How to deal with segmented circles and arcs


I am curious about what other users do about the segmented circles and arcs? I find this really inconvenient - I tend to forget to set the segment count to a high number, consequently intersections are not accurate, and eventually Sketchup loses the circle entity information. This is the biggest reason preventing me from switching to Sketchup for all my drafting.

Incidentally, if anyone is interested, I figured the number of segments n required to limit the error f (maximum offset between true circle and segmented circle) is given by n = pi x sqrt (r/(2f)). For a maximum error of 1/16 in. this comes out to 2pi sqrt D, D in [in].


this equartion was posted several years ago
N=pi()/ acos [(R-e)/R] This is based on the max error from the chord.
Note corrected typo


Interesting as I went to 360 for number of segments to get a better circle and wondered about the number ! ! Even went to 720 and flats got bigger . . At a 3.900 diameter circle using 360 segments gives me about a 3/16 flat area . . Why does it do that ? . . Actual measurement is 3.860 inches in diameter using the formula it says 93.5694 segments weird number


Using what formula? How did you determine the diamter? Why do you even need to calculate the number of segments in the circle? Just select it and look at Entity Info.

There is a limit to the number of segments you can have based on the radius of the circle. For a small radius such as what you are referring to 720 segments, even 360 segments is obscene overkill. You’ll just create problems for yourself trying to use those sorts of numbers. You need to consider the resolution of your 3D printer when selecting the number of segments.


The sketches attached show an example: Its a 4" dia circle with 8 and 16 segments respectively, compared to a circle with 198 segments (which is the maximum Sketchup will accept at that size). The equation can be rearranged to get an error estimate:
f = pi^2 R/(2n^2). For 8 segments this predicts a maximum error of 0.154" vs 0.152" measured. For 16 segments the predicted error is 0.0385" vs 0.0382" measured. The slight difference between predicted and measured error is due to an approximation to a cos term that appears in the exact solution. The approximation is the better the smaller the angle.


There’s a thread about this with some more information:


Okay will see how this goes with next round part I make . . down loaded the add on . .