What's new
What's new

What's everyone's preferred method of verifying 5 axis kinematic rotation point?

empower

Diamond
Joined
Sep 8, 2018
Location
Novi, MI
Our ibarmia has a cycle to probe the artifact and calculate the MRZP automatically, and i use it once a week or so.
i was wondering how some of you guys verify the alignment? machine a bore from both sides in a square, measure the mismatch in the middle with an indicator? examples would be sweet to see!
 
Ok. I'll Bite. On my setup, 2014 VF4SS with TRT-210 5ax. I just cut parts and check them on my CMM. Over time I have found a pretty good COR when the machine is mostly warm running parts. Who cares about aligned kinematics all I care about is green on my CMM report! :cheers:

I have done complex probing of an artifact and calculating COR and its completely worthless since my machine warms up and moves a few .001". I do know that my A axis and B axis are physically misaligned by .0005" from this probing routine so this is good to know I guess....

With a complex knutating head or some other weird kinematics I bet it gets even more complex to cut parts and comp things in as its not so straightforward as a simple 5ax trunion.

This is a very complex and deep rabbit hole to go down. Machining Motion and forces are different than taking points on an artifact.

If I had a nice a high end machine here's how I would establish machine kinematics. Mind you I am just spitballing here.

1. Grab Toolsetter Probe Calibration tool and set length with machine tool setter.
2. Warm spindle up for the required time and check for spindle growth. (Spidle Growth Measured and or accounted for!)
3. Grab Spindle Probe.
4. Mount Artifact to probe in easy to access spot.
5. Probe Artifact and establish location.
6. Use program to move axis throughout travels and check for repeatability and machine thermal growth.(Check Machine Repeatbility and thermal compensations!)
7. Use program to probe artifact for true geometric kinematic mastery.
8. Cut some parts. Put some force through the machine. Run for a shift or two
9. Repeat program to probe artifact for true geometric kinematic mastery.
10. Profit.
11. Back to step 1 when night shift smashes the machine.... :nopity::nopity::nopity:
 
The right way to find centerline is with a test bar in the spindle. On a table/table machine, finding the rotary center is really easy, just put an indicator on the table and find the spot in X/Y where the test bar is centered. Then finding the tilt center is like finding the center on a horizontal. You have to measure from the side of the bar, then to the tip of the bar.

It gets to be much more PITA with tables that aren't centered on each other, or a nutating head, but the principal is the same. I always have to draw it out and do some guess and check before it clicks.

I have found over the years that shops which find "center" by cutting test parts and checking them tend to have tool lengths that are way off, which turns into a total nightmare when trying to hold true positions or tricky 5 axis features. No judgement though - I have seen some amazing parts come off machines with the COR completely f---ed.

We just use 3d Quickset (automated alignment probing a ball) on our machines, but being the one guy who can actually find center on machines with suspect probing cycles (Haas and Matsuura in the last shop), has been really handy over the years. It's like being able to read G-code; not actually relevant most of the time, but a pretty critical base skill for somebody in the shop to have.
 
so here's what i was referring to. this is the part i cut to check/verify my MRZP (machine rotation zero point).
1709149083423.png


first off, i'll machine this feature and dial in my tool size until it measures perfect on size
1709149123261.png


then i'll machine this surface in 2 different orientations so they overlap, this allows me to measure how much my Z offset needs to be adjusted by
1709149174565.png

then these paths let me check and adjust my X offset
1709149222136.png

then these 2 allow me to check my Y offset
1709149256039.png

and finally i machine these 2 bores halfway from each end to see how well the bores line up
1709149288766.png


was able to dial in our ibarmia to ~3 microns mismatch following this method.
 
yes, so does our machine, yet i'm still having to make manual adjustments to bring it in closer.
I'd be curious to know if the cycle sucks, it's a probe issue, or you are chasing something else.

All of the above are potential issues. The reason I don't like cutting an artifact though is that you have potential for massive tolerance stack up. Putting an indicator on a testbar is pretty much the last word on locating where center *actually* is. Otherwise you don't really have a known entity that you are referencing against.

I do like your artifact though. It is well thought out, and definitely will provide much better results than the standard of just milling four sides on a block and adjusting until it cuts the right thickness across flats.

Who sells Ibarmia in your region? Ellison? I bet they have somebody who is really good at dialing that machine in by hand, and can walk you through the process.
 
One certainly does not cut a part to check kinematics. Use th3 bsr method. You said..'i'll machine this feature and dial in my tool size until it measures perfect on size' UFFTA
 
Last edited:
so here's what i was referring to. this is the part i cut to check/verify my MRZP (machine rotation zero point).
View attachment 430941


first off, i'll machine this feature and dial in my tool size until it measures perfect on size
View attachment 430942


then i'll machine this surface in 2 different orientations so they overlap, this allows me to measure how much my Z offset needs to be adjusted by
View attachment 430943

then these paths let me check and adjust my X offset
View attachment 430944

then these 2 allow me to check my Y offset
View attachment 430945

and finally i machine these 2 bores halfway from each end to see how well the bores line up
View attachment 430946


was able to dial in our ibarmia to ~3 microns mismatch following this method.
This is exactly what my calibration block looks like. Machine built in cycles never get it close enough. I have a spread sheet with formulas that take the measured test block data and output how much to adjust machine kinematic positions based on how far off it is.

Also, important to understand the relationship with tool length and probe calibration before going into any of this.

All the people here talking about manually calculating centers and all kinds of crap like that have no idea what it actually takes to bang out high mix 5 axis parts first go. It takes exactly what you are doing, making chips, measuring, adjusting, making more chips.
 
Last edited:
Also, this method you are using assumes 90 is 90 and everything between 90 and 0 is what it’s supposed to be. The Iba probably rocks that no problem but it’s not always a given and if you are chasing what seems like centers it could be angle issues. Always good to do a quick sanity check. Sweep table at 0, granite block column z, table at 90… all should be square.
 
I like to verify the machine kinematics with the probe and a calibration sphere after running the kinematic calibration cycles.

I basically probe the calibration sphere in a bunch of different rotary positions and measure the deviation between the calculated sphere locations and the actual measured sphere locations. Some positions have more deviation than others depending on the volumetric accuracy of the machine.

I find the accuracy of the axis center point calculation in the kinematic measuring cycles depends on how good the machine is and how the cycle calculates the center of rotation. If the machine has any tiny geometrical imperfections then the calculated center point will depend on the placement of the calibration sphere on the table and the rotary positions used in the calibration cycle.

Some calibration cycles calculate the center of rotation with the 3-point method. I prefer to use more measuring points and calculate the center of rotation with the method of least squares.
 
Base salary (40 hours)? Good for you. Seriously. You're a smart guy. That was first rate illustration and explanation there. You're easily worth that. It seems this industry does not want to match technically-minded problem solvers with appropriate pay.
yup, salary, 401k, full benefits, company stock options, 4 weeks of PTO plus bonuses. i love my job and have no plans on leaving any time soon.
 
I'd be curious to know if the cycle sucks, it's a probe issue, or you are chasing something else.

All of the above are potential issues. The reason I don't like cutting an artifact though is that you have potential for massive tolerance stack up. Putting an indicator on a testbar is pretty much the last word on locating where center *actually* is. Otherwise you don't really have a known entity that you are referencing against.

I do like your artifact though. It is well thought out, and definitely will provide much better results than the standard of just milling four sides on a block and adjusting until it cuts the right thickness across flats.

Who sells Ibarmia in your region? Ellison? I bet they have somebody who is really good at dialing that machine in by hand, and can walk you through the process.
speaking with HH apps guy, apparently they have 2 versions of kinematic calibration, we only have the basic one. not sure if thats it or not...
1709205943017.png
i know what you mean by tolerance stackup, but i THINK i did everything i could to minimize it.
the spindle probe was indicated to where my micron indicator was not moving at all as i spun the probe/ruby. then i verified that it probes accurately, verified the same with tool setter. then i started with calibrating the Z position of MRZP (length from b axis rotary axis to spindle gauge line), as this can and will throw you off when you rotate b to 90 to get your thickness cut. with that dialed in to within 2 microns, i then proceeded to verify x and y. it seems to work because once i dialed it in and cut the artifact again, the mismatch was almost non existent.

Ellison used to sell them, but apparently they split ways a few months ago, they still support it though. they dont really have anyone good with HH/ibarmia in this area, the best guy they have is in cali, i've worked with him before i moved to michigan, he really knows his shit. and we did have ellison fly him out here last year and he did a great job dialing in the geometry. i think the machine is pretty stable as is, 10-15 microns of movement in a big machine like this isnt unheard of.
 








 
Back
Top