To measure flatness of a plane is through a calculated result based ...

아래글은 3차원측정기 포럼에서 가져온 글입니다.

평면 측정 및 평탄도에 대한 의견을 주고 받은 것입니다.

참조하세요....

 

 

 Posted - 2/10/2005 11:00:32 AM     
--------------------------------------------------------------------------------
I have always been taught that the correct way to measure flatness of a plane is through a calculated result based on the average of several points on that plane (CMM). Now, I have a customer arguing with me that they are getting parts out of flat, and the way they are determining this is to take one point on the plane, establish it as zero, and then take several other points and dimension the location of each of those points from zero. If the point falls out of the plus or minus spec. which they had given to it, (flatness spec is 0.5, customer applying +/- 0.5 to point data), they are calling the part bad. I have never heard of this before, can anyone enlighten me on this, because I just can't see how you could get that notion from Y 14.5, plus, what if that point that was established as zero was not in the nominal position? I don't know, if anyone can shed any light on this, please feel free. I don't claim to be an expert on GD & T, but I feel like I know a little bit.

Thanks Guys.
 
kevinb

 Posted - 2/10/2005 11:33:02 AM       
--------------------------------------------------------------------------------
I believe its the diff. between the high and low point. That would be your form. As far as using the +/- 0.5 thing, "flatness" tolerance is as it specifies in the Feature control frame. There is no +, or Minus. JUST A TOTAL Zone (2 parallel planes sep. by 0.5mm) that all points(surface)must fall within.
I believe that's correct. (maybe not the wording
Kevin
Navin R.

 

 Posted - 2/10/2005 11:51:06 AM     
--------------------------------------------------------------------------------
What Kevinb says is correct. 1st off, measure a plane from the surface in question, and set your primary axis from that (not just zero point). The flatness is the highest point-lowest point. Depending on how large this surface is, you'll want to get a lot of points, not just a few.
Geoff P

 

 Posted - 2/10/2005 12:12:54 PM     
--------------------------------------------------------------------------------
Flatness is the condition of a surface having all of its elements (points) in one plane.

No datum reference is required. So if you have a callout of 0.5 then you’re looking at a zero to 0.5 tolerance zone. The calculation is a range between the highest and lowest point taken. Typically if the point you zero out is the mean then they are right at +/- half although not recommended. To zero out a random point on the surface you might want to apply +/- half the tolerance zone which would hold you closer to nominal. Let’s say you apply Zero to the lowest point. Now they are saying you can go -0.5 lower from an obvious highest point that may be 0.5 above your set Zero??? You allow a 1.0 range, makes no sense.
Click here for Flatness.xls
C.J. Andree

 

 Posted - 2/17/2005 2:32:00 PM     
--------------------------------------------------------------------------------
 I believe the best way to check flatness still is the old fashion way set your datum on a three point system and sweep a test indicator across the surface. This method should match the cmm results. The total travel of the indicator equals the flatness.
Justin Bishop

 

 Posted - 2/18/2005 2:58:44 PM     
--------------------------------------------------------------------------------
The way that I see it is that they are probing the face and then aligning. They then are using 1 point as reference. That is fine except they can't just zero any random point. They should zero the plane and then if they want to take points, just analize the spread between the points in reference to the zero plane. They can't zero any 1 single point. They would have to zero the average plane (I would suggest a massive amount of points). The way I see it, they are close but just not quite right
vlad

 

 Posted - 2/20/2005 9:11:12 PM     
--------------------------------------------------------------------------------
Flatness is tolerance of form. If you measure plane in three points, you are going to have a perfect plane, with form = zero! If you take four or more points, your sistem shall calculate you ''best fit plane'' and none of taken points has to actualy be in that plane, but they will be eavenly spread on both sides. Form of that plane will be your flatness and more points you take, closer to actual flatness you''ll get. In theory you can''t acheive the actual flatness because you will have to take countless number of points... This is the point where you have to compromise: find suficient number of points, depending on how important flatness of that plane is... If you decide to do it customers way, you are compromising way to much. Just in case, write a note that will say: that is done the way it''s done on customers request...
edited by - vlad on 2/21/2005 3:26:28 PM
Kevin Sipola

 

 Posted - 2/21/2005 8:30:03 PM       
--------------------------------------------------------------------------------
Also sounds like they're throwing coplaner in there also. Either that or they're doing it C.J.'s way but face up instead of face down. I've got parts for a customer that want .01 flatness unrestrained over 14" of 1/8 aluminum, 5" at it's widest point, near the middle .5". Oh and near the middle it's machined down to .040" thick. The CMM says it's within .003-.005"
It sags .005-.007" over three points. Duh! I'm checking wet spaghetti.
Jane Jiang

 

 Posted - 2/22/2005 12:20:10 PM     
--------------------------------------------------------------------------------
try to use Outer tangantial evaluation method for plane which needs meaure flatness.
Justin Bishop

 

 Posted - 2/28/2005 4:57:42 PM     
--------------------------------------------------------------------------------
I am having similar problems, but with parallelsim of an unflat plane to a flat plane. (see my post in general topics "parallelism") What is the the outer tangential method?
Miguel Oppus

 

 Posted - 3/14/2005 6:52:47 PM     
--------------------------------------------------------------------------------
Sometimes you have to educate your customers.
3rd Realm

 

 Posted - 3/16/2005 7:34:50 PM       
--------------------------------------------------------------------------------
Here is something you can try with your customer to illustrate the point on best method to verify flatness. place the surface in question at 45 degrees to the surface plate of your cmm. Your measurement should repeat no matter what attitude your plane is in. The alogrythim for flatness is quite simple and will automatically establish a nominal (average plane) and give you the variation of the points from this nominal. You don't need a coordinate system set! The plane function doesn't need it. Well, I hope it becomes obvious to your customer when they see no setup and the part at a weird angle that the math is correct and the CMM is usually the next best thing to an optical flat when it comes to flatness! Thanks for reading!

3rd Realm
Stephen Long
213.210.4756
Consulting, Robotics, Training