Accurately laminating curved pieces

[Bob Daniel]

Great Googly Moogly, you guys are awesome! Since 11 a.m. is, in a sense, my 5 a.m., I won't say much more right now besides THANKS!! Lots of detail to go on here. It started out as a hypothetical question, pretty much, but now I feel like I should go get the materials and get building!

Of particular interest was the idea of the offset caused by the kerf... never thought of that, that could've spelled trouble... I've be sure to leave lots of wiggle room...

[Bob Daniel]

Shawn, that's interesting. As I understand it (I think) then you're using the SAME edge shape to route BOTH the inside and outside of any shape, so no issues with making perfectly mating templates, there's just the one... this is fundamentally different than the other approaches discussed, I think.

This would be a handheld router situation, as you've drawn it, yes?

Not sure this hacked powerpoint pic makes sense:

Make a template of the edge (t) and screw it to a wide support (base) with a support piece (s) the same width as (t) about 1" from the edge of a. (s)'s shape is not critical as long is it basically follows (t) but allows bit clearance.

Using a flush trimming bit (r), mount the one piece of yr top (a) to the jig, and route its edge.

Then mount the mating piece (b) onto (s) and route it flush - making sure to keep the bearing riding along (t).

It then won't matter what shape or symmetry or smoothness (t) has, you'll create 2 perfectly mated pieces.

[mreza Salav]

Not sure this hacked powerpoint pic makes sense:

Make a template of the edge (t) and screw it to a wide support (base) with a support piece (s) the same width as (t) about 1" from the edge of (t). (s)'s shape is not critical as long is it basically follows (t) but allows bit clearance.

Using a flush trimming bit (r), mount the one piece of yr top (a) to the jig, and route its edge.

Then mount the mating piece (b) onto (s) and route it flush - making sure to keep the bearing riding along (t).

It then won't matter what shape or symmetry or smoothness (t) has, you'll create 2 perfectly mated pieces.

I am not sure I understand your steps correctly or not, Are you saying (a) and (b) will be the mating pieces? It seems to me with step one, you get a "copy" of edge of piece (t) on piece (a). Fine. But I cannot understand the rest.

One thing that I think is critical is to account for the thickness of the bit when making matching pairs otherwise the curves will be different.
Here is how I do/did it:

Suppose I have a piece "A" with a curve on the edge and I want to make a matching pair for it.

1) Since I don't want to change "A", first make an exact copy of "A" on another piece using a flush trim bit. Call this "B". so "B" is identical to "A".

2) Use a rabeting bit, and cut a rabit of, say 3/8" into the curve of B.
Then using a flush trim bit complete the rabet on B; so now the curve on B
is 3/8" offset from the original curve.

3) Take a wide piece of MDF (to make a pattern out of) and a flush trim bit which is 3/8" (which is the same amount of off-set created on B). You can replace all these 3/8" with 1/2" if that's all you have, or 1/4". Now follow the new curve on B to cut into the MDF; you get two curved pieces on MDF; one is identical to B and the opposite is the matching curve of A.

4) Use the MDF pattern and a flush trim bit to transfer it to whatever piece you want to be the matching pair of A.


The purpose of step 1 is to account for the thickness of the bit used in step 3.

I hope it makes sense now.

[Prashun Patel]

(a) and (b) are mating pieces.
By routing on the other side of the bit, my (b) will be the mirror image of (a). (b) needs to be clamped down, and you have to route in the opposite direction as with (a) else you'll be climbing the cut.

If you already have (a) cut, then you can use that as the template for flush trimming (b). You just need a support piece under (b) that's at least as thick as (a).

Yeah, this'd be only for a handheld router. I was thinking tabletop, where the pieces might be too large for an r-table anyway.

[Bob Daniel]

One thing that I think is critical is to account for the thickness of the bit when making matching pairs otherwise the curves will be different.

Wouldn't that only apply where the curves are tighter than the radius of the bit? Otherwise, shouldn't it just follow the curve? Isn't that the pattern bit's whole point, to match the pattern? Getting confused again....

[Derek Hansen]

Not sure this hacked powerpoint pic makes sense:

Make a template of the edge (t) and screw it to a wide support (base) with a support piece (s) the same width as (t) about 1" from the edge of (t). (s)'s shape is not critical as long is it basically follows (t) but allows bit clearance.

Using a flush trimming bit (r), mount the one piece of yr top (a) to the jig, and route its edge.

Then mount the mating piece (b) onto (s) and route it flush - making sure to keep the bearing riding along (t).

It then won't matter what shape or symmetry or smoothness (t) has, you'll create 2 perfectly mated pieces.

YES! Thanks Shawn!

Edit* Aww nuts Mreza is right.

[mreza Salav]

(a) and (b) are mating pieces.
By routing on the other side of the bit, my (b) will be the mirror image of (a). (b) needs to be clamped down, and you have to route in the opposite direction as with (a) else you'll be climbing the cut.

If you already have (a) cut, then you can use that as the template for flush trimming (b). You just need a support piece under (b) that's at least as thick as (a).

Yeah, this'd be only for a handheld router. I was thinking tabletop, where the pieces might be too large for an r-table anyway.

I am sorry but I don't think you'll get perfectly mating pieces this way.
You simply are not counting for the thickness of the router bit.
The radius of the curve on (a) and (b) will be different by exactly the thickness of your router bit. They won't fit perfectly.

If you think carefully about it you'll see what I mean.

[Bob Daniel]

I am sorry but I don't think you'll get perfectly mating pieces this way.
You simply are not counting for the thickness of the router bit.

I'm not sure I understand why. Seems to me you could use a two-inch radius router bit (if there were such a thing) and still the left and right pieces would match because they're cut using the same template. As long as any inside corners aren't smaller than the bit radius, then the bit radius is not even in play. Just like carving out a bowl with a bowl bit, the diameter of the bit is only potentially a factor in the corners, and only if the bit radius is bigger than the desired corner radius. In that example, actually, the bit radius will BE the corner radius, like it or not...

[mreza Salav]

I'm not sure I understand why. Seems to me you could use a two-inch radius router bit (if there were such a thing) and still the left and right pieces would match because they're cut using the same template. As long as any inside corners aren't smaller than the bit radius, then the bit radius is not even in play. Just like carving out a bowl with a bowl bit, the diameter of the bit is only potentially a factor in the corners, and only if the bit radius is bigger than the desired corner radius. In that example, actually, the bit radius will BE the corner radius, like it or not...

Imagine the piece (a) is a half disk with radius 2". If you use a 3/4" thick bit then the mating piece (b) will be a half circle with radius 2+3/4" which is clearly not matching (a).

[Prashun Patel]

I think you are right, Mreza. It's easiest to see yr point if the template were a perfect circle. (a) would have a diameter equal to (t), but (b) would have a diameter (a) + (r).

To correct for this, then I ask you, can you do it like inlay template kits?

First, you route (a) with the template on top, using a bushing with an offset equal to the radius of the router bit.

Then for (b), you remove the bushing, and route it like I said below.

This seems to work for inlays. Would it work here?

[mreza Salav]

I think you are right, Mreza. It's easiest to see yr point if the template were a perfect circle. (a) would have a diameter equal to (t), but (b) would have a diameter (a) + (r).

To correct for this, then I ask you, can you do it like inlay template kits?

First, you route (a) with the template on top, using a bushing with an offset equal to the radius of the router bit.

Then for (b), you remove the bushing, and route it like I said below.

This seems to work for inlays. Would it work here?

It would work, but the mating pieces you get, none is the copy of the original piece you started with (i.e. (t)).
What I said was, if you have a curved piece (t) and you want to find a mating piece for it then you have to following something like the steps I mentioned.
What you said (using router templates) gives you two mating piece none of which is a copy of the one you started from (t).
So if you had a 2" disk to start with you'll get two mating pieces each with radius 2+3/4', say.

[Van Huskey]

My way will NOT work correctly. It was very late and I saw some issues but couldn't completely think them through. The kerf is the killer. You need that "Atom Splitter" molecular kerf blade which I need to get around to inventing. Sorry for all the misinfo.

[David DeCristoforo]

Sorry...somehow I thought you were stacking pieces, not mating them. Templates and guide collars are the way to go for this. But here's another "trick" that's a bit "down and dirty". If you stack a couple of boards of contrasting colors and stick them together with some sort of temporary bond (hot melt, two sided tape, etc.) you can bandsaw a mild curve, separate the layers and the pieces will match closely enough to get a tight glue joint. You have to make the cut in one smooth motion so as not to alter the kerf. And if you are trying to follow a true radius, it can be difficult. Also, the tighter the curve, the more difficult it is to get the pieces to glue up cleanly. So like I said, a mild curve is best for this "technique".

[Prashun Patel]

What you said (using router templates) gives you two mating piece none of which is a copy of the one you started from (t).

Understood. I guess one could cut (t) a little small, or for the application, being off from the template by this much just might not matter.

[mreza Salav]

Aside from the method I posted in post #18 (which works; I have tried) the "other" trick David suggested works to some extend too.
To get an even better joint (when the two pieces are stacked up) I would use a scroll saw. The thin blades on scroll saw are much more forgiving than the blade on a bandsaw and the joint's will be a lot smoother too.