Evaluate Separate Curve scans in 1 characteristic

[[Mi...]]

My curve experience is next to nothing, but I've gotten a decent start.

I am scanning a punch that makes a gear, powdered metal forming. So it's not gear measurements I need but form.

I want to scan down lengths of the tooth flanks with 3D curves created from the CAD model, and report form deviation.

I figured out how to section the model, create the curve, scan and display no problem.

I can't figure out how to have separate scans reporting (CAD evaluation) at the same time without it being extremely busy by doing a form characteristic for each scan. I assumed I could recall the points from separate curves on the various teeth into a single one, but the form variation goes out by miles because it's joining the nominal lines together...which isn't right.

Recalling into FreeForm surface appeared to work but then the calculation time is 5+ minutes every click I do with that feature, and it was only about a 1/4 of the data.

I need to figure out how to have multiple, un-joined segments with a clearance move between each in a single 3d Curve, or recall multiple 3D curves into a single feature.

I've seen data displayed the way I'm trying to, so there must be a way but I'm struggling to find it. Here is the example.

[[Je...]]

Please sign in to view this quote.

.

Hey Mike.

It sounds like you're off to a great start. Don't let the idea of curves or freeform surfaces intimidate you. Although they can be evaluated with varying depths of complexity, they're just a collections of line segments.

I had an awesome Curve & Freeform instructor, Nick Janak, at Zeiss in Maple Grove, MN. Just like you're finding out, he recommended keeping curves as curves and not freeform constructions. Freeforms are acceptable for surface profile eval., but even then my freeforms are usually made from multiple recalled curve sections, so that I can evaluate each slice of the freeform. That's just my approach.

I was also taught to prioritize use of 2D curves over 3D curves except in cases of threads/involutes or when the customer specifically requests it.. The third vector can add unjustified error to the measurement.

Oh, and the line profile characteristic can give you some great plots in PiWeb. In your current example, I would use the "create features" menu window and then the "section" tab. After making slices, you can click on individual segments under the hierarchy tab in create features and then turn those into curves. There's more to that process, but that's the gist of it.

.

[[Mi...]]

Please sign in to view this quote.

Well physically measuring wasn't the issue, I was able to align, and measure the part. I believe I will need to use a 3d Curve for this. No rotary table either so there's that limitation.

Here is a screenshot, section created using a cylindrical section, at pitch diameter. I selected various flanks and created a 3D curve for each under the point set, from curves option.

Now I have a form characteristic from each 3D curve... which isn't all that nice to try to read, especially eventually I'll have more than one scan on these teeth.

I'm stuck at consolidating the separate scans into a single feature so they can be presented at one time.

Curve 1.PNG

[[Je...]]

.

Oh boy. Yes, those are definitely 3D curves. So, is the issue about how to display the characteristics and their flags simultaneously in the CAD window?

.

[[Mi...]]

Please sign in to view this quote.

Well ideally, I'm hoping to evaluate multiple scans as if they were 1 characteristic. So 1 min/max flag for all of them.

OR

The option to show the actual deviation of each individual point, but that got really messy even with only 25 points per scan.

OR a completely different idea from somebody more practiced with this type of evaluation.

[[Je...]]

.

I see. Perhaps some of the Space Point junkies could chime in here.

If the goal is to simply view a specific set of banners on-screen, then you can use a strategy like Tom mentions here:

viewtopic.php?p=45402#p45402


You can name a range of points (i.e. 1-10) or a pattern such as every 50th point, etc..

Otherwise, if the values of the various curves need to relate to each other, then you may be dealing with a freeform surface. The thing to remember there is that freeform surfaces always get their nominals from the CAD model, so your CAD needs to be accurate.

.

[[Ro...]]

Please sign in to view this quote.

How about creating a minimum and maximum result and using that in your CAD view?

Size/More/Minimum-Maximum result

[[Mi...]]

Thanks for the ideas all!

I ended up getting the solution elsewhere but I'll post it here.

I had to create a super 3D curve, from the same sections as the individual 3D Curves and limit the evaluation of the start and end of each segment, then recall the individual 3D curves into the super curve.

And it works! sucked, but it worked.

[[Je...]]

Please sign in to view this quote.

.
Woohoo! Nice work, Michael.

There should be a measuring hero award for craziest hybrid feature.
.

[[Mi...]]

Please sign in to view this quote.

I can't take credit for the method, it was suggested to me lol.

The worst part about this whole thing?

I do not have a rotary table...

85 teeth, need to scan about 150mm of length.

So basically will be repeating this process at at least 3 heights lol

[[Mi...]]

Please sign in to view this quote.

I can't take credit for the method, it was suggested to me lol.

The worst part about this whole thing?

I do not have a rotary table...

85 teeth, need to scan about 150mm of length.

So basically will be repeating this process at at least 3 heights lol

[[Ma...]]

Please sign in to view this quote.

I am using this method too - but remember, that missing nominal points are problem - they affect nominal spline which will result in some bad evaluations on beginning/end of original curve segment.

If you have model and it is possible, then play with nominals, sort them to make nominal spline like it would scan in one move and create new curve from model using nominal spline of all-in-one curve ( to make it selectable you have to make it non-theoretical curve with correct segment start/end point )

I can share some picture aid.

[[Mi...]]

Please sign in to view this quote.

I would greatly appreciate anything you would be willing to share.

I'm learning curve mostly by reading the manual so I'm eager for anything else

[[Ma...]]

When I was doing screenshots I found that problem is causing mainly 2D curves, but 3D curves are not making nominal curve that bad.

Here it is 2D curve - purple is nominal curve.
When I did that with 3D curve it was skipping gaps with straight line - not spline like on this 2D curve.

On evaluation tab you can select "Take segment gaps into consideration" which may help.

When I have some problems, then I select curve feature to see nominal spline, then I open CAD -> Create feature -> Point set.
Select "3d curve" from list and I click on shown nominal spline - this will make new 3d curve based on path given by nominal spline projected on CAD surface.

Actually I am using this method when I am making curve by clicking on model - this will make nominal spline go into material on bumpy model. After creating this wrong curve I highlight it's nominal spline, open creating feature window, selecting point set tab, 3d curve and selecting that wrong nominal spline - now it makes accurate 3d curve ( attch n. 2 )

[[Vi...]]

Please sign in to view this quote.

Hello Michael,

This is a awesome work around! I was wondering if you could elaborate on the limiting of the evaluation of the start and end of each segment. And why it's needed?

Thanks,
Vince

[[Gu...]]

Doesn't quite match the example above. But maybe useful information nonetheless

An example of how to display many curves in one plot. This view works very well if the curves are all in one plane.