The outer ring is made up of 4 blanks. The first one is shown below.
The blank is a rectangle in cross section with a trough cut down its length. The blank is cut oversize in width and thickness to allow me to “sneak up” on the final size.
The first challenge is getting a trough cut to the correct radius. I used a piece of fir as a prototype. I put saw kerfs on each side of the blank to hold the piece in my CNC.
I then cut the trough pattern.
It took several attempts to get the radius to match the core.
After I was happy with the radius adjusted the thickness to get my best guess for a piece that would cover one quarter of the circumference. Then I cut it to width and cut the length to give me 4 pieces. Then I did a test fit.
I came out pretty close. I’ll be able to dial it in better when I have 4 pieces to work with.
Next, I made four maple blanks. I again built them up from strips with alternating grain direction. Since I was going to remove material for the trough I left that part open.
With the blanks glued up I could start the CNC work.
This is where my unexpected challenges showed up. I had used the radius from my prototyping work. I took my first piece off the CNC and checked the fit. On the first end I had a near perfect match.
However, the other end was a different story.
This was a surprise and very confusing. There was no way my core radius varied that much. After several measurements I determined that the depth of the cut was consistent but the width was not. The left end of the piece was accurate but the right end was about 0.060” narrower.
I wondered if the piece was shifting. I set the piece back on the CNC and did a deeper cut after anchoring the piece more securely.
No change.
I then wondered if it was tied to cutting at the front of the CNC bed. I moved the piece to the center of the bed and made a third cut. I was running out of spare material at this point.
No change. The mystery was not going to get solved for this project.
I ended up cutting the other 3 blanks with the same error and sanded them to fit the core.
That took about 20 minutes for each piece. But at least I had 4 blanks with an accurate radius.
The CNC mystery is not yet solved. A future project has showed that the machine is not cutting accurate circles. That error is likely connected to the error I saw here. Internet research suggests that I need to adjust the pinion gear.
Not having a CNC, I figured what would I do to get that trough. If it was the correct diameter, a cove cut with the matching tablesaw blade would work, but what if it was not a common diameter?. Skew the cove cut, but then it wouldn't be circular (ellipsoid instead). Your last photo was the money shot. Get it close, then use the main part as a sanding block. Brilliant!
Without a CNC I would have set up a series of passes through the router table and then done the sanding thing. Given the issues I had, that might have been faster...
What size bit did you use on the CNC? Even 1/4" can deflect when cutting too fast or with unbalanced stress (as would happen when cutting the walls of the cove) or changes in wood density/grain direction. Nice recovery with the sanding solution, though.
I used a 1/2 cove bit, but I had the speed set slow - 20 inches/minute I think. The real tell came from when I tried to cut a circle. The top and bottom (in the Y-axis) have flats on them. Searching for this defect on line leads to suggestions of a loose pinion gear. The other bit of data I have is that I could measure very accurate positioning in Y when I did large or small moves. However, when I did a series of 10 small moves the right side lagged by ~0.060".
It's a ShopBot. I think it's rack and pinion. At least that's what the service videos make it look like. It's going to take a few weeks for me to get time to dive into it.
Ross - Since I bought it used that means it has been moved at least 4 times. I won't be surprised if the drive got shifted a bit.
