Design choices

cnc v3 1

Drawing in Solidworks is hard. Or well, let me rephrase that. Drawing in Solidworks is easy.
Deciding what to draw is imp-fucking-possible. Before I start, I have to decide the general layout; moving gantry or moving table, mill-type or router-type, single x axis ballscrew or double, material for the machine, linear bearing type, and so on. It just never ends, as soon as I’m sure I want to go down one path, I learn some more or testdraw it and realise it’s worthless. And when I’m finally done deciding everything, I have to decide on the dimensions, then draw it, then rethink every tiny part and redraw that. And you know what? Placing the holes takes ages. Literally ages.

Well, to describe what I finally ended up with. Looking at the above picture, what do we see?
First of all, it’s a moving gantry type design. This means it’ll get a large working area with a small footprint. The tradeoff is stability. The machine is designed for working with aluminum and softer materials, which means this should hopefully be stable enough to get a good finish on the cuts.
The second most important choice is what material to use when constructing the machine. The most popular materials are mdf, aluminum plates, aluminum extrusions and steel box section. It’s rather obvious here that mdf is the budget alternative, though it should be noted that by going the budget route at that point pretty much means you’re just throwing your money away. Not very budgety is it? Steel box section is by far the best alternative, but it does have some caveats, that were unfortunately insurmountable to me. It requires good tools to cut and drill, generally needs machine planing and you need to be able to weld to assemble it. It’s also very heavy, though that is also a benefit.

Aluminum on the other hand is easy to work with, and properly designed only slightly less rigid, while also more expensive. I tried to build with as much extrusions as possible when designing this, since that gives good stability at low cost and weight. As you can see, the base is made of 60×60 extrusions, the gantry 80×80 and the plate parts are 20mm 5083 alu. 5083 is slightly less rigid than the standard 6061, slightly less machineable, but less sensitive to vibration. Is the tradeoff worth it? Moot point, I could only get 5083, so that’s what I’m using.

cnc v3 2

The t-slot table is also alu extrusions, and as you can see it’s braced underneath by more 60×60.
One of the main problems with a gantry style router is the flexing of the gantry sides. I spent a lot of time figuring out a good solution to that problem, trying both raised base sides and bracing of the gantry sides. I think this is very neat, it gives great strength from the crossbraced alu plates, while allowing the z plate to slide in front and maintain full range of motion, while at the same time giving easy access to the bearing mounts.
For the rest of the machine, I’ve tried to keep everything with short lever arms, force inserts close to the fulcrums and crossbracing everything. You can see for yourself from the pictures, I assure you every tiny part has hours and hours of thought behind it.

Next up is the linear motion system. I’m using dual ballscrews on the x-axis, a must to prevent racking on wide machines like this one. The ballscrews are 1605. Arguably I could have used 20mm or even 25mm on the x-axis, but since I’ll stick to rather low speeds for alu cutting I wont approach critical speed for the ballscrew. Instead I keep a low inertia, allowing for fast accelerations when doing 3d work.
The bearings are another key point, usually deciding the stability of the entire machine. They are what killed both my other machines after all. The usual suspects when it comes to linear bearings are skate bearings, round supported/unsupported rails and square rails. I’m using square, the undoubtedly best option. Specifically, it’s HIWIN HGH20CA blocks, that are actually ridiculously overspecced for this type of work. However, the cost is the same as for the smaller versions and makes for easier mounting.

Looking at the z-axis, you can see me making some choices that are a bit outside the norm. Usually, you’d place the rails on the moving part and the blocks on the other side, this to be able to modify the stiffness of the machine for different jobs. I have elected not to do this, instead keeping the “overhang” short, and further bracing it with the massive spindle mount.
Funny story: the spindle mount is acutally a SMA50, a housing for a linear bearing LM50UU. The inside diameter of such a housing is exactly 80mm, perfect for a 2.2kw spindle. It’s also far sturdier than the usual spindle mounts, and very precise.
Back to the z-axis. This short overhang means I’d loose some range of motion, if I wouldn’t lower the spindle instead, which I did. This means I can still get an adjustable rigidity if I want to, while keeping the material needed low and the overall height of the y axis low. This is a key point for me, since I’m placing the machine under my desk, and I only have 790mm of clearance. Regardless, I don’t expect this part to be a weak link in the assembly, rather I think the deflection at max load is in the single digit um for this part alone.

I’m not sure I’ve covered the most important parts, so I guess I’ll have to edit this post when I can think of more things, but for now I’ll try to mention some of the weak points this design has.
First of all, the assembly is depending more than I’d like on the precision of the cuts when I order the aluminum plates. I’ve tried to avoid this, and get adjustment margins when drilling the holes, but I haven’t managed everywhere. This might lead to some inaccuracies, or more problematic, binding of the bearings and the screws. This type of linear bearing is very unforgiving when it comes to alignment errors, and I expect to have to adjust the mounting for days before it runs smooth over the entire length of the table.
Secondly, the rails have no protection against dust and debris from cutting. The x-axis ballscrews are covered by the e-chain rest, but the rails are not. I haven’t been able to figure out a good cover that wont restrict motion yet, I probably have to keep my dust boot on during all cuts. Ideally I’d also like to make an enclosure for the entire cnc to keep household dust away from the rails. That also has the added benefit of keeping razorsharp broken cutters from killing everyone around like a clusterbomb. That’s secondary though.
I’m also not superhappy with the stepper motor mounting, but space restrictions force me to do it like this. One benefit of mounting them flush on to the alu plates is that I’ll be able to run them hot, since they have a pretty decent heatsink mounted to them.

That’s it for now. I might edit this post with more info. Otherwise you’ll have to stay tuned for the next post. It’ll probably be either a description of the electronics or unpacking of the linear bearings.


CNC design evolution, or a history of bad decisions

Well, maybe I’m being too hard on myself. I was going to make a post to introduce my latest build, which is a new and superawesome cnc router. The idea was to do that by showing the path that led to the current incarnation, which means I had to start by actually reviewing said path. Looking at where I started and how I went from there, it sure looks like a long series of bad decisions or perhaps some sort of mental retardation.
But like I said, I’m being too harsh. It’s rather a testament to my complete and utter cluelessness of anything even remotely related to cnc, electronics or even basic building skills when I started.

Gather ’round, relax and enjoy while I tell a story of beginnings, endings and great revelations.
It all started in the beforetime, a long time ago. I was an avid World of Warcraft player, with no interest in building or creating anything. I was good at what I did, the best in fact. Perhaps that is what made me feel like a goal had been reached, a change was due. I had recently decided to switch path in life, away from the world of healthcare (2 years of medicine and 2 years of physiotherapy studies is what it took to make me realize I hate (sick) people) and to start studying electrical engineering instead. As an engineer I believed you needed to have a genuine interest in your field of study (which my classmates’ utter ineptitude and lack of interest in anything technological later disproved), and I started as so many others have done both before and after me; I bought an Arduino. Specifically, the Sparkfun Inventors Kit.

A new paragraph for a walk down a new path. Ever heard the song “I set fire to the rain”? I didn’t set fire to the rain, but pretty much everything else got a taste of the flames (of sulfuron). I managed to build one working thing with what survived that conflagration though, one of the few builds that actually got used extensively. Check it out under “ambilight clone” in my “old projects” page.
Anyway, the second item I intended to build was automated window blinds. I started by buying three stepper motors and drivers. That is also about as far as I got with those blinds. As soon as I got the motors, I promptly let the magic smoke out of all three drivers. I then got new drivers and started googling what a stepper motor actually was and how I should get it to move. You know what you get when you google stepper motor? CNC. That’s pretty much how much it took to get me completely hooked. I rushed to the hardware store and, standing in the isles, designed and bought the parts for my first cnc router. I didn’t have a clue what I was doing, and you know what? It was horrible. It’s truly remarkable that I even imagined it could work like that.
Oh, it did work. I even made a pcb with that first version. But really, rubber wheels as linear bearings? Check out the project log in “old projects” if you feel like you need a laugh.


CNC collage


I wont fill this entire post with history though. Suffice to say, I soon built another, this time with plenty of research and even cad modeling. That one was insane compared to the first one, I loved it. For about a week. Then I wanted more. I spent about a year learning, reading and learning some more, and last semester I started designing the third iteration. This time I took my time, and as you can see from the picture above, I went through several designs before I settled on the one I’m about to start building.
I think I’m going to split this up into two blog posts, and describe the actual design work in the next one.

But first, let’s talk a bit about the ones I scrapped. The first one, the mdf monstrosity, is the machine that I actually built. It has some strong points and several weak ones. The major strong point is that it was cheap. Secondly, the stepper motors are decent. As for weak points, I used mdf as building material. That could have been enough to actually cut mdf and things like that, had I used a stronger design for the z-plate and linear bearing blocks for all the axises. Instead, I used skate bearings for Z. That was really stupid, probably the weakest point in the entire build. The rest of it is pretty balanced, and I suggest you read about it in the designated project log if you’re interested.

The first design of my new one was pretty much a redesign of the mdf machine, with the same concept and layout but with the weak points I talked about earlier addressed. I even made some FEA analysis of it. It wasn’t very good.
The later designs progressively changed bearing types, stiffened the gantry sides, lowered the fulcrums closer to the work area, switched to dual ballscrews and so on and so forth. I can’t really summarize 5 months of constant pondering in a few sentences. Look at the pictures and ask questions, I’ll happily answer all of them. Next blog post should cover my final design in a bit more depth.

Working with the site

This is mainly a test post. Things are still looking real rough around the edges, but it’s getting there. Never actually tried web design before, and it’s a stumble in the dark. But you know what they say, being a web designer is like being a cyclist or a real estate agent. All you need to do to be one is to say you are one.

I’ll probably remove this post later.

Also, isn’t the calendar in the sidebar a bit hard to read? Can’t really tell what dates have links on them without hovering over them.