[designerbuilder]

Hi Everyone,

I've been lurking the cnczone forum for about a year and a half now, and this is my first post.

Over the past four or so months, I've been building a CNC Router of my own design. Over the course of the project, I've read a lot on cnczone and other forums, and what I've read has helped me iterate the design quite a few times, so thank you to everyone who's posts and discussions I've been reading.

My initial interest in CNC began many years ago, when I was interested in RC cars and planes, and wanted to design and build my own parts. With this current project, my aim is to build a machine that can:
- Machine aluminum and brass fairly well, though not necessarily at production rates
- Cut wood, acrylic, HDPE, etc
- Have precision of about 0.001" and accuracy around 0.005"
- Have a work envelope suitable for small to medium sized parts
- Be built for around $1000 including electronics, steppers, router, etc

Some of the constraints on my design were:
- Shipping to Canada can cost an arm and a leg
- I only have a drill press and hand tools
- I have a very small space to work in

I actually started designing this machine in the summer of 2011, but gave up during the school term (I'm a student) and started again in the summer of 2012. I've gone through about 5 versions of the mechanical design.
In the end, the design I settled for was:
- frame made of aluminum extrusion (not 8020)
- flying gantry
- 24"x12"x6" work envelope
- cold rolled steel rails with homemade ball-bearing trucks
- Bosch Colt for the spindle
- TB6560 stepper driver and 166oz-in Stepper motors from AutomationDirect
- 1/2-10 2-start ACME Leadscrews (from ENCO) with DumpsterCNC Nuts
- Thrust Bearings from vxb.com and homemade motor mounts
- LinuxCNC machine controller

At this point, I'm nearly finished the build. I had planned on starting a build log but never got around to it. So now I'll post what I have, a little at a time to keep it interesting.
First of all, the design:




You can see from these pictures some of the features of the design. CRS rails, homemade trucks. Flying gantry, aluminum square tube extrusions mostly for the frame. The overall size is 24"x36"x18". The base and bed are both made of two layers of 5/8" MDF.

The design of the trucks (which I will post more pictures of) is my own, and I haven't come across anyone making them like this so far. Mine are similar to momus and cncrouterparts trucks in the sense that they use skate bearings, but the design and construction are fairly different, and mine don't use set-screws.

I hope people on the forum find this thread interesting and useful.
I'll post more soon. Feel free to ask questions or tell me what I did wrong.

Ali

[litma]

just one simple question; having one screw on the y axis doesn't produce tilting of the X axis? Or is the duty not so heavy?

[designerbuilder]

There is a slight flex in the gantry on the side opposite where the x-axis screw is, but so far it has not caused any problems - I suppose the duty is not so heavy. The flex is not slop, and with small cutting forces does not hurt accuracy too much I think. It's probably around 5lbs/0.01" stiffness at the far end but I can't really measure it very precisely.

The reason I chose to have only one x-axis screw was that with the current size of the machine, I was able to get all three lead-screws from one 6' piece ($80). Adding a second screw on the x-axis would have meant another lead-screw ($80), another set of thrust bearings ($10), another stepper ($40), and another anti-backlash nut ($20), which would have put me over budget. I am planning on either adding this extra stepper in the near future, or else converting the design to a fixed-gantry moving table type design for the x-axis.

I had considered this racking/flex problem when designing the machine but when I was looking at the momus cnc design, I noticed that it also has a belt drive only on one side of the gantry, and uses CRS rails and I didn't find much complaint about flex on that machine so I figured it shouldn't be too much of a problem.

[designerbuilder]

For anyone else who's in the Toronto, Canada area, the best place I've found to buy metal is Ability Metals in Pickering. They're friendly and don't charge for cuts on small orders, if you pay cash they give you a discount. You can call ahead and have them get an order ready for you or you can just show up and they'll cut it while you wait. Their cutoff pile is well organized and is always full of good pieces, just make sure you wear closed-toed shoes. Their prices are much better (20-50%) than metalsupermarkets.


I began the construction of the machine with the linear bearings, which were the most experimental part. The rails are 1/4" thick cold rolled steel flat bar, which I had to file down the edges of to make sure there were no dents that would get in the way of the bearings rolling.
The trucks are made of two pieces of 1/4" wall aluminum angle. One piece (1-1/4" x 1-1/4" x 6") holds skate bearings on two faces, and the other piece (1" x 1" x 4") holds bearings on only one face, and attaches to the larger piece. I think the pictures explain better.

The skate bearings fit inside the slots I've cut in the aluminum angle. I cut the slots by drilling three smaller holes using the drill press and then filing out the uneven parts to make the inside faces straight.

I made holders for the skate bearing axles using some sheet metal (homedepot ducting section) and a little jig for pressing them into shape.

The holders hold the axles with bearings like so. The idea is that the axles should rest directly on the aluminum extrusion faces, which are assumed to be fairly flat and precise - much more so than anything I could machine by hand anyway. This makes the bearings along any direction line up, without adjustment screws, elliptical bolts, or setscrews.

This is the full truck assembly, without any bearings. The smaller piece of angle is bolted to the large piece, and the bolts go through holes that are oversize so there is room for adjustment.

This is two trucks with bearings, riding on a piece of CRS rail (this is for the z-axis). By clamping the two halves of each truck together, a tight fit with no flex or play can be achieved, again without any adjustment.

This is what a single truck looks like on the inside. There are 6 skate bearings per truck. I got the skate bearings from vxb.com, they were very inexpensive. They are R3ZZ 3/16"x1/2"x0.196", shielded. The axles are 3/6" CRS round, cut at home by a hacksaw.

[designerbuilder]

After the linear bearings, a month or two went by when I didn't do much. The parts sat around and collected dust, until I finally got around to building the gantry and putting the y and z axes together.

The y-axis is straightforward, just two trucks riding on a 4" wide CRS bar. The two trucks are part of the carriage, which also has two trucks for the z-axis. The z-axis has the rail as the moving part rather than the trucks. This makes for less parts and slightly higher rigidity I think, compared to having a fixed rail, and another cantilevered plate attached to the trucks to hold the spindle.

The carriage itself is made of just four trucks, and a 1/2" thick plexiglass plate in between them for shear stiffness. I would have used aluminum but the plate needed to have a slot routed out of it to make room for the z-axis lead-screw and I didn't have the tools for that job.

pictures = 1000 words:

This is the y-axis and z-axis together. Overall the assembly is very stiff and there is no play.

This is a closeup of the carriage, showing the four trucks and the plexiglass plate in between them. The 1/2" plate is actually two 1/4" plates that were welded together using solvent for acrylic.
The whole assembly slides very smoothly, especially with a little WD40 between the bearings and the rails.

This is another closeup. I think the trucks and carriage design was pretty good, at least it made for a small parts count and required only simple machining that could be done on a drill press. It was also very inexpensive.

I'm not sure how these trucks compare to the other kind of CRS trucks (momus/cncrouterparts, using setscrews), I've heard they also work very well.

[harryn]

Thank you for posting all of the pictures and explanation of your build. You are a really patient person to pull of that level of quality with hand tools.

I tried to do some filing of multiple holes in Al to make slots before, but was not able to make it anywhere near that nice. You really did a lot of thinking to stay within that budget and I am impressed.

Skate board + CRS setups are going to pretty much behave similarly. I used the cncrouterparts on an early test build and for some prototyping work. Their main advantages are: a) The build work is already done for you b) They are more accurate than many people (including me) can do by hand.

Your build results will be just fine.

IMHO, the main improvements that could be done with skate type wheel bearings in general would be to use one that are actually designed for this type of load, such as the "Swiss" skate bearings, but they are not cheap. As far as I can tell, ABEC style bearings are actually the wrong bearing for these projects, even though they are commonly used. I am still a beginner at bearing selection though, so I might be completely wrong.

I think it won't be too long before the flex in the gantry becomes painful for you though. Rather than add an entire additional drive mechanism to the build, there is a method that uses pulleys are wire to "force" the two ends to move together. This is the econo method and will help more than you might first imagine.

I think there is a thread called "make your gantry rock solid" or something like this that discusses it.

[designerbuilder]

Hi harryn,

Thanks for your kind words and advice. Building this machine by hand did take some effort, but surprisingly little time.

To cut the slots, I marked three holes along the centerline of the slot with a center punch, on at each end and one in the middle. Then I drilled out all three holes with increasingly larger drill bits, until I got to the final slot size (11/32"). At this point, I drilled the holes at the end of the slot first, then the one in the center, to keep the drillbit from sliding from the outside holes into the inside one.
For the first couple of slots I tried to just file down the remaining cusps inside the slot, but this was very slow and difficult. I soon discovered that using the 11/32" drill bit and holding the part really steady in a drill press vice allowed me to drill out the extra material in between the first three holes.
Since the slots themselves have no critical dimensions for the design, this worked out very well - the slots were straight and even through the part and were very quick to make.
I used liberal amounts of WD40 for all aluminum cutting and drilling - it makes things much easier and drill bits last forever. The same I found true for drilling in steel, and also that peck drilling helped a lot with small drill bits and that slowing down the drill press RPMs proportionally to the size of of the bit being used (slower for larger bits) also helped very much to keep vibration and noise to a minimum.

I don't know a whole lot about bearings and bearing types either, I simply chose the cheapest imperial sized bearings I could find, that were of the appropriate OD. I think that the ones I ended up with worked very well for this use, especially considering the non-industrial hobby grade nature of this machine. I'm sure there are plenty of shortcomings in the design and construction that make a slightly improper choice of bearings negligible.

Thanks for your tip about a wire and pulleys for the gantry. I had thought about this when I discovered the gantry flex problem, but I think I would rather rebuild the machine as a moving table design and have a fixed gantry. The gantry itself is slightly under-engineered, there is a little bit of shear/twist flex in the aluminum square tube member that makes up the main support of the gantry. It's too thin, and once I have the time and money to invest in fixing it, I'll also get an aluminum plate for the bed.

[designerbuilder]

One of the major challenges I had in construction was turning down leadscrews without a lathe. I ended up using an idea that I found on this forum, though I can't remember what the thread was called.

I used an electric drill with 1/2" chuck to spin the leadscrew, supported by two 1/2" ID bearings:

This worked out pretty well. The dumbells were to keep the whole thing stable and dampen some vibrations.
I used a regular metal file most of the time, though I tried other things like an angle grinder and various grinding stones. WD40 helped a lot to keep the file cool and helped cut faster too.
This was probably the most tedious part of the whole construction process. I would recommend to anyone who's doing something similar to just find a real machine ship and pay a little money to save your time and energy.

The results came out very well though. This is after turning down and tapping one of the leadscrews.

I used thrust bearings (vxb.com) and a helical coupler from eBay.

The stepper motors are Automation Direct 166oz-in. I chose them after much reading and deliberation - they were within the current limit of cheap TB6560 drivers, they were inexpensive, they had low inductance and inertia compared to cheaper eBay stepper motors that had higher holding torque, and the torque-speed curves were published. At the step rates I was looking to run them (calculated from the leadscrew pitch and my target speed of 100ipm rapids) they had more torque for the same drive voltage than larger beefier motors. They turned out to work excellently with the TB6560 driver, and I can get to 90ipm with plenty of torque left over. I can't try anything faster than this because the old P4 computer I'm running LinuxCNC on can't handle faster interrupts.

[designerbuilder]

If anyone has any questions or feedback please post, I almost feel like I'm putting so much detail in these posts that no one has anything left to ask lol

[the4thseal]

No, you are just doing a great job.

[kevincnc]

Very good work with limited tools and budget, keep it up!

[designerbuilder]

Thanks the4thseal and kevincnc!
I really appreciate the compliments. I guess I'll just continue with my verbose posting.

[the4thseal]

Please do.

[the4thseal]

Just remember, you will be able to say.......LOOK WHAT" I "BUILT.

[Kazik_Wichura]

Thanks the4thseal and kevincnc!
I really appreciate the compliments. I guess I'll just continue with my verbose posting.

Yes, yes. You have great ideas. I like your "lathe". It look and work fantastic.

Lucas

[megeshg]

great work!!!

[xairflyer]

Although it was a long time ago when I built my machine I had easy having the equipment on hand, compared to what you had to do well done.

[designerbuilder]

Thanks everyone for the kind words. To anyone who was interested, sorry I haven't posted anything in a while - I've been really busy with school.

This post is about the stepper motor mounts I built, which I think are a fairly nice design - made entire of one part, with very simple machining. I'll also talk a little about the electronics and software I used, with a little video of what the linear bearings and motion is like.

I had seen a lot of designs prior to building my own stepper motor mounts. Most of what I had seen involved four long bolts to mount the stepper motor to a frame of some sort. In my design, I used one section of 2.5" aluminum square tube to act as both a bearing block for the thrust bearings that constrain the leadscrew from sliding, and the motor mount.


This render shows the z-axis stepper mount and bearing block. On one face of the square tube section, there are four tapped holes for mounting the stepper motor. on the opposite face, there is a 3/8" hole (drilled with a forstner bit) through which the leadscrew is inserted. The leadscrew has two thrust bearings on it, one on either side of the bearing block face. The leadscrew is connected to the stepper shaft with a helical coupling (eBay).


This image shows the real thing. There is a lock nut which threads onto the leadscrew (threads cut into turned-down section with a die) to add preload on the thrust bearings, to keep the leadscrew from sliding back and forth, and to keep the axial forces on the motor low.


This is an image of the entire gantry and z axis, clamped to my desk.


This one is a bit blurry, but shows what's going on inside the carriage on the gantry. There is just enough clearance for the z-axis leadscrew and leadnut (dumpsterCNC) and you can see some of the bearings on the linear rails.

And now for a video:
[ame=http://www.youtube.com/watch?v=BKMQyanrmwA]DIY Aluminum CNC Router - First Motion - YouTube[/ame]
This short video shows the gantry and z axis sliding, and the machine's first motion. That's 90 IPM, 1/16 microstepping.


I used a TB6560 board from eBay (actually from aliexpress). It worked very well, and the guy who sold it to me was very friendly. Make sure you buy the red one that looks like this and not the blue ones that are out there - they have slow optocouplers, low power resistors, useless diodes on the outputs, etc and are known to violently fail.
The TB6560 is an integrated stepper driver chip from Toshiba, that can drive up to 2.5 amp/phase stepper motors. I read a lot and I'm fairly knowledgeable when it comes to electronics, and I found that these stepper chips, though probably not as reliable as Gecko drivers, cost a lot less and can deliver more than enough performance for my application. I run them at 75% of max current, at 12V from a ATC computer power supply. When this setup is used to drive low-inductance motors like the NEMA23 motors I have from Automation Direct, fairly high speeds are achievable, with no lost steps. The TB6560 chips are rated at 30V max (people say they explode over 28V though) so I still have to see how much gain can be had from using a 24V supply. I in fact designed a small boost converter to get 24VDC from the computer power supply, but when I etched the PCB the first time I accidentally did it backwards and I haven't gotten around to trying again.
I use linuxCNC running on an old 1.7 GHz P4 desktop I had lying around. I also tried a 3.0 GHz P4, but it runs far too hot and the fan makes as much noise as my router.

Next time I'll talk about the frame design and show some pictures of the actual machine. I hope anyone who's reading this finds it interesting and useful.

[ger21]

One of the major challenges I had in construction was turning down leadscrews without a lathe. I ended up using an idea that I found on this forum, though I can't remember what the thread was called.

I don't recall the name of the thread either, but I'll take credit for it. Your the only other person I've ever seen use my method. Sure, it's a bit tedious, but it works pretty good. Nice job.

[kwydjbo]

please, keep posting!

just cause we're not commenting doesn't mean we're not learning.

thanks for sharing your efforts

Thanks the4thseal and kevincnc!
I really appreciate the compliments. I guess I'll just continue with my verbose posting.