What am I missing about box joints?

[Don Mikulec]

I'm having problems cutting box joints in plywood panels about 14" wide. I'm using a dado head in a table saw with an home-made, adjustable, box joint-cutting jig. The jig has double glides that fit the miter gauge slots in the table. There is a sliding fence holding the pin. This fence is held by bolts to a fixed fence that rides on the sled. The hard maple pin is the exact width as the groove cut by the dado head -- 1/2". I set the jig to cut tongues about 1/128" undersized, as measured with a caliper. Depth of cut is 3/4". I've made numerous trial cuts in 3/4" x 8" pine to get the adjustment correct.

I just cut my plywood pieces, paired and offset by one finger width to start. The individual grooves and tongues measure correctly but the joint won't go together. After about 7" the cuts begin to grow out of alignment and they are off by about 1/32" at 14". What I don't get is how this can happen since the panels were paired when cut and are indexing off the same pin. I've checked the jig and can't find any major issues (e.g., a misaligned pin). Errors between the panels should not agggregate with the procedure I'm using. I'd appreciate any comments on where to look for the cause of the problem.

Don

[Myk Rian]

That 1/128" is getting you bound up.

[Roger Rayburn]

Yup. 1/128 inch x 14 cuts = don't fit anymore. Called tolerance stack up. the pins and tongues have to be exactly the same. Can get away with it over a short distance but it eventually matters.

[Alan Schwabacher]

When you say they are "paired", you mean they are clamped together so they can't shift, not just that they are cut at the same time, right?

[Richard Coers]

How clean is the plywood cutting? If it's fuzzy, you may be registering on the "fuzz" and throwing off the cut. Anything 14" wide is tough to do on a table saw. Too easy to wiggle around. I'd switch to another joinery method if I was you.

[Brian Brightwell]

To adjust for too tight a fit move the maple pin toward the blade. Move it away from the blade if too loose. Or maybe the other way around but you get the idea.
The block that holds my pin captive on my jig has an oversize holes so I loosen the bolts and tap it very very little.
http://i773.photobucket.com/albums/yy11/bebrightwell/DSC_0092.jpg
I used key stock for the indexing pin.
http://i773.photobucket.com/albums/yy11/bebrightwell/DSC_0096.jpg

[Don Mikulec]

Yup. 1/128 inch x 14 cuts = don't fit anymore. Called tolerance stack up. the pins and tongues have to be exactly the same. Can get away with it over a short distance but it eventually matters.

That is at the center of the issue - stack up. There is an error someplace that is additive with each cut. What I don't follow from your comment, though, is how one could ever cut a fingers that are undersized to the slot. If the fingers and slots have to be exactly the same, the fit will be too tight to ever allow glue in the joint. As you point out, if undercutting the width of the pins is the issue, one could never construct a box joint longer with more than 4 or 5 fingers. Do I have your point correctly? Has this been your experience, also?

[Don Mikulec]

When you say they are "paired", you mean they are clamped together so they can't shift, not just that they are cut at the same time, right?

Thanks for your comment, Alan. The panels were not clamped together but they were both registered to the pin: cut a slot, pick up the boards, move them over, and register the newly cut slots with the pin. The fit to the pin was tight.

[Don Mikulec]

Thanks, Brian. It's not that the fit is too tight; it's that there is an error that is stacking up and causing the cuts on one panel to shift relative to the other.

[Don Mikulec]

Cuts are clean. Dado blades are basically new. The only wear on them has been in setting up for these cuts.

[Rick Moyer]

That's why after two different attempts at box-joint jigs I decided on purchasing the I-box from Incra, developed by Sawmill Creek's own Alan Schaffter. While my homemade jigs were "servicable" I had to be very precise on everything associated with making the joints. A few thousandths make a great difference. The I-box is easier to set up and adjust as needed.

[Alan Schwabacher]

I think if you clamp the boards together, that may solve your problem: without clamping, errors can still accumulate. With boards paired and clamped, I've had no problem making 11" runs of 1/4" box joints using a similar simple jig and tight tolerances.

[Don Mikulec]

Thanks for the suggestion, Alan. I'll try that with the next cuts. I'm thinking that there is something else amiss with the jig, however. I drew out on paper a representation of what should be happening. I chose 15/16" pins to fit in 1" slots. Over five slots if the 1/16" difference were the source of the problem, the "error" would add to 5/16" and be quite visible. First I drew a representation of the two boards with the slots cut simultaneously and aligned. Then, I cut apart the paper and moved the top pins to mesh with the bottom slots.
252523 252522
I think this exercise shows that the undersizing of the pins is not the issue but correct me if I am wrong. I'm thinking that the only way for the slots to progressively drift out of alignment is for there to be a misalignment when the boards are cut; i.e., somehow the top board is not being cut the same way as the bottom board. Consistently. Either the fence is not square to the miter gauge slots or the pin is deflecting. Do you have any insights?

[Don Mikulec]

Myk, thanks for your response. See my latest response to Alan Schwabacher below.

[Brian Brightwell]

I made three jigs 1/4, 3/8 and 1/2. After dialing them in I have change nothing on them and they make perfect joints. I am not bragging, I am surprised. Do you use a router or table saw?

[Alan Schaffter]

I'm a little late to this parade. Please don't beat me up, but you should have purchased an I-BOX!

I-BOX aside for a minute, lets talk box joints.

First, I believe you have a construction or setup issue- it is possible to get accumulated error if your dado notch is wider than the index pin and you consistently slide or push the board so one side of the notch is against the index pin. If you change the side it rests against you will have erratic spacing and the joint won't fit either. The index pin MUST be the same size as your dado blade- not too loose or too tight. Typical errors - folks use calipers to measure the dado and the pin and introduce measuring errors or they cut the pin a tad too wide then widen the fence notch until the pin fits. That is backasswards. Use the dado blade to cut the notch then trim the entire length of the pin until it fits perfectly. Another issue is not restacking the dado precisely the same as when you made the jig, each time you use it, or you forget a shim, over-torque the arbor nut, leave dust between the blades, allow blade tips to overlap, etc., etc. Poor fitting joints can also be caused by a flexible index pin or loose bar in the miter slot.

Now as to finger size, that is controlled by the spacing between the index pin and the dado blade. While it is hard to do with a basic box joint jig it is possible to change finger size a couple of thousandths and still have evenly spaced fingers in the joint. In fact this a routine operation with the I-BOX (I had to add that :) ). As long as the index pin and dado blade (notch created by the dado blade) are EXACTLY the same width, you can change the spacing, usually by only .002" - .004" (yes 2 - 4 thousandths of an inch!!!!) for a tighter or looser fitting joint. Any more than about .004 and the joint will be too loose. The I-BOX has a micro-adjuster that allows you to adjust the spacing in .001 increments. With the I-BOX you can also completely de-tune the fit of a joint and make it crazy loose- that will leave gaps between each pair of fingers which you can fill with decorative inserts made from contrasting colored wood or (soft) metal.

So, yes you can make the fingers smaller and the joint will still "fit" but there will be gaps if the change is greater than a few thousandths of an inch, but there should be no cumulative error. See photos of various splined joints below:

Joint at upper right was made with 1/32" thick maple decorative splines. The I-BOX was set for a standard joint, then I dialed in 1/32" (one rotation) of micro adjustment (to loosen the joint):

http://www.ncwoodworker.net/pp/data/2404/medium/Pre-prod-jig-4.jpg

The top joint has 3/8" maple splines, lower one 1/4" thick walnut splines.

http://www.ncwoodworker.net/pp/data/2404/medium/Pre-prod-jig-5.jpg

In-progress splined joint (demo for the Klingspore show):

http://www.ncwoodworker.net/pp/data/2578/medium/IMG_43181.JPG

Another in-progress demo splined joint. (This is a special "Center-keyed box joint", hence the wider center finger/notch). The joint AND an extra set of walnut fingers used for the splines, were cut one after the other on the I-BOX using the same settings. The splines will be trimmed and sanded after the glue-up:

http://www.ncwoodworker.net/pp/data/2578/medium/IMG_43211.JPG

One of my favorite samples- Walnut with 1/8" aluminum splines. Easy to do! The I-BOX was set for a standard joint then I dialed in 1/8" (4 rotations) of micro for the splines which were cut from 1/8" thick aluminum bar stock. After the glue had dried I used a file and belt sander to trim the aluminum flush:

http://www.ncwoodworker.net/pp/data/2578/medium/IMG_4315.JPG

The I-BOX main setting knob (red) and micro adjust knob (silver). Each tic mark represents .001" of micro adjustment:

http://www.ncwoodworker.net/pp/data/2404/IBOX-knob.jpg

[Don Mikulec]

Thanks, Brian. I'm using a table saw. Did you use plywood or solid wood for the fence? I'm suspicious that my key is deflecting in the plywood fence I made.

[Don Mikulec]

Alan, thanks for your reply and confirming that the sizing of the pin is not the issue with stack up. Thanks for all the photo's that back up the words.

[Don Mikulec]

Thanks everyone for your inputs. I rebuilt my jig to eliminate play between the guide rails of the sled and the miter gauge slots in the saw table and to square the pin with the fence. First trial cut produced a perfect fit. in other words, errors were caused by misalignment in the jig such that fingers on one work piece were cut to a different width than those on the mating piece.

Don