Progressive alignment deviation

[Smallsvillanova]

Can anyone quantify the error one would receive when not properly doing a plane line Point alignment? For as far back as anyone can remember and from what I can see in our programs we have never done a progressive plane line Point alignment. The majority of our jobs are J Corner fixtures that are set up with a plane line Point alignment. During level 1 training I learned of progressively aligning with 2D features and i am trying to explain this concept to my colleague but I'm only met with resistance and being told that it doesn't even matter. Can someone help me better explain the sort of error that is introduced when not properly doing this type of alignment?

Thank you in advance.

[louisd]

You are both right, technically... Now end your pissing match and get back to work!
Plane
Align/Level
Line
Align/Rotate
Point
Align/Translate

is the same as
Plane
Line Point
Align/Level, Rotate,Translate.

UNLESS some of the following exist
1: Part fixture methods varies considerably. If this is the case, the progressive plane/level line/rotate, point/translate method is better, as it will locate each feature as it is measured. It will prevent nominal values from shifting on you further into the routine.
2: Part fixture method is dead-nuts consistent (good fixtures!) If this is the case, the plane/line/point align method is fine. your fixture location isn't going to move.
3: You don't have a tertiary datum available, or the part datum structure doesn't apply to this methodology.

[Smallsvillanova]

LOL believe me I would love for it to not be a pissing match. I was sent to level 1 training because I am being trained to be the programmer for the foreseeable future. I learned these principles in class and try to bring them back to work and put them to use only to be met with an argument with everything I suggest. In my eyes regardless of what he believes the error to actually be we should still be doing things best practice. Otherwise what was the point of learning this information? I'm simply trying to find the most reliable repeatable methods to do everything so I can rule those things out for later problems.

Additionally, I don't know what your standards are of consistent fixturing. Our fixtures vary in size and we don't have something like a set block that the fixtures get pushed against or anything like that we simply scale the edge of the fixture to the edge of the CMM table and write those values on the edge of the fixture and note them in the program then when the job gets put up the next time we scale the fixture that distance from the edges again and clamp it down. Is that sufficient?

[NinjaBadger]

If you do a manual alignment followed by a DCC alignment (of either type) any cosine error will be for all intents and purposes nullified.

[louisd]

yes, as long as you reach the intended goal of probing your part in DCC with vectors "normal" to the part surface,
and you effectively control all six degrees of freedom,
and you construct your alignment with level first, rotate second, and translate third
-your path to setting the initial alignment up, is absolutely moot.

[Smallsvillanova]

if I have consistent fixturing do I need a manual alignment? Like if I have a base with a J corner that I am putting on the table in the same place every single time do I need to run the manual alignment or can I just run the dcc since its in the same place is was last time, give or take a mm. I scale them from the edge of our table.

[Smallsvillanova]

That's great thank you. So what sort of error is introduced when you don't do an additional DCC alignment after your manual alignment?

[Smallsvillanova]

Louis what you mean by normal is that the vectors of my dcc alignment need to watch my manual vectors or i need to correct my theos to perfect?

[louisd]

If you do a manual alignment followed by a DCC alignment (of either type) any cosine error will be for all intents and purposes nullified.

Cosine Error is produced when your probe hit is not perfectly perpendicular to the surface ("normal to the surface"). It adds location error of that point, which equates to the result of pythagorean theorem delta between lines a and c, per the image in the linked calculator https://www.engineersedge.com/calcul...ution_menu.htm

[Smallsvillanova]

So that means i would need my theo vectors on my dcc to match the ones from my manual?

[Smallsvillanova]

sorry not to be a pest but could you provide some clarity on what you mean by normal? And training I was told that we were supposed to be correcting are vectors to theoretically perfect. Are you saying that I should not be correcting the vectors from the machine off of the fixture during my manual alignment, and that my DCC alignment should match those vectors?

[louisd]

Your manual hits will always be MEASURED with non-normal/have cosine error, to some extent. There's nothing you can do to improve this for your manual initial alignment hits. You can program your initial manual alignment hits with CAD in "Program Mode (CTL+F4, I believe as a shortcut)" but this will only locate your theoreticals relative to CAD when you probe those initial points. The MEASURED values will never ever be perfectly normal. that's the point of re-aligning in DCC, to isolate the MEASURED manual-hit cosine error. The measured values are what they are. don't override/correct them.

When in DCC for your initial DCC alignment, your THEORETICAL vectors need to be derived from CAD or typed in to be theoretically exact, relative to your part manual alignment. The measured values are what they are. don't override/correct them.

A normal vector is whatever IJK value is extracted to be perpendicular to the probed surface. if you have a flat plane, that's parallel to your A datum (Z level) your normal vector is 0,0,-1 (or 0,0,1 if it's opposing A datum). if you have a plane on the left its -1,0,0. These are normal vectors. If you have a non-planar surface, your vectors will be organic in value.