Thread Clocking and Lead In Chamfer Strategy Guidance

Here is an idea that I sometimes use when I want to measure the minor diameter of a thread. It may help for your experiment.

1) Take a series of points along the thread profile. You can use a scan or just a line with a bunch of hits.
2) Find the point that settles the furthest between threads.
3) Create a generic feature at the same level as that point
4) Align to that point
5) Measure the minor diameter

Below is a sketch to make it a little clearer. You want to have a probe radius that is small enough that it will get in between the threads a bit, but not so small that it will shank out. Just take a series of points that is long enough to cover one thread.



The code looks something like this:



Now in your case, you don't want a depth to align to, you want a rotation value. Let's consider a 1/4-20 thread. In this example you can measure a line along the thread profile starting at let's say 2 revolutions deep: .100". Make that line exactly one thread pitch long: .050" long. Take many hits. Let's say you make it 37 hits. You can then find highest point with the maxindex function. Each hit will equate to 10degrees of rotation. So if the first hit ( hit 1) is the maxindex, the root of the thread is at zero degrees at the face of the part. If hit 2 is the max, the root of the thread is at 10 degrees at the face of the part. 3=20deg. 4=30deg. You get the idea. You can use a simple equation that converts the maximum index directly into an angle value. ASSIGN/THREADANG=(MAXPNT-1)*10

From there you can construct a generic line in which to rotate to. Use some trig functions to fill in the vectors. I'd go into details for that, but I'm already feeling overwhelmed by the effort this will take.

If a 10 deg angle resolution is not enough for you, increase the number of hits. 73 hits would give you a 5 deg angle resolution. On a 1/4-20 thread that means over 50pnts/mm. More than I like to do even with an analog probe.

If you have an analog probe you also have the Center Scan feature that could be used to quickly find what I have been calling the maxindex. It will just scan along the thread profile until it nestles between the threads and finds the low point. I think this could also be done with an auto vector point using the 'Self Centering Point' measurement strategy. Once you have that low point you can do the math to find the rotation angle that leads to that low spot at the depth found. That would probably give you the best rotational resolution.

So, um, there's some ideas. The whole prospect doesn't give me warm fuzzies, but it can still be fun to talk theory. Please keep us updated on what you try and how it goes for you.

Here is an idea that I sometimes use when I want to measure the minor diameter of a thread. It may help for your experiment.

1) Take a series of points along the thread profile. You can use a scan or just a line with a bunch of hits.
2) Find the point that settles the furthest between threads.
3) Create a generic feature at the same level as that point
4) Align to that point
5) Measure the minor diameter

Below is a sketch to make it a little clearer. You want to have a probe radius that is small enough that it will get in between the threads a bit, but not so small that it will shank out. Just take a series of points that is long enough to cover one thread.



The code looks something like this:



Now in your case, you don't want a depth to align to, you want a rotation value. Let's consider a 1/4-20 thread. In this example you can measure a line along the thread profile starting at let's say 2 revolutions deep: .100". Make that line exactly one thread pitch long: .050" long. Take many hits. Let's say you make it 37 hits. You can then find highest point with the maxindex function. Each hit will equate to 10degrees of rotation. So if the first hit ( hit 1) is the maxindex, the root of the thread is at zero degrees at the face of the part. If hit 2 is the max, the root of the thread is at 10 degrees at the face of the part. 3=20deg. 4=30deg. You get the idea. You can use a simple equation that converts the maximum index directly into an angle value. ASSIGN/THREADANG=(MAXPNT-1)*10

From there you can construct a generic line in which to rotate to. Use some trig functions to fill in the vectors. I'd go into details for that, but I'm already feeling overwhelmed by the effort this will take.

If a 10 deg angle resolution is not enough for you, increase the number of hits. 73 hits would give you a 5 deg angle resolution. On a 1/4-20 thread that means over 50pnts/mm. More than I like to do even with an analog probe.

If you have an analog probe you also have the Center Scan feature that could be used to quickly find what I have been calling the maxindex. It will just scan along the thread profile until it nestles between the threads and finds the low point. I think this could also be done with an auto vector point using the 'Self Centering Point' measurement strategy. Once you have that low point you can do the math to find the rotation angle that leads to that low spot at the depth found. That would probably give you the best rotational resolution.

So, um, there's some ideas. The whole prospect doesn't give me warm fuzzies, but it can still be fun to talk theory. Please keep us updated on what you try and how it goes for you.

I've done the same exact thing with great success. However, I use the MININDEX and the MINOR diameter of the thread, as it's physically a shaper peak to probe upon, and therefore produces a more consistent maxindex value. Unless, the threads are rolled, and the minor has been drilled large (makes two minor peaks at rolled thread with a valley between rolled features)

louisd, that makes sense. I can imagine that finding the crest of the thread is easier than finding the low point between the threads. Also, interesting point about the rolled threads. I haven't had to measure rolled threads on a CMM before, but I have seen them have that two minor peaks on the crest thing you are talking about. I'll keep that in mind if it ever comes up.