Hi All,

Any help or clarification greatly appreciated on this one!

I have a dissagreement with a customer at present, concerning the method of measurement for symmetry. The attachement details the two methods that we each use to determine a result. In summary I believe that they should divide their result by 2 to get the true deviation from symmetry.

Does anyone where I could get any evidence (documents etc.) to prove which method is valid?

Many thanks,

Phil

Hi All,

Any help or clarification greatly appreciated on this one!

I have a dissagreement with a customer at present, concerning the method of measurement for symmetry. The attachement details the two methods that we each use to determine a result. In summary I believe that they should divide their result by 2 to get the true deviation from symmetry.

Does anyone where I could get any evidence (documents etc.) to prove which method is valid?

Many thanks,

Phil

Your diagram doesn't include the symmetry specification and doesn't identify the datum(s), thus it's not possible to tell for sure what you're asking. Symmetry controls are basically the same as for concentricity, and as such symmetry is often used when (true) position should be. As for evidence, what about ASME Y14.5M?

Having quickly looked at the .bmp attachment, I would say that your customer has obtained the correct value, but not by the method that I would use.

The assumption is that the 'T' member should be centralized between the two sides.

The distance from the theoretical centerline is 5 units. Much like profile, you would double this value to obtain the reported symmetry of 10 units. However, the shift you would make would only be half (i.e. 5 units)

It helps to think of symmetry as position contstrained to one axis.

Having said all that, I am presuming that you are referencing the ASME standard. Other standards may differ.

It helps to think of symmetry as position contstrained to one axis.
As I pointed out earlier, there's not enough information in the attachment to make a reliable call, but symmetry is analogous to concentricity, not position, so the error is taken as a direct measurement from the datum axis.

Hi All,

Thank you to JSW05 and Ron for the responses. I have added the tolerance to the image, but the image is meant only to represent the problem. On the particular specification in question the toleranced is referenced by a note requesting max symmetry of 0.15 to the centre-line (I have called the centre-line the datum line on the modified attachment).

I had always believed that a symmetry result (between two centre-lines) would represent the deviation from centre-to-centre (providing the distance that the feature would need to be moved by), although judging from your responses this does not appear to be the case...

Best regards,

Phil.

The tolerance zone for symmetry is oriented and located centrally to the median plane or axis of the datum feature. It is a total width zone within which the (ASME interpretation) median point of opposing or corresponding elements of the considered feature must reside or (ISO interpretation) median plane of the considered feature must reside.

If your worst median point (ASME) or median plane (ISO) is displaced 5.0 from the axis or median plane of the datum feature then it would take a 10.0 wide tolerance zone to contain that displacement.

I agree with Ron and your customer.

The tolerance zone for symmetry is oriented and located centrally to the median plane or axis of the datum feature. It is a total width zone within which the (ASME interpretation) median point of opposing or corresponding elements of the considered feature must reside or (ISO interpretation) median plane of the considered feature must reside.

If your worst median point (ASME) or median plane (ISO) is displaced 5.0 from the axis or median plane of the datum feature then it would take a 10.0 wide tolerance zone to contain that displacement.

I agree with Ron and your customer.

I think I misunderstood Ron's explanation. This is the correct interpretation of the requirement. I would still like to see the actual callout and datum on the drawing, however.

Not to belabor the point, but the way that symmetry error is measured is the key to understanding the reported result. In the configuration in question, measurement would be very difficult. It would require using diametrically opposed indicators moving along both sides simultaneously, and symmetry error would be reported at the point of greatest divergence of the indicator readings--the distance between those two median points, not the distance from the datum axis to one side. For this reason it's almost always better to use position. In addition to ease of measurement, feature size may be taken into account, while symmetry is always specified on an RFS basis.

Hi All,

I have attached a more complete representation of the specification to satisfy your curiosity JSW05. Thank you all for the responses so far, they have been very enlightening.

:thanx:

Hi All,

I have attached a more complete representation of the specification to satisfy your curiosity JSW05. Thank you all for the responses so far, they have been very enlightening.

:thanx:

Thanks. Just out of curiousity, how would you measure this?

HI JSW05,

I (previously!) would have:
1. Allign the sample and Zero Y-Axis on the datum line.
2. Measure the position of the two outer points (2 x Y-Axis positions).
3. Subtract the lower of the two results from the higher.
4. Divide result by 2 to obtain deviation from centre-line to centre-line.

I looked at this a couple of years ago and decided that the deviation result must be the difference between centre-lines, but it appears that I am probably the only person in the world who divides their result by two.

My own method makes more sense to me (well it would!), but I am clearly wrong, as fellow covers (such as yourself) with a great deal more knowledge and experience in this field have demonstrated.

HI JSW05,

I (previously!) would have:
1. Allign the sample and Zero Y-Axis on the datum line.
2. Measure the position of the two outer points (2 x Y-Axis positions).
3. Subtract the lower of the two results from the higher.
4. Divide result by 2 to obtain deviation from centre-line to centre-line.

I looked at this a couple of years ago and decided that the deviation result must be the difference between centre-lines, but it appears that I am probably the only person in the world who divides their result by two.

My own method makes more sense to me (well it would!), but I am clearly wrong, as fellow covers (such as yourself) with a great deal more knowledge and experience in this field have demonstrated.
Don't feel bad--it's not at all intuitive, and there aren't very many people who haven't misinterpreted a GD&T callout. I still question your method, though, as it seems that what you're measuring is position and and not symmetry. I think position was probably what the designer was after, but using your method, how can you evaluate all of the median points of the two features?

Phil P,

I can see either definition being logical. Perhaps this will help the other definition to seem logical to you as well.

If two features are symmetric, then you could put a mirror down the center line and the mirror image would overlap withe the actual feature on the other side.

In your case, if one feature is 1.2 below center, the mirror image would be 1.2 above center. This would be 0.2 from the location of the actual other feature at 1.4. The deviation from symmetry is 0.2!

Tim F

Phil P,

I can see either definition being logical. Perhaps this will help the other definition to seem logical to you as well.

If two features are symmetric, then you could put a mirror down the center line and the mirror image would overlap withe the actual feature on the other side.

In your case, if one feature is 1.2 below center, the mirror image would be 1.2 above center. This would be 0.2 from the location of the actual other feature at 1.4. The deviation from symmetry is 0.2!

Tim F
How do you reconcile this with the need to evaluate symmetry in terms of the ...median points of opposed or correspondingly-located elements of features? (From ASME Y14.5M). I think you're also talking about position, not symmetry.

How do you reconcile this with the need to evaluate symmetry in terms of the ...median points of opposed or correspondingly-located elements of features? (From ASME Y14.5M). I think you're also talking about position, not symmetry.

Good point. I should have said "center of a feature" or equivalently "median of a feature" rather than simply "feature" to be more precise. Then I think my statement is OK.

For example, assume for the sake of simplicity the center line of the part is x=0.

Suppose the top feature extends from 0.8 to 1.4, so the median is 1.1
Suppose the bottom feature extens from -1.0 to -1.2, which has a median of -1.1
These features would be symmetric in terms of either ...median points of opposed or correspondingly-located elements of features or in terms of the mirror image of the median point of the feature.

Tim F


P.S. I haven't really done this before. I'm just trying to fit my logic and Phil's logic to the quotations given from the ASME reference.

Good point. I should have said "center of a feature" or equivalently "median of a feature" rather than simply "feature" to be more precise. Then I think my statement is OK.

For example, assume for the sake of simplicity the center line of the part is x=0.

Suppose the top feature extends from 0.8 to 1.4, so the median is 1.1
Suppose the bottom feature extens from -1.0 to -1.2, which has a median of -1.1
These features would be symmetric in terms of either ...median points of opposed or correspondingly-located elements of features or in terms of the mirror image of the median point of the feature.

Tim F


P.S. I haven't really done this before. I'm just trying to fit my logic and Phil's logic to the quotations given from the ASME reference.

I think you're confused over the meaning of "median points." As you describe it, you are indeed checking for position rather than symmetry. In order to verify the latter, you have to measure the median points along the entire lengths of the opposing elements. In other words, you would have to have two indicators moving synchronously and in direct diametric opposition across the two surfaces (one indicator on each). The point of greatest divergence of indicator readings would be the recorded measurement value. Obviously, this is a colossal PITA and rarely done. As I stated earlier, the fault is usually with the designer not understanding the difference between position and symmetry in GD&T-land.

Thanks for everyones input, I now understand why the results should not be halved. The correct method of evaluation is clearly not practical with standard measuring equipment (such as projectors), but I do not understand why both sides (of the symmetrical feature) must be measured simultainiously, would you not also get the correct result by performing evaluation of each side independenly and then comparing the results?

Have a good weekend everyone.

Thanks for everyones input, I now understand why the results should not be halved. The correct method of evaluation is clearly not practical with standard measuring equipment (such as projectors), but I do not understand why both sides (of the symmetrical feature) must be measured simultainiously, would you not also get the correct result by performing evaluation of each side independenly and then comparing the results?

Have a good weekend everyone.

I've attached an illustration from the standard that might help. What's important is the median points of directly opposed elements. In other words, in order to calculate the median points correctly, the two readings must be directly opposite one another.

Hi JSW05,

I now completely understand the point that you were trying to make, that symmetry can only be measured between two parallel median points. Thanks for taking the time to explain. I have learned something new today, so that can only be positive.

Hi JSW05,

I now completely understand the point that you were trying to make, that symmetry can only be measured between two parallel median points. Thanks for taking the time to explain. I have learned something new today, so that can only be positive.
I'm glad the information was helpful. As I said earlier, there's nothing intuitive about it. It's one of those areas where using common sense can get you in trouble:frust: :lmao: .

Thanks JSW05,

I think I get it now. Symmetry involves finding the one worst point where the center of the pair of features is most off from the center of the piece. I'm not sure I would have defined it that way, but it is a reasonable definition.


Tim F

Thanks JSW05,

I think I get it now. Symmetry involves finding the one worst point where the center of the pair of features is most off from the center of the piece. I'm not sure I would have defined it that way, but it is a reasonable definition.


Tim F
Tim:

Symmetry and concentricity (symmetry's circular/cylindrical counterpart) are easily the most misunderstood GD&T callouts. The ironic thing about GD&T is that inspectors--and especially CMM operators--are much more likely to understand the requirements than designers and CAD people, so much time is spent in p***ing matches when the inspector can't figure out what was intended or how to measure it, and the CAD guy won't admit that he was wrong.