What's new
What's new

3d probing on machine

Milling man

Hot Rolled
Joined
Aug 6, 2021
Location
Moscow, Russia
Yesterday I had a discussion with colleagues about probing 3D surfaces on milling machines. Obviously, when examining hard surfaces inclined in two planes relative to the base planes of the machine, difficulties arise.
- the trajectory of the probe must be perpendicular to the surface at the point of intended contact
- the probe must have multiple calibration points and be calibrated on a sphere
- due to the many calibration points they should probably be stored in a file rather than in macro variables
- after the probe is triggered, the coordinates of the touch point must be calculated based on the specified angle of inclination of the surface and the base of calibration points

So, I started looking.
1. Renishaw. I didn't see any way to touch normal to the surface in their standard loops. It seems that this feature is available for the SPRINT scanning probe.
2. BLUM. The ability to measure surfaces of any shape by touching along the normal is directly indicated on their website.
3. Heidenhain. I didn’t see this feature in their measuring cycles, but the necessary features are available in their CNC systems.

A colleague also talked about the PowerInspect OMV program - an excellent software solution that allows you to create measurement programs on a PC for any surface.

If any of you know anything else about measuring complex surfaces on machine tools, I'd love to hear your thoughts and experiences.
 
I've worked with both Blum and Renishaw on these sort of applications, but definitely more time with the Blum team. Of course, my talks with them are specific to the Speedio, but the gist of it is:

1- You need probe hardware capable of hitting at off-angles. Blum is happy to do this with their optical TC52/62 probes. Renishaw doesn't seem to think the OMP series probes are really accurate for this - they want Renigade (improved kinematics) or Sprint (strain gauge) probes.

2- For simple point probing, your machine needs to be capable of skip signal feed commands at angles. Give it a G31 X1 Y1 F100 and see if it will even run it and what kind of results it gives.

3- Blum/Renishaw need to have ported their more advanced macro package to your control/machine to make this a possibility.

In talking to both, they seemed more comfortable skipping right to strain gauge probes for this kind of application, but that might be because we are starting to see a growing demand for processing complex 3D printed parts where picking up datums on organic + rough surfaces is required. For prismatic parts, single point probing may very well be sufficient.

Strain gauge probing is a totally different beast - instead of triggering a skip signal, the strain gauge probes feed force and position data to a proprietary Blum/Renishaw receiver that processes it and sends it to the control. Siemens and Haindenhain controls are essentially full-blown PCs, so that box can pipe data directly into them and run custom written applications to actually do things with that data. More traditional architecture, closed controllers (Fanuc, Brother, Haas, Okuma - even if they run an embedded Windows PC for the HMI) need to have a 3rd processor integrated into them that does all the heavy lifting and processes strain gauge results into a program that can be loaded into the main controller... so this stuff gets complicated sort of quickly.
 
The problem in 3d probing is, that in order to define the probe approach trajectory (which must be normal to hit surface) one must know the measured part geometry. But the reason you are going to measure this part is that you don't know it's geometry. Therefore, in addition of being goemetrical, the problem is philosophical too.
There are some technics which try to cope with this contradiction, but each of those known to me induce some "per definition errors".
Conclusion: Not everything in this cruel world is "measurable".

Stefan
Cogito Ergo Sum.
 
This can be done with Renishaw Productivity+. There's a stand-alone version, as well as a version that's built into Mastercam (additional license required.) There may also be versions for other CAM software.

The Mastercam version also lets you do conditional flow control; things like cut a feature, probe it, adjust the tool offset, and skim the feature if desired.
 
But the reason you are going to measure this part is that you don't know it's geometry.
I think usually the situation is a little different. We processed the part on the machine, and we know its shape quite accurately.
But, in general, I somehow solved this funny problem using standard Reni cycles. It was necessary to measure the width of the wall of a part unfolded in the XY plane. First, I determined the angle with the first cycle, and then determined the width with a cycle of measuring the width at an angle.
But still, more often the angle of the normal to the surface is known with sufficient accuracy to immediately measure the surface.
 
Nothing wrong with what you're trying to accomplish. Most people asking this question are wanting to sample a bunch of points off of some contoured part. They want to digitize the surface and output those points to CAD. They aren't using it for inspection.

If I'm pulling points off of a contoured surface for construction in CAD, within 0.10 mm is plenty close enough for most needs.
 
For that application, probing straight down in Z will tell you where the center of the ball is at each touch. Offset by the radius of the stylus tip, and you get the original surface.

That's typically how simple raster probing works - when you have a 2d or 3d line/sheet of points that were all probed with the same vector, you can extrapolate the normal vectors along the profile. When you have that you can offset the probe radius along the normal vector to derive the actual surface.

A colleague also talked about the PowerInspect OMV program - an excellent software solution that allows you to create measurement programs on a PC for any surface.

FWIW, I have PowerInspect, primarily for CMM programming, but I have used the OMV module in the past and intend to do so again at some point in the future. When I get back into that hopefully I will have more to contribute to this conversation!
 
FWIW, I have PowerInspect, primarily for CMM programming, but I have used the OMV module in the past and intend to do so again at some point in the future. When I get back into that hopefully I will have more to contribute to this conversation!
Oh, this is wonderful! If it's not a secret, how much did this program cost?
 
Oh, this is wonderful! If it's not a secret, how much did this program cost?

No secret, but it was a long time ago and I did not move it to subscription so I am stuck 2018 release. However that's fine as autodesk have done basically nothing to it since.

From memory, I think I originally bought it around 2014 it cost somewhere around £7500 in 2014, with something like £2000 annual maintenance.
 
@Milling man
You mentioned where you were looking but I didn't see your specific application, so are you simply investigating at this point?? I've used standard and 3d probing in NX and 2d in Gibbs, never got around to trying it in mastercam. I don't think mastercam probing is native, it's Renishaw. Anyway, sounds like previous posts have you pointed in a good direction going with software. I was just about to say you are at the point where you probably should step up your game. Software is definitely the way to go if it's in your budget. If you're going to buy software then don't bother hand coding anything and set up your posts properly so you don't have to do any hand edits. I recommend reading the manual from whoever you go with (free downloads), the Renishaw manual is very good, shows excellent examples, and just about anything you need to know on the back end of what's happening. I would go through the plethora of cycles so you know what you can do with it, and pay attention to the all-important calibration cycles, for Renishaw I believe you'll be needing 9804 so you can probe angled and 3D surfaces. Also thoroughly read through the list of optional inputs. It's an easy read, go grab it.
 
You mentioned where you were looking but I didn't see your specific application, so are you simply investigating at this point??
Yes, for now it's just out of curiosity. Several times in my life I had to solve similar problems, and I did it without tricks - touching in the XY plane, constructing the true position of the ball in CAD and, as a consequence, the true position of the surface. It just took an awfully huge amount of time :(
for Renishaw I believe you'll be needing 9804
Its main problem is that it calibrates the ball only in the XY plane. And for correct 3D measurements you need calibration on a sphere using many points.
 
Yesterday I had a discussion with colleagues about probing 3D surfaces on milling machines. Obviously, when examining hard surfaces inclined in two planes relative to the base planes of the machine, difficulties arise.
- the trajectory of the probe must be perpendicular to the surface at the point of intended contact
- the probe must have multiple calibration points and be calibrated on a sphere
- due to the many calibration points they should probably be stored in a file rather than in macro variables
- after the probe is triggered, the coordinates of the touch point must be calculated based on the specified angle of inclination of the surface and the base of calibration points

So, I started looking.
1. Renishaw. I didn't see any way to touch normal to the surface in their standard loops. It seems that this feature is available for the SPRINT scanning probe.
2. BLUM. The ability to measure surfaces of any shape by touching along the normal is directly indicated on their website.
3. Heidenhain. I didn’t see this feature in their measuring cycles, but the necessary features are available in their CNC systems.

A colleague also talked about the PowerInspect OMV program - an excellent software solution that allows you to create measurement programs on a PC for any surface.

If any of you know anything else about measuring complex surfaces on machine tools, I'd love to hear your thoughts and experiences.
I do on machine probing all the time with the Hermle and Heidenhain control. I program the points within Powermill, after that I import the results in Fusion. The end result is pretty slick, you do however need a 3d touch probe to get accurate results, and not all probes are.
 
We tested software from Hexagon a while back which uses this concept. You pick the measurement points on a 3d model, then the software will first calibrate those vectors on a calibration sphere before measuring the points on the part.

It seemed to have a lot of potential, we did run into some accuracy issues but I believe they were coming from our machine rather than from the software or probing system itself.
 
I do on machine probing all the time with the Hermle and Heidenhain control. I program the points within Powermill, after that I import the results in Fusion. The end result is pretty slick, you do however need a 3d touch probe to get accurate results, and not all probes are.
@Milling man
Yep. i.e. Renishaw 400 & 600 series > 40 & 60 series. The former series are high accuracy and sensitive, better suited for 3D shapes. The latter series are standard accuracy probes, intended for cuboid and cylinder checks.
 
we use Centroid probes w/Mastercam to capture cylinder head port surfaces as well as 5axis inspection. Works fine. We also have a CMM arm using verisurf - which takes a little practice to learn to use.
 
I've only done this on OMP400 (strain gage) probes with InspectionPlus F-4012-0519-AW software with the O9821 XYZ (vectored) approach. We got values fairly consistent with the CMM reports. You have to break the connection between the machining work offset and the measurement work offset.
 
Last edited:
Something else I was reminded of as I was talking with someone a bit ago; in order to do probing on a 5-Axis (in addition to the hardware requirements), you'll need to set some things in O9744(REN*FCS*TO*WCS)...

#30=23(MACHINE ROTARY AXIS CONFIGURATION)
(EXAMPLE 23=B+C)
(FIRST AXIS+SECOND AXIS)
(FIRST AXIS A=1 B=2 C=3)
(SECOND AXIS A=1 B=2 C=3)

and
(************************)
N30
(FCS*TO*WCS*CALC)
(ONLY SET WCS FOR TABLE TYPE ROTARY AXIS)
(ALL TOOL TYPE ROTARY AXIS = 0)
#1=0
#2=#5024
#3=#5025

#5024 = System Variable for machine coordinate position for the 1st Axis
#5025 = System Variable for machine coordinate position for the 2nd Axis

I've recently heard of a machine that had it's tilt axis reported in #5025 and it's rotary axis reported in #5024. I'd recommend setting your tilt axis at a position and your rotary axis at another position then write them somewhere to check. i.e.
#104=#5024
#105=#5025
or for those that like system variable names
#104=[#_ABSMT[4]]
#105=[#_ABSMT[5]]

Set Parameters as follows (Double check with your machine tool builder/dealer to confirm these are safe to change);
#3106.6 = 0 (DAK) - During G68.2 or G54.4 Absolute Position display is Program Coordinate System
#5400.5 = 1 (LV3) - Rotates MACRO variables to be read in active coordinate system
#11200.3 = 1 (WSK) - System Variables #5061-#5080 can be read.

This may not apply to the OP's machine but I figured in case it did... it's here.
 








 
Back
Top