CNC cut Hardwood Dowel Joinery

My first job out of college was for a furniture company noted for the metal framed residential furniture they made and sold. It was all designed to quick assemble.  No welding of metal parts. Clever engineering of steel connections made their furniture go together rigid and precise. One connection they used between small tubing into larger tubing was to taper the end of the small tube, then pull it tight into a slightly smaller die-punched hole.  This post is about a wood version of that connection.  

For this sample a three degree tapered end mill, a 3/8" diameter wood dowel, and a 1" diameter wood dowel are used. 

Taper the dowel and the hole

A 90 degree connection that wedges tight and stiff with no glue.  Draw tight with a wood screw. 

Pocket hole screw

Clean, precise, and easy to create.

The small dowel end was pre-drilled for the pocket hole screw. 

Pre-drilled for screw threads

Trapped tight in the tapered hole there is no chance the small dowel would split. The counterbored hole for the screw head could be covered with a wood plug. For a coffee table or end table a frame made from dowels using this joint could have the side frame pre-assemble and only require the end stretcher dowels to be screwed into the legs.  A simple wood frame that is rigid with clean joinery and no glue required. Hardwood dowels are available in many species. Imagine a tempered glass top resting on the frame to finish it off. 

Taper theory:   Wood dowels are imperfect. If they have been exposed to humidity changes after production then they are usually out-of-round by a few hundredths. Drill bits make round holes. A straight dowel end (likely a bit wider than thicker) into a straight hole drilled into wood is either loose in one axis, loose all around, or so tight you have to pound/press it into the hole. Pressing it into the hole will wear off the fat side or deform the hole to accommodate the oblong dowel end.  Use a CNC with precision down to 0.001" to recut the end to a "perfect" tapered shape and a "perfect" matching tapered hole. There is no abrasion between dowel and hole when inserting until near the bottom when the tapered sides start to press against each other. Tapping or pulling the dowel in the last 1/32" or so presses the sides together.  Uniform contact around the perimeter. Friction all around. 

Comments welcome!

4D

  

Whimsy in Design. Magnetic Appeal.

Whimsy

I started with a small table top made for a rejected project.  I added in a couple of 36" long 3/8" diameter dowel rods I've had for a couple decades or more. I cut them in half to be the legs.  Then I made 4 wood balls.  I mixed in some 3/8" diameter magnets left over from a previous project, and this is the result. 

I drilled holes for the dowel legs in the bottom of the table top using my CNC.  I wanted the dowels to angle out rather than stand straight up, so the CNC jig was tilted 5.8 degrees down from horizontal. 

On my Compound Angle Clamping Jig 

Rather than glue the dowels in, I embedded magnets into the bottom of the CNC drilled holes.  Since wood won't stick to a magnet, I inserted metal screws into the ends of the dowels. The biggest challenge was centering the screw heads on the dowel ends. A bit of filing might have happened to make them slip easily into the holes. 

Embedded Magnet

The holes in the top were limited in depth by the screws that held the magnets in place. Any deeper and the screws would have poked out the top.  An ideal design would have deeper holes to better brace the inserted dowels. 

For feet I made wood balls, 1.5" in diameter.  They need a 3/8" diameter hole in them.  They will get magnets in the bottom of the holes and as such also stick to the screws in the ends of the dowels.  For whimsy I'll painted one of them red.  Experience tells me the one red shoe will make the table more memorable.  

I gave my CNC rotary axis a chance to prove it could make wood balls.  

Five Maple Balls on my Rotary Axis

Five 3/8" Diameter Holes

One Ball, ready to trim. 

Four wood ball feet.  Ready for magnets and a finish.

Rare earth magnets hold all the parts of this table together.  You could replace the wood dowels with steel rods and skip having to add screw heads to the end of the dowels.  That would add considerable weight, sterilize the composition, and require paint to prevent the metal from rusting.  The wood dowels are just enough.  No more than needed to hold up the top. When whimsy strikes you can pop off the ball feet and replace them with perhaps a seasonal color alternative.  How about 3D printed shoes to slip over and magnetically stick to the dowel ends?

 

All projects survive longest if the parts have some finish on them.  I set about first applying clear finish to the wood balls.  No problem with the first three, but the 4th ball protested.  It tried to escape by rolling off the bench, but I caught it. It got so angry at the thought of joining the clear finish crowd that it turned red in the face.  In the end it got its way.  Meet the angry red shoe.  I may have to rename this post as "A Table with One Angry Red Shoe.  

One angry red foot.

The ball feet balance the composition.  An echo of the thick top. They recognize the thin dowels as official table legs. Although this little table stands up proudly, the thin dowel legs happily flex a bit, and a bump against the table top will reward you with a quick wobble dance. Not a table you'll want to sit on. Not a table for anything like a small TV.  Park the TV remote and maybe a coaster with a drink on the top. A frail little table such as this one needs a place out-of-the-way.  Where no one might accidently bump it.  Not in the open where predatory beasts in the house might find it. A more stout version with thicker and stiffer legs would eliminate the wobble, but also remove the joy the table elicits when it does dance a bit.  😉

Comments welcomed and appreciated.

4D

Wedged Tenon Joint for my Landscape Model Stand Design

Test Sample

My Landscape Model Stand posted here used a wide floating plywood tenon between the feet and the vertical legs. It was never intended to come apart.  This joint is one of a few different prototypes to use as an alternative connection.  It has to be snug with no flex when together, yet not glued so that it can come apart. 

Embedded Square Nut

The mortise was cut 1/16" deeper than the length of the tenon. That allows the screw to pull it in snug and not bottom out. A square 1/4-20 nut was embedded into the bottom of the tenon. The pocket for the nut was cut with a 1/8" spiral end mill. Metal threads being more reliable than threads cut in wood. 

A flat head 1/4-20 bolt pulls the joint tight.

When the wedged sides are pulled together there is no gap or play between the mortise and the tenon. Yet it slides in easily until snug. 

Together tight with no play.

Creating the toolpaths was tedious, and cutting the mortise and tenon sides took considerable time.  I realized after cutting this sample that I do have a 3° tapered end mill. It would have been quicker to draw up toolpaths for and using it would be much faster than using a 1/8" end mill and the fluting toolpath to cut sloped sides of both tenon and mortise. 

3 degree tapered end mill version.

Making the foot out of the same strip of wood as the leg is simpler than making a unique foot from wider material. The design requires a very stiff joint with no flex or wiggle when together.  This wedged joint drawn together with a machine screw satisfies that requirement. Using a stainless steel nut and bolt they won't rust when exposed to the moisture in the wood. I have seen that happen in previous designs of mine. 

Now that I have the details figured out and simplified, I'm making a knock down version of my model stand using Philippian Mahogany.  Light weight yet plenty strong for the application.  2" wide by 3/4" thick legs and feet.  A 3" wide shelf. Stainless steel nut, bolt and hinge at the top. This type of bolt rather then the flat headed bolt used in the test piece above. 

3/8" diameter Aluminum dowels hold the shelf against the legs although if I could find some stainless steel rod or tube already 3" long and 3/8" in diameter I'd use it. 

Test Assembly

A stainless steel hinge recessed flush into their sides holds the legs together at the top. 

Hinged together at the top

With the hinge at the top and the shelf in place the triangle created locks the geometry for a rigid structure. The light weight Philippian mahogany used makes the whole assembly easy to pick up to move about. Slide the shelf off and the legs will fold together for a 1.5" thick flat package that stores easily.  Standing up spread out several of these will nest together.  They can also be stacked, although at 5' tall each the stack would grow in height quickly growing  3" or so with each one added. 

A photo of the assemble design. Note the two black bolts near the bottom of the legs which hold the feet securely.

Stand.  Ing Tall. 

Update 3/1/2025:  The mahogany I used for this stand has some internal stress, and as such the shelf wants to twist and slip back from the legs on one side.  To help it stay in place I bought some rare earth magnets with a hole in the center for a #4 screw. 

Rare Earth Magnet

First to replace the aluminum rods with steel rods. Then screwing a rare earth magnet into the bottom of each hole. That will keep the rods and shelf firmly in place yet allow removing the shelf with a simple tug to overcome the magnetic pull. 

3/8/2025:  Now with some cherry stain and Minwax Polycrylic on it this latest version of my model stand/easel posed for a few pictures:

In summary, the wedged tenon drawn in with a bolt works well.  Of course that joint took my CNC to make, but my CNC is just another tool among many that contribute to the fabrication of my designs. 

Comments welcomed.  

4D  

Pseudo Tensegrity End Table

Held stable with only tension straps.

Most tensegrity projects have a rigidly connected post up from the base and another that projects down from the top.  They are connected by a simple tension cable. An unstable connection.  To stabilize this usually 4 more cables run from top to bottom at the outer corners. The end result is barely stable.  An unexpected side push will usually collapse such a table. 

My project starts with a rigid post that is not attached structurally to either the top or the bottom.  An unstable assembly. Both ends are rounded over and sit in a slippery socket at the top and bottom centers. The center post is threaded.  Right hand threads on one end and left hand threads on the other.   

Bottom and top of the center post rounded over.

Screw holes patched with walnut plugs.

Holes repaired

The ball ends of the post rest in sockets cut in HDPE that inset into the bottom of the table top and top of the base. 

Socket test.

A good fit.  Post spins easily in the slippery socket.

Hub nuts screw onto the post. They have slots around their edge. Webbing straps will slip into the slots and be pinned in place with a 5mm pin though a grommet in their end. 

With both RH and LH threaded hub nuts, a twist of the post will tighten or loosen the straps. I cut the threads using the threading toolpath available in Vectric.com's Aspire software and a side cutting V bit. I hook the straps to the underside of the top and the hubs in a slot with a pin dropped in a hole or slot.  The pin passes through a grommet in the ends of the straps.  The straps end in a slot perpendicular to the pin. 5mm x 5/8" steel pins.

Strap connection to hub nut.

Shelf pins pin the strap end in place.  I made a friction fit hole for the shelf pins so they fit snuggly and will not drop out easily.  It took some test holes to find the best size of drill bit to use.  A #9 drill bit provided the closest fit. 

The top has 4 perimeter slots for straps, and an inset HDPE section with a ball socket cut in the middle.  Socket test seen above.   

Underside of the table top.

The key to this project is how the straps are connected to make the project work. 

The non-threaded section of the post is where you grasp the post to twist it.  It might be easier to grasp and twist if it's shape was a hexagon or octagon cross section.  

For detail continuity the base is a scaled down version of the top. The contoured surface is on its top rather than its bottom. 

Base

I made both top and bottom inserts from some 1" thick HDPE I have, re-sawn to just under 3/4" thick, then milled down to 5/8" thick on my CNC. 

Socket inserts for the top and base.

A  test assembly.  Straps were installed with the nut lower on the post.  Once attached the post was twisted to lift the hub nut higher to tension the straps. 

Repeated for the top.

Repeat the straps and hub on the other end to hold the table top securely horizontal.

Making straps for the design I've realize matching strap sets is a hit or miss problem. It is easy to make two or more the same length.  As those in the long axis aren't the same length as those in the short direction, and all have to be a precise length to tighten up at the same time. Finding the length of a set to go with the other set is the challenge.  Making and installing the set for the long direction would be easier if they could be adjusted in length after installing.  Change them from being a fixed length with a grommet in both ends to an adjustable length with a grommet in one end. The other end wrapping around something to come back and attach to itself.  The attachment point should be adjustable to change the total length of the strap.  Something I'll look into should I use straps in future projects. 

The wood parts of this table need some final sanding and a finish applied.  I like tung oil on walnut.  This table does provide an interesting glance down at the base while in use.  More interesting than any typical table. It was an enjoyable exercise to design and build. 

Straps in tension make a stable assembly.

4D

 

Blind Double Dovetail Joint

I used this joint once in the past for a student's project. I stumbled across a need to use it again while iterating on an end table design. It allows dovetailing a board into the face of another board with no obvious entry point. Drop in.  Slide over and the board won't pull out. Dovetailed sections with a slight taper do the job. 

The taper of the dovetail keys makes them easy to insert but wedge up tight when slid over. 

Cherry 4" wide into mahogany face 

I'm toying with a way to lock the joint in place when assembled.  Of course gluing them would work. This is to connect them without using glue.  First idea would be to use a spring loaded pin that would pop out and prevent the joint from sliding back to remove. Another idea is a pop-up plate that blocks the dovetail section from backing up once passed. Conical springs behind the plate would allow it to be pressed down out of the way when the board is first inserted, but then pop up when it slides over.  

Drop in, slide over for a snug connection.

Should I come up with a locking idea it'll mean the boards will need to be sanded and a finish applied before final assembly as they won't come back apart. The fit can be tested before the locking pin or plate is put in place. 

I used a 1/8" spiral end mill to removed most of the area before the dovetail bit undercut the edges. 1/2"d 14° dovetail bit.  On the cherry board end a 1/4" spiral end mill was use to pocket cut the areas between the dovetail section. 

As the joint is only 1/4" deep it could be done on the opposite side as well. You could simulate a board that passes through another board.  Using my angle clamping jig and some careful layout the board could intersect at any angle up to 15 degrees or so from perpendicular. Rotate the joint intersection for a compound angle appearance. 

4D

Unfolding Flat Pack Coffee Table Design

Unfolded

This design is loosely based on a folding workbench table I came up with 20ish years back. Reduced in scale and revised in detail the table is 16" high when unfolded, 42 inches long, and 15 inches deep. Folded up it is the same length and depth but flat and 2.25" thick. No tools are needed to erect it from it's folded state or to fold it back to its flat state. The base consists of two leg panels that aren't attached to the top, two flipping panels that are attached to the top end of the legs and the top, and two diagonal braces that attach to the top and near the center of the leg panels. A twist lever locks the flipping panels to the top when the table is unfolded.  This is so the table can be picked up without wanting to start folding. 

Underside View

For the folding action to work a precise connection point between leg panels and braces must be found. Through some iterative drafting steps I've been able to get within .001" of where it is. To fold flat both the leg panels and the braces have an offset end. The leg panels are one inch narrower than the  table top.  The flip panels and braces are tucked under the top to minimize their visual impact on the stance of the table. 

Leg Panels

Flip Panels and Braces

When folding up the flip panels flip over and out of the way so the braces can lay down flat against the top. 

Folded flat.

The pivoting connections are the only challenge left to resolve.  The flip panels connect/pivot from the top underside and have to flip 180 degrees over where they connect.  The braces connect to both the underside of the top and near the center of the leg panels. Some clearance between pivoting parts and where they pivot from is needed to prevent wood from rubbing against wood as they pivot. 1/8" (3mm) Baltic birch plywood could be used as the pivot brackets extending from the bottom face of the table top.  Glued into a slot they would be a challenge to remove/replace if they ever fail though. Plate brass or aluminum could be used but would need a mechanical connection into the table top.  This connection deserves an original design solution. 

The leg panels have openings for the flip panel and the braces.  Making each leg from three sections assembled to capture the other parts would be the obvious strategy.  I need a way to mechanically do this so they can be taken apart should any piece fail. My intersecting binding bolt trick I used on the last two TV tray tables might work.  I'd like to figure out a way to have the action of bolting them in also snug the joint tight. Some sort of cam or lever action. There is a double end cam solution already out there used for knock-down furniture that might work.   

It wouldn't take much to modify the geometry to side table dimensions. Longer legs and a shorter top might require one leg to fold over the other to keep the folded length within the length of the top.   

My house is overrun with coffee tables so I doubt I'll ever build this design. It is parked here for future reference so I can clear the idea from my mind and let it fill up with newer ideas. 

CNC Cut End-To-End Finger Joint

I cut this joint as a sample to show my college furniture design class students. Originally just an idea I had to prove would work.   I haven't found a practical application for it yet, but that was true for every original joinery idea I came up with.  So far roughly 50% have found applications in student furniture projects. 

Stacked Sides

Each side was cut using the same vectors.  Cut inside the vectors for one half, and outside the vectors for the other half.  A -0.003" allowance was used for one side. This joint pattern can be extended for any wider width of  boards.  Done with a 3/16" diameter spiral upcut router bit.   The bit diameter was what specified the shapes and radiused corners..  From the outside the joint looks like simple interlocking fingers. 

Closed

Slid apart the secret begins to reveal itself.

Slightly Apart

When together the joint can't slide sideways or up and down.  Locked in two axes. 

Apart.

To cut the joint required clamping the boards vertically in my CNC frame.  

While the halves slip together relatively easy once glued the considerable surface area between the sides would make for a very strong joint. One option might be to make each side from a different wood. Split a table top making one half walnut and other half pecan perhaps?  

4D