Homebuilt CNC Router #1: A Journey

After posting my Woodworker’s CNC Router in the projects section, I received several requests to show how I made it. I debated whether to post something here or just send them off via a bunch of links. I decided to show you the journey that I took over several years and ended up with the machine that was posted. So here goes.

Do you need one? No.
Should you build one? Probably not.
Have I done anything useful with it? Not really.
Is it cool? Oh yeah.

In 2004 I stumbled on an Internet advertisement marketing a book “Make your own CNC”. I did not realize this was possible. It got me googling and I found a free set of plans call JGRO and a bunch of guys building them. I ordered some stepper motors and a controller board kit, and hit Home Depot. This is what resulted.

This is called a gantry style machine because of the overhead structure, as compared to a moving table type machine. It is a three axis machine, X is the long one that the gantry moves along, Y is going across the gantry, and Z is up and down. It uses roller skate bearings running on black gas pipe. These bearings were mounted in aluminum angle to form a V. The pipes were suspended with adjustable bolts at the ends. It used threaded rod for the leadscrews and homemade leadnuts made out of delrin plastic. I built it in one mammoth weekend.

I bought Mach3 (it was really Mach2 back then) as the controller software to take XYZ commands called G-code and turn them into step and direction pulses. I cobbled up an old PC and dedicated it to this purpose.

You can see the picture of what is being cut; a new router holder for the Porter Cable laminate trimmer.

Machine #1 was terrible, but I was one proud puppy. The machine flexed like crazy, it was amazing how much gas pipe will flex over a four foot unsupported span. It was incredibly slow and the leadscrew whipped like a jump rope. The backlash, or the amount of slop in reversing directions, was unacceptable.

This was quite early in the home CNC game and I could not find much on the subject, so I decided to attack each of these problems; flex, whip, speed, and backlash and come up with my own solutions. Enter the prototype for machine #2.

First attack the flex. The pipes had to be supported and some better construction techniques used. I turned the machine inside out, and supported the pipes with torsion boxes.

I used machine #1 to cut some struts for the Y torsion box.

I needed a way to keep the bearings riding on the pipe. So I came up with an adjustable plate that would press the bearings against the pipe by tightening screws captured with T-nuts to move the plate.

The assembled gantry with the Y carriage:

I added the same adjustable plate assembly to the inside of the gantry to ride on the X axis.

I mounted the gantry and the X pipes and checked for a smooth ride and that everything was parallel. A huge success. I could skim the clamped ruler to every corner of the table without an issue.

Now for some stress testing. To just add all kinds of weight I found a rather large motor and clamped it to the Y carriage, then repeated the skim testing.

It bound up. By adjusting the pressure plate on the bottom of the gantry, I was able to bring it all back to parallel and complete my testing. Yeah!

But the gantry was under a lot of pressure and the walls were bowing outwards. I would add stiffeners to the gantry walls later and pull them back vertical. The plate assembly was buckling a bit. It kind of wanted to rip itself apart, but it held. All in all, life was good.

So on to the Y carriage. I had noticed that it was falling apart with my stress testing. I needed the carriage to wrap around the Y beam and still be strong. I made several attempts at it and came up with gluing thin plywood to the sides of the structure and then using a laminate trimming bit to route out the inside, leaving a one piece skin. Strong.

And back to stress testing the new carriage. Success. Note the consistent vertical gap between the carriage and the beam. Yeah.

Now on to the Z axis. I chose to make it removable so I could play with it if needed. CNC guys love to play with their Z. I used polished steel rods with bronze bearings for the vertical rails to make the Z more compact and keep the router from cantilevering out too far. The picture is looking inside the Z from the back as it laying down. The two small blocks of wood have the bronze bearings in them, these are also attached to the face plate and the lead nut. So they go up and down.

I need to point out here what I did for leadscrews. Three problems need to be addressed; speed of the machine, the whip of the screw, and backlash. Backlash is a bad thing that occurs when you reverse the direction of the motors, which happens a lot. The leadnut must not have any slop in it or bind. I solved this by buying, you guessed it, anti-backlash nuts. They show up in several of these pictures. It is a plastic nut with a flange for mounting. They have slots cut in them, a spring, and a collar that form a very tight nut with just the right resistance.

To get the speed of the machine up, and the whip down, I went with fewer turns per inch on the leadscrews. Fewer rpms for the screws mean less whipping while kicking up the machine speed. This is a little harder than it sounds. These are ACME double start screws, which actually have two sets of threads. They are 8 tpi, but since the are two start, for every 4 turns of the screw the leadnut moves one inch. Why not just buy 4 tpi screws? Cause they don’t make them. Since you want about 6 to 8 threads for a nut to grab on to it would make the nut really long and unstable.

The next problem for the leadscrews in the hobby arena back then was mounting them and attaching them to the motors. What you want to do is ‘fix’ them to the frame of the router. A screw can spin upwards of 5 times faster without whipping if it is ‘fixed’ instead of just spinning in a bearing. The big boys have all kinds of tricks they do with turning down the ends of the screws, but I don’t have a metal lathe. So I mimicked their set up and came up with a method of using two bearings pressed into recesses either side of a plate that is firmly attached to the frame. Next on each side is a spacer and then locking collars. Then come Love-Joy motor couplers that have a piece of hard rubber in the center to allow a slight misalignment to affect things. These couplers also allow a transition from 1/2” screw to 1/4” motor shaft. This is then attached to the firmly self-supported motor. This is what those motor ‘towers’ are all about. Whew.

So here is Machine #2. Note the stiffeners glued to the gantry walls.

She then had the job of cutting out some larger router mounts.

For her big brother.

So how did it work? Not bad, not bad at all. A pretty tight machine. There are some things that I just wasn’t happy with. The gantry could rack or twist. I could hold the left side and push and pull the right side and it would move. The Y carriage could tilt backwards a little bit if I drove the router down quickly (I know, don’t do that). The whole machine was held together with tension and that just seemed wrong to me. It used radial skate bearings at a 45 degree angle and that also just seemed wrong. And my version was butt ugly.

I was posting this on another blog as I built it and quite a few people were watching and helping. Actually this has become a very popular design with at least two people selling kits of pretty similar machines. They will give the plans out for free. There are users groups now. It makes a Daddy proud. Here are CAD renderings of their kits. Google these names if you want to persue. See if you can spot any similarities.

LionClaw’s

JoesCNCmodel2006’s

————————————————
Next post, machine # 3
Steve

-- Stevethepeeve -- I'm no rocket surgeon