couplings using the center swashplate manufactured/sold by Mellow requires printing the other two baseplate components out of plastic. This is a failure mode in a high-temperature printer so Ultem 2300 (1010 + 30% GF) was used instead.

[!WARNING] The fit of the steel pins was too tight in the Ultem base plates. I did not have the patience to iteratively reprint the base plates with different slicer settings to loosen the pin bores. Instead I drilled them out (by hand like a cave man) with a 3mm drill bit. This resulted in several hundredths of an inch of skew when the assembled baseplates/pins were measured with a dial test indicator on a surface plate.

image

It's possible this wouldn't matter, but the goal is to reduce uncertainty where certainty can be achieved compared to other areas where resources are limited.

When the assembly was test-fitted on the printer everything checked out in terms of clearance.

20241017_192926

Solution 2 - Funnsor all-metal WS8 coupler set

20241027_125204

It has been some time since the Mellow couplers came out, and Funssor has produced a set that includes all three pieces machined from aluminum. This saves the hassle of machining my own from scratch.

[!WARNING] The pins on the Funnsor version were checked on a surface plate and found to have worse misalignment than the self-printed versions.

I noticed the pins had been glued in place to retain them. The adhesive appeared to be cyanoacrylate. Presuming that was the source of the misalignment I soaked the assembly in acetone which loosened the pins and dissolved the extra adhesive.

After placing the cleaned pins in their channels there appeared to be a poor fit in the the ball-nose channels. They were either machined poorly, or the tumbling process ruined their finish. The slots were too narrow for the pins in some cases and there was light visible between the pins and the rounded bottom of the slots.

[!CAUTION] The Funnsor plates lacked the M3 counterbores in upper base plates needed for each lead screw. It's not clear why this was omitted. It could be attributed to variations in the different versions MirageC79 created to suit different printer designs.

20241027_135109 marked

Workaround

Rather than modify the X-axis gantry thrust flanges again, I threw the base plates in the milling machine to drill some M3 clearance holes (3.6 mm) and used an M3 counterbore tool.

Clearance holes first:

20241027_142113

Counterbore 3.5 mm deep:

20241027_143021

Note the very rough surface finish in the pin channels. This operation was prior to identifying the poor pin alignment.

20241027_143621

Finished counterbores:

20241027_145843