EXTREME TOLERANCES REQUIRED
Prototyping can be a very valuable tool.
SHOP OWNER: Steve Strickland
LOCATION: Quitman, TX
The Six Board Burr puzzle-making project was a challenging one that demonstrates the need for testing, planning, and prototyping.
Frans de Vreugd, an architect in The Netherlands, discovered the central idea and Bill Cutler, a programmer in Illinois, did an exhaustive computer analysis. The idea is to have six boards intersect each other with a series of sliding moves in a certain sequence in order to free a piece. False moves are desirable for increasing the degree of difficulty in solving the puzzle so this was one of the selection criteria.
Another selection criteria was the ability to fabricate the pieces so only pieces that are "notchable" were used in the analysis. Pieces can also be "millable" but this is a more expensive technique. We wanted to just use a dado blade to make the notches.
I don't know how many potential designs were found (presumably thousands of them) but the 4 best were selected and built. I made 200 of this one, called Chequered Board Burr, in the summer of 2002 and it was shown for the first time in Antwerp in August.
We planned on splitting each board into two sticks, cutting the notches, and then gluing the two sticks into the final board shape. This turned out to be a good idea. However, the precision requirements appeared quite challenging. After all, we'd have to meet the tolerance envelope using four surfaces and a glue joint. This technique allowed us to use two species of wood to create a symmetrical color pattern.
I built a couple of prototypes and we played around with them. The tolerances were critical. The inside and outside dimensions had to be virtually perfect or the puzzle would either fall apart or lock up tight. The tolerance envelope of the stick size was only 0.002" and the tolerance envelope of the notches was 0.003". This was indeed, very tight!
We also discovered that the small glue joints broke easily so, at my suggestion, we adopted the technique of using splines. The splines made the joints quite strong but also added some additional challenges. The pieces were too critical to sand down so the splines had to be made perfectly. There was no way I could sand them down without degrading the geometry achieved on the table saw. The grooves for the splines also had to be perfectly located for the two sticks to join within tolerance.
The sticks were cut on a table saw 0.006" oversized and then double planed on all four sides removing 0.003" on each side. When planing a stick of only an 8" length, the trailing edge is about 0.001" larger than the leading edge. By planing from both ends, I ended up with a stick that was about 0.0005" thicker in the middle. This was well within tolerance. In addition, we used very sharp planer knives so that there was no need for sanding. I found that one of the original species of wood, Honduras Rosewood, wouldn't plane cleanly so I changed the wood selection to Black Walnut to meet this criteria. There was no planing problem with the Hard Maple either.
The wood was ordered and we brought it into my centrally heated and air conditioned shop and allowed it to fully achieve equilibrium with the controlled environment. I could not tolerate any wood movement so only the best quality wood was acceptable. Judith Mattart supplied the wood and it turned out to be of excellent quality. I didn't have to fight the wood.
I had to make a few revisions to my jigs. Everything had to be perfect for the 2,400 sticks (plus plenty of extras) to start off right. After some very careful micro tweaking, we achieved the desired precision. The jig held up very nicely and the tolerance envelope was only about 0.0015" variation in the entire batch. The planing went quickly and smoothly, the sticks were in great shape ready for notching.
The notching jig was a major challenge. I just could not cut a notch that large in one pass because the stresses were too great causing the dimensions to fall outside of tolerance. I decided to try cutting the notches in four passes and this test worked out pretty good after extensive revisions to my clamping methods. I resorted to using double clamps to hold the sticks firmly in place against the forces of the dado blade. The stresses were now flexing my side guide so I had to redesign the locking mechanism for that, too. I had planned to notch the sticks 24 at a time but reduced this to six.
Once I had a notching jig that would do the job, I made four of them, one for each notch position. In this manner, I could make a cut, change jigs, and make another cut until a finished prototype was available for checking the notched settings. We tweaked the settings until we achieved a good fit. As all the sticks were being notched, we checked every third stick with a dial caliper. This part went slowly but the results were very good.
After notching the sticks, they were glued together. I designed and built some positioning jigs for this, as there was no room for error in positioning the two sticks. We had to meet the tolerance envelope with this as well.
Next, I cut the grooves for the splines using a simple jig with adequate clamping. I did this with a full-kerf saw blade and had no problems with it.
Next, I had to cut the 2,400 splines. These had to fit the groove just right in thickness and had to be the perfect length. The width was set just a little undersized. The challenges of making very small parts like this to the tolerances required were definitely a challenge. A 1/8" x 1/4" stick is a hard little devil to clamp while cutting it on a table saw! I finally designed and built a jig that did the job. We cut and glued in the splines. About 15% of the splines were rejected so there was still some excessive variation. To deal with this, I made plenty of extras.
When we assembled the finished pieces, we discovered that in spite of all our precautions and careful measuring, we still had to do some "piece matching". Some were a tad too tight and some a tad too loose so we had to match things up for the best batch of puzzles. A few extreme samples were hand fitted with jeweler's files.
The puzzles have a couple of minor problems to solve. For instance, the rake angle on the dado blades leaves grooves in the bottom of the notches. In addition, using a regular saw blade to cut the groove for the splines, yields a notch with a "W" shaped bottom. The prototypes allowed us to show these defects to the client in advance and get them approved before production began. The job could possibly have been rejected without prior approval.
This project could not have been done successfully without the prototypes and revised drawings, paying close attention to every detail from wood species to the environment to micro-flexure in the jigs. This is one example where prototyping and testing of technique was essential. From the prototypes, we determined our tolerance envelopes at each stage, estimated our spoilage, estimated our labor requirements, and identified areas where additional jig development was required. We also learned all this before we accepted the job and ordered the wood. For some projects, this type of activity needs to be included in the quotes and planned for in your schedule. Without the prototypes, we'd have ended up with puzzles that didn't fit, broke apart and contained unapproved cosmetic flaws. With the prototypes, we ended up with puzzles that fit nicely, are very strong and a happy customer who is ordering more jobs. When a woodworker is taking on a new challenge, prototyping can be a very valuable tool.. . . Steve Strickland
CLICK HERE TO RETURN TO THE INDEX!
SEND US YOUR "SHOP SHOTS"
This is the place to share views of your shop, woodworking tips and methods,
mug shots, special tools or machines, finished work--you name it!
We prefer digital images via e-mail, but prints or transparencies will do. Include your name, address, phone number and a paragraph or two explaining the photo(s). Not every entry will be used, we reserve the right to edit for length and clarity, and we will not return photos.
P.O. Box 493
Springtown, PA 18081