A question of methods

[Homer-Jay]

I need some advice. I am just getting to know the program well enough to run the laser and the probe functions. I am having trouble getting my hands on cad files to use as reference models for machined parts. these parts regularly require tolerances of .001in. are reference cad files 100% necessary to accuratlely complete dimmensional layouts with the probe and laser. and if not which is the best way to go? i realize this is a fairly open ended question,and appreciate any feedback i can get.

[Admin]

Are you running into a problem getting CAD files?

Having CAD files makes the process 10 times faster/easier in my opinion.

[Homer-Jay]

yes, we have been making the same castings for a billion years. we are just recently machining them ourselves. on my green castings i was using the compensation point method. i would just pull the probe away as close to normal to the surface as i could, and reliably stay within .008in, which was fine for my purposes. but now that we are machining, this is of course, unacceptable. i'm going to go back to basics and make sure i'm calibrating the probe correctly, and doing the length checkout procedure. my next step is to start compensating to a plane. i'm not sure, but i believe that this will increase my accuracy over compensating by pulling the probe back.

[Jeff]

One thing you need to remember is the amount of error in the arm itself depending on how good the last calibration was. This error could be up to .001 in itself. The other thing you may want to try is a small plane using a compensation point in that area.

[Homer-Jay]

i could live with .001"    i'm still stuck at about .006 out, on a 1" gauge block. in the options menu, under compensation, using inverted seemed to work a little better when i probed the boundry of the gauge block. it improved my measurement by about .0015    i believe i am learning the tool, i am going to look very closely at my mounting surface, calibrations etc....

[Homer-Jay]

i re-did my caliibbrations and made sure they looked good, and changed the angle i was holding the probe. i was doing it wrong i guess. i was coming perpendicular to my measured surface, but when i come more parrallel like a bridge or gantry cmm would, i am getting measurements that are 1.0004 on a 1" gage block. if i have embarassed myself, then all the better if it will help someone else avoid similar mistakes. my next challenge is that it is easy to establish a coordinate system on a gauge block, not so much on my parts fun fun

[Homer-Jay]

actually changing the orientation of the probe greatly improved the accuracy of a point to point measurement, but when i probe a boundry, i am stilll .006" out on a 1" gauge block

[Homer-Jay]

i have a cuture of people who have given me a tool and expect results. there is nothing wrong with this, but i believe this software was meant to run with reference models, and i cant get them. at least until we get solidworks and i create them myself from our customers 2d drawings. it is frustrating and i appreciate the advise on ways to work around these issues

[jrayself]

Did you make up to the Hexagon class yet Homer-Jay?

[Homer-Jay]

not yet. was going to octobers, but got pneumonia. i'm planning on november

[Homer-Jay]

btw i figured out what i was doing wrong on my point to point and boundry measurements. or at least a way to get good measurements. if i measure the boundry of a one inch gauge block laying flat on the inspection table, i have to go off the left side of the ball for the right side, the right side of the ball for the left side, the front of the ball for the back....etc. in other words if i hold the probe in one position as i measure the boundry, i come within a thousanth. if i go around the boundry and pay no attention to how i'm holding the probe i can only get within like 6 thousandths. same thing for a ring gauge, except i was within like 6 ten thousandths. i am happy.

[Jeff]

Under Probing options, Device, Compensation Method what do you have the compensation method set to? It should be compensation point.

[Homer-Jay]

i have been using the compensation point setting to take linear dimmensions, and the probe orientation setting to do diameters.

[jrayself]

You sir, have discovered the issue with 6 & 7 axis manual cmm arms.  The more you rotate each encoder (each joint/elbow) the more error you can introduce in your measurement. 

If you don't HAVE to rotate or bend a particular joint to measure a feature, don't do it.  There are bad techniques in hard-probing.

Your deviations do seem a little high though.  I'm still leaning towards a calibration issue.  I haven't worked with Romers much but the Faro Arms calibration routine can be very misleading.  It will tell you the calibration passed, even if you cheated it or did something a little funky.

An example would be to calibrate with the arm's elbow to the left, then to measure a small part, like a gage block, with the elbow to the left and to the right back and forth over and over.  Your numbers would be all over the place from the error introduced by unnecessarily rotating every encoder.

[Homer-Jay]

i'll stay on task with checking calibrations. i'm somewhat inclined to live with the .001 error i have whittled it down to. i am generally seeing .0006 on a 2" ring gauge, and .001 on the boundry of a gauge block. i am inexperienced with cmm's and have been fighting this thing tooth and nail to learn the software, develope some type of technique, and develope a system to save my layouts for traceability. you can imagine how pleased i was when after months of having error of .006 to .01, the machinist comes over, holds the last joint of the arm still as he's measuring, and gets a perfect measurement. i felt like a moron. lol live and learn. oh and to answer your question about infinite arm calibrations, it comes with a length bar and you use it to check your arm. it doesn't tell you pass or fail, but gives you numbers in the manual about range/2 and standard deviation values you can accept. i havn't explored everything there is to explore in the calibration department though. i will undoubtably learn some new things.