Circular move trick??

[cmmguy]

I have been trying to make circluar moves work on plane measurements but have run into a stumbling block. I am 80% to having this figure out but am stuck and was hoping someone here can shed some light on it.

There have been posts in the past about circular moves only working on circular features, ie, circles etc... and wont work on planes. I have been measuring many circular face planes lately and many have been grooves which require lifting up out of the groove to move to the next spot. This works ok but is slow.

What I did was create a measured circle(not autocircle) using the diameter of the circular plane, and then edited the vectors for the touches to move down towards the face instead of towards the "diameter". This works good. The points on the plane are taken, the circluar moves work in the face groove, all is well so far.

The next step was to construct the plane from the points taken for the circle. I used the HITS[1..n] method and it calculated the plane but it is a ball radius from the correct surface. I tried both BF and BFRE and get the same results. I also tried constructing individual points from the hits and get the same results. I tried using the RAWHITS and then applying PROBECOMP but that didnt help. This is where I am stuck.

I could move or create another plane that is half a ball radius away, but that would defeat the purpose of keeping this simple. So this is just my little side project of figuring this out and was wondering if anyone had any ideas.

[Matthew D. Hoedeman]

Well, not matter what vector you apply to the individual hits, a circle will comp for the probe radially, not 'up and down'. As for keeping it simple, sound pretty simpe to me, all you have to do is make the offset plane for your final answer.

[djayne]

I would assume it has something to do with the vector direction of the circle definition. It is expecting a hit on one side or the other on your ruby but you are in fact taking the hit even with the shank of the stylus and PCDMIS is automatically adding the radius since that is what you asked it to do in the definition of your circle. - I think -

[djayne]

Wasn't fast enough - Matt beat me to it!

[cmmguy]

Well, not matter what vector you apply to the individual hits, a circle will comp for the probe radially, not 'up and down'. As for keeping it simple, sound pretty simpe to me, all you have to do is make the offset plane for your final answer.

Maybe it has been posted before, but how do I get the current probe radius?

[Tested]

I don't know if this would work or not but have you tried to do a polar vector comp?

[craiger_ny]

You could measure one point comped and one not comped (is that possible?) then do the math. I think Goodluck might be on the same trail.

[cmmguy]

I dont think that you can turn comp on and off in a feature. Maybe I can do the difference between rawhit and hit to get the comp value.

[craiger_ny]

Yes rawhit should be uncommped.

[#2#]

I have always just added points to my plane until the moves in between hits are short enough that they won't clip the part. I usually use autoplane for these types of planes and as you know you can't insert moves in between the hits on autofeatures. Of course, that does nothing speed up the process.

I've never tried circular moves outside of circles and cylinders. What about simply putting in move points at the same level (whatever prehit/retract is from the surface) but close enough together to avoid clipping the part? That would probably require quite a few movepoints.

This is one of the things they could add as an "improvement". For autoplanes with circular hit paterns an option to use circular moves would be nice. Of course as Matt would shortly start saying... They would be bound to screw up several other things in order to offer that option.

[#2#]

I have been measuring many circular face planes lately and many have been grooves which require lifting up out of the groove to move to the next spot. This works ok but is slow.

How have you been accomplishing this? Are you using movepoints or did you set up a clearplane and put a move/clearplane in between the hits?

[craiger_ny]

I use movepoints but I rarely ever use auto features at all. Learned features have a lot of flexibility and I have gotten so I can program with learned features without a part and without actually "teaching" them although I prefer to teach. I have found that using learned features without teaching to be buggy. Anyone else find it buggy? I find nominal hit values not being correct no matter what amount of editing I do (even if I edit the niminal hit value itself) it is not recognized for some reason. I should just use cad, it solves all and every issue using cad. I should not even be able to edit, just load a model and push a button................

[cmmguy]

How have you been accomplishing this? Are you using movepoints or did you set up a clearplane and put a move/clearplane in between the hits?

I normally use Move/Clearp inside my planes. and move up and over to the next point. What I am describing above is an experiment to try and use circular moves on these face grooves and like I said, I got it to work except for the compensation. 8 points on a circular face with circular moves around the groove to the next point, no intermediate moves. Works very cool. When I get the comp fixed I'll post the code. It is faster moving around the groove in a circular motion than lifting up, moving over, and then moving down.

I dont teach anything. I just type the code in.

[#2#]

I normally use Move/Clearp inside my planes. and move up and over to the next point. What I am describing above is an experiment to try and use circular moves on these face grooves and like I said, I got it to work except for the compensation. 8 points on a circular face with circular moves around the groove to the next point, no intermediate moves. Works very cool. When I get the comp fixed I'll post the code. It is faster moving around the groove in a circular motion than lifting up, moving over, and then moving down.

I dont teach anything. I just type the code in.

Yeah, I understand what you are trying to do. I've never tried it.

I have found though that using move points is quicker than using a clearplane. Even if moving the same amount. I have found with the clearplane, on my machine at least, that it moves up, pauses for a split second, moves over, pauses for a split second and then moves down. Inserting the move points to accomplish the same doesn't take as long because it doesn't pause.

Now you have got me curious and I'm going to have to try.

[cmmguy]

It probably depends on how many points you are taking. It is just easier to code the clearplane. The parts that I have been doing have a narrow channel and would require alot of move points to get around the inside the groove.

I just noticed how nice the circular move was and thought "what if". After all of this time, I just cant believe that there is no circular plane in PCDMIS.