alignments

[gage guru]

I need the help of the experts. Years ago when I was learning PC-Dmis the person who did the programming(of castings) would do alignments and then go back and re-align the part. Sometimes the same alignment would be done 3-4 times. I just want to see if anyone else out there does this and if so - why as I do not remember the why or the exact method of this.

Thanks!!!

[JamesMannes]

Yes, I do the same thing. I utilize Iterative Alignments(1 manual, 1 DCC), and then I do the blue-print datum structure. If the blue-print datum structure has datum off-sets for its' planar feature I will then redo this alignment. Likewise, if the my rotation features are a -B- cored hole and a single point called out on the blue-print. It makes for a more precise alignment. In AVAIL(Pcdmis legacy software), this type of re-alignment was called "vector-delta" alignment. In Pcdmis you could loop the alignment with arguements for a tolerance the datums must fall within. This can be accomplished through the use of the iterative alignment methodology, but I have seen some problems in using the iterative alignment methodology that makes me shy away from using it with my datum alignment.

[Hilton Roberts]

In the old days of PJM, if one did an alignment of a part that involved the rotation of a line, if the had its origin in a hole, the center coordinates sometimes moved. One sometimes did another alignment to reestablish zeroes in the hole. Could your mentor have been a person who learned programming with one of the early DEA languages?

H

[george frick]

ALWAYS- realign the part in dcc mode. this is one of the "musts" of cmm programming for accurate evaluations. as far as 3/4 times though- that is redundant as far as i'm concerned unless they are "new" alignments in the program.

[DuaneSiders]

I have done this on intake, cylinder head and water pump castings to "zero" in on the Datum Target points. Usually only two aligments would do it. We machined our own castings and we wanted to be sure we were hitting the same targets as the bullet nosed pins on the machining fixtures that contacted at the Datum Target points.

[Matthew D. Hoedeman]

I always do a DCC alignment after the manual alignment. Depending on your tolerances, you should too. There are several reasons for this, with the biggest one being probe comp issues. Pcdmis will only comp in the direction of the vector of the point and if your part/fixture/whatever isn't square to eaach machine axis, then you can not count on your manual probe comp being correct. A plane will comp right, manual or DCC but a line will not. For example, you are measuring the top of the casting and are going to use it for level. Well, it is 10 degrees out of square to the machine (for the example). Well, the plane will comp correctly, but the line you measure on the front of the casting for the rotation will comp the ball to the MACHINE axis, not to the plane you just measured. The deeper your touches on the side, the more off they will be. You can get around this by changing the way Pcdmis looks at the feature, change WORKPLANE to FEATURE and use the plane you measured. It will then comp for the probe in THAT plane, not a machine plane (You should remeasure it after you change this setting). Now, if you are looking at 1/4" tolerances, this may not really matter too much, HOWEVER, it is a BAD habit to get into. If you get sloppy on something with loose tolerance, you will tend to stay sloppy on other things as well, without realizing it. IF you do a DCC alignment after the manual, then you are about 10,000 X better than the manual alignment alone. Also, you may want even more DCC alignments, depending on the part/casting and how 'rough' it is, it may take a couple to fine-tune it down so that it is right. And, ITERATIVE alignments should ALWAYS run at least twice (measure all always) as far as I am concerned. There are too many possible areas of error to just do it once.

[Jan d.]

I try to always write programs that require only 1 manual point. This is a tremendous time saver. From that 1 manual point I try to find a feature that gives me enough info to find the rest of the part (which is sometimes hard). So my first DCC alignments are always FAR off. I find that in those cases, 2 iterations in DCC are the absolute minimum. I use typically 3 iterations.

If, however, the datums are machined surfaces AND I have a fixture that aligns them repeatably to the machine axis, I have run with just 1 iteration to save time.

On the other hand, I have dealt with non-machined cast datums and found that sometimes 3 iterations are not enough: my vectors were still moving.

My rules are: how well are your datums defined (machined or not), how repeatable is you part loaded and what are the tolerances involved. Run a few repeatability tests and convince yourself whether you really need all those iterations, but be prepared to find that you actually may need them!


Jan.

[bw_bob]

I will always do al least 3 alignments. This is what your zero,zero is. We have some datum schemes that are very close together and i will do at least 5 alignments. The better the alignment is the better the numbers will be.

Bobby

[rebeldude]

I do 1 manual, and 2 dcc aligns, then I come back and verify the datums with a very small tolerance for verification for the customer

[Doug Wells]

I repeat DCC alignments on 'Cast' surfaces 3 to 4 times when not using a fixture to stage parts. Remember that cast surfaces are drafted and repeating the alignments will 'zero' in on the specific unique point referred to as the datum target. If you repeat an alignment on a casting (not the iterative) you can view the grahics screen and see how the points are moving around as the touches occur in different areas of the drafted datums. I repeat alignments until the variation along the datum direction is within the stated 'Linear Accuracy' of our CMM.

Hope this helps,

Doug Wells

[gage guru]

thanks for all the feedback. our parts are not machined on all surfaces and the holes that we have are not always true. I have some parts that are giving me some unique results and I have a functional gage that is not correlating, but when I plotted the data the CMM showed some error that made me question what is going on with the CMM. The gage is worse and I have some work to do there, but I need to also hone in on the why of the CMM error. It is one of those things that was stated ' we have always done it that way' and now I have to look into if that way is the best.

We always have done a manual and then DCC alignments. I am thinking we need to look into re-doing the DCC 1 or more time s to assure it is sound. Just means more work, but it will shed light on many aspects of our parts/data etc.

Thanks again for the feedback and keep on beepin'........

[kbotta]

I may have missed it, but what type of work is it? You mention holes not being true...and gage not coorelating to CMM.
Are you using GD&T in your Pcdmis program's??
Kev

[Underspec]

Manual alignments first then auto feature alignments if possible is how I usually do it.

[Paul Sarrach]

Manual alignments first then auto feature alignments if possible is how I usually do it.

How can auto features not be possible?

[STANLEY]

It depends on the casting. I do iteratve alignment with meas all always, then 2 or 3 dcc alignments to zero in the datums. Most castings are rough and have draft on them, other times i do just 1 dcc alignment after the interative alignment. It all depends on the casting.