# Triangular kinematics Calibration Problems

I have changed my Maslow to triangular kinematics with a custom 45 deg set of arms

The problem I have is that if a cut a 5 inch test square it is 5 inches vertically and 5.2 horizontally.

I have done the cut 3 horizontal lines across the top and get those into a perfectly straight line that Bar has recommended for calibrations

SO any ideas on how to get my Square -> square…

have 93 firmware and GC

Thanks

Nope! but keep us updated. I am surprised to hear this, I always thought the triangular system was more accurate.

Just a guess, but I’d change how wide your machine width is set to. Change the number one direction (bigger for example) and see how it affects your square. I’m curious to see what you learn as I am also changing to the triangular setup.

You might play around with the triangular Kinematics in the simulator to get some ideas. Distance between the motors does sound like a likely suspect.

I have used the the distance between the motors to do the alignment that @bar suggested as the triangular Kinematics (TK) .

Cut horizontal 3 lines …

In my case 1" long
9 inches down from the top of the spoil board
20" left of centerline
on CL
20" right of CL

the middle line was 1/8 of an inch below the outside ones. If you go into the simulator you can see that varing the distance between the motors causes the middle line to bow up or down. So in my cause I adjusted the distance and after a couple tries I closed that down

But that did nothing for the vertical vs horizontal difference … it really just bows up or down the horizontal line.

I could not find anything in the simulator that spread the vertical lines apart while maintaining the horizontal line spacing.

1 Like

Oh forgot one piece of info… This last test box was pretty close the center of the machine vertically and horizontally.

A good puzzle, this. You’ve got a standard motor set, right? Could you share the cut file? Is the 5" square square or skewed? Does the error scale?

try playing around with the chain length error in the simulator as well (this
would be if you gave the wrong chain to bit distance)

That’s a good idea, the error is larger at the upper or lower extreme of the workarea depending on whether the error os long or short. That would be testable with two more cuts.

I have messed with chain length. could not fine anything that would cause a vertical and horizontal difference. Everything just skewed things

Let me run out to the shop and look at boxes in different places and sizes. Good Idea.

Back in minute

1 Like

Ok here are the latest tests

The gcode is about a simple as I could make it
starting at home
g0 x0 y0 (this line is really not needed)
g0 z-.3 ( using a v bit to be accurate so light cut)
g1 x10
g1 y10
g1 x0
g1 y0
g0 z.35

Home at 17 inches from left side (4x8 spoil board)
Home at 10 from bottom
10 inch square vertical 9 15/16 Horizontal 10

middle of board home
10 inch square V- 9 31/32 H- 10 1/32
square is not skewed

Home at 17 inches from right side (4x8 spoil board)
Home at 17 from top
10 inch square vertical 9 15/16 Horizontal 10 1/16
skewed so that the right side on the square is not 90 degrees with top … it drops 1/8 in 10 inches

these were measured with a steel rule as i did not have a 12" caliper in the wood shop
previous on 5" with caliper

On the simulator, the difference is greater over the vertical range, with a bigger error at the top center than at the bottom center. It looks to me like if the spacing error is positive, the right edge of the square on the right side of the workarea would slope down like you describe.
What version firmware?

1 Like

93 firmware and GC
Ben