Following the CNC bootstrap, which was an unqualified fail, I was left with the choice to abandon this venture and save up the 1000 USD++ cash to buy a kit in China or else try and address the failings. As you will surmise from the image above, I opted for the latter.
The big learning from the CNC bootstrap is that the secret of CNC machines is in their rigidity. The structure has to be as rigid possible and the play in the bearings has to be minimal, i.e. in the micrometer range. The CNC bootstrap had aluminium rods and the bearings were fitted with duct tape and tie wraps. The wooden structure wasn’t rigid at all. All this resulted in a cutting precision of about 10 mm. In the workshop that would be described as “complete absence of precision”.
I spent a lot of time on eBay to decide which parts I should be investing in. There was a very tempting offer for just over 300 USD to get all the linear bearings and rails and a set of ball screws from a workshop in China. But that was a bit like buying a kit rather than design my own machine so instead I went minimalist and I opted for a set of linear bearings and rigid rails from an outfit in Germany (cnc-discount) who delivered a set for 55 euros including shipping in a few days only.
Since the x-axis is where it all starts (alpha and omega stuff), I decided to rebuild the entire CNC from scratch. Since precision was a requirement now, I designed all the pieces on a computer first and cut the jigs from paper printouts which I used in cutting and drilling the components.
The y-axis was also very wobbly in the CNC bootstrap. It was made of two 8 mm aluminium rods that were simply too flexible. I replaced them by steel rods. Since I saved 6 linear bearings from the original x-axis, I decided to used 6 bearings for the y-axis (rather than 4). I also scouted around for a strip of hollowed wood that could act as a seat for the bearings. I found a wood beading that was perfectly matched and decided – for better or for worse – to retain the tie wrap solution. This is what the Prusa 3D printer design uses as well though in fairness, 3D printers don’t carry as much strain and weight as CNC routers do.
Another weak point was the spindle mount. In the CNC bootstrap I had just tie wrapped them to the z-axis plate and it moved all over the place. I needed a proper aluminium mount bracket but I had neither the aluminium nor the tools to make a good bracket. I went for wood instead and was surprised at the rigidity this produced.
Another ‘innovation’ is that I back-to-backed the y and z-sledges. I haven’t seen this on any of the CNC routers that I have studied but it makes perfect sense: it increases the z-clearance significantly and – at least in theory – it shouldn’t have any impact on rigidness.
Below you can see the final rig before testing.
If you look carefully, you will see that the x-axis screw is on right side of the router only. I didn’t want it in the middle since that requires an additional ‘table’ above the screw which takes away another 5 cm of z-clearance and adds calibration issues since no table is ever perfectly horizontal.
I had prepared a few test files for cutting fractals (what else!). But in the end I used the ‘Hello’ file for comparison.
OK – it is still crap. But on inspection, the precision is about 2 mm. That is 5 times better than the bootstrap. Yay!
But it isn’t in the micrometer range, which is what a decent CNC router should produce. I need to gain another factor 10.
The biggest culprit of the 2 mm play is the asymmetric x-axis screw. I need to add a seconds stepper motor and screw on the left side of the rig. This will almost certainly reduce the error from 2 mm to less than 0.5 mm. Now we’re talking.
Another source of error is in the y-axis. Although the steel rods are a huge improvement over the aluminium rods, they are still not rigid enough. I am considering replacing them by the sort of rails and bearings that I used for the x-axis. These are as rigid as is possible and the bearing play is minimal.
Lastly, my spindle is too simplistic. It is just a hobby gadget that I once bought for drilling PCBs with sub millimeter drills. It isn’t suitable for a CNC router. So I need to buy a powerful spindle.
The last element I’m pondering over is the z-axis. It works well enough but I feel that sooner or later it will need to be replaced by a solid aluminium & steel assembly. Not sure I want to do that yet.
I haven’t mentioned any of the electronics, such as the stepper motor drives or the CNC controller (Arduino). These are working fine although the stepper drivers need some tuning since the motors make horrible buzzing sounds. The cabling needs to be sorted corrected and once the CNC router is working correctly, I will design and build an integrated controller & stepper motor driver. But I need the CNC to cut the PCBs’ so that is for later.
To be continued.