Bolt Hole True Position

[Gail Plush / Amcast Automotive]

This might not make sense, but here is what I have (had). In the past I had a bolt hole true position program. I know this seems simple, but the program was written in DOS, and I could never decompile the program to figure out how this guy made his calculations. It was more than simple 2D true position. I think he made some sort of radial adjustment to make the holes fit in the "perfect" circle (if that was the problem. This program also output a "scatter plot" that showed you where the holes were inside of the tolerance band. It also gave you an angle and magnitude of the entire bolt pattern. Let me back up, you type in the 2D position dimensions that the CMM gives you, bolt circle diameter, and tolerance, and it spits out the angle and magnitude of each hole, and the entire pattern. What this does for a simpleton like myself is, tell me what is bad, i.e. one hole, or the entire pattern. You can get similar information off of the CMM, but this is much cleaner and user friendly for people like myself. Our QA department upgraded to XP SP2, and guess what..... Does anyone have anything like this? [email protected]

 

 

Thanks,

 

 

Gail

[Tom Szelag]

I don't have anything like that, but I could probably program it. Though it would be in MATLAB or something similiar which I doubt is all that common in machine shops. Or maybe over a weekend I could just write all the math down.

[Gail Plush / Amcast Automotive]

I would be VERY interested in seeing what you come with. Does my topic make sense? Do you understand what I'm after?

[Tom Szelag]

Pretty sure. If I'm right...

 

You take a part that's been made. Do some co-ordinate measuring on it to pick up the XY locations of holes on a bolt pattern. You enter what the position tolerance should be according to the print. Computer calculates the true diameter and angle of the physical holes, and then can tell you whether or not the whole pattern is OK, if just one hole is off, or if the whole thing is shot.

 

Correct?

[Gail Plush / Amcast Automotive]

Yes, He also does some sort of ordinate rotation to see if you can make them fit. I'm just not sure how. That isn't as important though.

 

 

Gail

[Gail Plush / Amcast Automotive]

I've reviewed it, and I think he is rotating the pattern, as I said before. Typically, when you measure the entire bolt circle, you would measure "hole one," and then calculate everything else off of that. I think he "finds" the pattern, and rotaates the "tolerance" circles to make them fit (if possible). From a fit and function standpoint, that is the right thing to do.

[betts]

Curiosity is killing me, so I gotta ask. What size are these holes, in what BC and what is the tolerance?

[Gail Plush / Amcast Automotive]

The holes are ~ 15-16mm in diameter, and the circle varies from 100 to 165mm.

[Gail Plush / Amcast Automotive]

Whoops..Tolerance of .25mm

[Gail Plush / Amcast Automotive]

bump.

[betts]

Forgive me Gail, I just don't live in a production world, or atleast not yet, so I don't spend much time on repeat processes.

 

Since we build molds (for the time being) most things are 1 to 12 quantities, smaller molds.

 

We are looking in to doing more production type work, as the mold building margins "ain't what they used to be".

 

I am assuming you are checking production run parts, and everything you and Tom are talking about makes logical sense, I was just curious and wanted to learn a little.

[Gail Plush / Amcast Automotive]

No problem. I think I've found the problem, but I have no idea how to decompile a DOS executable.

 

 

Gail

[DC Gorn]

You have a CMM operator telling you that giving you magnatude and vector of a feature (any feature) from nominal is hard? You need either a new operator or a new CMM.

 

All above should be doable in a wizard for your CMM tho you may have to do a program for each number on the BC.

[Tom Szelag]

Think I'll give programming it a shot anyway. Good excercise in least-error squared curve fitting and flow quality control.

[Gail Plush / Amcast Automotive]

DC, It is actually a little more complicated than that. Tom, I have the current program, if you want to see it. Shoot me an e-mail.

 

DC,

 

It will actually tell me whether the "pattern" is bad to itself, or bad to the original feature. As you can imagine (especially on old machines), this can be a very useful troubleshooting tool.

 

I have been looking for a way to decompile the executable. It was written in basic, in 1984!

 

Gail

[DC Gorn]

True computer geeks are found here:

http://www.sharkyforums.com/

 

You could use the Cad out to output true DXF info into a Cad system and review it there. But you need a sampling not just one cut.

 

If you have a decent CMM you can program in basic or C++ right in your CMM program. Why hand input all the info?

[Gail Plush / Amcast Automotive]

This needs to be done several times per day. It also needs to be output exactly as it is today. Not that I'm reluctant to change, but I am reluctant to task my people with any more to do or learn, at this time. I have their plates full!!! In the long term, we are evaluating methods to output to a script file, etc. My CMM programmers are extremely talented, so I'm sure they will get it. Now, just isn't a good time.