Heart-shaped holes

[pernilla]

Hi.

I'm having some issues with our Mistral and it's been going on for a while. Sometimes (randomly I think) when I measure holes I get this weird result, almost heart-shaped and one of the points are in opposite direction. After measuring, I can restart the program without touching or moving anything and the hole is nearly perfect.

I've tried measuring different size holes, slowing down between points, adding more points, fewer points, changing start/stop angle, different probes... I don't know what else to try?

Has anyone else seen this before? When I changed the start/stop angle the arrows did not "follow".

Would it be possible to add some kind of filtration? This particular hole is normally measured with 30 points.

The roundness can sometimes get a deviation of 1.200 mm when normally it's about 0.005 mm.

grafik2.png

I've also had a similar problem when measuring flatness.
It's measured on a pretty big surface with a lot of points and sometimes (also randomly) a group of about 10 points on a small area, gets these huge red arrows (always up from the plane, Z+). But when I test the machine to try and see what happens, the result is always nearly perfect.

Any ideas?

[Schlag]

Vector issue from the CAD ? Before I start any program I will " FIX SURFACE VECTORS " as step 1. That's all I have for ideas....

[pernilla]

Thanks. But in my test-holes I haven't used CAD.

[JEFMAN]

COuld you describe the probe and the tip ?

[DAN_M]

One thing to check...

If your feature is an autofeature, "F9" it and then make sure you don't have something incorrect selected in the dropbox I highlighted. Usually people leave it at "NOMINALS".

picgrouppp.PNG

[pernilla]

It's not an autofeature.

[DAN_M]

sorry =/ i tried

[pernilla]

This is the one I've used, D3L50. But I've seen the same problem with D4L50.

D3L50.jpg

[JEFMAN]

TP20 or TP200 ?
What module ?

[pernilla]

It's a TP200.

[JEFMAN]

Can you change the body ?
The TP200 is a little fragile, if you screw it without the little tool, you can unstick the strain gauge which is inside...
Or just use the kind of "patafix" which is delivered with to clean the contacts...

[pernilla]

The TP200 is only about a year old and I had a guy from Hexagon replace it. I clean the contact surfaces before measuring with "Puty for cleaning" from Hexagon.
And also, probe verifications doesn't show anything out of the ordinary.

[Schlag]

If its a regular Measured Circle and you are creating it "online", I would only take 4 hits ( front back / left right ) to create it and then just change the # of hits to whatever you want to and say yes when it asks you if you want them equally spaced. This isnt a fix but a workaroud. If you are seeing this on all different kinds of features from time to time and you manually create them on a part I am voting on a hardware issue.

[pernilla]

That is exactly how I've done it. I always want them equally spaced.

[Mike Ruff]

It looks like you are getting one bad hit on the entire feature... Is this hit off by the same amount as your prehit/retract?

[pernilla]

It could be one bad hit, or one hit that didn't take.. But why are there more than one red arrow in that case?

And is there a workaround?

[Mike Ruff]

There are more than one red arrow because of the best fit

[pernilla]

Here are some of the tests I've done. As I wrote before I've tried many different settings. This is a refix-plate with 24 holes, diam. 20 mm. My alignment is a simple plane, line, point.

Measured with 30 points:
1.jpg

Measured with 9 points:
2.jpg


Measured with 60 points:
3.jpg


But these are just 3 measurements, I have a lot more and about 50% of them shows that all diameters are perfectly fine, with deviations between 0.003 - 0.007 mm (roundness).
The largest deviations are just as Mike Ruff suggested, about the same amount as the prehit/retract (1.5 mm) on roundness but on diameter 0.03 - 0.04 mm.

[VinniUSMC]

The most notable thing is that it appears to be random. If you run the same part with the same # of hits 3 times, I'm assuming that the holes that are wonky would be different each time. Then, you can eliminate program and part from suspicion. Watch very carefully as it runs. It should be noticeable when it takes a bad hit.

Also, note that it can happen twice in one hole (two times in your pictures). So, the problem seems to be with the probing system or machine. Have you tried running different speed/acceleration? Do you have a different TP200 you can try?

[pernilla]

I have seen it making a sudden stop before taking the hit, and in a few cases I've seen it skip that hit entirely and move on to the next. But even when it appears to take all the hits, I can still get these weirdly big deviations.

I've tried slowing down speed and acceleration but to no avail.

No, I would have to buy a new TP200 to try. And I would do it if I was certain that's the problem.

[JEFMAN]

Another question :
What is the diameter of holes in front of ball diameter ?

You can also construct a circle from circle hits, then filter it...

[pernilla]

Sorry, I don't understand the question... What holes? And by ball diameter you mean the probe?

How do I extract the circle hits to construct a circle that can be filtered?

[William Johnson]

It could be the TP200. I had one start taking air hits when rotated at certain angles. replaced it and all was good.
It took having Hexagon sending service tech in to say that he could not find the problem. They went thru everything , updated the controller firmware ect.. 2 full days here.
After they left defeated I discovered that when I put on my new RBE TP200 all was good again.

[pernilla]

Did your problems also occur randomly, or was it every time you rotated to those certain angles?

[JEFMAN]

Sorry, I don't understand the question... What holes? And by ball diameter you mean the probe?

You're measuring holes, so I ask about their diameter. And the ball diameter is the tip diamleter.
Sometimes, this kind of problems can appear if you use circular moves with a short difference between hole diameter and ball diameter.

How do I extract the circle hits to construct a circle that can be filtered?

Construct a circle (Constructed feature / circle), select the measured circle, (CIRC1 for example), add .HIT[1..CIRC1.NUMHITS], and select BFRE. Then, select OUTLIERS/ON,3.

[pernilla]

Now I get it. So the probe diam. is 3 mm. The holes I've been testing are 20 mm diam. And I have not used circular moves.

Thanks, I'll try that!
What does the number 3 set, at the end of your example?

[JEFMAN]

the outlier level : 3 allows to keep 99% of hits. You can also try 2.8, 2.5...

[pernilla]

Thanks for all of your great ideas. This problem is not yet solved but I got around it by using filtration and I have yet to test a TP200 from another CMM to see if that could be the problem.