I've got a real problem with the DEMON changing the number of hits in a scan, programmed offline vs. being ran online. My real issue is pulling hits from these scans for intersect points on short surfaces. It there a way to lock this up so the hits are what I programmed it to or do I need to get my program changed. Picture a small piece of pie (each side approximately 7mm long) with chamfers on the 3 corners at 0.65mm to 0.7mm. Running the program offline produced exactly what I the CAD/print shows and I've been beating my head against the wall (run this through Inspect so I saw no data) tryin to get this figured out.
maybe "defined" for execution mode? that's about all I can give you, and I'm not sure what it will do to the path, will it stay a scan scan or a stitch scan.
Neil that just sucks (pardon my attitude) but if I define a scan around "my pie" offline and when I run it online it changes what it does based on the deviation of the surface. I find it odd that when I tolerance the form (best fit profile tolerance) of "my pie" it is within 0.02 to 0.050mm of CAD nominal and the DEMON detects this deviation and change the whole scan. This is not acceptable.
I see the number of points different you have posted elsewhere, and that seems extreme to me.
However, What Neil says is true in mCosmos and Calypso as well, as I have those and PcDmis here scanning in all three. If you want two hits per mm (.04") and your part has a perimeter that is 2 mm longer than the model, that would be an extra 4 hits.
With the delta you describe, and my best guess about actual size, I'm thinking you go from 23mm perimeter to 23.2-ish (based on the stated profile error if the part is bigger, 22.8-ish if smaller), so at 2/mm that would not even be one hit. I think you said something like 12-15 hits different.
What Neil describes, based on what I can guess, is not what you are experiencing. That doesn't make Neil wrong in what he said, though, as he doesn't know how many hits are missing, nor did he know your profile deviation.
In all three softwares, I can extract features so the points used are dynamic based on the actual scan (what Neil is suggesting), or I pump the density so that a few extra or missing hits don't wreck me on discreet point selection for constructions (what you appear to be doing).
I would never be able to live with 15 hits missing, though.
I can say, in Calypso, once, I had to re-select points after running a part online after programming offline because they were WAY off. Once. It is a machine we bought for a product family of 8 parts. Seven parts went from offline to online flawlessly. One part I had what you are having.
I didn't bother with the why, I changed my point selection and then it has worked ever since. Some artefact that the software saw in the model that isn't real, maybe, a seam that the software followed into the solid and back out that skewed my point count? I have no idea.
My point there is, I changed my point selection and it worked ever since. Can you do that?
If you run 10 parts, do they all have the same point count and distribution that this first part had? Maybe a single fix and you are good to go forever?
From the way mCosmos makes scan sections, I can visually see that the software follows seems into the model sometimes, and there is a method to trim that crap out. Maybe that is happening to you here, or something similar behind the scenes, and a one time adjustment will get you on track.
I see the number of points different you have posted elsewhere, and that seems extreme to me.
However, What Neil says is true in mCosmos and Calypso as well, as I have those and PcDmis here scanning in all three. If you want two hits per mm (.04") and your part has a perimeter that is 2 mm longer than the model, that would be an extra 4 hits.
With the delta you describe, and my best guess about actual size, I'm thinking you go from 23mm perimeter to 23.2-ish (based on the stated profile error if the part is bigger, 22.8-ish if smaller), so at 2/mm that would not even be one hit. I think you said something like 12-15 hits different. What Neil describes, based on what I can guess, is not what you are experiencing. That doesn't make Neil wrong in what he said, though, as he doesn't know how many hits are missing, nor did he know your profile deviation.
In all three softwares, I can extract features so the points used are dynamic based on the actual scan (what Neil is suggesting), or I pump the density so that a few extra or missing hits don't wreck me on discreet point selection for constructions (what you appear to be doing).
I would never be able to live with 15 hits missing, though.
I can say, in Calypso, once, I had to re-select points after running a part online after programming offline because they were WAY off. Once. It is a machine we bought for a product family of 8 parts. Seven parts went from offline to online flawlessly. One part I had what you are having.
I didn't bother with the why, I changed my point selection and then it has worked ever since. Some artefact that the software saw in the model that isn't real, maybe, a seam that the software followed into the solid and back out that skewed my point count? I have no idea.
My point there is, I changed my point selection and it worked ever since. Can you do that? If you run 10 parts, do they all have the same point count and distribution that this first part had? Maybe a single fix and you are good to go forever?
From the way mCosmos makes scan sections, I can visually see that the software follows seems into the model sometimes, and there is a method to trim that crap out. Maybe that is happening to you here, or something similar behind the scenes, and a one time adjustment will get you on track.
I ended up reprogramming those scanned surfaces to lines as I had a mill down waiting on me. I may mess with it at another point in time but was hoping to scan it and, pull the points and do the constructions.
Hi Dwade, I see your comments here and on the other place - I think maybe you're being maybe hasty in your criticism, and if I'm honest a little unfair.
I've read both threads through, and you don't mention what setting's you're using, which makes it difficult to fully consider what may be going on. You're saying it's BS but there might well be a valid explanation for what you're seeing, and it might serve as an opportunity to learn some of what's actually happening during the scanning process.
You do mention that you have 10 hits per 0.65 chamfer, this suggests a point density of ~15 points/mm? That seems like quite a high density, small changes in length of scan pass will therefore result in a greater change to the overall number of hits.
You also say that you have a profile result of up to 0.05 (i.e. possibly +0.025 per side) - For simplicity sake (and I know it's not quite right) but lets add 0.025 to each end of each side of your profile. this would equate to an increase in perimeter of 0.3mm which would equate 5 points straight away.
Also you don't mention what stylus you're using, perhaps the offline hits per mm is calculated by the cad surface perimeter, but the online hits is calculated by the distance the tip centre travels (for such a small shape this could be significantly larger).
I guess what I'm saying is, rather than just call BS, give us the relevant information, and together we can work out what's happening and work to resolve it.
Because it's what I'd term fairly intricate geometry, there may be things we can do to compensate - I'm thinking have a really high hit density, but then apply a very small linear filter (instead of NULLFILTER) to ensure a predictable spacing of hits around the feature.
Can you let us know:
Current Line 1 Tech type (NULLFILTER, LINEAR or VARIABLE).
Current point density (on the settings tab of the scan dialog)
Comp methods? I assume Probe Comp, but are you using CAD Comp as well?
Hi Dwade, I see your comments here and on the other place - I think maybe you're being maybe hasty in your criticism, and if I'm honest a little unfair.
I've read both threads through, and you don't mention what setting's you're using, which makes it difficult to fully consider what may be going on. You're saying it's BS but there might well be a valid explanation for what you're seeing, and it might serve as an opportunity to learn some of what's actually happening during the scanning process.
You do mention that you have 10 hits per 0.65 chamfer, this suggests a point density of ~15 points/mm? That seems like quite a high density, small changes in length of scan pass will therefore result in a greater change to the overall number of hits.
You also say that you have a profile result of up to 0.05 (i.e. possibly +0.025 per side) - For simplicity sake (and I know it's not quite right) but lets add 0.025 to each end of each side of your profile. this would equate to an increase in perimeter of 0.3mm which would equate 5 points straight away.
Also you don't mention what stylus you're using, perhaps the offline hits per mm is calculated by the cad surface perimeter, but the online hits is calculated by the distance the tip centre travels (for such a small shape this could be significantly larger).
I guess what I'm saying is, rather than just call BS, give us the relevant information, and together we can work out what's happening and work to resolve it.
Because it's what I'd term fairly intricate geometry, there may be things we can do to compensate - I'm thinking have a really high hit density, but then apply a very small linear filter (instead of NULLFILTER) to ensure a predictable spacing of hits around the feature.
Can you let us know:
Current Line 1 Tech type (NULLFILTER, LINEAR or VARIABLE).
Current point density (on the settings tab of the scan dialog)
Comp methods? I assume Probe Comp, but are you using CAD Comp as well?
Probe diameter.
There have been responses on both here and the "other side" referring to point density and I had it set high because of the size of the "chamfer flat surfaces". With those surfaces being effectively less than 1mm long I thought that the more points taken, the better my results would be.
This is my default setting for this section of the program. This is a new design for a knee and the customer is concerned with the form of everything involved with this cam lock. We had asked for a form and location profile callout and they didn't like that idea but chasing the small flats was a PITA when I was pulling the scan points because they were varying in numbers which is what I did not know. In my mind with the surfaces being machined to within 0.020-0.050mm of CAD nominals what could go wrong.
After discussing it with one of the other programmers here we both decided to stick with individual line measurements partly because the 3rd programmer would have really struggled to comprehend what was being done and why. It has been an eye opener for us here and I hope that it has enlightened others.
