Uncertainty in finding axis locations for 5-axis

[Bob W.]

I am setting up a probing routine to determine the axis locations (B and C) for populating parameters for WSEC. This will be done via macro every time a pallet comes into the machine when justified by the job requirements. I was curious if anyone had done tolerance analysis or statistical studies on this sort of thing to determine how accurately they can be located based on probing measuring tolerances, positioning tolerances, probe repeatability, etc...

 

My machine is a horizontal with a rotary on the B-axis so the C-axis physically comes in and out of the machine during pallet changes. I am trying to eliminate the C-axis shifts due to pallet change repeatability, etc... I also want to track the B-axis movement due to thermal changes in the shop, machine condition, etc...

[Guest]

You just probe a known position and/or angle write to the WSEC variables and be done with it.

[crazy^millman]

Dump the values to a place on the control or dumb them to an external device or something. SPC them and then you have what you are after. People doing SPC to check for machine maintenance do other things to help track potential machine problems.

[Guest]

In your "Common to Lathe System/Machining Center System User's Manual" around Chapter 15 in the "Compensation" section the exact variables to write to are there.

 

Remember, you can comp for all 6 degrees of freedom (3 linear and 3 rotary)even though you only have 5 (3 linear and 2 Rotary). The control will compound the two rotational axes you do have and create the axis that you don't physically have.

[Bob W.]

Remember, you can comp for all 6 degrees of freedom (3 linear and 3 rotary)even though you only have 5 (3 linear and 2 Rotary). The control will compound the two rotational axes you do have and create the axis that you don't physically have.

I was aware of that but the C-axis moves required to account for A-axis skew are so large it would make me really nervous. I don't believe I could get an accurate simulation from Vericut. I have done some sensitivity studies in CAD and where the points are measured to determine the axis locations have a pretty big impact on the resulting accuracy (4X). Measuring the tooling ball in three locations with a total span of 120 degrees results in 4X the error compared to the same measurements taken at a total span of 240 degrees.

[Guest]

Simulation programs (i.e. Vericut, ICAM, etc...) are not good at estimating what the machine is really going to do when WSEC is active. They generally do a great job with the linear errors but the rotary stuff is an entirely different beast.

[crazy^millman]

Simulation programs (i.e. Vericut, ICAM, etc...) are not good at estimating what the machine is really going to do when WSEC is active. They generally do a great job with the linear errors but the rotary stuff is an entirely different beast.

 

That is where the control emulators have to start doing the heavy lifting.