I very recently completed rebuilding the y and z axis of my homebuilt CNC router. However even as I was finishing that build, I knew the z-axis design was no good. After quick testing I could see I was right, it ‘worked’ but really wasn’t much improvement on the mk I design. This weekend, I rebuilt it again using the ideas that I developed when I rebuilt the y-axis. It now works great.
First some history.
The original design of both the y and z axis was based on the same idea. two identical metal plates would form either side of a carriage, the threaded rod would pass through a tapped hole in the centre of each plate and thus they would be driven. At the ends of the metal plates I drilled a hole which mounted two 90 degreee brackets at 45 degrees to each other. In the brackets I bolted skateboard bearings. The linear rails were made from lengths of square section, mounted parallel with corner the of the square section facing the other rail. The carriage plates fit between the two with the 45 degree mounted bearings then running flat against the faces of the rails.
This design allowed both linear travel along the rails, and provided a clamping effect to stop lateral movement. At least that was the idea. The problem with this design is that it really required a lot of precision all over the place. The first issue was that the tapped holes that the threaded rod passed through were not perfectly aligned, and the rod could not easily pass through both and still turn. Over various iterations I first enlarged one hole, then both and replaced driving effect using a nut mounted in a wooden block.
The theory of the clamping effect, also failed to operate perfectly, it was too hard to adjust all the bolts to be tightly secured and clamping with sufficient force. The slight mis-allignments in other things meant that there was always a small amount of wobble in both axis.
To be fair this first design worked enough, and I was able to use the machine to make a few things, including parts for the mk II z-axis. However it was apparent in usage that the machine was too prone to having the router tip pulled off course, there was too much twist and wobble in the design and this caused backlash effects, particularly under lateral cutting loads. Drilling operations were mostly find and accurate, however the z-axis would visibly rock when moving up and down. This has the effect that when deep in a cut, retracting the router would wobble the cutter back and forth into the walls of the cut, not terrible, but far from ideal.
The Mk-II
I wrote all about the Mk II in a previous post. I focussed on the lateral wobble issues, and realised the machine worked well enough to help me build a new z-axis. The idea was formed on 4 ‘layers’, a back plate, a front router mounting plate, and 2 carriage pieces that would clamp driving nuts inside, along with bearings at 4 corners. 4 bolts would run through each layer clamping everything together and forming the axels for the bearings.
Basically I hoped that the carriage would be held tightly between the two, now square on, linear rails, and the back plate and front plate would simply slide up front and back flat surfaces, resisting flex in the +/- x direction. The whole thing would be nicely centred on the rails and the z-axis motor would get a nice straight alignment.
I also realised that in order to mount this new z-axis, I really needed to make serious adjustments to the Y axis. So having cut the pieces for the z-axis, I started thinking about the Y axis.
Here I came up with a much more elegant idea. I had been thinking a lot about my (in)ability to get a high level of precision, so my design allowed for that fact. The other post has all the details, but effectively it relied on lots of m8 threaded rod, and lots of nuts to allow positioning.
The Mk-III
No sooner had i tested the Mk-II than I knew the z-axis needed to change again. The rocking when moving up and down was still an issue, and if I tightened things enough to resist wobble, the back plate and front plate added too much friction for the motor to overcome. Again it worked enough to produce a piece, but by now I already knew I could do better.
I knew I needed to make space to mount bearings behind, and in front of the z axis rails. the way I had them mounted there was not quite enough room for a bearing to run behind. so out came the dremel, and I cut away the y-axis wood platforms, to make room for a bearing. Fortunately there was enough ‘waste’ wood to cut away without compromising the y-axis.
with space cut, I could make up some threaded rod with bearings mounted at the appropriate spacing for the z-axis rails. I had the rod left over from making the y-axis. The initial theory was to use the same kind of platform I had in the y axis, with a bearing mounted on a right angle bracket to run along the sides. However, in a classic case of design by limitations, the remaining lengths I had were not long enough for that idea. So instead I realised I just needed a way to mount a threaded rod at right angles, which I could put another bearing on. For this I could use a small amount of plastic (not sure what kind, its black and used for machining, I got given a box of off cuts a few years ago) drilled with two holes at right angles to each other, one above the other.
this lead to something that looked like:
This design looked great, it used very little room, required only a very simple machining operation from me to make the plastic blocks, and the rest was just a question of tightening nuts in the right places to space bearings around each side of the rails. I could re-use the central 2 layers from the mk2 z-axis to run between the rails, and mount to the new runners.
This means I have bearings running tightly against every surface of both linear rails. providing a nice tight, wobble free linear motion.
In one of those happy accidents, the long bolts I used for the outside edges of the runners had the same spacing as the slots I had cut in my router mounting plate. So they passed right thorough and allowed for easy mounting by just bolting it on. All of this added about 25mm depth to the axis configuration accommodate the bearings running along the front, but I’ve easily gained in overall rigidness of the structure.
The new mechanism slides up and down very smoothly and in a test I was very happy with how it performed.
So with the mk III z-axis and the mk-II y-axis running great, it’s time to turn my attention to the mk-I x-axis. It is now possible to cut a test circle that shows a nice smooth direction change on the y-axis, and decidedly un-smooth transition on the x-axis. I’m hoping that by applying some of the ideas I’ve developed on the other two axis, I can get the x-axis running better and start to really turn my attention back to making things.