Wednesday, February 9, 2011

A New Road to Flatness

Sooner or later, after tearing themselves away from the internet and maybe making an end grain cutting board or two, most woodworkers wind up building a workbench. There are a zillion variations on the theme, but one thing that every workbench needs is a top that's flat and likely to stay that way.

The Old Ways

To make flat work surfaces, the traditionalists glue up mondo slabs of wood, grab their hand planes and go to town. Other folks glue up the same ginormous slabs, then flatten them with routers and sleds. Both of these methods involve quite a bit of work, and the router method in particular makes a mess and a half. Worse, because wood is wood, it will move over time and it's a pretty good bet that a thick wooden top that's flat today won't be so in a year or two or five.

A completely different approach is to abandon the thick, heavy slab, and build a torsion box instead. The result is a stiff, lightweight structure that will remain stable over a long period of time. That stability is great if the torsion box comes out flat to begin with. But if something goes wrong during the build and it ends up with a bend or a twist, about the only fix is to start over from the beginning.

A New Way

"If you can't make it perfect, make it adjustable." I don't know who first said that, but it works for a lot of things, including yet another way to make a flat work surface.

Suppose you have a square piece of 3/4" MDF, say one foot on a side. It's going to be flat, and it's going to be stiff and stable enough to remain flat over that small area. But a panel of the same material as big as a workbench top will sag under its own weight, and will not be acceptably flat over its entire area unless it is supported by, um, something flat?

Sounds like a Catch-22. It is, almost, but not quite. Here's the trick. If the panel can sit on a series of regularly-spaced supports, and if those supports can move up and down, then it's possible to adjust the elevation of the panel by a small amount at each support position. That in turn makes it possible to tweak the panel until it's almost perfectly flat. I've done this with both a workbench top and a router table, with very good results.

The Support Mechanism

To make the adjustments easy, I used two 1/4-20 bolts at each position, as shown in the figure. Both bolts are threaded into T-nuts that are embedded into the rigid frame of the supporting cabinet. The bolt on the left is the one that actually supports the top. It bears against a metal disk (a penny, actually) embedded in the top so that the bolt doesn't dig into the MDF. The bolt on the right simply holds the top down and in position against the first bolt.

Both my workbench and router table have pairs of bolts like this arranged in a grid with the pairs spaced about twelve inches apart. Here's what one of the pairs looks like in real life, viewed from the bottom:

Adjusting the Supports

It's easiest to adjust the bolts in two stages. Start by installing all of the bolts that come up through the frame to support the top. Then select a bolt somewhere in the middle as a reference, and use a carpenter's level to adjust all of the other bolts so that they are at the same height as the reference. If you have a good level and do this carefully, the tips of all the bolts will define a surface that is not only flat, but also level.

Next, set the top on the support bolts and install the hold-down bolts. Tighten them until they are snug, but not really tight. Then, using a straightedge for reference, adjust the bolts so that there are no humps or dips in the surface. To raise the surface, loosen a hold-down bolt slightly and then snug up the corresponding support bolt. To lower the surface, loosen the support bolt first, then snug up the hold-down bolt.

This sounds easy enough, but it can be frustrating if done in a haphazard fashion. A methodical approach works much better. Here's one procedure that works well. Assume that your bolt pairs are arranged in three rows of three, as shown in the picture.

Start by placing a straightedge along the diagonal from corner A to corner I. Then adjust the bolts at A, E, and I so that there are no gaps between the straightedge and the top surface. Next, place the straightedge along the diagonal from corner G to corner C. Now, without changing the bolts at E, adjust the bolts at G and/or C so that there are no gaps. Finally, without changing any of the previously set bolts, adjust the bolts at B so there are no gaps when the straightedge is placed along the A-B-C edge, adjust the bolts at D so there are no gaps when the straightedge is placed along the A-D-G edge, and so on for the bolts at and F and H.

The procedure for other arrangements of bolts is similar: first set the diagonals, then without changing any of the previously-set bolts, adjust all the bolts in the intermediate positions between the bolts that have been set.

4 comments:

Paul-Marcel said...

Hi idea, Russ and a nice start to a new blog! Look forward to more ideas. Excellent detail, too.

rmac said...

Well, Hi idea to you, too :) Thanks for the encouragement.

dbird said...

Neat idea. I've been looking for something like this to build a (What?) router table. I'll tell you about it the next time I see you at BarrosGeeksPizza.

BTW, after you followed the whole procedure, did the AEI and GEC diagonals still measure the same?

rmac said...

@dbird: I hadn't completely worked out the procedure when I set up my router table, so I can't say for sure. I was still in flounder mode, and it took me a few tries using different approaches before I eventually got it figured out.

Even with a perfect approach, though, I think you would still see some interaction between the adjustments because the supporting structure is never going to be 100% rigid. Every adjustment you make to the top is going to deform the supporting structure a (hopefully) smaller amount, and that will throw all the other adjustments off slightly.

The support structure for my router table is a fairly substantial affair made from particle board, and I designed it with rigidity in mind for the reasons we're discussing here. The top is an inch-thick piece of MDF with plastic laminate (Formica) on the top surface. Given that setup, and the reasonably close initial state that's established by the carpenter's level, I can't imagine that you would need more than two or three iterations of the adjustment procedure to get the top as flat as it was going to get.

Post a Comment