ARK Rocker

The ARK Rocker is concept piece exploring the intersection of modern manufacturing technology with timeless craft techniques.

Every part of this rocking chair was made was made using digital fabrication equipment and computer modeling software through Autodesk and the Pier 9 workshop. But each part was also taken directly from hand crafted lines and forms - either sculpted by hand and converted to a model by 3D capture, or carefully drafted and then vectorized and extruded into a solid model.

With this piece, I was hoping to examine what new possibilities for craft are emerging as we become ever more digital. We are using technology to relate to each other, and to shape the world around us. My hope is to harness this power to remain intimately connected with the things around us by preserving the magical quality imbued by handcraft, rather than create a computerized landscape of forms and objects. By tying CNC (computer numerical controlled) machining and powerful software to traditional hand shaping and drafting techniques, I'm personally seeking to deepen my own connection to materials and to craft, but also provide new outlets for craftsman to create work, and for designers to employ craft at scale.

The name ARK is taken from the story of Noah's ark, but also references the both the function of the rocking chair to hold the sitter and take them away. A rocking chair for me is all about that motion, and the opportunity to leave the bustle of daily life and rest, or read. Just like the gentle motion of the ocean can provide - minus the storms in Noah's story. It's also drawing on the fabrication methods used to make it - from the sensitive manual techniques to the ultra precise digital ones - and to take the story of that progression full circle.

In this Instructable, you will find an outline of my process, from conceptual sketches to design development, through to an overview of the refinement and fabrication stage of roughly 50 unique CNC machined parts that come together to create the ARK rocker. I won't go into technical depth, but just provide an overview of the process and final product.

If you want all the dirty details, you may in the near future find sub-Instructables for each part of the process linked throughout. Venture there at your own risk. You were warned.

Thanks for reading - and enjoy!

1. In case you just want to jump to the end, the vanilla version is on my website.

My initial excitement for ARK was in creating merging craft and technology in a new way.

I wanted to play with traditional hand-sculpted wood chair designs like those produced by Sam Maloof in a radically different direction. I wanted to see how differently I could examine the same themes of craft using digital techniques.

With the precision offered by digital means, I wanted to use mechanical fasteners to replace traditional wood joinery. ARK can be taken apart and reassembled time after time, making it much easier to move and keep for a lifetime. By no means is the point of ARK to be flat-pack, but at the same time, why shouldn't modern tools enable us to make our keepsakes and heirlooms more flexible? Somehow, it seems like technology has made our objects more disposable.

For me, so much of the craft tradition is rooted in producing heirloom-quality goods. And I wanted to explore what the modern version of heirloom could look like.

Even in early sketches, the aesthetic I have pursued throughout has been to incorporate organic, hand-sculpted shapes into an ultra-precise design, so I am playing with natural contours and how they meet with hard, artificial lines. Like the concept for ARK - an exercise in contrast.

As soon as I had a rough idea of what I wanted to say with this piece, and some of the basic forms I wanted to use, I dove right into thinking about what processes would enable me to create the most tension and contrast visually, as well as conceptually. I was seeking to create an object hybridizing my experience as a designer and fabricator to create something the most original object I possibly could.

I decided very early on to use mechanical fasteners as the center of the design, which mostly means 10-24 and 1/4-20 machine screws (small but strong) and dowel pins for constraining geometries.

Also early on I decided on two basic workflows that would guide the design process.

First, I would hand-carve the seat and back of the chair at 1:4 scale in order to create a model to 3D capture using Autodesk's 123D Catch. Then, this model could be blown up to full scale, cut on the 5-axis DMS router from MDF, so the shape could be refined by hand to create the most interesting contours and best fit to sit in. Then the model would be recaptured with these changes, and ready to machine on the DMS out of hard maple, leaving the rough shape intact, with only subtle hand finishing to do afterwards. This economizes the process, and makes it easy to produce multiples of a sensitive, sensual object.

I chose to use a cantilevered design as a testament to the engineering behind the tools that would be used to make the chair - a technical contrast the traditional hand-crafted component. I was excited to make the most of the unique tools at Pier 9, like the OMAX waterjet, CNC lathe and CNC mill. The materials were decided early on to be wood and metal - one for the technical and one of the sensual - and I chose maple and aluminum for their strength, light weight, relatively low cost, contrasting colors, and commonness. I didn't want exotic materials getting in the way of my voice as a designer.

Before making any lines, I measured some of my favorite chairs to get a good starting point for seat height, width, and seat-to-back angle. The book The Measure of Man and Woman is also a favorite reference of mine, which I used to make sure my numbers were in the right ballpark.

First, I laid out the underlying geometry of heights and angles. Then I sketched some rough forms over the top of this, added rough dimensions of wood that would create those pieces, and continued to refine the forms from there. At this point, I was thinking of using aluminum on the outside of the of the wood - an idea that would change over time.

At the end of this process, I was left with a solid image that I used to create vectors for starting to make parts. I would use this sketch for all the parts of the sides (arms, legs, rockers) of my chair, and just cut the same joints into the sketch to define

The seat and the back were shaped entirely by hand, and then references would later be added to allow them to interface with the edge pieces (arms, legs, rockers).

I was interested in using software to make the process of sculpting faster and more repeatable, so I started with a huge time saver: 1/4-scale in a cheap, soft material - MDF.

I brought sketches into the shop that included just the main outer dimensions that I wanted for each piece, to make sure that I was working at the proper scale and aspect ratio. Since I was 3D capturing this data, it needed to be a fairly accurate starting point.

Carving the two pieces took just a little over an hour - a huge savings compared to sculpting the whole thing from solid stock, let alone from hard maple.

Once the shapes were done, I painted them blue for higher contrast, and captured them using Autodesk's 123D Catch.

I started moving 3D captures from mesh to Inventor solid models in the first project of my residency. This workflow has become very comfortable and efficient after some practice. I won't go into the details again in here, but if you're interested in diving into the weeds, check out the how-to Instructable I created here. It goes into detail on how I moved captured mesh data through various softwares into Inventor so I could make mechanical features in it and eventually CNC machine it on the DMS.

This step shows some screenshots from how I worked with the rough capture data (which even includes the rubber floor mats! Nice!). I split the mesh in half around the better quality data, smoothed the contours and reduced the facet count. Then exported it, monkeyed around in other programs, scaled it up to full size, and brought it into Autodesk Inventor HSM.

There, I was able to create a really clean reference center and add references, tabs, and stock material to prepare CNC tool paths. With this full size model, I was ready to cut the seat and back full size. Full size prototypes would allow me to further sculpt the details of the shapes, and refine those shapes around ergonomics that I could test in real time.

This process was almost entirely reliant on the the CAD and CAM that I referenced in the previous step. With the model and toolpaths finished, I just sat back and waited for my parts to be cut.

Each part is rough machined on the top, the scalloped with a big ball end mill to leave a smooth surface finish, then flipped and the roughing/finishing process repeated on the underside. Aside from leaving a nice reference to make sure the machine knows exactly where your part is after the flip, there's not a whole lot to it.

Once the seat and the back were finished, I zipped off the tabs on the bandsaw and prepared to sculpt them again.

I did make a small change to the seat that I had made at 1/4-scale by hand. After I worked on the back, I decided I wanted both parts to have a clean geometry on the non-sitting side, to up the contrast between the sensual and the mechanical. So I added an Inventor extrusion to the hand-made mesh, which you can see left a really clean underside on the seat.

It was super gratifying to take the parts of the machine and work them by hand again.

Obviously, my hand skills are superior to my CAD skills. I guess some people can do as well or better at making nice forms digitally than manually, but for me, sculpting a shape by hand is the ultimate.

The shapes were really close to what I wanted - the quality of the captures was quite high. But given the full size of the parts, I was able to be even more aggressive with material removal, especially being able to test changes for feel as I worked.

I only sculpted half of each part to save time, since after the next round of 3D capturing I was going to mirror the better half on the centerline anyways. I put this half up in the air for better camera angles, and captured the parts again.

At this stage, I went back into CAD for a good, long while, working through the design of all the other parts that would determine the final geometry of the references machine onto the bottom and rear of the seat and back.

As an aside, I love the details in the fourth picture. What you see in the very foreground is not actually frantic chiseling - it is the CNC tool attempting to match the mesh data that it sees in the model that it is cutting. What's so remarkable is how similar it looks primitive human carving, despite being the absolute furthest thing from it. I wish I could have gotten consistent results with this process and I would have left the whole top surface of the chair like that!

In this step, I took my drafted image of the profile of the rocking chair, and used Illustrator to overlay vectors onto that drawing. This way, my own lines could be transferred into the digital world.

I really like the tactility of draft by hand. I get in a flow, and have a different feel for proportion and how to shape a spline than I do with a mouse in the computer. It's also nice because using overlays as I did, you can make changes gradually, responding to what you like or don't like about the previous version.

I made a couple small changes to my lines in Illustrator, then exported a DXF to import into Inventor.

Once these lines were in Inventor, I used the same master sketch to create each of my parts. I simply made all the lines into construction lines, and created the necessary geometry for joints off of these master sketches. This process was fairly straight forward, since for fabrication purposes, the profile is essentially 2D, so all these parts were simply straight extrusions off of the master sketch, with just some details for joinery.

At this stage, I also started to think about how to create the geometry I was modeling. Which machine to use, what tools would be able to create the features I modeled, and where the part would need to be flipped. It was really important to minimize extreme tool requirements and number of part flips since there were so many parts to make.

The side panels from the previous step are angled relative to each other, which made my life really complicated. Part of what I wanted to play with was letting the computer and digital fabrication take up the slack on the math side of things, where they perform really well. But it still proved immensely complicated, since the parametric modeling needs angles to be 0.0000 etc degrees in order to properly mate components in an assembly. Which I needed to do in order to make sure all my parts fit just right.

The rockers were angled at 2 degrees relative to the center plane, or 4 degrees relative to each other. Just enough to give the chair a dynamic look. Adding to that, the legs were each splayed outwards at 8 degrees. Add to that the stack up of the various angles between the joints and it became really hairy to try to figure out what I needed to do to the seat and back in order for them to make any sense to my Inventor assembly. The problem with the seat and back was that I had 20,000+ triangular facets, but no reference geometry, let alone projecting compound angle shenanigans onto it.

So I slowly worked through this step by first importing my latest captures of the seat and back. I matched them in scale to the ones that I used scaled up from the 1/4-scale captures, since they fit perfectly. Once again, I brought these through all the programs and then into Inventor as a solid. I created some basic references that were as best as I could get to center and bottom, and just went from there. I added material in some places to create a neat, well-defined surface to replace a mesh, and cut away material in other places to leave clean geometry. In both cases, I was extrapolating in real-time to what a CNC tool would do to the shapes I was making, and didn't worry about sloppy edges that would be easy to clean up by hand as I finished the pieces.

Before I went completely nuts on the model, I made a really rough mock seat of just plan extrusions to just see what the compound angles added up to for certain references, which is why you see me scribbling extremely long decimals for angles... So that I could take these numbers into my own model, tweak my own angles, and then test the full set in the model. In order to make sure it was perfect, I let the two legs float apart, constrained by artificial references to each other - in reality they are constrained to each other via the seat. But I mated one to the seat, and the measured the projected distances on the opposite arms and created reference geometry that replaced the artificial constraints I set up to hold things in place. You can see snippets of this crazy process in the photos on this step.

In the end, this bizarre, iterative process left me with a slew of reference planes and weird angles, but everything fit just right. Even Inventor was happy. I'm sure this is not the best way, however. Please, for the love of something you love - share better workflows with me if you have them!!!

All these crazy angles were well suited to a CNC project, since they can be "easily" made by a computer. Or at least, the machine has no harder a time with 45 degrees than it does with 14.66457 degrees, in most cases. Especially with the DMS which has 5 axes, compound angles and complicated reference geometry is easy to make.

In the next step, you'll see where this takes the design.

I've been wanting to do some work around metal brackets for furniture... Yet this application was so seamless with the concept I didn't even notice that I was doing it. I considered sand casting some of the brackets, and so early are striving to incorporate cast geometries. Casting is great because you can make multiples without waste from just one pattern, but at the scale that I needed, CNC ended up being a more straightforward process.

In the sketches, I'm trying to find visually pleasing forms that related to the other shapes in the design. I decided to use brass inserts threaded into the wood as supports for machine screws to hold the brackets. I used flats and set screws to hold the different brackets together. The designs also are seeking to incorporate reference geometry for CNC machining into the form for the design. By adding a surface that is parallel or perpendicular to a key angle that I needed to machine, I avoided having to make special angle blocks for myself.

I sought to create simple mechanical constraints - and concentricity is such a good one. Circles are easy to machine, and automatically give you X/Y registration for parts in just one feature. This also allowed me to match diameters on the CNC parts I was designing to tubing that I bought from McMaster to connect the rockers and the seat to the back. It also meant that I would need to use the Haas lathe as well as the CNC mill to complete the project - a fun and time consuming diversion. Totally worth it, of course.

All my designs were done with the CAD for my fasteners on the next screen over, to make sure that everything was going to fit just right. For me, rapidly moving back and forth between "nuts and bolts" thinking and conceptual or aesthetic design really helps me draw out the cleanest connection between my design intent and the forms I create. Its a really literal workflow between form and function.

My goal was not to have the brackets scream out. They are certainly present - neither hidden nor celebrated. But I wanted to give them my full design attention, to make them pleasing and subtle.

With the brackets finished, all my geometry was locked successfully into place, and I was ready to begin fabrication.

WIth the design finished, it was time to get the stuff to make my chair.

I purchased all my hardware and special metal bits from McMaster-Carr right away, so that I could work out any hardware kinks in real-time. This was mostly just fasteners, including 1/4-20 set screws, 10-24 bolts, and brass inserts, as well as 5/8" tubing for connector pieces and brass rod to cut into dowels. All the aluminum for the CNC machined brackets were available as stock material at the Pier, thankfully.

The real fun was going to Macbeath and picking wood. I ended up choosing maple - a warm, light tone - to contrast the blue-ish aluminum. I wanted to keep the theme of contrasting elements, and I felt that aluminum and maple would pair well together. Since each is a simple, classic material, it wouldn't be too extreme a combination, either.

I ended up selecting to large, long board to take back to the Pier, which allowed me to book match various sections of the chair (the arms, for instance) while still giving me creative room to create a nice lamination for the seat and back. I had contemplated using small dimensions boards that required less work, but opting for bigger stuff gave me much more flexibility. I was also looking for generally clear, light boards that seemed nice and true without many knots to work around.

Before I made the first cut, I sketched out where each part of the chair would come from. I took into consideration where I wanted to match pieces or create continuous relationships. For instance, the arms are cut from the same slice of wood, side by side - so on the final chair the sitter is between the cut line on the two blanks. The legs lead directly to the arms, and the arms to the rear seat supports. Similarly, the sitter is between the matched faces of the legs, rear supports, and rockers matched like the arms. These little details create a subtle visual continuity between the pieces.

Once the layout was done, the work was straight forward. I rough cut the blanks well oversized, and then set about milling the wood four-square. That means the all the edges of the cross section are perpendicular to each other. First, I jointed the a rough edge of each piece, then jointed an adjacent face to be perpendicular. Of course, before any of that, the jointer must be carefully set to exactly 90 degrees by square or digital angle gauge. With two perpendicular sides, I used the planer to mill the opposite face square to the clean one. For now, I removed the minimum material, in order to leave myself room to match parts. The last step is to clean the final rough edge on the table saw, which is the perfect tool for the job now that the first three edges are “square.” From there, I started segmenting the matched sections of the boards, and then planing these to the proper thickness, and cutting them on the table saw to the correct length and width. Obviously the wood will move a little bit, but taking good care at this stage will mean my matches will end up nicely, and that on the DMS CNC router I would be able to get consistent results.

I milled the seat and back portions with great care, since there was a number of factors to consider.
First, I wanted the parts to have a relationship to each other, so the panels from the seat align with those on the back. I also added details to the lamination with face grain intermittently interrupted by edge grain (no end grain, thanks!) which required more care.

But the two most important factors I was keeping an eye on were wood expansion and saw angle. By alternating the grain arcing upwards and downwards symmetrically from center, as the wood expands and straightens out, the changes will offset each other, resulting the truest possible lamination. Further, I flipped mating faces in the lamination to be cut on the saw so that any tiny deviation in saw angle would be slightly acute on face and slightly (equally) obtuse on the mating face. This is a subtle distinction that a precise angle gauge can solve, but I still think its best practice. Theoretically, you could do a lamination several degrees off and still get away with it using this technique!

Once the boards were milled properly and matched as I wanted them, I cut a couple extra pieces to serve as holding points for the DMS, which I added to the outsides of the lamination. In the glue-up, I was aiming to get a light but properly slick surface on both faces to be glued. It’s important to move quickly so that the glue is still flexible when clamps are applied, so the right pressure can be added to make sure the glue-up is completely straight.

I was excited to see nice oozing of glue from the seams - a sure sign of a tight gap and ample glue.

The process of making the rocker slats to laminate was very time- and dust-intensive.

After milling the profile of the outer dimensions like the other boards, I had to add a taper to the outside of each rocker. I wanted this taper to give the chair a more three dimensional, dynamic feeling. I marked and cut rough on the bandsaw a two degree taper, and then use the disc sander to get it nearly perfect. At that point, I devised a crazy workholding scheme that allowed me to use the short two degree wedge to cut the taper on the first rocker. I clamped the taper to another square-edged board, and clamped them to the rocker. Using a low profile board I was able to avoid the clamp with the fence, and still have a solid reference edge to guide the cut. I slowly worked up to my line.

Once one rocker was tapered, I just clamped the second one two it and used the first as my jig.

With the tapers cut, I slowly ripped 1/8" slats off one at a time. I should have swapped to a ripping blade to leave a finer cut and minimize waste material... Next time. By the last cut, it was seriously hairy there was so little stock material left.

With the slats cut, I used the DMS to machine a precise set of rocker buck pieces for me to clamp the slats to for gluing. I used dog-bone joints to make a fixture for them, and left large holes to allow for clamps to hold at even intervals.

Once the jig was ready, I had to make one last series of cuts in the slats. I ripped long strips in each slat to create a slot for the waterjet cut aluminum flange that runs along the rocker. Thankfully, the kerf of the table saw matched my material thickness (no accident!) so I ran the saw blade up to my line and carefully backed it out.

I lightly sanded each slat to make sure the gluing surface was great, then started applying glue to each face... Just enough to make them slick. Nice to see them oozing out the sides.

I used a small piece of aluminum to space the slats to the thickness of my eventual flange, just to be sure there was room.

With the rockers well in the works, I came back to the DMS to mill the legs, arms, and back supports.

I chose to do the entire piece in the DMS, but in retrospect, I should have rough cut each piece on the bandsaw with a little extra material left behind, then just put it on the DMS for the finishing touches. There was tear-out in some areas that I suspect I could have avoided by doing this.

Nonetheless, it was great practice working with wood, as well as using the DMS to joinery, especially the 5-axis capability to make angled tenons!

Fixturing and toolpaths are critical to the success of this and the next series of CNC steps in this Instructable. I will search around a little for good references and write sub-Instructables on the nitty-gritty details of each part if I feel there is a need. Let me know if there's something you'd like to see.

The idea was the same on each. Drill and mill pockets and holes on the inside face of each part and cut half a tenon, then flip it and mill out the shape. The shape could be tough because of tear-out, so I ended up making a small knife wall with a couple light passes before really trying to hog anything out. My woodworking teach would be proud.

The operations for milling the seat and the back were not much different than the prior steps on the DMS. However, each piece represented over $100 in wood and the two were matched together from the same board. So it was high stakes... And there were a few errors that I had to work though, which I'll mention at the very end of this Instructable.

Essentially, the milling pattern was to cut the contoured side first, using a large flat end mill to clear most of the material. Then, I followed that with a ball end mill to leave only small scallops that would be easy to scrape and sand away.

The bottom and back (respectively) on my two parts had all the complexity of compound angles in milled pockets, so the 5-axis DMS was critical to make that happen. I milled away the bulk of the material, then machined reference planes on the bottom side (a feature of my design, to contrast the organic top surface), all before adding the pockets for brackets.

From here, I would sculpt the shapes by hand, so I was happy with anything close to the final shape to work from.

Once all the other work on the DMS was finished, I went back to the rockers and brought them into the machine to make the mortise for the leg, and to machine pockets for the rear crossbar bracket. These operations were nerve-racking because of how much labor was invest in the rockers, but otherwise quite straightforward.

Since I always planned that these would go into the DMS, I designed flat sections on each end of the rocker for me to clamp in the vise.. Without a flat section to reference, alignment for machining would have been a nightmare!

These were some of the most complicated little parts to make for the entire chair.

The first part in the photos on this step was particularly hard. It has a significant angle between the axis of the rod (which was also the axis of the lathe) and the milled interior cavity that would house fasteners.

The workaround for that part was to mill the top surface rough, but mill as best I could clean edges on the bottom and sides. Since the bottom as at an obtuse angle relative to the axis of the lathe, there was room for me to work a flat endmill do the length of the bottom surface in 0.005" increments, leaving a neat surface which was at a weird angle - but precisely the angle I needed for the cavity that I needed to hollow out. This meant that I could take the part over to the Haas mill and finish machining it over there, where I could use the clean edges and bottom to load it precisely into the machine.

The second part was easier, since the pocket I needed to clear was perpendicular to the lathe's axis of rotation. The Haas lathe can perform radial drilling and milling, so I simply used an endmill radially to act as it would in a 3-axis vertical CNC mill.

For both parts, once the appropriate machining was performed on the bracket, the dowel was turned like a normal lathe, which grooving and OD turning tools. Then indices were added for set screws. The nice thing was that I could run an entire part exactly the same, just leaving out the indexed surface and the final parting off, and then run those separately depending on which part I needed.

The last stop on my CNC machining tour as the Haas 3-axis vertical mill. My bread and butter. It was a nice treat to have dealt with the finicky Fagor controller on the DMS, wrapped up my trials on the Haas lathe, and just crank out parts on the mill.

I took lots of photos of the process, so I'm going to mostly let them do the talking. I will add a couple things, however.

A big time saver with as many parts as I needed to make on the mill (I think 20) was to use the edge of the vise as a reference for my X-home coordinate. By holding a parallel to the edge of the vise and carefully pressing my material against it, I could get a highly repeatable X-zero value just like with a vise stop, except I didn't have to worry about crashing into a vise stop - which I didn't have any room to avoid on some of my tool paths. I'm sure machinists everywhere are rolling in their graves.

In general, I used a 0.005" stepover with a flat end mill to machine any angled surface really quickly. A ball end mill always runs into height issues with tangency putting the bottom of the tool below a given surface - I didn't have room for those shenanigans. By making everything with a really small stepover, sandblasting made the marks almost invisible. But to a trained eye, there are still there, and then it just highlights a pretty cool feature of the chair - all the custom CNC milled hardware!

I would mill the top of my part from the stock, and create clean references on my part within the stock, so I could flip and machine away left over stock on the backside. With the stepover angle machining, I also created references for weird angles that my design required. These angles became like built-in angle blocks.

The last little trick I used was to use precisely machined parallels to changed the Z-zero height of my work coordinate system, so I could swap a parallel, apply the offset to the machine, and run the next program. All my parts had coordinate systems on the bottom left back of the part, which made it really fast to keep changing parts with almost the same coordinate system. Except for the angled bores, which I did in a batch at the end, since I chose to reference them in a different way... I'll stop now and write a sub-Instructable later. This step is getting close to becoming one as it is.

At this point, I've entered 2D heaven, fabrication wise. I feel like I've crossed the Red Sea and all that's left is a little red sand and a lot of water in the Omax waterjet.

This thing cuts parts really clean, really straight, really right... I have almost nothing to report on this step, except that waterjet cutting is amazing when the machine isn't broken, despite our incredible shop staff's efforts to keep it running basically around the clock. Turns out machines really hate water and sand. Go figure.

My parts got cut.

As much as I would have loved to say at this point it was glue and go, there was still a bit of work left to do.

Aside from cutting the slots for the waterjet cut flanges to rest in, I needed to clean up the joints from the DMS.

I can't be quite sure whether our machine is slightly out of tram (when the table is not level with respect to the spindle's Z height) or my fixturing with CNC strap clamps (the bolt-y kind) create a slight deflection from parallel. In any case, I needed to square off the edges of my joints. Fortunately, I noticed this while taking the parts of the DMS, so I undersized the mortises I cut so that I could shave the tenons square and then down to size.

I used the table saw with a simple tenoning-type jig to just shave a hair off of each face of the tenon until it fit perfectly in the mortise.

For the aluminum slots, I used a ripping blade (kerf is less than 1/8", compared to 1/8" stock aluminum) to make my cuts cleaner. By using a smaller blade, I was able to make a cut referencing one side of the wood, then flip the piece to the other side and widen it but keeping it perfectly centered. I used a test piece to get the width just right (hammer tight), and then cut the real thing.

Brass pins were used in many places to hold the wood and aluminum together, although the fit was so tight, they are mostly redundant. But that means the joints will last better even as the wood moves over time.

This was a fun process. Super manual, and fairly quick. Basically, by carving in smaller scale and lighter materials during the design process, I set the DMS up to just hog out the material I didn't need, leaving the sensitive fine-tuning (the fun part) for me at the very end.

The other nice thing about this process was I used the same scraper shapes to shape the MDF prototype as I did to shape the maple final version, so all the shapes that were in the model made sense now that there were back in the physical world. This tactile part of the process was so gratifying after all those hours standing nervously over a humming CNC machine.

Again, the contrasting elements in this piece. Striking to me as the maker, that's for sure.

Now that all the pieces were fabricated, I could try putting them all together.

I spared the glue for later just in case there were any issues, and I left a couple angles still to trim up just in case there was any problems in the test.

I had to make a custom drill bit in order to get the brass inserts to fit. They are easy as pie in softer woods, but in hard maple, they just weren't going in. I had to widen the hole to an X-size drill almost 1/32" bigger than recommended, and put a flat on the end so it wouldn't cut deeper than I wanted and puncture a part, leaving a flat bottom to the hole. I also had to file down the threads on the brass inserts, because they were so tight. I did this by using a drill with a socket head cap screw to hold the brass insert while I held a file lightly in contact in the appropriate direction. Worked like a charm.

Everything seemed to fit really well, but there was a degree of looseness without glue that left me still feeling anxious even after the test. But it was time to charge for the finish line!

With everything set for gluing and assembly, I sanded all the parts to 220 grit or higher (up to 400 for touch points). It was much easier to keep crisp edges and get an even surface finish to sand before gluing, even knowing full well that more finishing would be require after glue.

I debated finishing the pieces with oil at this point too, but for reasons that escape me now, I chose not to.

This felt like jumping off a cliff. There's really no coming back from a bad glue-job.

I was so tired at this point that I hardly even noticed, although I think my body was able to produce one last surge of focus-inducing adrenaline. I test fit and clamped dry each joint before applying glue. I added glue to the inside side-grain faces of the joints, and barely even touched the end grain, since it is so bad at holding a glue joint. I opted to keep these faces clean, so that I didn't have glue spilling out the edges.

Everything went smoothly. Go figure. Sometimes things work like they should. I've just learned never to count on it.

With the glue dry, I added brass pints to the necessary points, locking in all the geometry I needed. It felt like a really nice extra guarantee to have the pins keeping shear forces from building up against the joints. Given the cantilever design, I was really grateful that I had the aluminum flanges for extra bending strength as well. Which is, of course, what inspired them in the first place.

And that was pretty much that!

All that was left was some final delicate touch up work, filing pins flush, scraping off a little stray glue, and rebuffing the contact points to a super-smooth 400 grit. With everything smooth and even, I applied Danish oil.

I did three coats. I wiped on the first, sanded the second coat in with 400 grit to create a slurry of wood dust that would fill in the grain to make a smoother surface. Then I wiped on the final coat.

It was a super fun treat to have the last step in the process be screwing in a bolt, and the chair was done.Very anticlimactic, although I will say when the head reached the face and stop turning, I had a pretty epic feeling of exhausting sweep over me. What a good feeling.

I did have to clean up a couple insert threads along the way, swapped out for shorter set screws, and had to widen a couple through holes to get them to fit. But all in all, things went smooth.

The funny thing is actually that as I was putting on the last bracket, thinking I was all done, I found it to be milled at the wrong angle, which I didn't notice during the test assembly because the lack of glue left it loose enough to hide the error. I'll mentioned a few more snafu's in the last step of this Instructable, but it did feel like a fitting end to have to hop back on the Haas mill for an hour to remake a part, right when I thought I was done.

Some projects just fight back, all the way to the finish line.

I have no words to put here, except thank you for reading, if you have indeed read this far.

Outtakes on the next and final step.

Yours,

Xander