I've been quite busy over the summer, and with the lovely weather I've been less inclined to spend any time locked away in my workshop with no windows. However, I have been designing and thinking about my CNC. In my previous experiments with my machine I found a few issues, which I was keen to fix. The first issue was in the Y-axis, the motion was not smooth. Moving back and forth it would sort of pulse in the direction of travel rather than slide smoothly. It was also not a great alignment of motor/shaft/driving nut, so some parts of the workspace were more reliable than others when it came to moving the y-axis.
The z-axis had a similar pulsing problem, but also the mounting for the router had some issues. I had originally just hard screwed the rotary tool in place on a piece of mdf, and that was only really supported at the top end where it attached to the z-axis sled. When I tried to make the machine cut faster, what I found was the increased forces caused the mdf to bend where it was unsupported, and also flex laterally allowing the cutting bit to cut off course. I had this happen inall the materials I tried and once pulled off course the job is ruined.
So I set out to redesign my whole approach to these two axis. When I first had the idea, I was working on the principal of 2 bearing mounts, at 45 degrees to each other, riding along the corner of a square section linear rail. This pattern repeated at the four corners of a sled. In theory this gave linear slide as well as holding against lateral movement. But the reality was that whilst that could have been possible with perfectly engineered pieces, my actual construction had too much inaccuracy and there was wobble that I could not eliminate. I was also working with a very limited budget and tried to achieve what I wanted with relatively few bearings.
Now that I stopped to rethink my design, I was prepared to spend a little more (though still no where close to the cost of a 'real' axis.) I also wanted to minimise the precision requirement, or rather, make it easier to adjust precisely.
I came up with an idea i sketched out in my maker's notebook
basically there are 4 threaded rods, joined by a series of mdf platforms spaced along their length. each platform mounts some right angle brackets that hold bearings that will run along either the top, or bottom, of the square section rail. The threaded rods are spaced so that I could also put a bearing on the rods, and they would run along the front and back of the square section. Such that there is a bearing running on each side of the rail to hold tightly against any wobble. Every part, the platforms and the side bearings, are held in place by nuts on the threaded rod. These nuts can be adjusted precisely up and down to make sure each piece is aligned and at the right height. Once positioned, they can just be tightened in place.
In addition the top mdf piece and the bottom one, extend out to the front to support the vertical rails of the z-axis. The only accuracy required in this was to get all the holes in all the mdf platforms aligned. This is simple, since you can just clamp them together and drill through in one pass. The main issue was making sure the spacing was correct toallow the side bearings to be snug against the rails, not too close, and not too far. However, worst case I figured i could widen the holes and use the nuts to clamp things in position at the optimal point. As it was I'm happy with the spacing and everything holds really tight. The y-axis carriage slides back and forth with no wobble. I was also able to mount the driving nut on top of one of the platforms, aligned for about the middle of the carriage.
For the z axis I had a different idea, here I wanted the rails to basically clamp a sled between them. a hole passing through the centre of the sled contains the driving nuts perfectly centered. I actually machined the bits for the z-axis on the CNC machine with its original setup. In order to keep the overall depth of the axis as slim as possible, there are no bearings front and back, there are some strips of acrylic set to press up against the front of the rails, hopefully allowing a fairly tight fit, whilst also permitting the whole lot to slide even under some pressure. I used acrylic so that I have the option to lubricate the rail to reduce the friction if necessary.
The last part of the new z-axis was to provide a well supported mdf base with slots to allow interchangeable router mounts. I have started with a mount for my bosh palm router. This means I have an easier ability to switch between big cutting jobs with the powerful router, and more detail work with the rotary tool. This mount isn't too easy to switch, since it does need to be clamped hard in place. but the option is much better than the previous screwed in place hack I had.
All told I'm very happy with the y-axis, it runs really smoothly and has no wobble. The z-axis is less great, the vertical rails are not quite perfectly parallel, and that is allowing a small amount of lateral give, which is magnified at the tip of the router. I will need to work to correct that. However it is still a substantial improvement on the previous designs. Whilst still being very low cost.
One significant factor in all this has been the motor couplers. I have been using some 3d printed ones, but they simply aren't perfectly aligned, and this seemingly small inaccuracy at the coupling point causes all kinds of issues. I have replaced the y-axis coupler with one I machined myself from a machinable plastic. Basically I drilled half way with a 10mm bit, then the rest of the way through with a 5mm bit. Then drilled and tapped two small holes to allow tightening screws to hold the shafts in place. This gives a perfect alignment, but obviously has no give if other things are out of alignment. This seems to work well, however after all the trouble I've had with couplers, I've bitten the bullet, and ordered 3 real metal couplers. I now see how much of the difficulties I've had come down to the misaligned couplers. So perhaps they are worth the money.
The other thing I'm taking the time to do, and the reason I started with the mount for the larger router, is to make sure the bed of the machine is perfectly flat with relation to the router movement. Previously I struggled with height difference from one end of a piece to another. So this was first on the list of things to do. Along with making sure I have a better way to mount work pieces than just screwing them down.
Here you can see the machine with the new axis, flattening the work area, and starting to machine slots through which I will be able to use bolts.