2D Curve scan troubles

[[Pe...]]

Got a question I’m hoping y’all can help me with. I am trying to scan a profile in curve and import it into ACCTee software to be able to pull some measurements from.
I am using the newest version of curve and the newest version of Calypso on our Duramax. We do not have any CAD files so I am digitizing 2D curve scan from a machined part (I know it’s not ideal, but it’s the best I can do). I have tinkered with most all the settings and I have mostly ended up with two results. The first: I get a decent scan but end up with weird spikes at the front and/or the rear of the scan and what looks like overlapping points on the radius’s (kind of looks like scribble).
The second: I get a real smooth nice scan but it distorts the scan and makes the dimensions way out.
Filters don’t seem to help, best fit blows it way out of shape. I’ve tried doing it as one, two, three and four segments. I had my speed set at 1.0 and the step at .007. I’ve tried lead in and out of .25 and tried shaving points off start and finish.
I have tried 3mm and a 5 mm probe and I have tried pushing it through and pulling it through the profile. Ultimately getting similar results.
I am new to curve. I have the manual and read the digitize section. I’m missing something but I just can’t put my finger on it. Attached are two pics. 1 being the scan I mostly get, and it is what I got with the above settings. The other being what the shape should look like as it was traced on our contour machine.
Any nudge in the right direction would be greatly appreciated

8C3240C5-B640-4AD6-B658-EF3A2D231D45.jpegE672EE6D-4CA5-4081-9756-1E042BFC4151.jpeg

[[Ow...]]

Don't have much time to answer and I've never used curve on a Passive scanning head like the Duramax uses but, even though it will be much slower, change it to take single points to remove any filtering issues.
Make sure you select to turn on the stylus correction checkbox before you begin the unknown contour.

[[Ro...]]

If you're digitizing a curve, I would suggest that after you scan it the first time you turn the actuals to nominals and repeat the process until you have no significant difference in your actuals between scans. The nominals are determined by the first scan and don't change unless you do that. This will clean up your approach vectors and give you the greatest accuracy, IMO. It may or may not have anything to do with your issue. Just some info if you need it.

[[Pe...]]

Please sign in to view this quote.

Wow. This was painful. It did go extremely slow, but it didn’t do any better. I gave a lot of weird points. Every time the probe made contact, it shimmied, which in turn added a bunch of random points.
Also, maybe I’m missing something else, but the stylus correction didnt appear to do anything.

[[Pe...]]

Please sign in to view this quote.

Sounds logical.
How would I turn the actuals to nominal?

[[Ro...]]

Open the feature, center pulldown menu above the actuals/Nominal Geometry Manipulation/Actual to Nominal

[[Pe...]]

Please sign in to view this quote.

I’ll give this a try tomorrow... thank you, I’ll post back.

[[Ri...]]

Please sign in to view this quote.

Most likely your probing dynamics is set too high. DuraMax and O-Inspect should be set to 20-30% for standard probes. You will need to set it lower for smaller probes.

[[Pe...]]

Please sign in to view this quote.

Could you elaborate? Which dynamics?

[[Ow...]]

They've made a lot of changes to the XXT passive scanning heads over the years (TL1, TL2, TL3) and I believe the Dynamics should be set differently depending on which one used and the type of probing used(?).

You can change the probing dynamics by going to the Measurement Plan Features Editor and selecting probing dynamics in the drop-down menu.

From the Calypso help-menu:
Probing Dynamics- Important for avoiding overshooting with soft or very thin styli.
• The standard is 100 %. For particularly long and thin styli or softer material, we recommend values between 10 and 50 %.
Note: The probing speed is considerably reduced by a changed measuring force activation only with values below 50%. Correct probing is not guaranteed with values below 10%. We recommend to carry out tests to determine the optimum value for the respective application.

[[Pe...]]

Please sign in to view this quote.

Please sign in to view this quote.

This looks like it may have done it. It definitely got rid of the scribble, I still am getting the random spikes at the end. If I trim off points or not, add pre/post travel it’s still there. Weird...

[[Pe...]]

Would any of you have an idea about the random spike at the end? Even with lowering the dynamics the spike is still there. I have added & lowered the point count, tried all the different %'s in the dynamics screen, adding pre/post travel and trimming points in the evaluation window.

[[Ma...]]

I'm usually masking a few points from start and end - depends on speed and number of points or width - i take min 0.5s to 1s
I'm masking it in eval page almost at bottom there is an option to select. I have no clue what's with exported points.

[[Ow...]]

I'm assuming you are recording 3 times more points than nominals in the strategy and have the filters turned on in the curve feature? If so, you might try opening up the outlier elimination filter and choose to "Include Adjacent Points" and set it to maybe 2 or 3 and see how it affects the result.
Typically the pre/post travel helps eliminate them but, like I said, I've never used a passive sensor using curve.

[[Pe...]]

Please sign in to view this quote.

I need to try the filters again. I have them off right now because it did nothing helpful. Also I should try the outiers again. I have not tried either since lowering the dynamics.

[[Pe...]]

Please sign in to view this quote.

So, here’s another question.
I know how to Digitize the curve. I know you adjust the point count by the spacing and length by leaving it as an unknown segment. The question is: how do you record “3 x more points than nominal “?
And to be clear... it’s better to have less points when digitizing?

[[Ma...]]

You can get 3x more points only by scanning digitized curve with edit in strategy/ num. of points or point distance.

I have active head and even with ditigizing forged pieces i don't get many spikes.
Digitizing at 1mm/s and point density each 0.05mm.

I prefer high density. You can always reduce points after. But as someone mentioned - it's good to create new curve from scanned just digitized curve - contour is now known so you can get better results.

I.E. : changing nominals - from actuals to nominal

[[Pe...]]

Please sign in to view this quote.

I must be missing something. I’ll digitize a scan and get let’s say 200 points, depending on the spacing, etc.
When I go back in and edit the point count, it reverts back to the original count when I click ok.
On screen it sometimes shows the additional points but it doesn’t show in the points list or the export.

[[Ma...]]

Maybe wrong place to edit points 😉

[[Pe...]]

Please sign in to view this quote.

Ths is the same spot I try to change them. I have also tried in the Features Editor, but both revert back.
This is got me going nuts.

[[Ow...]]

When you change the number of points to record, IT WILL NOT change the amount of nominals in the point list, just the amount of points to be recorded per nominal.
Even if you right-click on the cad screen, select properties and choose to show all nominal points, it will only show the nominals, not the amount of points recorded.
To show the amount vectors being recorded, select to display measured points and it should show 3 times more green vectors (if you set it to) than nominals.
In my picture, it doesn't show 3 times more green vectors than nominals because I haven't re-run the curve feature yet.

[[Pe...]]

Please sign in to view this quote.

OK, gotcha. Let me ask you this.
Down at the bottom of the segment window is the start and stop points. If I changed the points like you did to 1500, the end point should default to the 1500 after I scan it, correct? And if not default, I should be able to at least change it to 1400 or whatever, correct?
I ask because I have done as you have said (display measured points) and it will show a ton of the green vectors but they would all be jumbled close together and would still only show the original 500 in the export, the point list in the strategy window and in the end point box there at the bottom. And this is without having anything in the evaluation window excluded.
Like I said, I must be missing something, and I am sure its something small.

[[Ro...]]

Negative. The Start and End points are the nominal point count from the points list. They don't change unless you delete/add nominal points.
The number of points list where Owen showed you to change the amount is where you put in 3x or more the amount of nominal points. You can change this to whatever you want. 3x nominal is recommended.

If you want more nominal points either recreate the curve or modify the nominals and increase the point count. I almost always recreate the feature if I want more points, as opposed to letting Calypso guess what nominals to fill in for the added points.

[[Pe...]]

Please sign in to view this quote.

Gotcha.
I am getting good scans for the most part. I am still getting some scribble at the sharp downward turn and the random spike at the end. The scribble I blaming on the searching because the sudden drop. I'm going to try to make a couple segments and make that segment super slow and see if that helps. The severity of the spike varies but it is almost always there. I need to play around with the lead in and out.
I get better looking scans when I use less points, however, I am importing these scans into ACCTee for some really fine measurements and that software really needs the extra points to be able to run that program.

[[Ro...]]

Sometimes it helps to reverse the direction of the scan. It's not uncommon for there to be anomalies at the ends of the scan as the machine ramps up and the probe settles. You can exclude the rogue points from analysis if you think they aren't representative of your actual geometry. Another thing to consider is probe size. The bigger the ball, the more it will smooth the results. That may or may not be a good thing, but something to think about.