Tip Qualification

[jon_dav_uk]

This may have been covered before but I cant find anything on it so here goes.

We currently have 5 CMM's in the shop, 3 with SP25M's c/w FR25 toolchange racks. On the machines with SP25 and multiple tips the qualification takes upto 2 hours due to the variety of parts hence the amount of angles and tips required.

I currently force all machines to qualify all tips daily which may be a bit overkill but I dont trust half of the operators who I rely on to load clean deburred parts.

As a halfway house and a method of reducing qualification time I was considering some kind of daily check and dependant on results either force calibration or not, does anybody else do anything like this ?

I know another factory who simply measure a point on a part with as many angles as can access it. If the deviation between all angles is less than a parameter they have set, they accept the machine does not require calibration.

Whats everyones thoughts on this ?

[Gregl]

Get a master part,

right after a qual measure it. and keep the results.

measure it the next day and if the results are within a acceptable limit you should be good.

[Matthew D. Hoedeman]

This may have been covered before but I cant find anything on it so here goes.

We currently have 5 CMM's in the shop, 3 with SP25M's c/w FR25 toolchange racks. On the machines with SP25 and multiple tips the qualification takes upto 2 hours due to the variety of parts hence the amount of angles and tips required.

I currently force all machines to qualify all tips daily which may be a bit overkill but I dont trust half of the operators who I rely on to load clean deburred parts.

As a halfway house and a method of reducing qualification time I was considering some kind of daily check and dependant on results either force calibration or not, does anybody else do anything like this ?

I know another factory who simply measure a point on a part with as many angles as can access it. If the deviation between all angles is less than a parameter they have set, they accept the machine does not require calibration.

Whats everyones thoughts on this ?

I do NOT like that one at all. I do NOT see how it can prove or dis-prove if the head was rotated to the column. I mean, if they are, as an example, measuring a point on the table with all the angles, then if the head were rotated in the column, that will not show up. I would want a 3-D feature, like a sphere, HEY, that's a calibration! If you are going to take to time to 'check' the calibration, you might as well calibrate! I don't trust anyone else either. I calibrate each day, but only as I come to those probes. So, if I am checking JOB#1 today, but not JOB#2, then the tips for JOB#2 will not get calibrated, until the NEXT time it comes up to get checked. I do NOT 'anticipate' up-comming parts. When they tell me they are coming up, I will calibrate the tips (unless they have been calibrated that day and no crashes/changes have been made.

[jon_dav_uk]

Cheers Matt I had the same thoughts regarding what the point measurement would prove just needed some backup for when I start my rants as to why we need to calibrate daily. The s**t some people throw on the CMM table in here then have the cheek to come in and ask why they have .050mm variation between measuring the part on two different days amazes me. I generally run my hand across the primary datum and see how much crap it gathers, turn to them and say there is your problem only to have them reply....ahhhh that won't matter. I need a beer

Whats your thoughts on a master part as suggested by Greg. I have stacks of scrap turbine blades which I could use as master parts but not sure whether the amount of features and the amounts of tips I would need to cover a complete check would be any quicker than a qualification

[Underspec]

I have a program that asks the operator which tip they want to check. I wrote it using the cal sphere itself. I am currently using only 5 angles per tip, A0B0 and the 90s. It spits out location and diameter of the sphere based on the qualified A0B0.

[jon_dav_uk]

Underspec I am presuming your Sphere is resident on the table for you to do this ?

Due to the parts I measure I need to remove the sphere every time after qualification.

[WayneKnazek]

Just a thought. If you want a fairly fast check, but a good one . . .

Write a program that finds a point when A0/B0 probe is placed over calib sphere. Operator positions tip above ball and hits. "Let'r Rip".

In DCC . . .

Tip finds top of ball. Sets Z origin. Move Z origin to minus 1/2 sphere size. Measure 4 or 5 point circle at approx half way (equator) of sphere. Set XY zero. Now you know about where sphere is.

Measure a 5 pt sphere at XYZ zero. Establish XYZ zero. Now re-measure a 9 (or 11) point sphere at XYZ zero.

Establish XYZ origin one last time.

Now measure a sphere at XYZ zero with each tip. It takes less time to measure a 9, 11 or 13 pt sphere with a scanning head than it takes to calibrate! (That would be for your TP20 module) Plus you can scan a shpere (for the other scan tips) for measurement in 12 to 15 seconds, instead of 3 1/2 ro 4 1/2 minutes to scan sphere in a calibration.

Measure with all tips. Determine tolerance tip can be out before needing to calibrate. Report
only OT dimensions.

If you get a report, all you need to do is calibrate the reported tips.

If you're not looking for "perfect" results, as in, if you're within 0.0005" to 0.0010" no need to calibrate, then . . .

Do the same as above, but just measure the sphere using 5 pts for each tip angle.

Instead of a 2 hr calibration, you'll have results in maybe 10 minutes.

If say, 3 tips need calibration, that's another 12 minutes. You just saved an hour and 38 minutes.

[jon_dav_uk]

Just a thought. If you want a fairly fast check, but a good one . . .

Write a program that finds a point when A0/B0 probe is placed over calib sphere. Operator positions tip above ball and hits. "Let'r Rip".

In DCC . . .

Tip finds top of ball. Sets Z origin. Move Z origin to minus 1/2 sphere size. Measure 4 or 5 point circle at approx half way (equator) of sphere. Set XY zero. Now you know about where sphere is.

Measure a 5 pt sphere at XYZ zero. Establish XYZ zero. Now re-measure a 9 (or 11) point sphere at XYZ zero.

Establish XYZ origin one last time.

Now measure a sphere at XYZ zero with each tip. It takes less time to measure a 9, 11 or 13 pt sphere with a scanning head than it takes to calibrate! (That would be for your TP20 module) Plus you can scan a shpere (for the other scan tips) for measurement in 12 to 15 seconds, instead of 3 1/2 ro 4 1/2 minutes to scan sphere in a calibration.

Measure with all tips. Determine tolerance tip can be out before needing to calibrate. Report
only OT dimensions.

If you get a report, all you need to do is calibrate the reported tips.

If you're not looking for "perfect" results, as in, if you're within 0.0005" to 0.0010" no need to calibrate, then . . .

Do the same as above, but just measure the sphere using 5 pts for each tip angle.

Instead of a 2 hr calibration, you'll have results in maybe 10 minutes.

If say, 3 tips need calibration, that's another 12 minutes. You just saved an hour and 38 minutes.

Wayne

Will have a try of this when I get the time and see how I get on. Only problem I see is where you mention I can only calibrate the tips that are reported as being out. I need to remove the sphere from the table every time I calibrate so this is not an option for me, its either all tips or nothing.

[Pat Ahern]

I worked for an areospace company making investment casting turbine blades and vanes of all sizes for 16years. Our 10 production cmms were Sheffield/Giddings & Lewis, running MeasureMax, with PH9 probe head and TP6 sensors. Every cmm program was correlated, with the correlation being documented, before being released to production (thanks Pratt, but also very valuable, as some programmers weren't as good as others). Every production program had a Master casting that would be run as the 1st and last piece. The cmm operators (not programmers) would use "Master" as it's serial number, sending the program to open the Master casting file which contained it's actual dimensions, and compare those values with the ones it just measured. If all measured values were within .xxx (typically .002 inch, but we could have used any value), a message would pop up stating the Master casting passed. The measured data would be collected and exported to the .csv data file with "Master" being it's serial number. This way we could double check (if needed) that the Master was indeed run, and we could do stat studies to see if the cmms were repeating. If the Probe tips were not calibrated (or still good from last time), the Master would not pass. Having the Master casting pass both the 1st and last run, meant the cmm stayed acurate (as acurate as we needed it to be at least) during the running of the production castings. Plus the operator could run the Master at any time they felt something might be wrong. Master still passes, it's all good, keep running production. Master failed, stop running production, something is wrong.
The use of Master castings did NOT replace calibrating. It was just another step to verify accurate data was being collected. Our systemwas to run "Calibration programs", that would only calibrate the wrist angles needed for the casting program that was to be run. Then run a "Fixture program" to verify that the fixture was good to use (no itterative alignments needed, the fixture tol was +/-.0005 inch on the 6 point nest). Then run the "Casting program", making sure to run the Master casting 1st and last. We ran 1000s of parts per week, 24/7/365. These cmms never stopped. Was our system perfect? Of course not. But I never got a call at 2:00am either. Lots of luck.

[jon_dav_uk]

Cheers Pat and everyone else for the info. Definitely have enough here to at least go and do some trials and select the method that works best.