Restricting Degrees of Freedom

[[Za...]]

Hello all, I'm still learning a lot of this stuff on the fly, but I want to be able to explain to somebody why restricting degrees of freedom changes the results.

For example, we have a part that I just checked that checks the true position of a bore to a scan of a true position pin that's threaded into a 1-1/16-12 internal thread.

Using the program as is, the callout fails. However, if I restrict X and Y on the TPP scan, the true position comes into tolerance. Can somebody help explain this? I've tried looking for a good reference online, but I haven't seen anything that explains what is actually happening when I do this.

[[Cl...]]

Typically you would constrain the location of circle segments, or radii that are less than
180 degrees to get a more accurate size, and constrain their size (radius) to get a better
location. Your FCF datum's will be your "constraints".

[[Za...]]

Please sign in to view this quote.

I'm getting a full scan of the cylinder as well as the TPP. Why would the results be so much different with constraining X and Y?

[[Jo...]]

i.m.o lots of bad things going on there. First of all, Primary Datum is a circle. Second is any undefined datum structure reverts back to alignment. You can't just leave it empty. Third, is by constraining X,Y you are effectively telling it that the measured feature is in nominal position.

[[Jo...]]

Also be aware, threading a pin in and measuring the cylinder above the position can introduce projected position error if not Perpendicular to datum

[[Za...]]

Please sign in to view this quote.

The primary datum is the 1-1/16-12 Pitch Diameter, so I'm scanning a True Position Pin. Should this be a cylinder? I've never seen it done any other way than as a circle, so that's how I've been programming these.

As far as the undefined datum structure, what else would I use and why? The callout only asks for A
I partially understand what you're saying about constraining x&y since the location is at 0,0, but why would constraining x and y not leave the result closer to zero?

Thanks for the input, this is the type of answer I was looking for 🙂

[[Za...]]

Please sign in to view this quote.

How else would you recommend getting the pitch diameter? Most of the callouts that I see that reference threads, allow us to use the minor diameter scan, but this one specifically calls out the pitch 🙁

[[Jo...]]

Please sign in to view this quote.

We use a tool that machines ISO port profile all as one, so all the features are concentric. Then come back with thread mill and machine threads. So the logic may be with that in mind. I would check the same part at least three times, removing insert each time to see repeatability. Typically the Trupos inserts have enough pin to measure as cylinder. Lastly try helix scan with small probe that can land inside thread face without shanking. I doubt you have position error. More likely filter or projection issue. With Datum on center I'm not understanding the amount of error.

Without the primary datum being able to level it is using base alignment. You could also make an secondary alignment on the spot face area, centered on port. That would insure projected error isn't introduced.

[[Me...]]

Since the callout is only to -A- you must use a cylinder to check it per print. Calling out the PD is problematic using a CMM. Unless the TP Plug fits perfectly snug on the PD you will have slop, which basically nullifies the whole point of the callout. Once you tighten it down it will use that slop to reposition itself.

As stated before, constraining the geometry is telling Calypso that it's perfect in the axis or/radius you check. That's why your result is better.

I'd measure the TP Plug as a cylinder, getting as far apart between ends of it as possible on the cylindrical portion. Either that or scan the threads themselves as a helix, since you're not necessarily seated on the PD anyway.

[[Ke...]]

Unfortunately, to be a "print legal" evaluation, you can't evaluate the thread using "any method necessary"
Good eye noticing that it is called out differently than the other other prints.
What type or Pos Pin are you using?

If you are using a split-type Pos Pin, these are made to provide full contact on the pitch diameter, and would give the most accurate results. Unfortunately, these can not be accurately re-calibrated, as they must be inspected before the split is machined; so I would put it on a "disposal" frequency, rather than a calibration frequency...

If you are using a solid type: (what I usually use, because they are "good enough" for my tolerances, and they can be re-calibrated)
I would definitely do a study to check repeatability (as previously mentioned), but I would use two separate approaches. First: measure multiple times, and re-insert every time using the same seating pressure. This will tell you how repeatable it is, when used with great care.... next: measure multiple times, and re-insert using lighter-to-firmer seating pressure. Noticeable differences in the the two methods will indicate if/how much "slop" is present. I believe that if there is "slop" that (if the part is formed properly, and I'm presuming that this is all machined) when the Pos Pin is tightened, the PD of the thread will seat against (not perfect but close)the PD of the part, and give good results. If the thread is not perpendicular to the threads and there is "slop", then over-tightening the Pos Pin will skew the Pos Pin (not seat properly), and the results will be skewed, as well.

Just my 2 cents (a lot of typing, for little value added 🤣 )

[[Za...]]

Please sign in to view this quote.

I have little to nothing as far as machining knowledge goes, but I am almost positive that we also machine these ports as one as well... which is why I was confused how the TP could be out of tolerance.

Please sign in to view this quote.

Scanning the TP as a cylinder did the trick! It brought the result down to about half of the allowed tolerance. Why did that solve my issue though? Does Calypso "best fit" a cylinder's centerline while a circle could show as skewed? I'm just trying to understand why the results are so different from circle to cylinder.

Please sign in to view this quote.

We use the solid threaded inserts instead of the split style as the tolerances are pretty reasonable for most of the parts I inspect. I think I will eventually get to testing these True Position Pins at some point. I'm pretty swamped right now, so this will be a good thing to do the next time we slow down for a bit.

Thank you everybody for the input.

[[Me...]]

Please sign in to view this quote.

Because I circle has no spatial vector. Think about when you create an alignment. The spatial input determines what to level your alignment to, i.e. the direction all your other origins are aligned perpendicular to. It's the starting point of building an alignment, like a flat surface plate you put your height gage on. When you create a True Position characteristic, you're actually creating an alignment based on what Datums you input. If you use a circle, it has no spatial vector, so it reverts back to the Base Alignment. If your BA is using something else, like the plane on top, it will determine the spatial direction from that. If your top plane and the bore you're checking aren't perpendicular to each that will add error, because it's not looking down the axis of -A-, it's looking perpendicular to the plane.

A cylinder has a spatial vector, so when you use that it's looking down the axis of -A-, as it should per your print.

[[Za...]]

Please sign in to view this quote.

Awesome explanation, thanks again 😃

[[Gr...]]

When I was first learning how to use Calypso's awful, unintuitive, and poorly explained characteristic alignment system, I found it helpful to check my numbers with a Polar_2d characteristic.
If true position = 2 * Polar_2d, it was correct.
It only worked if the drf was primarily based off of a nice plane, and the feature could be reduced to a point, but it helped me catch errors in my understanding.