Symptoms of model problem or progam issue?

[[Me...]]

So.. not sure where to begin, we have a giant hunk of cast aluminum, and I'm having some issues with hole positions, aka almost all of them. I know the pictures are a little goofy, but I tried to fit everything in as few pictures as possible. So in the picture labeled, New from Scratch, I re-loaded the model, and set just my alignment, and allowed calypso to orient to the model automatically (offset 22.5 degrees, and origins in space originally, did not use cad model translation), using the features in the picture (tried to fit it in one). So the large plane for my X direction/spatial, vertical plane for planar, and the hole for the other 2 origins. I went to test a simple threaded hole, and in the model location from scratch, the nominals look normal (adheres to model nicely), except when it measures it hits me with scanning optimization, ok so the actuals are off I'm guessing, as seen in the actuals picture..

Now this is where it gets a little confusing, and the answer may be pretty simple, I've just spent so much time back and forth my brain has rattled out my last remnants of goo..

I had originally made a program for this part, and ran it through with all the scanning optimization errors, thinking it was a problem with my alignment (instead of using the hole as origin from picture below, I was using a plane perpendicular to the pictures spatial/x origin, as my Z origin, and the pictures planar as my Y origin to compensate, take a breath for a second.) So I said ok, I will change my origins to a hole, because I've heard more features can help orient better, separating my planar and Y/Z origin, to the way it is now, and that is where picture number 1, the old program using from scratch alignment comes into play. The nominals have been shifted, and now there are no scanning optimization errors, just the occasional probe scratch the side wall from being so close to the edge (leading me to believe the holes are actually out of position). I know it's a lot to take in, and I'm fairly confident the holes are out, but I'm not sure why the model would change the nominals, just by changing a few alignment features. Also, I am not sure if the model is messed up or where to go, so just looking for any thoughts. Thank you.

Also, from the base model, all the features used in all attempted alignments appear perpendicular to each other, with no inherent angles. Perhaps a more simple question would be, if I copy a program and change the alignment, is there a possibility it will use the last known actuals as new nominals?

Old Program using from scratch alignment.pngNew From Scratch.pngModel Location from Scratch.png

[[Am...]]

I am not sure why the nominals would change with a different alignment, but I think you're right about the features being out of position. Open one of your features and right click the CAD model view, turn on "show actual points" if its not already on. How do the points line up with the model? you should be able to see easily on a circle if your points are off center and going into the model. Also id double check the qualification of the probe, make sure something hasn't come loose since you started. how many probes are being used in your alignment?

[[Me...]]

Hi Amel,

Thanks, the actual points follow the nominal position line for the most part as opposed to the model side, I don't think the probes are loose, and the newest method of checking I was able to use a single probe position and check all features, no odd position issues or deflection or anything.

Actually, the more I look at it now, it is odd that the nominals are identical to the actuals in the copied program, which makes me believe more it was a more a bug with Calypso, or something I setup weird. Also, this whole thing started because I was getting not only the scanning optimization errors, but also massively out of tolerance true positions with another set of holes (but that's an entirely different model/print issue as well, I have no basic values for the hole positions, so I don't know what the nominals are supposed to be haha..only angle positions to a common reference point, which are good, but no positions to any other individual reference).

[[Am...]]

Yeah I thought it was odd that your actuals were dead on across the board, I have seen my fair share of weird bugs in calypso it very well could be the case here.

[[No...]]

Dave, I admit I probably didn't fully understand the core of your problem (too much information for my morning brain I guess), but I stumbled across your statement in the first post, that you allowed Calypso to orient the model and did not use CAD model transformation.
How did you do this? By probing the real part? Or by defining the features on the model? And if so, was dispersion on or off?
If you allow Calypso to orient the model according to your alignment, it's essential that your actuals at that moment do not in any way deviate from your nominals (or even better delete them if possible). Otherwise Calypso will orient by the actual values (as it would on the real CMM) and this will almost always cause trouble. That happened to me when I programmed my very first part in Calypso and it almost drove me nuts. It took a very long time (years) until I had fully understood what had happened there and was able to position my model correctly. Since then I NEVER EVER allow Calypso to orient the model by itself.

[[Me...]]

Hey Norbert, most of the time I use the cad model transformation but I can't remember why I didn't this time, I think because I was trying different alignments, unsure how I was initially going to set it up, but I created the alignment by selecting the features and allowing calypso to rotate it, and where is the dispersion option (which means I probably didn't use it), and what is it's function? Also, is there an easy way to import all the features and characteristics I've done so far to a new template so I can try using the cad model transformation method?
And for the record I don't think I had any actual data when I initially did the alignments (except in the case of a portion of my question, where I think it did use the actuals as reference)

I remember being on the phone with Zeiss when they showed me how the model could be oriented using the alignment and I said "Wow neat". Guess it's not so neat haha.

[[No...]]

The dispersion function is in the measurement settings. It adds random noise to simulated actuals (points) to add a bit more realism. But it can also interfere with certain other functions, automatic model transformation being one of them.
Newer Calypso releases show you a big yellow warning sign in the CNC start window when dispersion is on, but of course it's only used in sim mode, not during real measurements.
Importing the features and characteristics to another measurement plan should be simple copy&paste. You can copy with refernces, so all referenced elements are copied over too, but if you copy everything anyway, you can just C&P all features and chracteristics seperately, as references are by name only.

[[Ro...]]

This may or may not be of any help.........
I've had issues in the past when i healed my model and for no good reason selected "build geometry".
I had a large batch of similar parts to program, probably 70 or 80 programs in total, and when we started running one set of parts there was a profile that was out by roughly .010 inch. all hard checks were good (luckily the profiles were on easy to measure surfaces) so i saw my cad models were oddly skewed by .005 inch from side to side. I sent it up to engineering to check the cad models they received from the customer and they reported that the cad models on their end were ok.
I reloaded the cad models and checked them, they were good, when i healed the model again with build geometry selected the models were again skewed.
And now i'll never ever ever ever use build geometry ever again.

sfgdjhklsfgdjksfgdjhksfgdjhklsfgdjk.JPG

[[Me...]]

Ok good to know, thanks for the info, and yeah Roberto, Build Geometry just sounds scary, I don't generally use that, or stitching.
So I'll take a look, still going with the holes are most likely out of position but just thought I'd learn something.
Thanks again

[[Ri...]]

Please sign in to view this quote.

I would shy away from using words like "never" and "always".

There are times were Build Geometry is needed, I have seen this with IGES formats from time-to-time.

I would say that most times you shouldn't need to perform Build Geometry, but there are times when it is needed, and I would take it on a case-by-case basis.

[[Me...]]

Ok so I tried Norbert's suggestion, I started fresh healed with simplification and correct boundary curves selected (because it was default) and used the cad model transformation to rotate and slide my center near my datums used for alignment, then I extracted the features and set the alignment to them as need be. The nominal values changed considerably, which was a little concerning, but after running the program, some quick threaded holes I checked for simple center positions are out a similar distance...however they are out in the opposite direction, which is driving me crazy. So I uploaded a picture of my alignment numbers, and if I had to guess perhaps the angle might be an issue, but I am not sure how this is referenced, but I only had to rotate the model 22.5 degrees to come to 90 degrees of how the part is loaded, if anyone knows more about this thanks, also wondering why the nominals are so different as well (default extracted hole position is .300 difference y and z, metric, between the old auto aligned way, and the new healed way..). The only thing left I can think of is that my original planar plane is not long enough to produce a repeatable angle orientation, as its a very small cut (incomplete part), controlling over a much larger overall size, but why would the nominals be different still

Test3.png

[[Ri...]]

Interesting.

A question - how do you extract features? Do you use "Extract Features" or do you use the "Define Points" method?

[[Me...]]

Please sign in to view this quote.

I use the built in extract features button, never tried defining points

[[Ri...]]

OK. Just checking.