Constructing An Arc From Part Of A Scan

[PlasmaGuy]

It's no fun searching for information on constructing an arc when the search function limits you to a minimum 4 characters (searching for '*arc*' seemed to work though). But I digress...

What I am trying to do is construct a best fit circle based only on part of the data from a linear closed scan of a circle. I have a section that I need to skip.

I actually need to compare the results of two best fit circles - one with the questionable area (an uneven section on the circle), and one without. So if I can avoid actually doing additional scanning and just utilize the scan of the entire circle it would help a lot.

I'm having a hard time understanding all of the possible settings under the 'Scan Segment" option for constructing a circle.

First, the 'Discard Start Maximum' and 'Discard End Maximum' boxes. From reading the help file it looks like I'm entering a range of points to skip. So if there are 160 points of scan data, I could skip, say, points 60 to 100? Also, if I want to disable the 'Discard Tolerance' option, should I leave it at 0, or enter some ridiculously high value that will not filter out anything?

The other questions I have are concerning 'Approximate Start Point' and 'Approximate End Point'. Does this control what direction/order the data is used to construct the circle in or can I assume that the order the data points are read is in the order/direction of the original scan? (Edit: I don't see how these two points could control the direction...direction could be either way.)

I have been able to play with the settings and gotten results that look promising in the report window, but I can't think of any way to verify that the data is being used in the way that I think it is.

I have entered Start and End points, but I'm not sure if the circle is being constructed in a clockwise or counter-clockwise direction, which could potentially result in skipping a lot more points than I would like.

Thanks in advance for any help,


PG

[MrComment]

Heres one way you might try:

After you have scanned your part.
Create a copy of the scan below the original. (Use copy and paste)
With the copy you can then turn the scan into individual points without destroying your original. Construct your circles from the points to verify the results your questioning.

[PlasmaGuy]

Heres one way you might try:

After you have scanned your part.
Create a copy of the scan below the original. (Use copy and paste)
With the copy you can then turn the scan into individual points without destroying your original. Construct your circles from the points to verify the results your questioning.

Hmmm...now I'm confused. In my mind, the scan data is already a bunch of points. I've used those points to construct a circle and reported circularity. Now I simply want to use some of those same points to construct another circle and report its circularity and then I want to compare the two results.
My questions are about exactly how to extract only the points I want.

Am I missing something fundamental about scan data?


PG

[Underspec]

I'm currently running programs and haven't played with it yet. A few things come to mind. In the scan menu there should be a "TO POINTS" button which should generate the points you want. Another thing is that you can dimension point info through the insert/report. These will probably help you out i bet.

[PlasmaGuy]

I'm currently running programs and haven't played with it yet. A few things come to mind. In the scan menu there should be a "TO POINTS" button which should generate the points you want. Another thing is that you can dimension point info through the insert/report. These will probably help you out i bet.

I looked into the "TO POINTS" feature and I ended up with a large number of individual points...but only about half of what I expected for a given scan of a circle. This was a LOT of lines of code just for one circle and I have many circles I will need to make this work with.

I will check out the insert/report bit you mentioned, but it seems like working with individual points could be very time consuming.

Thanks for everyone's suggests so far.


PG

[PlasmaGuy]

workaround

Well, I couldn't figure out how to do what I originally set out to do (construct an arc from parts of a scan)... If I ever get the particulars on that type of construction nailed down I'll post it here.

In the meantime I've just added in some extra linear open scans that only scan part of the circle.

[Alexander]

I found a good solution for this type of analysis, developing a program called CurveAnalyzer (interfaced to PC-DMIS) having many options for "profilometry" problems. Naturally it is always better to handle just one software, no question..., but to my opinion it is also good to have open mind looking for the best solution. I was induced to develeop it just because of some functions that I missed in PC-DMIS, so that the idea is not to substitute existing functions, but just adding new and more efficient ones. If interested you can take a look at it at http://ams3d.astrei.com.