A Small Pottery Tool
This is a small pottery tool I made for my favorite potter. It's a very close copy of a tool that's available commercially, though I forget where from, and it's used for carving a lip at the base of a pot as it spins on the wheel. It's curvy, so I made it as a 3D carve of a model made with Rhino. Lately in my 3D carving I've been ignoring the fairly primitive multi-sided carving features in PartWorks 3D in favor of manually flipping the model over manually and generating two separate sets of cut files. It's too bad; multi-sided carving would be easier if the tool were more helpful. Using a jig is really helpful for this; the tool is nestled up against a jig in the photo below. You just have to measure the dimensions of your stock very carefully so that when you flip it the part stays centered.
The Tool With the First Side Carved
The wood is 3/4" maple, and the fine pass was done with a 1/4" ball-nose cutter, which is just small enough to fit in the notches on either end of the tool, which is roughly 2" x 1" overall. Total cutting time was about 10 minutes on my BT48 Shopbot. Here's the final result:
The Finished Tool
This is just as it came from the Shopbot, I did no sanding (joy!). You can see a faint cross-hatching in the photo; this is real. There are two finishing passes on this with the ball-nose, each at 45 degrees to the long edge, and 90 degrees to each other. The lines aren't that visible in person, but they do seem to make the tool a little easier to hold onto.
We'll see how long this tool lasts - those sharp points had broken off on the original, and probably will do so eventually on my copy as well.
Boomerang
This is actually the project I wanted to do first on the Shopbot, but my 3D-fu wasn't up to the drawing for a while. Not that it's a complicated drawing; boomerangs are actually pretty simple. In the end, I drew the cross-section of the wing, and extruded that to form one blade. Then I spun 3 copies of the blade around a point, and build a cylindrical hub to hide the mess where the three blades come together. I used Blender for all this.
The Beginning of the Boomerang
This was a 3D carving, so PartWorks 3D built the cut file. Cutting it took 45 minutes or so, because the only ballnose cutter I had at the time was the 1/8" one. I think it'd take about 10 minutes with a 1/2" cutter. It came out looking like this:
Boomerang Done Carving
Here's the wing cross-section. This isn't a calculated airfoil. I followed a rule of thumb: make the thickest part of the wing about 1/3 of the way back from the leading edge. Since I was drawing this with splines, the rest of the curve was automatically smooth. You might notice that the front of the leading edge is undercut. This is a one-sided carve, so I had to do that with the sander after the cut was done.
Boomerang Side View
I did have an advantage; this isn't my first boomerang. I built another one using a plywood blank, a rasp, and sandpaper when I was about 10. It was the first thing I ever made out of wood, and it even worked. This one does too. It's a nice beginner's boomerang; relatively big and slow, so it's easy to catch. One of these days I'm going to get my hands on some Finnish birch plywood, and build some lighter faster ones, with more interesting shapes.
Boomerang & Shopbot
If you're curious about boomerangs, Wikipedia is a good place to start, though the flight diagram they have there isn't exactly clear. Here's another site with some better information on how to build one. Boomerangs are a lot like kites, in that they're enjoyable and relaxing, and there never seems to be time to relax and enjoy them.
Wooden Gears
After I discovered Clayton Boyer's incredible wooden clock designs, I realized that there was a hole in my life that could only be filled by some wooden gears. So I whipped up a quick Rhino drawing of a couple of different sizes, taped some plywood down on the Shopbot's table, and hit start.
Wooden Gears
Gears are everywhere, of course, but these days they tend to be kept out of sight, except in watches, where they are an ornament in an ornament. Real working gears evoke something else altogether.
These take the Shopbot about a minute each to cut. They do work, though on the next batch I think I'll adjust the tooth profile so that they mesh more completely.
I'm not sure yet what they're going to be for. I don't think a clock is the answer. Maybe later. I've got some teflon bearings and a length of axle to experiment with while I figure it out.
A Simple Table
Everyone needs tables, but so many of the worlds tables are expensive, complicated, or boring. Some overachieving tables are all three. So here's a cheap, simple, and...well, it's still a table, let's not go overboard.
Ghosted Table Rendering
Like Pinocchio might have done, it all began with a quarter sheet of half-inch birch plywood. And a Rhino drawing. Then the plywood spent some time with the Shopbot. A few minutes later, it emerged, with some really attractive arcs cut into it. (As an aside, watching the Shopbot cut arcs is a wonderful thing. It almost seems alive.)
Table Parts on the Shopbot
The resulting three pieces of plywood slide together into a table after a little sanding. The slots in the underside of the top hold the tops of the legs in place. (The glue also helps.)
Table Before Finishing
For a finish, this table seemed to call for something a little unusual. So I mixed up some aqua blue aniline dye with some distilled water, and gave that a try. This stuff is very easy to apply - it just wipes on - and since it's water-based there's no need to wear my usual finishing respirator. I also used water-based poly to finish the piece, so I put on a coat of shellac between the dye and the poly to keep them from getting too friendly. As it turned out, two coats of shellac might have been better - I still picked up a fair amount of the dye with my foam brush while applying the poly.
Table Leg Detail
But it came out looking even better than I hoped. The dye doesn't obscure the grain, it just makes it look like blue plywood.
Finished Table
Finished Table, Dramatically Paired with a Chair
Beer Boxen
Homebrewing beer leads to a lot of bottles that need storing. Sure, you can stow all those bottles in cardboard boxes in some out-of-the-way corner, but then it'll just look like you haven't unpacked completely from the last time you moved. This might be nice if you actually haven't unpacked completely from the last time you moved, since it could lead to scenarios like this: "Say, where's that Oscillation Overthruster I used to have in the old place? Must be in one of these boxes I haven't unpacked yet... Hey, a beer!"
Sure, you're thinking, I can see how finding a beer would be a much better way to spend a Sunday afternoon than another zany trip to Planet 10. But my wife/girlfriend/pet/exotic carnivorous plant is already unhappy about the boxes I haven't unpacked yet in our new place. Adding more to the pile isn't going to increase my quotient of domestic bliss.
And you're right. Besides, homebrew deserves better. For some reason once I'd reached this point in the reasoning process the next thing that occurred to me was something along the lines of, "I wonder what a dwarvish beer case looks like." This may be the result of a childhood steeped in multiple readings of the Hobbit, a circumstance for which I make no apologies. I'd do it again if I had the chance.
The Completed Beer Box
This is a 2D design. The first time I drew it I used Shopbot Partworks, which is a renamed version of Vectic VCarve Pro. It's really CAM software; it's missing some features you'd want for drawing designs from scratch. I found this out while I was drawing the design. The Vectric software is really good, as CAM software, and it does have drawing tools, so I wanted to see how powerful they are when used with something a little more complicated than what I'd used them for already. They're just OK, it turns out. The snapping tools are weak, and once you combine a couple of primitive shapes into a more complex one you lose the ability to change the parameters of the primitives. This makes editing the design more difficult than it should be.
I cut some test parts using the first design, but then I had to buy some new plywood. Plywood thickness is dimensioned somewhat nominally, and different brands are never quite the same. The new plywood was .03" thinner than the old, so I had to redraw some things. This should have been easy, but it wasn't, so I just decided to redraw the thing with some real CAD software. I decided to try out QCAD, which is free software. QCAD is supposed to be a clone of some old version of Autocad. I've never used Autocad, so that doesn't help me much. This second drawing came out well enough after I'd climbed the learning QCAD learning curve a bit. I cut the second design on the Shopbot from these drawings. One thing about QCAD: it doesn't lend itself to experimentation like Sketchup or Blender do. Partly that's a function of my still-low position on it's learning curve, but one of the things that's really fun about 3D software is how easy it is to try different things.
What? Empty?
The design is pretty simple: 24 bottles are separated by 1/4" plywood panels, each one of which is notched halfway, and interlocked to make a grid. There are slots cut 1/4" deep in the insides of the outer panels to hold the grid in the box, and slots cut in the bottom of the case as well. The bottle grid is glued into those slots on all sides, and I used biscuits to join the sides and base for added strength.
The outsides of the side panels are V-carved. It takes the Shopbot about 45 minutes to cut the long side - the mountain design is pretty complicated. The short side, complete with some slightly original dwarvish verse, takes about 10 minutes.
The hinges on the lids are a mistake. They're hidden barrel hinges. I found them while I was combing the aisles at Rockler, and was smitten. The hinge mechanism is completely hidden inside the barrel of the hinge until you open them, and then it folds out in a way that doesn't seem quite possible. The problem is, the plywood I'm using really isn't thick enough to hold them. I'm going to have to find an alternative for the next version of the beer boxen.
A Ghosted Rendering of the BeerBox Version 2.0
Then along came Rhino3D, and I redrew the whole thing again, in 3D. That's a story for another day, but I'll leave you with a ghostly rendering.
Shopbot Project #1: The Bitten Nameplate
For some reason, when the Shopbot was new and I was imagining projects to build, the first 3D project that I ended up building was a replacement nameplate for my cubicle at work. With a bite taken out of it.
Wait, what? Well, I don't know exactly where that came from. The imagination is a strange place. Sometimes whoever works in there sends up some very odd proposals.
It was clear from the outset that this was a 3-dimensional project. Everything you can put on the Shopbot's table is 3-dimensional, of course; but when all the cuts you need to make are the same depth (usually this means all the way through the material), you can call it a 2-dimensional project, and use 2-dimensional CAD programs to produce your drawings. If the cuts have several different depths, but only one depth per cut, you can call that 2.5D, and still use the same tools. When your design has more complex surfaces not defined by simple cut lines, you're in 3D land. The design for this object was a rectangle with rounded corners, a thin raised border, and a raised name in the center of the top surface.
A Blender Rending of the Nameplate Design
I was going to need some 3D software, so I fired up Blender. Drawing the nameplate took several attempts. In the end, I drew its shape using Blender's curve tools. I started with one of the corners, which I drew as a bezier curve, and then adjusted the handles to give it approximately the diameter of a quarter. Once I had that done I made three copies, then moved and rotated them to the other three corners. Then I joined them with straight lines, converted them from curves to meshes, and extruded them to the right thickness. All this took...much longer than it seemed like it should. 3D software is like that at first. It's like that at second, too, and for quite awhile after that.
I now had the nameplate shape, with a flat top. Adding the thin raised border was next. What I probably should have done was to copy and scale the outer curve before I extruded the nameplate, but the best way to do something in Blender doesn't become clear until you have used it quite a bit. After more manual adjustment of the raised rim than I really expected to have to do, the drawing was complete, and just needed a bite taken out of it.
I built some virtual teeth - really just an approximate outline of teeth extruded so that it was just taller than the model. Cutting a bite out of the model should have just been a matter of running a boolean subtraction of the bite object from the nameplate. It wasn't. I hadn't realized yet that the new points introduced into the mesh where the two objects intersected would need to connect to other points in the mesh. The mesh that formed the nameplate was very sparse, while the intersection created a large number of new points. So the result was a lot of small triangles in the mesh, where lots of new points connected to only a few old ones. I subdivided the mesh on the nameplate a couple of times, and tried again. The results were better. I came away from this with the feeling that there must be a better way to do this in Blender, but I still don't know what it is.
Exported to an STL fileand loaded into Part3D, which produced the cutting programs for the Shopbot.
The Nameplate after Shopbotting
Next: a plank of black walnut, which while featureless to the naked eye, already contained the nameplate that had just emerged, all dripping and numerical, from Blender. I attached this to the spoilboard with screws - I hadn't yet learned the trick of holding down parts using reinforced carpet tape, which you can find at your favorite hardware store. The rough pass used a 1/4" straight end-mill; the finishing pass used a 1/8" ball-nose cutter. This took about 20 minutes on the Shopbot. Watching the Shopbot is really fascinating, especially when it's doing 3D carving - the piece really seems to emerge from the wood, and one little line at a time, since the cutter is usually rastering along the long axis of the piece. I set up tabs in the cut file to hold the nameplate in place after it was cut loose from the surrounding wood. If I'd used carpet tape as a holddown, I could have omitted the tabs.
Black walnut and tung oil go together like chocolate and scotch. The walnut gets nice and dark as soon as the oil hits it. After just a little light sanding - mainly to remove some tufts of wood still attached to the raised letters - and a couple of coats of tung oil, the result was pretty good. I then made another in my series of mistakes: shellac.
The Finished Nameplate
My intention was to seal the oiled wood. This was unnecessary; once the tung oil dries it seals things pretty well by itself. But I didn't use dewaxed shellac, which was a bigger mistake. It left a noticeable and unattractive sheen on the surface. I did a little more sanding to remove that effect, but it wasn't possible to get it all off, especially around the letters. It's not very visible under normal light, fortunately.
It's been hanging outside my cubicle for almost a year now. Surprisingly few people have mentioned the bite taken out of it.
Building Things with the Shopbot
For the first real Shopbot project I wanted to do something a little challenging. The Shopbot can cut fantastic two-dimensional objects, but to me the really fascinating stuff is 3D. And I've always been a fan of learning to run before you can walk. It's faster, and after you fall down you often get a nice view of the sky.
Other than existing in 3-dimensional space most days, I have no particular background in 3D. I've drawn some things with Google Sketchup, but that's pretty much it. The Shopbot comes with some very good software from these guys for cutting 3D drawings into material, but not for drawing anything. (Technically the software that comes with the Shopbot is CAM software, and what I needed was CAD.) So before I could draw anything I needed to find some something to draw with. Sketchup was one possibility, but the free version can't export files in any format that the CAM software can read. The Pro Version can, but it's $500. That's actually very cheap compared to most 3D CAD software, and Sketchup is a pretty impressive tool, so I may go that route in the future. There's also Rhino, which I've heard is amazing, and even though it costs $1000, I'm planning to try it out at some point.
Trying to choose software like this - software with steep and long learning curves - when you're new to something is a bit of a crapshoot. As a beginner you don't even know the right questions to ask, and there's the feeling that you may invest hours and hours learning a particular tool, only to discover that it's missing some crucial feature further down the road.
In the end I settled on Blender, which isn't even CAD software, really. Most people seem to use it for building 3D models for games, and animation. It does have the huge advantage of being free, though. And then I stumbled on this excellent tutorial, written by someone who was using Blender to design parts for his CNC machine. That proved that the software was capable of what I wanted to do with it.
When you first launch Blender it opens a strange-looking window that fills the whole screen. There are buttons, a few menus, and a grid, and a pink box in the middle. Clicking anywhere on the grid causes this small red and white circle to move around. Clicking on the buttons often does nothing at all. Opening the menus reveals a bunch of commands, most of them unfamiliar.
If you persevere, and find the right tutorials, you will eventually discover that the pink box is actually a cube, when you learn how to rotate the 3D view. Continue on, and you may find out how to build objects more interesting than cubes. I'm now far enough up the learning curve that it doesn't feel completely vertical anymore. But it took awhile.
After a few tutorials, I successfully followed instructions well enough to build a penguin. (A model penguin, not a real one.) I exported this into every 3D format Blender can write, and loaded up the results into the Shopbot's CAD software. Several formats worked, but the only one that was trouble free was STL. (I thought the more widely-known DXF was going to be the winner; it is certainly supported by more applications, but the DXF version of my model had some problems.)
The CAM software places your model inside some material whose dimensions you specify, and then walks you through setting up a number of cutting passes to produce your part. It's aware of the dimensions of the router bits you're using, so the instructions it produces will pass the bit over and around the model without cutting into it. You can simulate the cutting process within the CAM software to check for problems. When you're satisfied, you export the cutting instructions to one or more files, take them to the Shopbot's computer, and load them up there. When you finally hit the big Start button on the screen, your part emerges from the material as if it had been there all along.
If it works. There are a number of things that could go wrong. You might put a hold-down clamp or screw in the wrong place and hit it with the router bit, or enter the wrong feed speed for the material you're cutting, or the bit you're cutting it with. A lot of these mistakes can end with broken router bits, which is an expensive deterrent (the Shopbot uses solid carbide router bits; most of them cost about $20). There is an exciting-looking red shutoff switch on the side of the Shopbot, which you can hit to stop the action if something doesn't look right. I run air-cuts above the material before doing the real thing. This catches mistakes like setting the wrong XY origin on the table.
Now that I had a workflow, it was time to make something.
Connecting the Shopbot
The Shopbot Assembled and Connected
The arrival of the Shopbot was followed, like a clutch of SYN packets, by a lot of connecting. The 3 HP router this Shopbot uses for teeth produces prodigious quantities of dust, which if not collected will accumulate into ever-larger dunes, eventually burying your home town for millennia, much like this.
So I needed a giant vacuum. Unfortunately, giant vacuums are by and large not any more interesting than normal-sized vacuums, so I'll save the details for another post, maybe after I've exhausted the fascinating subject of pocket lint. Oh, wait. Someone already did that. Oh well. At the risk of causing groan-related injuries to my audience, I'll just say that this vacuum really sucks, as giant vacuums are wont to do. It's connected to the Shopbot with a 4-inch hose, the end of which is connected to a dust collector encircling the router bit. The whole thing acts a lot like a black hole when turned on, and like a black hole eventually the bag needs to be emptied or there might be a Big Bang of some sort. It's also stupendously loud, but that doesn't really matter, because so is the router. With both of them running at the same time it feels a little like standing inside a jet engine watching magic tricks. A little.
Then a computer was needed to run the show. I had one lying around gathering dust, and it seemed perfectly happy to go from gathering dust to producing it. The Shopbot control software runs on Windows, which is a shame, in a way, but it's probably safer: it seems pretty unlikely that Skynet is going to evolve from a Windows machine, even if it is connected to a large robotic cutting machine. Skynet's slow brother is a possibility to watch for, I suppose. I think his name is Microsoft Bob.
And then everything needed a connection to power. The vacuum cleaner (you remember, the one I said I wouldn't talk about any more) uses 12 amps. The router uses 15 amps. The Shopbot itself, which has three rather large servo motors, can use up to 10 amps. The computer gets to fly the green flag in the group, since it only uses 2 or 3 amps. All this is only interesting because it means I needed to plug these things into three separate circuits. There are now extension cords running all over the basement to reach those circuits. At some point I'll need to have an electrician put some outlets by the machines.
Because of the ongoing necessity of going to work most days, it took a couple of weeks to get all the pieces connected together. And then, after plugging all the plugs and clamping all the hoses, I turned it all on.
The Shopbot thunks powerfully as its servos power up. The giant vacuum cleaner roars like a cyclone, which it pretty much is. The router just screams. The computer might as well be floating through space once it has finished saluting its ancestors from the 1980's by emitting the traditional post-POST beep.
I wear earplugs, because it's so hard to find a decent ear trumpet these days.
At this point I commenced with Shopbot Project #0, which consisted of surfacing the ultralight MDF spoilboard on top of the moving table on the Shopbot, using a router bit designed for this purpose. It was fascinating to watch, even though all that was happening was that the cutter was shaving about .05" off the top of the spoilboard, in an inward spiral. Not as fascinating as watching the shopbot work on a real project, but the story of project #1 is for another post.
ShopBot Arrives
Programming computers is good fun, but sometimes it's a little too abstract to really satisfy on a visceral level. So geeks invented robot wars and other pastimes that occasionally graze the public consciousness. Me, I've always been fascinated by computer-controlled fabricators of all kinds, and only partly because understanding them will be an advantage during the inevitable Robot Uprising. So about a month ago, after an impressive demonstration by the delivery truck driver of Things You Didn't Know a Pallet Jack Could Do, a large crate arrived in my driveway. It was 7 feet tall. It weighed 1100 lbs.
The Mysterious Crate
Inside was a machine I've been kind of obsessed with for a couple of years: a ShopBot CNC Router. Think of it as a 3D printer, or a router with a USB plug. You plug it in to a computer, and a little notification appears in the usual place, something like "ShopBot connected". After which, if your concerned about the Robot Uprising, you keep a close eye on that computer, since it's now capable of building all sorts of Skynet components. Especially the wooden ones.
Until that fateful day though, it works more like this: you spend even more time than usual locked away in front of a glowing monitor, keyboard clacking, mouse moving fitfully, MC Frontalot blaring on the speakers, until you've created a suitable 3D model of something you've been imagining. Then you take the model downstairs, load the file up on the ShopBot, put some wood down on the router table, and hit start. Some minutes later, your 3D drawing has been translated into the physical, analog world.
ShopBot Unboxed
Of course, it's not actually that simple - there is technical information to be grokked, feed rates to be calculated, and 3D drawing software to wrestle to the ground. Stay tuned.