Best scan selection

Could use a patch scan depending on how much of the part you need to check

Couldnt you do 1 linear open per "cross section " ? Selecting all the surfaces you want to cover and then adjust your "CROSSINGS" # in the lower right corner on the 'Execution" page ?

I need to cover top to bottom of the teeth (roots and radii included) Patch does not seem to want to work. Linear Open is what I've been trying to use, but it will not allow me to generate the points I would like along these sections, it will only allow me 3 hits, and I can't use that 'calculate' button unless I begin with atleast 5 hits. Working on trying to hand plot all the points for the scan but would love to find an easier way of dealing with this.

U need to SELECT all the surfaces you are going to cover , It will only show you 3 hits, start , direction, and END. I am a bit new to this myself and just beginning to understand so if anyone wants too add here or correct me please do. Imagine a "PLANE" that goes thru all the surafaces you need to scan and every time your "path" crooses that plane is a "CROSSING". Once this is all set, then goto GENERATE, and it should show true path with all the points.

Use LINEAR OPEN scan, click the START, DIRECTION, and END POINTS of the scan on the cad model.

Now comes the FUN part......

Your scan will wander all over UNLESS you make it a LINE, as in one axis needs to have the SAME number for all three points. You can change this maually by clicking on those points in the list in the scan dialog box. Once you have them all at the same value in one axis, you will need to tweak the VECTORS in the vector section of the scan dialog box. Make them all 0 or 1. Don't worry about what it will do, you just need to do it for the 'setup'. Once you have the 3 points and the vectors tweaked, THEN you click on GENERATE (which is what it was called for DECADES until some 'brilliant' coder decided to change it, not exactly SURE what the new name is, it may be calculate). Pcdmis will then go and generate all the hits (thus the original name of GENERATE!!!!) with all the correct vectors. When you do this, you will see the points being generated on the cad model. They will look like they are NOT on the cad data, but don't worry, they will be, the graphics show CENTER OF BALL path until it is all generated. Watch it generate the hits, some times Pcdmis will flip to the back side of the surface and go back to where it started on the INSIDE, if you see this, you will need to cancel and change the various tolerances for scan generation. If it generates it all correctly, then you will need to UN-MARK the measure now (YOU DO NOT WANT to measure it from the scan dialog menu). SET nominals to MASTER and set EXECUTE to normal IF you are using a touch trigger probe. If you are using a scanning probe, then you need to set it to FIND NOMS and RELEARN. Once you have this set, CREATE the scan (without measuring it) and close the scan dialog box. Repeat for as many 'strips' of scans as you need. Of course, it gets more complicated if the scans you want to make are not on the "peak" of the thing in an axis (like at 45 degrees) since the single axis number thing won't work.

Use LINEAR OPEN scan, click the START, DIRECTION, and END POINTS of the scan on the cad model.

Now comes the FUN part......

Your scan will wander all over UNLESS you make it a LINE, as in one axis needs to have the SAME number for all three points. You can change this maually by clicking on those points in the list in the scan dialog box. Once you have them all at the same value in one axis, you will need to tweak the VECTORS in the vector section of the scan dialog box. Make them all 0 or 1. Don't worry about what it will do, you just need to do it for the 'setup'. Once you have the 3 points and the vectors tweaked, THEN you click on GENERATE (which is what it was called for DECADES until some 'brilliant' coder decided to change it, not exactly SURE what the new name is, it may be calculate). Pcdmis will then go and generate all the hits (thus the original name of GENERATE!!!!) with all the correct vectors. When you do this, you will see the points being generated on the cad model. They will look like they are NOT on the cad data, but don't worry, they will be, the graphics show CENTER OF BALL path until it is all generated. Watch it generate the hits, some times Pcdmis will flip to the back side of the surface and go back to where it started on the INSIDE, if you see this, you will need to cancel and change the various tolerances for scan generation. If it generates it all correctly, then you will need to UN-MARK the measure now (YOU DO NOT WANT to measure it from the scan dialog menu). SET nominals to MASTER and set EXECUTE to normal IF you are using a touch trigger probe. If you are using a scanning probe, then you need to set it to FIND NOMS and RELEARN. Once you have this set, CREATE the scan (without measuring it) and close the scan dialog box. Repeat for as many 'strips' of scans as you need. Of course, it gets more complicated if the scans you want to make are not on the "peak" of the thing in an axis (like at 45 degrees) since the single axis number thing won't work.

+1
In addition to what Matt said, there is so much going on with patch scan and tweaking (i STILL think it is an art form) that i went and bought bigger dunce cap to go with my artist's easel.......

Matt, excellent advice, and I just got the whole area coated from 1 effort using linear open scan. But it leaves me with a question. I know I've read you posting about these items a few times already as before posting this I read other peoples questions with scanning.
With respect to Exec Controls: Why 'RE-LEARN' as opposed to 'DEFINED'?

I suppose I am biased with re-learn because it wreaks havoc in our validation system, but if re-learn was used in the first run, could it then be swapped back to defined after and still run the same?

Just as an fyi, we use SP25 technology on our PC-Dmis software cmms

Matt, excellent advice, and I just got the whole area coated from 1 effort using linear open scan. But it leaves me with a question. I know I've read you posting about these items a few times already as before posting this I read other peoples questions with scanning.
With respect to Exec Controls: Why 'RE-LEARN' as opposed to 'DEFINED'?

I suppose I am biased with re-learn because it wreaks havoc in our validation system, but if re-learn was used in the first run, could it then be swapped back to defined after and still run the same?

Just as an fyi, we use SP25 technology on our PC-Dmis software cmms

RE-LEARN is only used for SCANNING PROBES (continuous contact or laser). If you do NOT use re-learn, then it turns into a touch trigger type scan, even with the scanning probes, it will NOT make the smooth, continuous contact scan anymore, but individual points, like you get from a touch trigger probe. If you have a touch trigger probe, you do NOT use re-learn.

Well, I see I will have some playing around to do. Thanks for all the help folks!

I will probably be back with questions later on.

With respect to Exec Controls: Why 'RE-LEARN' as opposed to 'DEFINED'?

I'll give this a shot but I'll warn you that I am to 100% positive that I understand all of the nuances. Somebody please correct me if I make a mistake.

For many scans, but not all, there are 3 options for execution mode. Normal, defined, and relearn. As Matt mentions, normal will cause the probe to behave as a touch trigger probe regardless of whether it is or not. If you have an analogue probe you generally never use this.

Defined will take the path that you have defined absolutely literally. It will follow through those points and use those vectors for the comp no matter what. It does apply an offset into the part to load the strain gage and will make fine adjustments as it travels to maintain the correct offset force but it will not follow the part if the part deviates by more than the standard offset amount. I believe the offset is in the settings editor but it isn't something that you normally would have to change. This type of scan is very predictable and will do a great job if you are dealing with machined surfaces that are in the ballpark of where they are supposed to be. If you have parts that deviate by several millimeters from nominal you will likely run into some trouble here and see the probe scanning the air.

Re-learn will try to follow the defined path but will constantly react to feedback from the probe and will attempt to follow the part as it deviates from the path. It will record the actual path that it measures and use the actual recorded vectors from the probe for the comp. This will give you much better luck if your part deviates significantly from the CAD model.

Here's where it gets a bit fuzzy for me: Does it relearn the scan every single time it executes or does it only relearn it the first time it executes and from there on it uses that recorded scan as the nominal? I try to only use defined scans so I'm not totally clear on the details of how relearn works.

Note that if you are programming offline and you execute the scan offline with relearn on it will simply relearn the exact same path that is already defined so you won't notice any difference. The difference only comes into play on a machine with imperfect parts.

On top of this you also have "find-noms" which will take the measured data and attempt to best fit it to the CAD model and repierce to get a more accurate comp vector. Keep in mind that the tolerance set here is from ball center so you want to enter the expected part variation plus the radius of the ruby. If you have a 3mm ruby and you only give it 1.5mm find noms tolerance you are effectively giving it no tolerance at all.

Make sure that you have both probe comp and cad comp turned on if you are using a CAD model. CAD comp won't turn on automatically and it is absolutely essential to getting accurate measurement data. I learned that one the hard way.

You also have the "use best fit" button available in the find noms area. This will add in a short iterative cycle where it will perform a best fit, repierce the cad, perform another best fit, repierce the cad again, and perform a final best fit that will be used to generate the comped data. Theoretically, this should give you your most accurate comp vectors if the part varies significantly from nominal. I don't usually use it but plan to do some experiments with it soon to see if it will help tighten things up.