In January 2009, I completed a coffee table I had been building for over twelve months. The idea first came to me in mid-2006, when I saw a beautiful coffee table at Crate and Barrel - for about $700. I looked at it and thought to myself, "I can make that."
It took a while to get around to it, but in late 2007, I roughed out my design for a strong-but-light table based on a hollow-core "torsion box" design: 1/4" thick oak-veneer plywood sandwiched on top and bottom of a 1"-thick lattice of poplar, with the upper sheet covered by a 1/8" layer of decorative oak parquet strips.
I needed a large, flat surface on which to build the coffee table. So before I started, I had to build an assembly table. That was enough of a project in itself that it merited a whole separate web page; click here to read all about that. The assembly table went together fairly quickly though, so after a month and a half - on New Year's Day 2007 - I was ready to start work on the coffee table.
The first step was assembling the poplar lattice onto which the plywood skins would ultimately be glued. Starting with 1" poplar planks, I ripped them into strips 1/2" wide for the internal ribs, and 3/4" wide for the perimeter of the whole lattice. 25-inch long ribs were to run the width of the table, and they were to be spaced out by pieces about 3.5 inches long. These short pieces varied in length a bit, so I measured each one with a dial caliper and sorted them by length:
the number on each piece represents the thousandths of an inch portion of each measurement; thus, the rightmost piece is 0.037" longer than the leftmost piece. Why sort them like this? Well, these short pieces sum up to determine the total length of the table; if they happen to stack up wrong, the left and right edges of the table could end up visibly non-parallel when done. The sorting allowed me to pick and choose which pieces to use together to make sure I ended up with a nice, "square" table when done.
Once sorted, the short pieces were pre-glued to the 25-inch long ribs:
The 5"-wide planks are where the legs end up being attached. The large holes in these planks are for weight reduction, shaving off perhaps a couple of pounds. The shiny metal pieces on the left plank are threaded inserts:
These inserts accept 5/16" bolts which will secure the legs nicely. Tangs on the inserts dig into the wood to prevent them from spinning when tightening the bolts. This is important, since these inserts are inaccessible once the plywood skins are glued in place.
After pregluing the short crossribs onto the 25-inch ribs, it was time to start gluing up the whole lattice. Here the short overhang is glued onto the outboard side of one of the wide planks:
That outboard rib on the left is 3/4" wide and forms the left edge of the table (not including trim, to be added later). The red object at bottom is a six-foot-long spirit level, functioning as a straightedge against which to assemble the lattice; this ensured that I wouldn't end up with a crooked lattice when finished. The boxes contain five pounds of lead each, and help to hold the wood pieces flat against the assembly table while the glue dries.
More ribs being glued into place, with more weights and clamps:
Once the interior part of the lattice was glued up, the 3/4" wide ribs forming the front and rear edges of the table could be added. I drilled dowel pin holes in the leg attachment planks and in the left and right perimeter pieces to assure strong joints. Here, the first one is already installed, and the second one is in (at front/left) ready to glue:
Here the second piece has been glued in place, held firm with nice straight oak planks and pipe clamps to assure a good, straight edge while the glue dries:
Finally, the completed internal lattice:
Without any skins on it, the 27" x 62" lattice was very flimsy and fragile. But that was about to change.
Anybody who has worked with drywall can tell you what peculiar, interesting stuff it is. The interior is crumbly plaster with hardly any tensile strength at all, and the paper on the surface could easily be crumpled up if it was all alone. But put the two together, and you end up with a building material that is surprisingly resistant to bending loads.
My table was to function on the same principle: a fragile interior lattice holding two tensile skins apart from each other, creating a lightweight structure that resists bending loads.
The skins were cut from sheets of 5-mm oak-veneer plywood. The stock sheets had sat in my basement for a couple of months before it was time to use them, and so they weren't exactly flat at this point. Here's the lattice, sitting on top of one of the sheets:
This warpage was of little concern. Again, the plywood sheet was not very strong by itself. Once I applied glue, all that was needed to make everything sit FLAT was a whole lot of weight:
With such a large area, it took pretty much everything I had - dumbbells, lead wheel-balancing weights, buckets of blasting sand, a drill press vise - to make sure the whole thing laid completely flat while the glue dried. But with a dead-flat assembly table underneath it all, a good result was virtually guaranteed. Indeed, once the second skin was glued in place, the finished sandwich laid completely flat. It was not quite as stiff as I had expected, but more structure was to be added later, in the form of wide oak trim that would help bear some of the load.
The oak-veneer plywood I used was manufactured with the intent of providing a ready-to-finish surface: the top ply was a paper-thin layer of oak, pre-sanded smooth and ready to accept stain and varnish. But for such a big area, I didn't want just a plain field of woodgrain; that would be make for a dull, uninteresting table. Instead, I decided to glue strips of oak in place to form a pattern.
Because this surface was going to see a lot of use/abuse over the years, I decided not to use thin "veneer tape" which was available from woodworking supply stores. Instead, I took a couple of 1" thick oak planks and ripped them into oak strips, 1" wide and 1/8" thick:
The pattern I had in mind was a simple diamond, growing outward from the center of the table. I divided the table into quadrants with my long spirit level and a carpenter's square:
And that's when the real work began. Working outward from the center of the table, each strip had to be cut to EXACTLY the right length to butt tightly against both the level and the carpenter's square, AND against the previously-glued-in strip. The "exact length" requirement got eased once I worked out past the carpenter's square; then the pieces could simply overhang the edge of the table by a bit, to be trimmed flush later on. However, I still could only glue the pieces in one at a time, with weights holding it down. In this shot I'm about halfway done with the first quadrant:
With enough strips in place in the first quadrant, the carpenter's square has been removed, and I've started adding strips in the second quadrant (at right). Notice the weights being used to hold the strips flat against the tabletop while the glue dries.
Occasionally a stubborn strip refused to butt up snugly against the previous strip, in which case some extra persuasion was required. Here I've completed two quadrants, and in the third quadrant (at lower left), I've got a piece of angle iron clamped in place to force a strip to butt up tightly where it belongs:
With 128 strips required, gluing them down one strip at a time was agonizingly slow. Typically I was able to glue down two or three per evening, requiring a couple of months to to achieve the desired result:
When I was finally done, I was horrified to discover that my once-flat table had somehow become warped. The corners of the coffee table were raised up by perhaps 3/16" from the surface of the dead-flat assembly table:
I finally realized that the moisture in the glue had caused the oak strips to expand slightly; then, as the glue dried, they shrunk, pulling the top surface of the coffee table tight, making the corners rise up. Thankfully, there were tricks I could perform to make this thing sit flat again.
One of those tricks was the edge trim. The trim is a structural element: made of oak, 1.5 inches wide and 2.4 inches tall, it would be very rigid. With the table core clamped flat while the trim was installed, this would help to remove the warp.
Rather than buying heavy oak timbers, I built each piece of trim up from a pair of 3/4" thick planks. Here, one long piece of trim has already been assembled (notice the glue that has dripped down its sides), and the planks for the second piece are drilled, dowel-pinned, and ready to be glued:
Once the glue gets spread on the joint face, the clamps (waiting on the floor) get put in place:
The trim is clamped to the assembly table while the glue dries, ensuring that the finished piece will be perfectly flat when done. A sheet of plastic wrap underneath the trim prevents overflowed glue from sticking to the surface of the assembly table.
With the four pieces of trim glued up, I ran each one through the table saw twice, ripping a thin layer off of the top and bottom. This got rid of the squeezed-out glue and left a dead-even edge, no overhang due to misalignment of the original planks:
Before attaching the trim to the table, the edges had to be cleaned up. The plywood skins overhung the edges of the lattice, and the oak strips were overhanging the plywood, too. All had to be trimmed flush to provide a flat surface onto which the trim could be glued.
The dowel holes, as well as residual glue here and there, meant that a flush-trimming router bit would not give an acceptably smooth/straight edge. I also didn't want to chip out any pieces from the oak strips I had worked so hard to install. Instead, I flipped the table upside-down and used a hand-held circular saw to clean up each edge of the table:
I used one edge of a sheet of MDF as a straightedge to guide the saw. With a high-quality fine-tooth blade cutting upwards through the oak strips, the quality of the cut was excellent; I knew the trim would glue up tightly, leaving no visible gaps on the tabletop.
With four clean edges on the table, I flipped it over and began drilling dowel holes. On previous projects I tried using commercially available self-centering jigs for this purpose, but found them to be lacking; they often wobbled, and the drill bushings they used were oversized, leaving a lot of variability in hole placement. Instead, I built my own drill jig for this project:
The jig is made from two heavy pieces of aluminum bolted together. The clamping flange has a sight hole for aligning a mark on the jig block with a pencil line on the wood; the jig block has a pressed-in hardened drill bushing. Since I use the same drill jig on the table and on the trim, the dowel holes on each piece end up at exactly the same distance from the top edge. Well, not exactly: owing to minor variations in wood grain, there's a little bit of variation here and there, but far less than I would have had with any of the other jigs I've seen. That means less material to be sanded off later to ensure a perfectly flat tabletop.
With dowel holes drilled in the left and right edges of the table, I was ready to attach the first two pieces of trim. I already owned several pipe clamps that utilized 5-foot pipe lengths. But the table was actually slightly longer than that, so I actually had to buy extra 2-foot pipes and pipe couplers to make the clamps long enough to hold these pieces in place. Here's the first piece of trim being held in place while the glue dries:
These short end pieces of trim projected out past the front and rear of the table. Before attaching the front/rear trim, I had to saw these pieces flush with the front and rear edges of the table to form one large glue joint. Again, I turned to my circular saw, along with a straightedge to guide it:
I didn't want to remove any of the poplar core between the plywood skins, but I had to just barely shave it so as to ensure there wouldn't be any "step" onto the end of the already-installed trim. This required careful positioning of the straightedge via a dial caliper, and repeated passes with the saw, moving the straightedge 0.005" at a time, until my pencil marks on the poplar got just barely scraped off as the saw passed by.
Before the first saw pass, with pencil marks plainly visible (around dowel pin hole), and maybe 1/8" of trim overhang:
After several passes, with the pencil marks just starting to disappear near the bottom, and trim cut perfectly flush with the existing face:
For installing the trim on the long front and rear edges of the table, additional dowel pin holes were required to solidly mate the long trim to the short pieces of trim on left and right edges. Instead of the 1/2" dowel pins I had been using, I used four 3/8" dowel pins at each joint:
Because the top edge of the trim was later to be rounded, I couldn't place a dowel pin too close to that edge; thus, three 3/8" pins near the inboard face of the trim, and one 3/8" pin near the outboard face.
The masking tape on the long piece of trim (laying on top of the table in the above photo) shows the part of the trim that will underhang the table when installed. Since glue squeezed out of the joint when I installed the trim, I put the tape here to make it easier to clean up excess glue; once the glue had squeezed out (but before it dried), all I had to do was peel the tape off.
Here's one of the long pieces of trim being glued in place, using every damn clamp I had:
Notice the line of glue which has drooled out onto the concrete floor below the glue joint. One of my pleasant surprises early on in this project was that wood glue does not stick to my concrete basement floor at all once it's dried. A quick poke at the edge of each blob of glue with a chisel, and it pops completely free. So I wasn't worried about all the glue on the floor; my priority was making absolutely sure that the joint was completely sealed with copious amounts of glue, no gaps at all. Mission accomplished!
With the second long piece of trim glued in place, the assembly work on the tabletop was done (the legs got dealt with later; see below). Here's the tabletop upside-down, showing the leg bolt holes and the trim underhang:
And the topside:
With the basic tabletop built up, it was time to smooth out the rough spots and sharp edges. First up? Round the corners. The wide trim allowed a generous 1.5" radius on the corners without cutting into the lattice core, or even getting into the dowel pins that were holding the corners together. However, the 2.4" tall trim was a smidge too tall for my jigsaw to cut all the way through from the top. Instead, I ran the jigsaw down one side, with its baseplate tilted at a 45-degree angle from the blade:
With the bulk of the wood sliced off, the rest was easily removed and rounded with a coarse-grit belt sander and an orbital sander:
By the way: for dust control, it's hard to beat a shop-vac attached to the orbital sander, as you see in the above photo. It can make the sander a bit unwieldy, but in my mind, that's a fair tradeoff to avoid blanketing the entire basement (and everything in it) with ultra-fine sawdust.
The oak strips on the top surface were not all perfectly even; their thicknesses varied just a bit, and they didn't all get glued down absolutely flat. Plus, as mentioned earlier, the edge trim didn't sit perfectly level with the oak strips, either. They all needed to be sanded down to an even, flat surface with coarse-grit on the orbital sander. This was a roughly 45-minute process that left my hands rather tingly (in spite of vibration-damping gloves), but had to be done. Of course, it also had the added benefit of removing the dirty, grungy surface, and exposing the virgin wood beneath. See if you can tell which side has been sanded in this photo:
Finally, I was ready to run a router around the edges to create a hand-and-foot-friendly piece of furniture. The bottom edge of the trim was to function as a grab rail, for pulling the table towards you. I had a 1/4" radius roundover bit lying around that worked well for that. The top edge? Well, I wanted to be able to comfortably rest my feet on it, so I needed a substantially larger bit:
This is a 1.25"-radius roundover bit, purchased just for this project. The overall width was 3 inches, and so I had to actually make a special baseplate for the router holder, with a center hole big enough to accommodate that monstrous bit:
With lessons learned from previous projects, I hung plastic sheeting from the ceiling around my work area and stuck the shop vac's hose inside to ensure no dust-laden air escaped to rain down on the rest of the basement. During the removal of all this wood, I wore a full respirator, without which I probably would have been gagging on all the sawdust thrown into the air. A brief shot immediately after finishing:
Check out all the sawdust on the table, the floor - and in the air. Looks like it's snowing!
The blue box in the foreground is a variable speed controller. Sort of like a heavy-duty dimmer switch, it allowed the router to turn at a somewhat lower RPM, preventing excessive surface speeds with that huge bit that might otherwise have burned the wood. If you look at the corners of the table, you can see where I did in fact burn it a little bit, but not much; that's nothing that can't be sanded out later.
Although the trim helped remove most of the warp from the tabletop, there was still a little bit remaining in the central part of the table. I was already thinking about strengthening/stiffening the tabletop a bit more, and seeing this little bit of remaining warp pushed me over the edge: I decided to add some stiffening ribs to the underside of the table.
Simple additions, just two 1"x3.5" planks (that's 1" actual, not nominal), each with a few dowel pins on either end anchoring it to the solid poplar planks in the lattice, and a continuous glue joint along its face:
The right rib is up on its side, showing the dowel pins and holes.
Again, lots of weight to hold things flat, plus something extra:
The dowel pin areas at the ends of each rib needed to be clamped, but their locations (far from the edges of the tabletop) meant a single clamp couldn't be used. I had to bridge each end of the stiffening rib with a short piece of 2x6, using two clamps to pull it down tight against the assembly table. The central part of each rib was persuaded to lie flat and straight with a whole bunch of weight, as you can see in the photo.
With the two stiffer ribs installed, the buildup of the tabletop was done, and it was time to turn my attention to the other part of the table...
Click here for Coffee Table, Part 2
Questions? Comments? Email me!
©2009, Mitchell P. Patrie