Adjusting CAD to Machine Axis

[MUlissi]

Hey folks,

I'd recently asked about setting up a part to run based on a readpoint or one point touch off.

I'm already familiar with the fact that it has to be squared up evenly, and I've got this working pretty well for the most part, but I've found another wrench in the gears with some of my parts.

If the Datums are lined up nice and neat on the X Y and Z axis, merely using Part setup in the Setup Options menu (F5) to tell it how the part is sitting on the CMM works fine.

Similarly, if I had any surface, features, or edge to line up along an axis, I could line those up somehow, but I've already had a few parts that don't play nice.

Is there a way I can tell PC DMIS to orient features along a machine axis after starting with a Readpoint, before I've actually measured anything, so when it moves from the readpoint, it can correctly get to each feature needed for the DCC alignment?

An example of what I've found is Datum B and Datum C are similar to the Before picture according to the CAD model and it's axes, but I've got the part set up like the After picture on the CMM, because it's the best way to orient and secure the part on the plate.

I know I can determine the nominal angle to any of the axes by plotting a line between datum B and C and checking its angle. I assume there's a way to offset this angle to instruct PC DMIS on where the part sits compared to the machine axis.
machineaxis.png

Thanks in advance, you guys are incredibly helpful!

[Matthew D. Hoedeman]

Have you tried adding a "bogus" DCC alignment after the read point? As in, use a BIG prehit/retract and points VERY close to your read point to make a first DCC alignment? I'm saying like 10mm prehit/retract and points within 10mm of your read point position, then once you have this 'bogus' (very rough) DCC alignment, go on to the real DCC alignment?

[MUlissi]

I knew that was an option, but I was hoping there might be something cleaner before I resort to that.

[Matthew D. Hoedeman]

if you can't 100% control part positioning, that is about the best way to locate. If you CAN locate, but can't 'make square', you can, in the alignment you make from the read point, add a ROTATE_ABOUT 'Z' axis, and give it the angle.

[MUlissi]

Yup, I'd mentioned that in the original post here too. I'd found that and looked around for a place to use it. I can't do a regular alignment with this, as it would pull the model out of car body alignment, but it could be used elsewhere.

[louisd]

I often have upwards of 60+ alignments in my routines. Sometimes 4 or 5 initially, just to get 6dof of the component really locked in. Whatever interim alignment you define, can always be recalled re-defined, or reverted back, after you get your desired measurement or result.

[NinjaBadger]

MUlissi - yes you can use this method and retain car body coords

There's a little known (but rather useful) setting in the Setup (General tab) called 'Force alignment in car body'

With this on, you can create a readpoint, origin on it (A1), measure your first circle, origin on it (A2). Do your theoretical rotation to measure the slot (A3). Measure the slot. Recall A2. You can the align through circle and slot and maintain car body coords. (Like an iterative would).
You CAD origin will remain where it is (ish - the circle you've origined on will be your datum (i.e. it will have zero location deviation) but it's nominal (and measured) values will be the theoretical values from the CAD.

[louisd]

It seems you are not quite grasping how to correlate the location of the CAD, to how it's placed on the machine, if i am understanding your concern correctly.
1- Open a new routine.
2- Import CAD.
3- Make sure the tetrahedron of the CAD relatively aligns with machine coordinates as it sits on the granite
----(seems like you have all the above figured out)-----
4- Then, click on "Program Mode"
5- Click a point in the graphic display window (offline/on the computer NOT WITH JOGBOX/MACHINE), approximately where you want your readpoint to be. now press "end" on keyboard a point has been made, you will delete this in a second.
6- Insert an operator comment, asking them to move to your readpoint location (describe where on the part!), and press OK when ready.
7- Press CTL+G to create a readpoint. Manually replace the readpoint theo's to match the xyz values of the point you just made above (or precisely where you want it relative to CAD).
8- Delete that first point.
9- Press CTL+ALT+A origin XYZ to your readpoint.
10- Save routine. execute. move machine to your first readpoint when the command pops up.

Now, to program your initial alignment
11- Change your mode to MODE/DCC.
12- Add a clearplane command (clearp/zplus,5") should be prudent
13- Add a move/clearplane command.
14- still in program mode, and still clicking on CAD virtually within your graphic display window, click your alignment features/points to start your first alignment (NOT WITH JOGBOX/MACHINE), pressing "End" to construct each of your features.
15- Align to them Level, Rotate, Origin (or iterative)
16- Save routine. execute.
The CMM has now perfectly located your part, and has placed the CAD model precisely on it. Try programming your routine from here.

[Kp61dude!]

I used to use this a bunch back when I was programming for operators. You can create a bogus alignment like it was mentioned before but before I ever wrapped my mind around unlimited alignments rule I used to:

1. Import CAD
2. Translate CAD to tryhedron
3. Import virtual CMM
4. Move CMM to CAD.
5. Kicker. Tryhedron doesn't have to match CMM axis (usually rotated a bit off about Z axis). F5 | Part/Machine | Adjust... | Rotate about Z
6. Read point sequence.
7. Part routine sequence.

Difficult to explain in bullet points but hope it helps, that is if I understood you correctly.

[MUlissi]

How do you translate it to the tryhedron? That's something I'm unfamiliar with in PC DMIS.

I'm pretty sure I understand what it means, just not how to accomplish it in the software.

[Kp61dude!]

2. Translate CAD to tryhedron
This is your normal: Operation | Graphic Display Window | Transform
I probably just said it the unconventional way: "translate CAD to tryhedron" instead of the normal "move your CAD" or "shift your CAD".

[MUlissi]

Thanks for all the responses. I'm going to see how these work.

In the meantime, I've also found that I can use that previously mentioned angle in the Setup Options menu under Adjust, using the 'Rotate CAD about <X, Y or Z> axis:"

Worked like a charm.

[Kp61dude!]

Got your alignment working? Or rotating your CAD worked like a charm?

[MUlissi]

In answer to Kp61dude! , so that I can include a picture:

Adjusting the 'Rotate CAD about <X, Y or Z> axis' to tell PC DMIS how the part is aligned on the machine worked like a charm.

I had a part whose base plane and Datum B were on different planes, with the Datum B on a plane -30° rotated from Y.

I told it to rotate -30° about the necessary axis (I can't remember if it was X or Z at the moment) and was then able to put a readpoint in the center of the Datum B hole and run in DCC after that without any other adjustment.
.
In the following picture, the picture on the left is how the part was sitting in virtual space, and the picture on the right is how I set it up on the CMM.

In a more complex scenario, I could rotate it about 2 axes to level the desired plane, or at least get it level enough to pull off a rough DCC alignment to get it perfect for a final alignment run.
rotate CAD about X,Y, or Z.png

This particular part was small and simple enough that I could spare the time and there was low enough risk that I could experiment a little.

[Kp61dude!]

Good! glad I was of some help. Just remember the probe is moving about your -30° theoretically. You always want to account potential load errors eg. how far off you're from machine axis like ±5° or so. To test this just misload your part in increments of 1° or so until you see a misprobe, use extreme cation. If you're always the operator don't assume you'll be there forever... build it dummy proof for the next person is a pretty safe philosophy to stand by.

[Quality ish]

Have you tried adding a "bogus" DCC alignment after the read point? As in, use a BIG prehit/retract and points VERY close to your read point to make a first DCC alignment? I'm saying like 10mm prehit/retract and points within 10mm of your read point position, then once you have this 'bogus' (very rough) DCC alignment, go on to the real DCC alignment?

^^^ This^^^.

It's not really a "last resort" thing. NOBODY is going to be able to square a part up perfectly every time. This gives you the best chance of making sure the program will run without an "unexpected hit" or a failure to hit within the expected distance.

I'll often just take a small sample of hits within a couple inches on all three planes in relation to my readpoint, and align to that. Then the CMM has a better idea of where the hits are supposed to be.

[MUlissi]

Good point. I've been thinking primarily in terms of people having the good sense to use the documentation I'm creating (detailed pictures and comments) but I guess there is a good chance someone at some point might just try and wing it.

For what it's worth, everything I've been working with since having started utilizing the readpoint to start a program has had Datums that are pretty close together, so it doesn't have to travel a long distance, increasing the chance of deflection from an unexpected or miscalculated angle.

I'm currently working with a part that required 3 angle corrections to tell PC DMIS how it's sitting on the CMM, thanks to it's car position. I've got it going after having miscalculated an angle the first few times, and accidentally correcting an angle I didn't realize I had to correct at first.

[BIGWIG7]

have you tried using 2 readpoints?