Heidenhain 640

I am getting conflicting advice, does anyone know if you can combine a C axis rotation with a 3 point probing basic rotation for part alignment?

5 axis Fidia guy said:

I am getting conflicting advice, does anyone know if you can combine a C axis rotation with a 3 point probing basic rotation for part alignment?

Click to expand...

Is that on a Hermle ?

I assume you are talking 5 axis (three point) not 4 axis ?


I should be getting some answers to these sorts of questions by the end of the week (hopefully); there are scores and scores of "lurkers" that actually know the answer to what you are asking here.

[That discussion about what a 530 control can or can't do on that GF machine was an eye opener (allegedly); made me feel like I'll never "Trash talk" a Fanuc control ever again (in spite of awkwardness it can actually "Do" stuff that is says it can "Do" "On the tin", and more besides. Even HAAS can do 5 axis in process probing / WPEC etc. ].

That is on a Hermle, during training they say it can be done, however watching a youtube tutorial the guy very specifically said it won't work. It would be the "3 point" probing option to skew a workpiece. Can it be done with a C axis rotation already in the preset table would be the question.

5 axis Fidia guy said:

That is on a Hermle, during training they say it can be done, however watching a youtube tutorial the guy very specifically said it won't work. It would be the "3 point" probing option to skew a workpiece. Can it be done with a C axis rotation already in the preset table would be the question.

Click to expand...

Mathematically and digitally it can, but trying to find the rationale for why not (without context) is going to be interesting.


I'll let you know if I manage to square the circle on this and maybe put the question to the Germans directly.

Ok Thanks !

Eric

Mathematically,

If you only use X and Y coordinates for work piece orientation (like rotation and translation) then two points would be the minimum to yield a solution. In such a case a trunnion or other would have to be set into a 0,0, orthogonal position.


HOWEVER if you have the ability to sample more points in three dimensional space with XYZ values then you certainly can locate position and orientation of the part in 3d space. I.e. when a trunnion and C axis is tilted/ rotated.


I can show the math for that + methods that produce more accurate results using more redundant points rather than a "Minimum" solution. [There is a specific mathematical technique to do that and converge on the correct transformations from what are known as three dimensional affine transformations]. (I built software around that for photogrammetric determination of position and orientation of camera scenes from control points to extreme accuracies.).



It will be more of case of WHY Heidenhain (allegedly doesn't do that... ) I assume their engineering "Math" is as good as mine.


Maybe the training statement / youtube video was out of context/ minor mistake / mis-speak ?

There was a German video and also an English video, the presenter was clearly German and maybe he misspoke, I rewound it several times to make sure I heard him correctly.

5 axis Fidia guy said:

There was a German video and also an English video, the presenter was clearly German and maybe he misspoke, I rewound it several times to make sure I heard him correctly.

Click to expand...

I think it's this...

Two points with only XY coordinates then trunnion and C axis best at known 0,0 angular position.

IF you have three XY points (control points / probed points ) then you can rotate the C axis but Trunnion has to be at it's non tilted position.

IF you have (minimum solution) three XYZ points in true 3d space then you can locate / determine position and orientation of a part in 3 dimensions with a tilted trunnion and rotated C axis*. (mathematically / digitally that is possible).

Having a mini tomb stone on a tilted trunnion where you have to locate second ops parts is not that unusual and I find it hard to believe that HEIDENHAIN does not have a provision for that.


_____________________________________________________________________________________________________________


* Notes: (like full three dimensional WPEsC, (Work piece error setting compensation in three space) BUT better to use more points and redundant points well placed to yield a stronger solution that grants greater positional certainty**. (I think sometimes these things fall flat when folks use more traditional points to pick from their parts that yield almost indeterminate/ weird and wild solutions versus what the "Math" would like. ).


For using redundant points more wisely to gain better accuracy one of the challenges for that would be to write and create a nice interface that is easy (for anyone)to use to accomplish that (otherwise things could go pretty haywire/ unstable mathematical solutions)… That might be why in some cases these controls keep things simple but I feel sure that Renishaw has done something in this direction ?

__________________________________________________________________________________________

** Mathematically special techniques have to be used as normally there is not a direct (stable) solution, so (what is called) iterative least squares techniques are used using close cousins to what are known as linear regressions.. However I have been to academic/ technical / high tech conferences where someone will present a paper and it's clear they are trying to re-invent the wheel (unknowingly) and also do it really badly and not use the proper techniques... Sometimes stuff just doesn't cross into different disciplines and remains very specialized. Get's a little embarrassing / awkward... Like Yeah that was solved in 1956 and (properly). Admittedly some of these thigs are a dark art.