Autocone not working as intended (I think)

[HxSwartwood]

Good afternoon,

I have attached a screenshot of said cone that is giving me issues. The issue is that, when measured, it angles this cone as shown in the screen shot. When it is measuring it however the machine is not measuring it even close to what this is showing. It is measuring normal to the part. I have an overall height that proves that the part is not weird as well. I have tried everything I can think of to get this cone to work, but no matter what I do, it gives me the same. I created this program offline and the cone lined up like it should then. I have checked for shanking or anything else that might cause errors. Below I have posted the code for the actual cone itself in case I am missing something simple.

SOLVED: Added more points per level (30 points per level) and the cone calculated correctly.

CONE_ID =FEAT/CONTACT/CONE/DEFAULT,CARTESIAN,IN
THEO/<0,0,0>,<0,-1,0>,1.5,-21.5,469.574
ACTL/<-0.213,0.004,-0.168>,<0.005916,-0.9995804,-0.0283565>,2.299,-33.214,469.769
TARG/<0,0,0>,<0,-1,0>
START ANG=-40,END ANG=220
ANGLE VEC=<1,0,0>
SHOW FEATURE PARAMETERS=NO
SHOW CONTACT PARAMETERS=YES
NUMHITS=7,NUMLEVELS=3,DEPTH=3,END OFFSET=9
SAMPLE METHOD=SAMPLE_HITS
SAMPLE HITS=0,SPACER=0
AVOIDANCE MOVE=NO,DISTANCE=10
ONERROR=NO,READ POS=NO
SHOW HITS=YES
HIT/BASIC,<184.984,3,-144.525>,<-0.7879432,-0.0130896,0.6156087>,<184.942,2.999,-144.494>
HIT/BASIC,<233.97,3,19.099>,<-0.9965995,-0.0130896,-0.0813517>,<233.611,2.994,19.068>
HIT/BASIC,<159.096,3,172.612>,<-0.6776739,-0.0130896,-0.735246>,<158.91,2.995,172.41>
HIT/BASIC,<0,3,234.748>,<0,-0.0130896,-0.9999143>,<0.001,2.999,234.758>
HIT/BASIC,<-159.096,3,172.612>,<0.6776739,-0.0130896,-0.735246>,<-159.243,3.002,172.775>
HIT/BASIC,<-233.97,3,19.099>,<0.9965995,-0.0130896,-0.0813517>,<-233.965,2.999,19.1>
HIT/BASIC,<-184.984,3,-144.525>,<0.7879432,-0.0130896,0.6156087>,<-185.194,3.003,-144.688>
HIT/BASIC,<-184.935,7.75,-144.487>,<0.7879432,-0.0130896,0.6156087>,<-185.132,7.754,-144.639>
HIT/BASIC,<-233.908,7.75,19.094>,<0.9965995,-0.0130896,-0.0813517>,<-233.909,7.75,19.095>
HIT/BASIC,<-159.054,7.75,172.566>,<0.6776739,-0.0130896,-0.735246>,<-159.199,7.752,172.726>
HIT/BASIC,<0,7.75,234.686>,<0,-0.0130896,-0.9999143>,<0.001,7.749,234.687>
HIT/BASIC,<159.054,7.75,172.566>,<-0.6776739,-0.0130896,-0.735246>,<158.854,7.745,172.349>
HIT/BASIC,<233.908,7.75,19.094>,<-0.9965995,-0.0130896,-0.0813517>,<233.536,7.744,19.062>
HIT/BASIC,<184.935,7.75,-144.487>,<-0.7879432,-0.0130896,0.6156087>,<184.891,7.749,-144.454>
HIT/BASIC,<184.886,12.5,-144.449>,<-0.7879432,-0.0130896,0.6156087>,<184.829,12.499,-144.407>
HIT/BASIC,<233.846,12.5,19.089>,<-0.9965995,-0.0130896,-0.0813517>,<233.449,12.495,19.053>
HIT/BASIC,<159.012,12.5,172.521>,<-0.6776739,-0.0130896,-0.735246>,<158.819,12.495,172.31>
HIT/BASIC,<0,12.5,234.623>,<0,-0.0130896,-0.9999143>,<0.001,12.498,234.637>
HIT/BASIC,<-159.012,12.5,172.521>,<0.6776739,-0.0130896,-0.735246>,<-159.162,12.501,172.687>
HIT/BASIC,<-233.846,12.5,19.089>,<0.9965995,-0.0130896,-0.0813517>,<-233.836,12.499,19.089>
HIT/BASIC,<-184.886,12.5,-144.449>,<0.7879432,-0.0130896,0.6156087>,<-185.078,12.503,-144.598>
ENDMEAS/

[RIDER]

Do you have an analogue probe? Can you try the other measurement strategies, like concentric circle scans? With this cone being such a small angle, you may need lots more points for the software to accurately solve. I can't know for sure. Just throwing ideas out there.
Is void detection turned off?
Your machine seems to be traveling to the correct Y locations and your vectors look correct. How does the data look for other features on the part using the same probe angle?
What size ruby are you using?
Was the probe angle calibrated?
What are the T-values of random vector points taken inside the cone?
Have you tried measuring a line up one side, paste with pattern around the part and constructed a cone using the lines?

No idea. Your code looks correct.

[HxSwartwood]

We do have a scanning head, but we only have 1 so that would lock this part down to 1 machine. If nothing else works I will definitely try this. I am going to try to put like 30 points on each level and see if that helps.

Void detetction is off (I have had so many fits with it that my first check lol)

We dont seem to have any issues on other dimensions. Everything else is constructing correctly and looking good.

5mmx75mm_CFprobe (we use it so i can reach down into that cone without shanking.

Everything is calibrated and correlates nicely

As for T Values I am not sure. I have never actually had to use those values for any of our parts.

Vector point cone: Yes this is usually how we measure cones due to issues with autocones in the past, but it did the exact same thing.

Thanks!

[UKCMM]

Your code looks good so in theory it should work, but there are a number of things to note.

The cone diameter is large and the wall looks thin so distortion could be an issue.

The cone angle is very shallow again distortion/form error will have a large influence on the computed cone if the distortion is large enough to
move the cone half angle of 0.75 to 0.0 or even reverse it.

The cone length is 21.5 but the measured length in your code is only 9.5 is there a reason ( probe shanking ), would recommend using the max
possible length and the largest tip ball to allow for physical alignment error and distortion of part.

[HxSwartwood]

Thanks for the suggestions! The reason I had the depth so shallow was I was getting even worse angling of the cone when going the full length. I am going to try to dump a bunch of points to see if the DEMON can resolve it then.

Thanks!

[dwade]

Using autocone sucks at times and when you end up having nightmares over it you need to check it with another method. Try measuring it with vector points and then construct the cone from those points and see what your results are. The DEMON works great most of the time but other times will cause you to lose sleep and drink copious amounts of alcohol trying to figure out why it isn't doing what you're telling it to do.

[HxSwartwood]

This is actually normally how we do cones. I tried this and got the same results actually. I'm slowly getting to the drinking over cones

[dwade]

Another method that I have found to work well is to use open linear scans moving up the angle then constructing the cone. Scans measure faster than single vector points.