wierd arcs

[Programinator]

Today i got a file step file from a customer and programed the part with x2 mr2 and ran it in the machine.

 

Inspected the part and realized that the outside dia was out of round by like .02

 

i found out it had to be the arc in the geometry because i redrew the arcs and it worked fine...

 

the wierd thing is that when i analyze the arcs before i changed it they were fine.

 

everything looked good in verify...

 

anyone see this before???

[OTOMMYO]

sounds like you had bad geometry the 1st time

[Josh T™]

Did you create the wireframe in MC?

[Surface]

I'm curious how your post outputs arcs. Do you use R or I,J. Do you break arcs?

[Programinator]

the wire frame was already there. but the thing is when you analyze it there is nothing wrong its a perfect circle. well it is 2 arcs together to create a circle. 2 half circles.. if you know what i mean. but they both have the same center and they are both the same arc size. so it should be ok. you would think.

 

the only thing different is when i delete the 2 arcs and make new geo it is a full circle. but that should not matter should it?

[crazy^millman]

Well that could be your problem when it posted the code. I would like to see the code that is posted. If your post does not break up quadrants and posts it as a 180 degree arc and then posted the other one as an 180 degree arc it could have been a problem somewhere that the machine had a problem with the R if you used it or the IJK if the intersection were off even by only .0001. I have seen this cause weird things on machines when there is just the smallest miss match on intersections.

 

HTH

[Surface]

Ron, that was my thought. It's rare that it happens, but a tiny rounding error caused a value of a 180deg arc to mismatch .0001

 

Been there/done that.

[crazy^millman]

Yeah I have had my fights with that over the years. I always break in quadrants just because of that. It makes for longer code but less problems in the long run.

[Programinator]

my output is in R not I's and j's. and how do i go about seting if to be in quadrants.

 

Monday im going to output code with the old arcs and then do it with the new arc and see what the diff is in the code.

 

if there was a setting that would do away with that problem it would be great.

 

I almost didnt check the o/d on the part just assuming that it would be good. cuz mcam ussually puts good code out.

 

got lucky that i cought it. but now im puzzled as to why it did that.

 

must find out

[Programinator]

if i closed all arcs and then did findoverlap chook do you think that could help

[Surface]

Shawn, I don't think it would. I don't think there was anything wrong with the geometry. It's my guess that the problem was caused by a rounding error, then multiplied by the manor it was posted.

Try this. Create a half circle, one inch diameter. Now create a two-point circle 1.0001 diameter and choose the two endpoints of the first circle.

The centerpoint will be significantly different.

 

It's a rare occurrence, but perhaps the circle was 1.00005 and MC rounded half of the circle up and the other half down.

[Programinator]

Surface,

 

Would that really make it .02 out of round though. I can see how that would maybe make it slightly out of round but not .02

 

I will try it and see what result i get

 

Thanks

[SIG]

The exact thing happened to me 3 weeks ago.

The cause was my arc geo was not broken at 180deg and I was posting w/ R value.

SIG