Best Approach to Measuring True Position on Threads

[[Da...]]

What is the best approach for measuring true position of threaded holes? I typically scan a cylinder using 4 lines which has been working well. The Zeiss cookbook mentions “Circle path with pitch”, which I am a little unclear on what this means. I have a UNC threaded bore size of 5/16.

[[Ch...]]

Cylinder with helical path that matches thread pitch

[[Me...]]

You'll match the Gradient/Slope Box to your thread pitch. Just pay attention however, for me at least Calypso is goofy, and changing the revolutions/target height, will automatically calculate a new thread pitch number, so you may have to re-apply it if you adjust those values.

Thread.png

[[Ow...]]

Nice article on this topic attached

Threaded_Holes_metrology conf 2011.pdf

[[Da...]]

Thanks everyone for your response! Owen the information you uploaded was exactly what I was looking for!

[[Ri...]]

Measuring the position of a threaded hole or shaft on a CMM is about as tough as it gets and it probably shouldn't be done at all. That being said, there are a few different strategies that work as well as is possible with a CMM. First of all, on a ID thread you are trying to measure the Minor diameter and on a OD thread you are trying to get the Major diameter. Trying to measure the PD on a thread with a CMM is a fantasy. The first question to ask is how the thread is created. Is it formed or cut? It makes a difference. The next question is how much tolerance? The answer for that will mean taking a very time consuming strategy versus a quick and dirty strategy. MMC? All that good stuff. Lines in a cylinder works well and I've found 5 lines versus 4 and lots of points for each line. Use the Maximum Inscribed evaluation for a ID thread and a Minimum Circumscribed for a OD. You should get a accurate, repeatable answer. The downside of this strategy is that it takes the longest amount of time. A helical strategy with lots of rotations, and here's the counter intuitive element to this helix, if it's a right hand thread, run the helix in the left hand direction. If it's a left hand thread, run the helix in the right hand direction. Remember, you are trying to capture the most interior or most exterior points on the thread form. You are not trying to measure the PD. As with the Lines on a Cylinder strategy, lots of data and use the Maximum Inscribed evaluation for an ID and Min Circumscribed for a OD. A helical path for a Circle is useful for those threaded hole position with a lot of tolerance. 4 pts and LSQ evaluation and follow the pitch. This will give you a decent answer for position but the size result will be useless. If you have MMC as part of your evaluation, use the Line path or Helical path strategy and the Major/Minor diameter tolerances will give you the bonus.

[[Ca...]]

I am in this disjunctive now. Most of case I use the Lines on cylinder and sometimes we repeat some of Minors with other strategy (Helix) because with the Lines on Cylinder we get an undersize dimension but not with Helix.
Is there a way ---like a condition programming--- to change the strategy in runtime from Lines in cylinder ?

[[Ro...]]

I do threaded holes all day long, every day.
I messed around with helix paths and different strategies of cylinders, but i've found plain old circles works really well for me.
Now, most of our threads are very small, 1/4-20 would be very large for us. most of our threads are 2 to 4mm helicoil holes.
.005" (.127mm) typical tolerance.

[[Da...]]

Simple thoughts to this problem:

Since a thread is made of at least two, sometime three tools (centering drill, core hole drill, thread tool), there are usually three things that can go wrong: the centering drill and the core hole drill don't line up exactly (I've had a lot of co-workers who were unable to check their tools for run-out after installation), the core hole deviates from the ideal position (maybe too high feed rate or blunt edges) or there's too much burr on the minor diameter. Or there's a combination of things.

If you want to measure a thread correctly, you will have to measure the flank diameter, which is almost impossible to do on a CMM (without a plug), especially on smaller threads. Any other measurement (ID for a inner thread or OD for an outer thread) is just an approximation and gives you information of the minor diameter (inner thread) or major diameter (outer thread), NOT the thread itself, which could be skewed in comparison to the other diameters.

Scanning a thread helically looks nice, but you're still not scanning the thread itself.

If you want the least bit of information of the thread itself, I'd recommend to scan the thread with axial lines, and then use the inner tangential cylinder. These points at least ever so slightly grazed the edges of the flanks if you use the smallest eligible ball size for your task (attention to a possible shank-out or too much burr on the thread!).

For everyday measurement of the ballpark positions of threads though, I use eight touch points on one or two levels (depending on the length of the element). The measurement uncertainty is just enough to satisfy my curiosity.

If you're unsure, run a study with different techniques on the same holes and validate your results.

[[Ri...]]

Please sign in to view this quote.

viewtopic.php?p=42490#p42490

I use this method if I'm going to do the linear line scans.

If I can, I prefer the helical scans though.

[[Ja...]]

I have a part where the .750-16 external thread is Datum -A-, and a .250-28 thread is being checked for True Position, Projected at 5.75". We are using a Tru-Pos gage for the .250-28 thread.

The issue is the .750-16 threaded section is just .24/.26 long so I get 3-4 threads max. I have tried several methods, helix at pitch , 3 circles, from 5 to 8 lines, reverse helix, etc. I cannot get consistent values for the .750-16 thread on the same part rerunning it 4-5 times each method without touching the part.

I tried doing Self-Centering with the helix paths, using a 1.5mm probe, but it does not follow the helix. It just runs at the start distance without moving more than a few tenths in the Z axis, instead of the .200 distance I have setup. If I remove the Self-Centering, and change nothing else, it does perform the helix move. Any suggestions would be helpful.

In the picture below, we cut everything to the right of the red line in one go, that includes the .750-16 threads and the .250-28 threads.

[[Mi...]]

I recently fairly successfully had to do this, on a female thread... but I think it should work on a male too. I have extreme doubts about this working on XXT, XXT can do some basic Self Centering but I don't see it working with this. Even if this method isn't 100% correct it was very stable and that was the most important thing in my mind. The project was/is destined to be ran on the shop floor so we needed a solution that did not use highly misplaceable Tru-Pos gauges.

-Use a fairly large ruby. I used a 3mm ruby on an M4x0.7 thread - smaller was less stable
-Self-Centered helix path, as many wraps as you can fit.
-Slow speed, high point density (I did 2.5mm/s with 0.01 Step width)
-Increase Self Centering Force
-Self Centering Settings like this:
-Increase point masking time interval (trial and error, you want it to mask everything from before it drops into the thread)


Alternatively, you could take a large ruby and scan lines on your cylinder. Creative Outlier settings like 1 Inside part, 3 Outside part and 3 iterations (or something close to that) should let you fairly reliably catch only the major of that thread.

[[Ma...]]

Please sign in to view this quote.

Hi Jake - which head you have?

I am worried about (M) used there on thread - how will you accomplish this?
With rotary table and L probe with really small diameter you possibly can scan that.

[[Ja...]]

I am using a VAST XXT-RDS head on a Contura G2.

For the (M), its not the best, but I am using the Diameter from the scan of the threads.

[[Ch...]]

I always post this when people ask about measuring on threads. It is an old lunchNlearn that covers most of this.


It might not be recommended, but I think you can get this with an XXT.
I recommend using helical path on a cylinder set to the thread pitch.
Then filter the crap out of the cylinder.

DO NOT USE SELFCENTER if threads are too small. Maybe don't use self center on the XXT at all.

Your features are small enough and tolerance large enough; I think this should be doable.

[[Ri...]]

Please sign in to view this quote.

You are going to have a lot of issues inspecting this. The Datum axis is pretty short, and you are going to run into projection error because of this. There is also a projection tolerance zone callout, so you are going to have to project your feature axis 5.75" off the surface. I believe Calypso will also project your Datum axis up to your feature projected tolerance zone, so the projection error is just going to compound.

[[Mi...]]

Please sign in to view this quote.

Genuinely curious, why bother with a helical path if you're not going to attempt a self center and follow the thread spacing? In my head it makes more sense to scan straight lines then filter the crap out of the data to evaluated based on the the major/minor diameter only... It might even work well enough to allow you use to the MMC bonus which my original suggested method won't do.

[[Ch...]]

Please sign in to view this quote.

The idea is to still follow the thread pitch. The entire idea is that sometimes self-centering gets buggy on small threads. This strategy yielded comparable results, but without making the cmm bug out.

When using scan lines, I have had aggresive filters chop off the peaks. But honestly, I doubt there would be much difference between scan lines and helical scans.


A few thoughts. There are lots of potential sources of error and assumptions made when scanning threads. Using different strategies and probes will give you different results. I suggest doing repeatability studies and generating some sort of Max Permissible Error for whatever strategy you use. Then compare this MPE to the tolerance and hope it is small.

For one part measured on threads, I did a large Intra-laboratory comparison with our affiliates. The same parts were shipped to 5 labs in 4 countries. All results were compared. If I can get in a shareable format, then I will post on here.