3 Point Plane VS 4 Point Plane

[simcityguy]

Heres my problem. The part I'm working on has a flat machined surface, it 12x12 inch's. Theres cores about every 2.5 inches around the edge's all the way around. I need to check to see if the part is concaved in-between the cores so I picked up a 3 point plane in-between each core then did an alignment and set Z to the plane. I then took single points on the area I just set Z to and my numbers were all over the place. It was like they were going down hill from core to core. So Then I did a 4 Point plane and my numbers looked much better. Now I'm wondering if I did something wrong in the programing.
3.7 MR2.

[Matthew D. Hoedeman]

You should never use the minimum number of points required for a feature, it will always create a feature with perfect form, always, and you have no way to check to see if you have made a bad hit.

As for checking to see if the part is concave between the cores, I would measure a 4-hit plane, at the corners of the block, set zero to that, then go and check the points between the cores.

[tking]

A 4 point plane will best fit and when you align the vector will tilt to match it. Look at the 4 points of your plane, you'll find 2 at zero, 1 at plus x.x and the other -x.x (same magnitude opposite sign), make sense?

TK

[simcityguy]

A 4 point plane will best fit and when you align the vector will tilt to match it. Look at the 4 points of your plane, you'll find 2 at zero, 1 at plus x.x and the other -x.x (same magnitude opposite sign), make sense?

TK

Thats what I thought, just wanted to make sure. Someone here (coughing Engineer coughing Engineer) was trying to tell me different and wanted 3 point planes.

[Jan d.]

Depends on the accuracy you are looking for. If you are trying to find a few 0.0001", then you will be in a lot of trouble by just taking a few points. If you try to find a few 0.001", you can use very few points.

The issue is that your process has a certain uncertainty. It comes into every single hit and (presuming it is random in nature, I know, I know, that's not always true) your measurement can vary +/- that number. Typically, for the machines listed in your Profile, the uncertainty can be as high as 0.0001", maybe even higher (depending on stylus config, maintenance of the machine etc etc). So if you have 3 points, the uncertainty of the measurement will NOT be averaged! It will be in there full bore! The more points you take, the more averaging you get and the better you plane will be defined.

But then again, if you try to find a few thousands, who care if you are off by a few tenths.


Jan.

[#2#]

Thats what I thought, just wanted to make sure. Someone here (coughing Engineer coughing Engineer) was trying to tell me different and wanted 3 point planes.

I had to explain to an engineer one day that there was no possible way a cart with two wheels and one leg could rock. He wanted me to check to make sure.

[Hilton Roberts]

I had to explain to an engineer one day that there was no possible way a cart with two wheels and one leg could rock. He wanted me to check to make sure.

Great analogy!!!!!!

H

[simcityguy]

Ok thanks guys for the refresher, just needed to reboot my brain cells this morning.