Sliding Dovetails for Top
2010.05.02
I previously described how I used a combination of dowels and wedges to hold the four sections of the top in place. Well… humidity is a powerful force. Despite banging on those wedges like a chimpanzee (to paraphrase Dire Straits), over time the sections shift and twist, undoing all the work I put into flattening the top. Also, while I was flattening the top, working along the grain, sometimes the sections would slide with the plane. So I really needed some way to prevent lateral movement like that, and hopefully minimize the rising and twisting movement as well.
I decided to try stopped sliding dovetails. I would attach the male part (the tongue? the key? how about rail) to the leg assemblies, and cut the slot across the bottom of the top sections. I would not want the dovetail to show on the front of the workbench, hence the stopped version.
I set up the four sections of the top on sawhorses, upside down, and clamped them together. I spent some time flattening the areas that rest on top of the leg assemblies. To speed up the process, I started with my belt sander, then did some work with my jointer plane.
2010.05.13
I used 3/4″ red oak for the rails. I ripped them with the Milwaukee circular saw set to 14°. Then I used the saw to make the first pass on one of the sliding dovetail slots. Of course I could not cut all the way to the final stopped end. Then I realized a serious problem with this method. Cutting that side was pretty easy. But to cut the opposite side of the slot, I would have to start the circular saw above the wood and plunge… at exactly 14° and in such a manner as to end up snug against the guide rail and then plow on from there. Yeah like that’s gonna happen. Much safer to use a handsaw.
2010.05.20
I used a section of 2x6, ripped along one edge at 14°, as a guide for the saw. I clamped it in place, along with another scrap as an end stop. The saw I used is a Simonds miter saw with a 24″ blade and 11 teeth per inch. I put some tape on the blade for a visual depth stop.
To keep the saw tracking straight, I used my left hand to push the blade against the guide about halfway along as I sawed with my right hand. I had to stop periodically to clear sawdust out at the stopped end. Drilling a hole down there first probably would have been helpful. I bashed a clearance hole with a mortise chisel after a while. I had difficulty keeping the saw level throughout the cut, so that I first bottomed out at the near end and then had to take the rest of the cut down to match. It was a slow process; according to my photo timestamps, each cut took me about half an hour. This old saw probably has not been sharpened in a long time though.
I started clearing in between the saw cuts by using a mallet to lightly drive a chisel held almost horizontally, running up the slot. This took less than a minute to clear out about half the depth, or more in some places.
Then I unboxed my Veritas Router Plane, a tool designed for jobs like this. I was amazed how fast and accurate it was. I thought I might have to take micro passes, but decided to try hogging out somewhere around 3/16″. No problem. The only issue I encountered was the chips piling up in a logjam because of the angled walls. I found that if I first went down the center with the saw or a wide chisel, then when the router plane went through, the chips would break into two smaller pieces that could clear the narrow top opening… at least more often. This was a minor issue though, really.
To set the final depth, I put the router plane on top of the two rails and then dropped the cutter until it touched the table. I then backed off just a little. I made a 14° sanding block to touch up the side walls a bit.
I intentionally cut the slots slightly narrower than the rails I’d made before, so I could plane those down to fit. I kept taking passes on the angled sides and then test fitting in the slots. Somewhere in this process appears to be where things went a little astray. As you can see, it’s hardly a piston fit, though not super loose either. I think I accidentally changed the angle of the sides as I planed.
I put the front section of the top in place on the leg assemblies, using the dowels as before. Then I slid the rails in until they hit the stopped ends, and secured them to the upper stretchers with countersunk screws.
I slid a couple more sections on, and added another screw to each.
When the last section was on, I tapped the wedges in place, and then re-planed the top. We’ll see if it holds.
2010.07.28
The sliding dovetails helped, but did not entirely eliminate the shifting problems. So I purchased some long lag screws. I used a forstner bit to drill a countersink for the screw head and washer, and screwed each section of the top into the one in front of it. I used about four screws for each section. With the top pieces now locked together as one element, I no longer needed the wedges and dowels to hold them together. I kept the sliding dovetails. Since they are stopped sliding dovetails, they keep the top from pushing towards the back, and they keep it from moving laterally as I use my handplanes along the surface.
If I were going to do this all over again, I would use the lag screws from the beginning, and also devise a different means of locking the top in place on the legs… possibly building vertical tenons into the legs. It would need to be substantial enough to resist the forces of rough planing and clamping something in the leg vise.
2012.12.06
I recently went back to the two dowels. However it did not solve a problem I had with the sliding dovetails: when I clamp something in the vise that is no deeper than the thickness of the top, the clamping pressure pushes the workbench top back, pushing the front face of the top out of alignment with the front face of the leg. It is, at least, less pronounced with the dowels than with the sliding dovetails.
(Repeating this from an update on the dowels page) If I ever build another workbench (egads), I will probably use the sliding dovetail tenon shown in Roubo’s illustrations, and detailed in Chris Schwarz’s book. This locks the front of the top and legs together so they move as one. Without glue, it could theoretically be knocked apart for transport if needed.
Vise Chops