Circle Accuracy

[Chuck]

I have a Spar that has a ever so slight curve or radius. Its about 20 inches long. I would imagine the diameter to be really huge. I'm trying to get a accurate diameter on it. I'm doing a linear open scan on it and then I'm trying to constuct BF circle from the scan and I'm getting some really bizzare results out of this world. If I measure it normally by taking hit points I get bad results. Is there any way to get a accurate diameter on it? I don't have CAD. Running Ver. 3.2063 on a Validator.
Thanks.

[Andrew Gardner]

Good luck - If you know the theoretical center point, then you could do a polar radius. Are you in aircraft coordinates? If so, about all you can do, if you have surface data, is to dimension point data ( T value).

[cmmguy]

A template might be your only option unless you have some fitting software.

With fitting software, you can take all of the data points and fit them to a theoretical radius of the correct size. It is sort of what Andrew was saying except without the theoretical center point it is very difficult.

[Paul Barriage]

Try using circles around the spar, as opposed to the scan. 3, 4, 5,....20

Then change working plane to view this arc. Create a circle form the centers of the circles. BTW, when CAD only shows me the two end circles, I use thickness to get different dimension points on a rod. A cylinder scan may not give the definition necessary to create the arc.

[bob mappes]

it sounds like your part has a very short section of the diameter......
that being the case, you don't have much to work with.
I have been told that you need a minimum of 90 degrees of arc to get any accuracy at all.
Less than that and the software "sees" a line not a radius

good luck

[bboyd]

I have been told that you need a minimum of 90 degrees of arc to get any accuracy at all.
Less than that and the software "sees" a line not a radius

good luck

I have tried this a number of times on some different machines, and have found that you need at a min 30-40% of the dia to get anything remotely close (+/- .010). The way discussed earlier, working from the center and dimensioning each point individually is more accurate.