The core of a torsion box is a grid made up of many thin strips of wood. For my shelves, those strips are 1/2″x3/4″. I machined the strips for the core using my table saw and thickness planer to ensure they were dimensioned uniformly. Then I moved to my tablesaw to cut the half lap joints using a dado stack.
Installing the dado stack, I tried to be lazy and not remove the insert to the right side of the saw blade (a sliding table saw is quite different than other table saws in this regard) and found that I could put a 1/2″ dado stack on the arbor and get the arbor nut to thread on enough. I was happy about that, and proceeded to get ready for a test cut. However, when I closed the hinged blade cover, I found that it contacted the blade and the saw will not run if the cover is not closed. I could probably have rigged up something to trick the microswitch into thinking the cover was closed, but that wouldn’t be very safe. Furthermore, I soon realized that my sliding table wouldn’t clear the dado stack. Okay, Chris, stop being lazy – you’re wasting your time.
So I removed the dado stack and took out the table insert – it’s attached to the side of the table casting with four machine screws and lock washers. I also had to remove the riving knife which only works for 10″ blades. I also removed the spacer on the arbor to allow the dado stack to sit further onto the arbor.
I reinstalled the dado stack. I am always careful to position the teeth of one blade in the gullets (spaces between the teeth) of the adjacent ones. Doing this protects the brittle carbide teeth and ensures that the cutters stack properly. In order to set my dado stack to be precisely as wide as my stock is thick, I built the stack slightly undersize using the two outside blades and inside chippers, then inserted shims between the blades as well to intentionally make the stack too wide. Then I made a notch in a scrap and tested the fit of one of the strips that would be used for the core. As expected, the fit was loose. Here’s the trick: I then fit shims into the gap until I had a good fit. The width of the shims inserted represent what I needed to remove from the dado stack to get the perfect fit. In my case, a 0.020″ and 0.005″ shim filled the gap, so I needed to remove 0.025″ of shims from my dado stack. Thanks to my friend Charles for showing me this trick.
I removed the necessary shims and then worked on adjusting the depth of cut. For the half laps, I needed the blade set at exactly half of the stock thickness. Since the core’s grid is formed with strips on edge, I actually needed to set the blade to half the width of the strips. I set the blade a little low and made one rabbet cut, then flipped it over and moved it over slightly to make the second rabbet cut. Since the blade was too low, the two rabbets left some material in the middle. I raised the blade a little and tried again until it was removed. Once I had the blade height set, I did a final test fit.
Then it was a simple matter of cutting all the joints. I set the stop on my sliding table’s crosscut fence and gang-cut the strips six at a time. To assemble the core, I brought all the pieces to my workbench. Using a small scrap of wood and my deadblow mallet, I fit all the pieces together. They would go together with hand pressure, but the mallet makes it easier. The scrap of wood protects against damage and also helps ensure that all pieces are set as deep as their neighbors. Once they were put together, I checked them for square and flatness. To glue the joints together, I used Chair Doctor glue which has a viscosity similar to water. I simply added a healthy bead at each joint and let it wick down into the joint. After a few minutes, I flipped the core over and applied glue to the other side. This was much easier than knocking the joints apart, applying glue, then putting it all back together.