Horizontal Accuracy Drops off out From Center

I have been trying to dial in the calibration of my machine by running a few tests and I am struggling with the horizontal accuracy away from the center.

I cut a series of 5.5 x 5.5 inch squares in different locations on the frame. Below is the height, width, and deviation (height - width) in various locations all measurements in inches:

Bottom center: 5.460, 5.480
Top Left:      5.512, 5.436
Bottom Left:   5.485, 5.325
Bottom Right:  5.450, 5.400

The sides on the bottom seem to be the worst. The width of the bottom left is 3/16 off. I didn’t cut a square in the top center, but I did all of my calibrations there so I would expect the accuracy to be similar or better than the bottom center.

Anyone have any ideas? Using the simulator, I don’t see an error that would show up so prominently at the sides particularly at the bottom sides, yet not appear at all in the center.

I didn’t see the sled catching or sticking at all on the bottom sides, but I suppose anything is possible.

I also suppose, that this could be highlighting an error in my balanced linkage maybe one of the setups isn’t centered on the bit just right?

OK, I think I have an idea of the issue.

I think what I am seeing is the result of the router being driven by the rotating bit. I think the bit rotates in a clockwise direction. I am doing conventional milling, so the left side of the square is being cut on a downward path. If I understand the forces correctly the router is being pushed to the right as it falls.

If you look at this photo of the part, it seems to strongly suggest that this is occurring.

The right hand side is mostly straight.

Tomorrow I will try climb milling to see if that helps. I suppose I could also decrease the pass depth and feed rate (i was using about .18 and 10in/min) which might help a bit too. Finally, I could move the arms farther out, how far have others moved their’s and still seen decent results in the top center?

Any other suggestions are welcome.

try adding weight to the sled, or tilting the machine closer to vertical.

the chain tension is low in the bottom corners, so you will run into more
problems there.

you can also try lowering the feed rate, if the problem is that the bit is
pushing the machine out of position, letting it stay in the general area so that
it falls into the area it’s cut to cut more (until the chains are tight) should
show a wider cut than normal, but the router should cut the right place (towards
the outside)

and if you are using the linkage instead of the stock chain setup, make sure you
have the kinematics set to triangular.

someone should also double check that the triangular kinematics accounts for the
wrap of the chain around the sprockets properly, I probably won’t get a chance
to for a couple days.

Ooooo, good theory! I would also check to see that the router bit is not free to “wiggle” left and right in the router. I know that I see this type of behavior when I have the clasp on my router loose so that the router can wiggle in the sleeve.