As my previous router was a huge flop, and I still wanted gloriously milled PCBs, and as Homofaciencs shows that you can make a working CNC from even the crappiest of materials, I collected my courage and tried once more. I now knew what the two biggest problems of making a cheap DIY CNC were the linear guides and a precise way to move the whole thing over said guides. I was inspired by this article about how to build a cheap CNC router for the least amount of moneys possible, just as I intended to do, and seeing how his problems coincided with mine. Moreover after reading an article about how another person again failed with exactly the same type of CNC, which had some tips about possible improvements, (which I sadly could not find anymore) made me realise that I simply could not build a okish quality CNC router without at least buying some decent quality lead screws and that I should ditch the floating gantry idea.
So what next?
After a quick search on Ebay, I landed on using a simple lead screw for the z-axis and some GT2 timing belt for the other two axis. This should be precise enough to get my end goal of milling PCBs while still being sturdy enough to also be able to get some wood cutting done. Next is the linear guide problem. The slop was mostly due to the fact that I could not get the 45 degrees cuts right with my current tools and carpentry skills. Moreover every slight deviation in length in two parallel wooden pieces would reflect in a linear guide that would not be linear. Instead I decided that I would use my local construction market to cut as much identical pieces of wood using their much more easy to use and precise panel saw.
Back to the drawing board.
Keeping these two points in mind I made another design for the wooden CNC machine. This time using Fusion360 instead of Solidworks due to its integrated CAM compatibilities.
The frame and consists of a single wooden panel of 60×56.4cm (60cm minus 2 times the panel thickness), nine 60x15cm (or for the americans out there the 23.62” by 5.91”) pre-cut wooden panels and some other shorter 15cm wide pieces of various length cut from some extra 60x15cm panels. The sled for the y-axis will rest on the large panel with the previously used bearing system. However this time no 45 degree grooves are cut for the aluminium L-profile. Instead the profile just rests on top of the panel and is attached with several screws. The x-axis sled uses the same aluminium profiles as before. However this time the bearings are attached differently, such that it would be relatively easy to detach the sled. The z-axis uses two 8mm iron rods as rails together with four linear bearings and is semi-permanently attached to the sled of the x-axis.
Enough talk, lets build it.
Lets start with the base. Two sets of the 60×15 panels are butt-joined together with some M4 bolts. This creates a sturdy frame which is exactly square. Next the bigger panel is screwed in the middle of this frame in such a way that the sled of the y-axis is flush with the top of the base.
Next the gantry is installed with some over-sized holes on the bottom in order to be able to adjust the gantry such that the z- and x- axis are square with the y-axis. Afterwards two L-profiles are mounted on the inside of the wooden H-profile and the sled for the x-axis attached. As I “forgot” to add an easy way of attaching the sled inside the H-profile (aka I was too lazy to think of a good solution), I had to use some wood clamp trickery instead.
Next I attached the sled of the z-axis to the two iron rods and added a stepper motor on top. After that I attached a spindle to the z-axis and finished the sled of the y-axis by securing the aluminium profiles and installing two stepper motors. As a last step I added a timing belt to the x- and y-axis and put a waste board on top of the sled of the y-axis.
Lets test it!
Next is the most fun part the first run. However before testing, I wrapped everything up by connecting the stepper motors to an Arduino running the grbl software. Finally I put in a random drill bit and ran a piece of test code on a piece of leftover wood of my failed first attempt.
IT WORKS! However it still needs some work in order to become truly an useful tool. As it currently neither has homing or automatic height probing. Moreover there are no end stops and the most important button of all; an emergency switch when disaster strikes. Maybe I will post an update in a future post somewhere in the Christmas break.