What's new
What's new

On Machine Probing

kayrate

Plastic
Joined
Jul 11, 2008
Location
Alabama
Just thought I would start this thread to get some opinions using machines to buy off parts before we remove them from their fixtures. I've been doing some extensive testing on our machines and we are nearing the required level accurracy Vs repeatability. Our machines are quite large so it doesn't make sense to use a CMM, parts are too big. So the next best this was to use laser trackers but we are finding we can get as good reults or better using the machine and a 3rd party software that takes over the controller.
Anyone else using this scenario?
 
Not using that scenerio, but there is a trend I've noticed of using a cmm
program like pcdmis in your machine computer, with renishaw probes to measure your parts while on the machine.

.02
MJB
 
Probing on the machine

We are using a version of DMIS+ as Siemens calls it. The machines are qualified to run production parts and then certified using 6 sigma approach to allow us the confidence to accept the data we receive. The cycle time is greatly reduced as well as utilization of resources for inspection. Although we are still in the "project stage", we are quickly moving forward. It's hard to argue with the numbers.
Just wondering if others are using this approach.
I feel I'm working on an island so far, not too many folks believe you can use the same machine to cut the parts and then inspect with it as well. Next step is to move to Renishaw’s new strain guage machine probe system to see if we can get better numbers. So far, we are seeing less that .0005 repeatability and less than .001 accuracy. If anything we are learning more about the machine capabilities and have good confidence that they are performing as they should prior to running production. I've included a machine volumetric and axis check program to verify everything is working properly before running production. It checks all 5 axis as well as the working volume.
 
kayrate
I don't have the application for it but always though it would be great on large stuff like you said. How do you check the working volume of the machine?
Sorry about the "no clue" question.
 
I'm using (3) 2.0 dia Baltec spheres, 2 at the bottom of the volume and a third at the tallest dim we will be machining and measuring. Then we get a base line and compare that to data collected by a Leica laser tracker. We also are using an artifact that was measureed on our CMM and compare data to that, although the artifact is no where near the size of our parts we will be inspecting.
As I said, we are just now getting comfortable with the repeatability of our axis and volumetric checks
Sorry for the long pause in the reply.... Vacation in Jamaica!
 
Maybe you guys can answer this relating to probing. A number of shops here probe parts after machining. Results are always fantastic- parts are 'perfect' within 1 or 2 microns for hole positioning and size for example (these are small watch sized parts).

Often however, when assembling, rather significant errors are sometimes evident and when checked these errors can be .01-.04mm for a tolerance callout of +- .002mm. That's a lot!

Seems to me if the part is probed on the machine that made it, in the same setup, any errors in precision inherent to the machine will be duplicated, giving essentially perfect results. Would make more sense to me if the part were rotated 180 degrees and then probed. If suggested this to a couple of shops but get a blank stare.

Could very well be that I have no idea what I'm talking about :crazy: , but to me it makes sense. I have zero CNC experience, never used a CMM; I just know that the protoype parts made on my jig borer often seem to go together better than the production parts that follow.

The machines are all late model Swiss or German high precision CNC's specifically designed for the small work involved.

Sidenote: I'm not looking to discuss why the machines are not holding the tolerances- just the technique in probing which seems to be blind to the errors.
 
Maybe you guys can answer this relating to probing. A number of shops here probe parts after machining. Results are always fantastic- parts are 'perfect' within 1 or 2 microns for hole positioning and size for example (these are small watch sized parts).

Often however, when assembling, rather significant errors are sometimes evident and when checked these errors can be .01-.04mm for a tolerance callout of +- .002mm. That's a lot!

Seems to me if the part is probed on the machine that made it, in the same setup, any errors in precision inherent to the machine will be duplicated, giving essentially perfect results. Would make more sense to me if the part were rotated 180 degrees and then probed. If suggested this to a couple of shops but get a blank stare.

Could very well be that I have no idea what I'm talking about :crazy: , but to me it makes sense. I have zero CNC experience, never used a CMM; I just know that the protoype parts made on my jig borer often seem to go together better than the production parts that follow.

The machines are all late model Swiss or German high precision CNC's specifically designed for the small work involved.

Sidenote: I'm not looking to discuss why the machines are not holding the tolerances- just the technique in probing which seems to be blind to the errors.


Renishaw gets around this in their factory by using "artifacts" in the machines.

Say you're machining part number 123. Well, in the machining area, you need to have an already pre-machined copy of part number 123. One that's been verified on a CMM, and guaranteed to be good. When the machining cycle is finished, you check artifact with the probe, and compare the newly finished part against the artifacts already known dimensions. This way, you're not checking the machine's accuracy against the newly machined part, you're checking a proven part against the newly machined part.

The system works pretty well once the bugs are hammered out. Virtually every Renishaw product in the world was made and inspected in this manner.
 
Omp

Yes, an artifact comparison is the best way, I agree. We use a (now) Siemens product called Valisys (old name) that only uses the controller for movement only. It uses DMIS to run the machine like a CMM and we are getting great results so far. Really nice since we can input GD&T callouts and recieve GD&T data. Our probelm with the arifact rule of thumb is, our parts average 5m dia and the forgings cost usually $60,000.00 plus (US) so we have to use other methods besides the artifact. By doing a volumetric check and an axis check to verify all is working correctly we are trying to cut out the variables. Right now we are using a Leica Laser tracker to compare data, along with six sigma techniques to shw we are more accurate than the laser tracker.
 
Screwmachine - 1st, yes, if the machine has some systemic error that is not accounted for, it will mislead in exactly the way you describe.
Hence the use of artifacts.
I don't know that you actually need a 2nd part, rather you need a 2nd thing of dimensions vaguely related to the part, with known properties in the right directions.

Also - temperature matters. Are the parts at 20C or some other speced temp when they are measured on the machine? Are they at 20C or some similar speced temp when assembled?

Finally - has anything else been done to them - heat treating? plating?

As for Kayrate's question - I would guess a number of shops that do large relatively low volume stuff in big fixtures are doing this, and mostly not talking about it (you do see it written up in the trade press from time to time.) High volume shops maybe don't want it in the cycle time. For some parts, other kinds of inspection may be required anyway.

The think I wonder most about is temperature consistency in the machine tool. Of course, if the artifacts and other references are of similar materials, you could likely correct for that.

Finally, CMMs aren't magic, they're just a different machine with different systemic errors and a lot of calibration work done on them.
 








 
Back
Top