Been digging through my K40 during the rebuild and several things struck me.
The design of the y rail is good in that it uses only one linear slide rod on the right and only a vertical “capture” on the left. Geometrically, a line and a point define a plane. This arrangement forces the Y axis to determine the line of motion of the X axis, with no possible binding from misalignment of the right side.
In theory at least, only the joint between the X rail and the left side of the Y rail controls orthogonality of the X and Y axes. It’s still possible to cause racking by misaligning the left and right Y belts, but then nothing is perfect.
The Y linear rail and the captured right side of the X rail determine the plane of movement of the laser head, assuming that the X axis rail is straight. It’s not, but within the limits of roll forming sheet aluminum, so it’s probably OK for the intended purpose of a K40, engraving rubber stamps. The Y linear rail is pretty straight, coming from a well established industry to create non-binding, straight linear motion. You could find a high precision replacement and sub it in there if you wanted really precise Y motion.
The Z height of the X axis is fixed on the left side by its sliding attachment to the left side Y rail. The right side of the X axis is fixed in height only by the “capture” rollers on the frame assembly, so it’s Z height is a variable, depending on the straightness of the frame extrusions on the right, but critically on the degree to which the frame is aligned to the Y rail. If the right side of the frame is not parallel to the Y axis rail, then the X axis will vary in the Z direction as the head moves along the X axis.
So getting the outer frame aligned with the Y axis rail is critical. Focus and apparent power will vary over the cutting area if the frame on the right side of the X axis is not parallel in the Z direction with the Y rail. This additional complication cannot be tweaked out by the mirrors, and is probably one reason that mirror alignment can never be perfect.
The laser tube is not mechanically coupled to the X-Y motion excepting through the outer case. The outer case is not rigid, so moving the laser to a different surface will cause some misalignment of the laser tube to the XY path, magnified by the distance the beam travels. So beam wander is worst at the front and right of the machine, and especially at the front-right corner of the available area. To a smaller degree, the mounting of the first external mirror to the case does the same thing.
In an ideal setup, the machine would be made with as rigid a frame as possible, and the frame would include mountings for the laser tube, mirrors, X and Y rails. The frame ought to be isolated from the outer enclosure with compliant mountings to keep external forces on the case from twisting the frame. This would obviously cost too much for the price point of the K40 laser, but it’s nice to know.
A (reasonably) rigid frame would allow shimming of the XY motion rails to get better orthogonality. The K40 as is will always be a compromise and need tweaking into acceptable accuracy. But we knew that.
The X axis motion of rollers on a rolled support member is OK-ish. It’s in a precision class comparable to careful hand work with pencil and rulers. But it’s easy to improve. A Harwin-style MGN12 linear guide rail on top of the X rail and some tinkering with remounting the head gets it to being much straighter, at least in terms of error per unit length. It doesn’t do much for curing the twisted-frame issues above, but it will cure a lot of the minor wobble and backlash issues on X without a lot of redesign and rebuilding.
The K40 really needs a good air path. As it sits, it has an air path, air coming in through the power supply cabinet and the bottom hole as well as the cracks and seams. But the stock air “path” isn’t all that conducive to removing combustion products. If the outer case were not so important to what mechanical rigidity the K40 possesses, it would be tempting to put a ~1" slot across the front of the case near the focus plane and perforate the frame somehow in front to allow the exhaust fan at the back to pull air easily across both the top and bottom of the work area. Lack of a rigid and low-cross-section front frame makes this problematic, too.