This is the Mantis9 PCB mill. It’s the first time we’ve featured the project, but it’s already well known by some as it keeps popping up in the comments for other CNC mill projects. Yes, it’s made out of wood — which some frown upon — but we’re happy with the build instructions and the especially the price tag (parts as low as $85).
We did feature an earlier revision of the hardware back in 2010. Subsequent versions changed the frame to use an open-front design, but it’s the build techniques that saw the biggest evolution. The problem was getting the holes for the parallel rods to align accurately. In the end it’s a simple operation that solves the problem; clamp both boards together and drill the holes at the same time. A drill press is used for all of the fabrication, ensuring that the holes are perpendicular to the surface of the boards. From there the rods are given some bronze bushings and pressed into place. Only then are the platforms secured to the bushings using epoxy. This is to ensure that the bushings don’t bind from poor alignment. We think it should end up having less play in it than other builds that use drawer slides.
Check out a PCB milling run in the clip after the break.
http://www.youtube.com/watch?v=nlDpZl-QIAA
[Thanks Harvie]
I spent about three months trying to build one of these mills using the instructions provided. This is much, much more difficult than they make it seem. I was never able to get the ways perfectly parallel, and the axes jammed even when I upgraded to self-aligning bearings. Very, very frustrating few months!
Not saying its not possible, because it obviously is, but people should be aware that you should only really attempt this sort of project with extensive prior experience in shop skills and CNC systems. With a laser cutter and a manual mill, this project would probably be not too hard, but if you try to take their words at face value, you may have problems. Tried contacting the Mantis design team via e-mail several times and never received any replies :(
I was able to make one of these and it works great! Just finished a PCB, actually. I wrote a bit about it here -http://toasterbotics.blogspot.com/2012/03/cnc-part-2.html
I have yet to post pictures of PCBs but they come out beautifully. It was made with a sliding miter saw and a drill press.
Try a draw runner build, no need for pillar drills and should be reasonably accurate depending on what you want to do.
It’s perfectly obvious how there could be alignment problems with this build. For example, if the parts aren’t cut to near exact proportions with good right angle cuts, the frame of the machine can become stressed and tend to skew into a diamond shape instead of a rectangle, and the distance between the rods shifts, jamming the bearings.
Or, if the end pieces don’t sit flush on the bottom plate, or the bottom plate isn’t straight then the rods will twist like the two bones in your forearm when you turn your wrist, and again they go out of alignment.
Despite the claims, it’s actually quite difficult to build this with hand tools, without a device to make precise cuts.
I waited the whole time that he would blow on the board so I could actually see something.
What I’m not terribly sure about is why people dislike wood?
Yeah its probably a bit harder to get solid angles and it’s more prone to warp. So probably a bit more calibration over time.
But looking at the parts this is something I could probably actually walk down to my local hardware store and purchase a bulk of the non electronic parts.
I’d much rather be running one of these than sporting the laminator and chemical etchant that I am now :(
Rigidity and repeatability. This little mill is probably fine for light projects, but in general wood is a poor material for milling projects because it’s too flexible.
Since the wood doesn’t appear to be sealed you may even run into problems with a change in humidity.
Choice of wood type is up to you… If you google a bit, you’ll find that there are Mantis builds from different kinds of wood…
Why not aluminum plate? Or is the whole point simply keeping the cost to a minimum?
Seems the other advantage is customization and not being locked into a manufacturer for your choice of replacement parts, since you essentially are the manufacturer.
Same can be said for the disadvantage. You’re the one fixing issues with it when it quits.
Go for it :-) Mantis from aluminium seems like an interesting idea :-) you can even make wood Mantis and replace it for aluminium part by part… I think some parts can be replaced by aluminium instantly (eg.: lower part of Y axis table).
For me wood is interesting because if you have some new idea, you can screw it in place almost instantly (while adding new screw to aluminium plate is bit of work).
Also calibrating of Mantis is done simply by milling off few layers of table which makes table perfectly aligned with mill. While milling into aluminium table is probably not worth it…
Also maybe you will prefer some of Manti’s brothers:
http://mtm.cba.mit.edu/
I guess you will like this one:
http://web.mit.edu/imoyer/www/portfolio/pcbmill/index.html
I would do this in aluminum.
People hate wood because it takes weeks for the moisture to normalize, then it screws up alignment. This happens every season change.
What I want to see is diy builds using plasticrete/concrete. Big fancy mill manufacturers that don’t use cast iron use this. It is very, very ridgid, excellent vibration dampening, can hold a ton of weight, and is easy to work with (just one precision mold = a bunch of frames).
The only thing stopping me is space and a budget to get the kit necessary to do steel. It just doesn’t feel fun until I can mill my own engine from scratch. Gotta pirate them cars ya know?
Would be more than adequate for my current level of projects. 3 hole pad vs milled? Give me the parts list already!
Double sided would be nice. Sounds like another headache to worry about though.
double sided isn’t an issue – you can fixture the table with a couple pins to go through holes in the PCB to index it. When you flip it you can keep it indexed that way. Otherwise, set the edge up against a set of reference pins to keep it squared up, then use any one hole you just drilled as a reference point. Use it to zero the machine for the second cut after you flip and set it up against the pins. You drilled the hole, so you know where it is, right?
imho it’s possible to make double sided PCBs same way as it’s possible to drill them after milling…
1.) drill the PCB
2.) change drill bit for mill bit
3.) realign and mill
4.) turn PCB to other side
5.) now it’s little bit tricky to realign
6.) mill other side
Wood isn’t a great material for larger machines, but for this its actually has some advantages.
The dimensional change due to humidity is trivial for a machine of this type and size, and there is far more error by other parts of its construction technique.
Rigidity is more than enough as well. In larger machines and heavier cutting, wood has a tendency carry instead of damp vibrations, and it leads to issues like tool chatter and destructive harmonics. Its not an issue of gross deflection as much as micro oscillations. Just not a problem here either.
All in all, a very nicely developed machine so far. They have come a long way since the initial prototypes.
Honestly, I wouldnt care if it was made out of banana peels it is a nice build.Judging from comments it sparks imagination.That alone is worth sharing.I think it’s pretty cool.
I would not want to mill a layer off aluminum or wood. Is it possible instead to use 2 layers and insert feeler gauges in between as shims? Have to see the calibration process to understand I guess.
Using a fly cutter is easy, and is the simplest method for truing up surfaces and ensuring parallism. Not as hard as it sounds, its a $15 tool you need anyway, and maybe 2 minutes to set it up and cut.
What really grabs me about this is that it represents an easier price-point at which to enter the world of milling. Commercial mills are painfully expensive, home-built mills using all metal builds are still quite expensive, home-built mills using wood for the structure and metal only for the moving/frictive parts are suddenlty under $100… for something that a newb wouldn’t be sure if they really wanted to get into or not, $100 is much less of a risk than the other price points.
And, if you’re careful and precise when cutting/drilling/etc… you’ll get a project that’s just as good for nearly all hobby purposes. If you want to scale up later and work on grander projects, then you’ve already got your feet wet and you know where your weaknesses are. If you never need to go further, then you’re out a much smaller sum and you could probably sell it to a maker friend for your break-even cost or maybe more (since they wouldn’t have to build it).
If nothing else, the experience of making the unit would be a great way to practice your fine-control skills… measure 5x, cut 1x, level *everything*, recheck level, learn how to index, recheck level, etc…
Now if you’re gonna rush into it and then toss it because you don’t have the patience to do it right the first time, then you’re screwed at pretty much everything in life except maybe water-baloon fights… but then you probably wouldn’t have even read the article if you’re that flakey.
So, if you’ve ever wanted to try your hand at milling/cnc/___-bot-ing/etc… this would be a great place to start, and it’s financially attainable for just about anyone. Articles like these are why I continue to love HAD!
Looks prettier than mine, but the need for a pillar drill is pain, will right up my build some day.
https://plus.google.com/photos/112081584624304379263/albums/5727343013732477297
I’ve got my Mantis about 90% complete at this point. Rather than wood I used MDF and sealed it very well to keep the creepage to a minimum. I really liked the elegance of his design so I went for it. Instead of building the spindle, I made mounts for a Dremel flexshaft handpiece so the motor is mounted remotely. I also tried to make a few things removable as his design is almost entirely epoxied together so maintenance may be difficult.
I have a nice drill press and a deadly accurate tablesaw so making the frame was a snap. I ended up redrawing his DXFs to get some better dimensions and push the Z travel a little higher because I want to do some 2.5D stuff beyond PCBs.
So instead of going with a dimensionally very stable material (plywood, used in aviation in days pays for its durability strength and low creep) and used glorified cardboard, that no matter how well sealed will never have the same strength?
Sorry, this might come off a bit crude, but I just have to vent this. Why do people use MDF for this sort of thing? It’s not wood, should not be qualified as wood, and should not be used for anything dimensionally constrained!
As for the overall design: speaking as an engineer, this was obviously not designed by someone with engineering training. One of the 2 guides should only be constraining movement in the rotation around the other guide. (IE 2 flat plates bearing against the front and back of the rod. Like this |o|. Then parallel-ness matters much less and the whole system becomes easier to adjust. Using bushings on parallel guides is a beginners mistake.
(For reference, check out the design of pro-systems. You will always find one of the guides is only contrained in a single direction and free to move and rotate in any other)
MDF cuts more accurately than wood and is more dimensionally stable. It literally sucks up shellac, curing oils, or varnish so when finished it’s very durable. It’s also heavier, so it adds mass to the final product, which in a machine like this is a plus.
There are plenty industrial machines that use 2 linear bearings (rail or shaft) per axis and both are constrained. I’m not sure where you get the idea that one has to be floating.
From the article..
“Its kind of cool that a few simple manual operations and some epoxy can trump the accuracy of a ten-thousand dollar CNC machine.”
o.O
I get that idea from being trained as an engineer and the rules of designing mechanical systems! And I’m pretty sure you WONT find any machines constrained in both axes. (For instance lathes use a prism bed on one side and a flat surface on the other) One of those bearing or pillow blocks might LOOK like it’s fully constrained on those machines, but it most likely has slot or oversized holes to allow play in most directions.
Some very very high accuracy machines using lineair ball-bearing rails have a double constrained design, but the accuracy needed to get that to work is within 1000ths of millimeters. Far beyond the reach of wood or a normal hobbyist.
There simply is no reason for this machine to use 2 fully constrained linear bearings per axis. It’s completely over-defining the design.
MDF cuts more accurately because it is softer and has no fibers. That doesn’t make it better. Aluminium cuts easier and more accurate than stainless on the average mill. That doesn’t mean I want to use aluminium for everything! And just BECAUSE it sucks up stuff so easily makes MDF unsuitable. The effect of sucking up shellac or laquer or whatever is already a very slight alteration of dimensions.
all this talk of wood warping, i seem to remember a past submission of similar nature using plastic cutting boards as it’s primary support structure. i wonder would this be a better material than wood? i think it was like 3/4″ polyethylene or something like that.
–>justdutch, is solution below much better?
http://web.mit.edu/imoyer/www/portfolio/pcbmill/index.html
It is a lot better, yes. Notice how the second guide shaft on each axis has a slot and is only constrained in the z-direction but free to move in the x and y respectively. There’s some bits I would personally have designed different, but that’s just me being a bit difficult in some areas.
This looks pretty cool. I know, it is not easy to make these things. Takes lot of time and effort. Yours looks really fine, thanks for the post and the video.