Router Box Joint Tuning

[Patrick Baney]

Hi

I've put together a dual router table designed to quickly cut alternating width box joints. The design was inspired by a Cosman video some time last spring. It does what I ask of it very well...but not perfectly. Before I say more, here's the setup:

492138 492139 492140 492141

The backside of the fences have an inlaid t-track holding bolts through the sled, allowing me to slide the fence laterally to set the gap between the cutter and index pin. I set the gap on each side to the cutter diameter of the opposite side. The routers are currently set up with 1/2" and 3/16" bits, both spiral upcut whiteside.

So the issue, on both sides (both routers), the start of the cut is 0.002-0.004 wider than the exit of the cut. Put another way, the sides of the inside of the notch are not parallel. This happens to a greater degree with the 3/16" side than the 1/2". This occurs regardless of cut depth. As low as a .05 cut depth will produce non-parallel inside walls. The pictures below display this:

492142 492143

Here are a few things I've addressed while trying to sort this out:

The index pins are rock solid and not flexing. They interface the workpiece squarely.
The sled has virtually no play.
The bits are sharp and true.
The routers are solidly mounted and square to the table (note the tape shims...miserable process). The MDF being compressed by the bolts has been saturated in thin CA glue.
There's a lot to be said about router & collet runout, but I don't think those would cause my issue.
The fences and table are flat, and square to each other.

Router speed seems to have something to do with it, worsening with some speeds vs others, but I haven't found a magic rpm that eliminates it. The cut in the pictures above appears very clean and is basically tear-out free.

Anyone have an idea of what's going on? Am I asking too much from this setup? Generally, this is an issue that I can live with, as the glue swells the joints enough to fill the gaps, but they have caused some grief on a few occasions.

[Cameron Wood]

Seems like accumulated play/vibration would be at work.

Have you tried running a groove against a stationary fence to try to distinguish if the issue is more the router, or the sled?

Tried with different type of bit?

Reinforcing the lateral setting of the sled?

Add a more solid base & clamp to bench?

[Tom Blank]

Ensure the calipers are square and not on a slight diagonal. They also need to be at the same depth in the cut or bit run out will affect the width measured.

[Lee Schierer]

Since your bit is climb cutting on the way into the work piece, I suspect you are getting some deflection in the bit and the "virtually no play" combined is enough to give you the 0.002" tapered cut you are seeing. It is wood you are working with, so you are certainly close enough.

[James Pallas]

Rigidity of your whole set up would be my guess. Try going extremely slow with your cut. The initial contact is high torque. Flex in the bit, table top, weight of the router motor acting as a pendulum.
Jim

[John TenEyck]

Spiral bits always gave me problems cutting box joints on my router table. Straight bits cut, well, straight and the joints fit together w/o fuss.

John

[Stan Calow]

Yes, I prefer straight bits for this purpose too. Too much torque on spiral bits wanting to twist the workpiece.

[Steve Demuth]

I would add that any router bit, even a spiral one like you're using, is prone to chatter a little as it enters a cut. I am not at all convinced that any wood or composite based guide groove and follower can be tight enough to completely eliminate the deflection or chatter that Lee suggests is your problem. Two thousandths isn't much movement, after all.

I have a very similar setup to yours, except that it's on a cast iron table, and I use an adjustable metal miter gauge bar as the follower. I still get a thousandth or so delta on cuts, but to be honest, as Lee says, this is wood, and a few thousands are not an issue for the joints.

[Thomas McCurnin]

I generally have fair results from a router table box joint, and the whole box joint set up is better done with a table saw and a dado blade. That said, I made a dedicated router table just for 1/4" box joints, and permanently mounted a cheap used DeWalt trim router to the base and made a sled with a 1/4" pin. Works great for drawer boxes and other applications. I like it because I don't have to fool with set up, but pull the router table from storage and I have box joints in a few minutes. No set up.

[Patrick Baney]

Thanks for the input, everyone. Doesn't seem like there's a single quick thing I can look into to sort this out, more so just general shortcomings of the design. Very interested to see that some prefer straight bits to spiral.
Two thousandths is certainly good enough for wood, but that’s really the lower bound that I see. Average is probably 3 or 4, with a max of maybe 5 or 6. Often at least the ends of my boxes are very hard exotics. This kind of play with wood that doesn’t want to flex or swell much has caused a bit of grief for me. Splitting hairs though.


I generally have fair results from a router table box joint, and the whole box joint set up is better done with a table saw and a dado blade. That said, I made a dedicated router table just for 1/4" box joints, and permanently mounted a cheap used DeWalt trim router to the base and made a sled with a 1/4" pin. Works great for drawer boxes and other applications. I like it because I don't have to fool with set up, but pull the router table from storage and I have box joints in a few minutes. No set up.

Yep, for some reason I don't have a lot of patience for fiddling around with a dado stack, or frankly machines in general. I'm a lot more hand tool focused. With this dual router setup, once set, I can cut the joinery on a small box in minutes.


Since your bit is climb cutting on the way into the work piece, I suspect you are getting some deflection in the bit and the "virtually no play" combined is enough to give you the 0.002" tapered cut you are seeing. It is wood you are working with, so you are certainly close enough.

I was being somewhat pedantic when I wrote "virtually" because there's no felt wiggle in the sled, but technically there probably is. Though, to offset this I apply pressure to the same side every time. Same amount of pressure? who knows...


Seems like accumulated play/vibration would be at work.

Have you tried running a groove against a stationary fence to try to distinguish if the issue is more the router, or the sled?

Tried with different type of bit?

Reinforcing the lateral setting of the sled?

Add a more solid base & clamp to bench?
Thanks for the troubleshooting ideas here. If I can bring myself to dig into this more, 1 & 4 is where I’ll start.

[Steve Demuth]

I agree that the table saw with a dado stack is a better cutting implement for box joints. But the setup time is greater, and in my case, given the way I work, I generally want my table saw (I only have one) free when working on pieces that require box joints. On the other hand, I've got jig for my router table that is adjustable for 1/4, 3/8 and 1/2" box joints that takes just a minute - drop in the right bit, adjust the height, set the jig for width, and cut - to set up.

[Thomas McCurnin]

Steve, what kind of jig do you have?

Don't you still have to either swap the pins or widen them, then run test cuts for a tight fit and height?

[Steve Demuth]

Mine is a shop made jig, made from solid core plywood, mdf and uhmw poly. Yes, I have to swap the pins when going from one size to another. I hardened the pin slot with CA glue and then recut them, so they are very stable. I set the height using bridge type depth gauge blocks (and the indexing of the router lift, to back up or down from a standard height, if my stock thickness isn't dead on). I do make test cuts for any given setup, but generally don't need to adjust. If I do, the fact that it's a router table bit, and the lift has index marks accurate to 1/128" makes it one shot adjustment.

ETA: I went out to the shop to look at the jig. Forgot in all this about the zero clearance insert. I do have a slot for a replaceable backer behind the router bit. So if I'm changing setups, that would also typically be replaced. I might have been guilty of exaggerating how quickly I can switch over - but it's certainly quicker than set up on dado stack.

Box joint jigs don't have to be fancy (I've seen some builds of jigs on youtube that boggle the mind in complexity and adjustability), but as you obviously know, good results do require attention to accuracy and stability. Like a miter sled or jig, any variance from dead-on is fatal to good fit and finish.

[Patrick Baney]

UPDATE: Thanks again for the insight, everyone.

Switching to straight bits solved my issue. Quite surprising as I thought spiral bits were superior in every way. My cuts end up +.002 over diameter due to runout, but this is easily adjusted for by the jig.

[Jim Becker]

Patrick, in a sense, spirals do offer a lot of advantages, but not to every kind of cut because of the dynamics of how chips are removed and how material is physically moved by the cutting action, etc. The former can affect quality of cut and the latter affects workpiece holding on a CNC for sure, but those things can apply on a router table or hand-held, too.) There are times, as you've noted that a straight cutter (or one with a minor shear but still straight cutting) will do a better job. 'Glad it worked out for you!