Z-axis design has reached the lead screws! The gantry frame will constrain one end of the the X-axis lead screw in all degrees but rotation, and the opposite side will probably just include some axial preload for the lead screw support bearings. To avoid overconstraining the carriage, the lead screw nut needs a bit of radial compliance. I also want to preload the nut axially, so on the far side of the radial compliance I also need to support a second nut and add axial compliance with some means to adjust force. And it all has to be quite small!

Here's a first test:

The top nut gets a pair of rotationally offset flexures to provide some radial compliance, while the thickness of the flexures should keep it pretty stiff axially. The lower nut mount adds some axial compliance via a bunch of fairly low aspect ratio beam flexures; note that its base is attached to the same bit of material that mounts the upper nut. I'm hoping to take up any radial misalignment between the two nuts by adjusting the mounting bolts; the alternative is to give one nut radial compliance and the other both radial and axial compliance on separate pieces, but that gets even larger. I'm printing this flat from the top nut side (oriented like the first picture) so the arc-shaped flexures are in the preferred direction; not ideal for the lower flexures, but a preview of the slice suggests that the layer lines are going in the right direction:

I tend to print with ~5 wall loops, so the inside layer of the flexures seems okay:
Some support material, obviously, but it should be pretty easy to clear out. The lead screw nuts I sourced are M3 tapped so I just provided clearance holes, along with larger openings so the heads would clear during assembly. This is an 8 mm pitch, 4 flight lead screw, so it should assemble nominally as long as the face-to-face dimension is a multiple of 2 mm. I intended to make it short by 0.2 mm to apply a nominal preload but it came out to 24.8, probably after last-minute modifications to an earlier operation. That is okay, as it will let me play around with preload a bit more using shims, or just by letting the bottom nut float a bit.

This worked pretty well! The extra space between the base of the preload nut and its support didn't matter, since all that nut needs to do is exert axial force against the lead screw. This could be simplified and shrunk down quite a bit since I really don't need to rigidly mount the preload nut to PLA. I like being able to adjust the preload with a hex driver, and could always add a jam nut to keep the setting.

I added a bit of lithium grease and was able to make backlash quite difficult to detect by manually pulling the lead screw out, but this required quite a bit of tightening. I was able to snug the preload bolts down two full turns and could still rotate the lead screw (albeit with some difficulty), so I may try thickening up the flexures a bit so the preload is stiffer. I need to add a bit of clearance to the bores that hold the nut bosses, perhaps a few hundred microns. And generally shrinking will be important, as it's too big for the Z-axis right now:

Given flexure directions and assembly challenges, this part will bolt onto the back of the Z-axis using a few bolts from the front and maybe with some locating features. That helps with eventual Y-axis design too, of course, since I can re-use the part on the bottom of the machine!

Other unrelated minor update: I ordered a pile of 40x40 ENIG-coated FR4 coupons.