FRTZF TOP calculating 2

[SeasonLee]

Please advise if it is correct on the FRTZF tolerance of position calculating, without any reference datums on the lower segment.
Thanks

Season

 

[SeasonLee]

chez311
Thanks for your reminding.

Season

 

[3DDave]

pmarc - separate, as in it's own FCF.

The question was why wasn't LMC in the FRAME? I read the pasted on note; I didn't realize I needed to parrot it back.

I suggested that the ONLY way to accept such a part might require a large rotation for evaluating the two segments of the one composite frame. That is, if one made a physical gage, the datum feature A could sit at a significant angle to the axes of the pins, which means that the pins would have to be long enough to accommodate that orientation across the width of the part. While an academic curiosity, it would be an inspection nightmare. OTOH this is exactly the stupid callout that should result in suppliers making conformant parts that force the buyers to spend a ton of money to inspect. Too bad suppliers never do and so this bad callout gets onto drawings unchallenged or even supported.

chez311- a number of hall effect package application notes/data sheets state that the sensor is in the 'center' of the chip. Without the data sheet for the particular chip and probably a lot of magnetic simulation it isn't possible to determine if the tolerance scheme is worth anything. Since it is specified as a note and notes are allowed then it's up to the QA guys to figure out if the way the chip is described on the data sheet is compatible with the callout on the drawing.

 

[chez311]

then it's up to the QA guys to figure out if the way the chip is described on the data sheet is compatible with the callout on the drawing.

Seems like thats a job for the designer, not QA. Regardless, I still believe as it stands the centerpoint/axis is poorly defined - not that it couldn't be done with just a note, I just don't think "APPLIES TO CENTER" is good enough.

if the whole pattern of sensors shifts as a block it will increase the difference in pickup location as the entire group drifts off axis, so I fail to see the benefit to a composite tolerance.

Assuming the centerpoint is more clearly defined, I think a multiple single segment control would be more appropriate with [A|B] in the lower frame to allow rotation of the pattern but not allow it to shift off axis. I don't know of a good way to allow simultaneous expansion/contraction of the radial 13mm basic dimension (dilation of the circle having basic radial dimension of 13mm on which the centers lie) - perhaps this is enough allowable tolerance that the additional freedom by allowing this dilation is not necessary.

 

[pmarc]

greenimi,
chez311 got my point correctly. Thank you chez for describing it nicely. Notice, greenimi, that when I made that comment about the chips allowed to freely rotate about their centers, I said "even if we assume that the drawing clearly defines what the "center" really means".

pmarc - separate, as in it's own FCF.

That is still not clear to me. Separate with no datums or separate with reference to A|B(M) or separate with reference to A only?

I suggested that the ONLY way to accept such a part might require a large rotation for evaluating the two segments of the one composite frame. That is, if one made a physical gage, the datum feature A could sit at a significant angle to the axes of the pins, which means that the pins would have to be long enough to accommodate that orientation across the width of the part. While an academic curiosity, it would be an inspection nightmare.

Really? That would be the only way to accept such a part? Besides... would you really like to use a physical gage with pins to check the FRTZF requirement - the requirement with a tolerance value defined at RFS or LMC?

OTOH this is exactly the stupid callout that should result in suppliers making conformant parts that force the buyers to spend a ton of money to inspect. Too bad suppliers never do and so this bad callout gets onto drawings unchallenged or even supported.

Too bad designers don't use this kind of callouts more often (when functionally reasonable). Too bad suppliers often don't realize that this kind of callouts gives them more freedom in making the part. Too bad inspectors often don't try to make their life easier in the first place by checking this kind of callouts as if A was referenced in the FRTZF (because if it passes the requirement with A added, it will definitely mean that it passes the datumless requirement). Too bad words such as 'stupid' are used in a conversation between two MVP members of the forum in a hope that this will help to win a debate. Really too bad.

 

[3DDave]

pmarc [pos|dia.25(L)|A] on a line all by itself, if wall thickness is so important.

Good point - it's going to take hours to check RFS and LMC on a CMM for holes that are only there for decreasing the weight of the part. Makes it a perfect choice to allow parts through that cannot be handled with a simple optical comparator, but will require a skilled CMM operator to see if it can be accepted.

The smarter activity is to make drawings such that part are simple and cheap to inspect while also accepting what will work.

 

[3DDave]

chez311 - the designer of the board is in no position to better define what the center of the component is than the person who designed the component. Without the datasheet any callout of any kind is a shot in the dark.

The pattern can be defined with separate radial and angular position tolerances, allowing a significant rotation about the circuit board axis of the chips as a group, with smaller angular differences between them, while limiting radial travel to a narrow band. At least with separate FCFs for radial and angular, one could use faces or widths of any number of features on the chips.

If free rotation of the package about the invisible hall-effect element is to be allowed, and estimating the center isn't, then an electro-magnetic test is required as no FCF will be suitable to describe the required performance. Again, without the datasheet and the full magnetic analysis, there's no way to see that any method is suitable. This isn't the first drawing to put mechanical tolerances on an item that cannot be validated using only mechanical measurements.

 

[pmarc]

pmarc [pos|dia.25(L)|A] on a line all by itself, if wall thickness is so important.

Does "all by itself" mean that this would be the only position callout applied to the pattern of lightening holes?

Additionally, I never said the wall thickness was so important. I just said it was more important than the general orientation and location of the pattern relative to the datums. And for that very reason I did not originally use LMC in the tolerance frame on my illustration, but RFS. I could have also used MMC, if you like, because the whole point of this example was to show that there could be valid reasons for not having datum references in lower segment of composite FCF applied to a pattern other than of coaxial holes.

Good point - it's going to take hours to check RFS and LMC on a CMM for holes that are only there for decreasing the weight of the part. Makes it a perfect choice to allow parts through that cannot be handled with a simple optical comparator, but will require a skilled CMM operator to see if it can be accepted.

It would be also much easier from inspection point of view if, instead of using secondary datum pattern at MMB, the side faces were defined as datum features B and C. It would be even better if the holes (as relatively irrelevant for function of the part in assembly) were located with +/- dimensions. This would make the part the simplest and cheapest to inspect. Should I change the drawing accordingly?

The smarter activity is to make drawings such that part are simple and cheap to inspect while also accepting what will work.

I would say even smarter thing is to make drawings that in addition to accepting what will work, will not to reject what might work. Your proposal of adding A to the callout does not work that way.

 

[mkcski]

I am coming to this thread a little late and have not read down through all the posts. Let me add my two cents: The note next to the FCF says "applies to the center". A rectangular sensor are not a single feature of size and has no "center" (like a cylinder). The hall sensors are rectangular and are composed of two FOS - two sets of parallel planes (widths) As I see it, the composite position can be applied to either or both of the widths but not the center (the intersection of the center planes from both widths)

Certified Sr. GD&T Professional

 

[3DDave]

pmarc - it was one and only one control. I re-did the drawing before seeing the request for even more clarification. You could have posted a "like this" which would have been simpler.

"Spacing between holes ... to protect wall thickness." That seems important if a note was added to emphasize it.

Did the original creator of the 47 holes drawing include an explanation of how it would be economically inspected? I've tried to find it.

See the attached, an alternate lightening hole drawing for maximizing flexibility. I thought about adding the (L) but I'm too GIMP'ed out. Not sure why it ever needed a reference to A in the first place.

The plus-minus, et al, is not easily gaged, so there is no clear advantage. Were the holes all at MMC a fixed gage would take seconds to confirm the top level features and a second would gage the lower segment. A couple more gages with floating spring loaded go/no-go pins would take care of the hole sizes. They could be inspected nearly as fast as they could be lifted into an out of the fixtures.

If that's a production drawing of yours I would like to see the engineering requirements analysis that determined the tolerancing scheme and the tolerance values as being optimal.

I did have one case where such a relaxation was requested by manufacturing. The CNC operators were too lazy to clean the chips from the fixtures, so they wanted the top level of the composite FCFs to have huge diametral values and then got rid of the lower datum references, but were good with really small intra-pattern tolerances because the CNC was consistent and it made it look like they were performing well. Too bad it was in a way that didn't make the product any better. It's weird to see a diametral tolerance of 0.250 from datum simulators based on targets but diametral 0.005 from hole to hole. Small wonder they had fit problems at higher levels of assembly. Too bad for engineering that we didn't include 'chips on the fixture' and they had done 8 hours of machining on the pile of $25k each castings before checking. First article was fine - no chips on the fixture. Great fun redesigning a bunch of other parts to fit the defects and writing paper to accept garbage. Not that I'm still bitter.

(GIMP is just enough like Photoshop to seem familiar. But enough not like it to be a punch in the face for the simplest things. Worth not paying the Adobe tax instead.)

 

[3DDave]

mkcski - what is important is the location of the hall-effect element buried in the epoxy shell. The most likely indicator of it's position is the lead frame which is what it is attached to before the epoxy is injected. The makers of the chips sometimes refer to the center of the chip as the location of the hall-effect element - so it's one electrical industry standard way to refer to it.

Of course, the chip guys like to mix things up - attached is one that has one plus-minus dim from the epoxy and the other from one side of a lead on the lead frame. A pox on all their houses.

 

[pmarc]

I am not sure what the point of your last post is. It seems like by suggesting the alternate scheme you are admitting that it makes perfect sense to use FRTZF with no datum references.

All feature control frames that I used in my made-up example were taken from Y14.5 and modified in Paint. But the numbers and material modifiers were not really the point of the example. Again, the point was to show that there could be valid reasons for not having datum references in lower segment of composite FCF applied to a pattern other than of coaxial holes.

If you don't see this message, then I am afraid this whole discussion will take us nowhere.

 

[Kedu]

It would be also much easier from inspection point of view if, instead of using secondary datum pattern at MMB, the side faces were defined as datum features B and C. It would be even better if the holes (as relatively irrelevant for function of the part in assembly) were located with +/- dimensions. This would make the part the simplest and cheapest to inspect.

Sarcasm? Or you are dead serious?

Too bad words such as 'stupid' are used in a conversation between two MVP members of the forum in a hope that this will help to win a debate. Really too bad.

Wow!!

 

[3DDave]

pmarc - not admitting that at all - it was reductio ad absurdum. Since you feel that getting rid of datum feature A is a bad thing when I do it, then that suggests you cannot support getting rid of it except to advance an argument, which would be great if it was a deliverable product, but it isn't. I haven't argued that it's not a legal thing to do, just something that is useless in actual parts/assemblies, and one that even the guys doing the illustrations stayed away from.

It's possible to drop all datum references for even moderately complicated parts, and it would allow manufacturing more variability, though it becomes nearly impossible to validate that all variations that are acceptable under those requirements are actually usable. Cutting degrees of freedom is a compromise that can result in otherwise usable parts being scrapped, but I see that as a tiny chance as the fabricator should keep within the process limits based on the drawing as-is.

The argument that this costs more is not supportable. The extra variability is only available for completely open setups with hand-held tools. Once one has indexed the part to datum feature A per the upper segment, the parts are already indexed for machining per the lower segment. To do otherwise takes a significant extra effort. To carefully align to the virtual condition of B is an extra effort, but the feature-to-feature variation is a matter of the precision of the machine used to produce the holes; it could certainly be a punch or gang-drill, which would control that variation.

 

[pmarc]

Kedu,
That was sarcasm, of course.

3DDave,
Like I said, I am afraid this discussion will take us nowhere, mostly because we seem to believe in different philosophies of how and what for (G)D&T should be used. I did not comment in my last reply on your proposal of using [pos|dia.25(L)|A] as "one and only one" control for the pattern of holes, but to me this just nicely proves the difference. By applying the [pos|dia.25(L)|A] callout, the location relationship between the pattern of holes and the datum pattern B becomes unnecessarily and significantly tightened comparing to my original proposal. I don't believe in this kind of (G)D&T. If I did, I would probably never have to use any composite callout in my life.

Also the fact that you keep repeating (in this discussion and in some past threads) that a part used in an example is not a deliverable product and therefore the example is questionable is not an argument, in my opinion. It is really not that difficult to imagine a real-life application for the concept I was trying to picture here. And who knows (unless you want to say that you know) how many other real-life applications could take advantage of that approach.

And finally, it is kind of ironic to see you using argument that the concept is useless because "even the guys doing the illustrations stayed away from", while in the parallel thread you are trying to convince me that the figures used in the standard "are only to be an aid" for the text. Well, the text of the standard explicitly states in para. 7.5.1.(b).(1) that the lower segment of composite FCF do not have to contain datum references and it makes no distinction whether it applies to coaxial pattern or flat pattern or possibly radial pattern. And just for the record, it also clearly says in para. 1.1.4 that "The absence of a figure illustrating the desired application is neither reason to assume inapplicability, nor basis for drawing rejection."

 

[mkcski]

3DDave:

mkcski - what is important is the location of the hall-effect element buried in the epoxy shell. The most likely indicator of it's position is the lead frame which is what it is attached to before the epoxy is injected. The makers of the chips sometimes refer to the center of the chip as the location of the hall-effect element - so it's one electrical industry standard way to refer to it.

I understand your response about how the element is in the center of the shell. But Y14.5 states "7.2 POSITIONAL TOLERANCING Position is the location of one or more features of size relative to one another or to one or more datums." The center you are describing is not a FOS as defined in paragraph 1.3.32.1, so you cannot direct the leader to the center and apply position to it. It must be applied to either the 3.00 and/or the 4.40 width features, which are FOS.


Certified Sr. GD&T Professional

 

[3DDave]

mkcski - Ok, that's a logical reason for not using a symbolic method to solve this problem, but it does not further the needs of the original drawing creator. It is meaningless to apply it to the external features because those don't control the item that should be controlled. At this point I go with what I have hinted at before, that there is an electrical performance specification without any FCFs at all.

Still, it's to wonder that it is a center plane, an axis, or a point, which have no size, that is controlled as derived from the feature of size. So maybe the writers of the standard could be persuaded to recognize that a similar idealization is fair use in some future version so that this case can be managed exactly as this drawing suggests it could. But not today.

 

[mkcski]

3DDave: Got it. Thanks. GDT can't be used for "everything".

Certified Sr. GD&T Professional

 

[chez311]

The pattern can be defined with separate radial and angular position tolerances, allowing a significant rotation about the circuit board axis of the chips as a group, with smaller angular differences between them, while limiting radial travel to a narrow band. At least with separate FCFs for radial and angular, one could use faces or widths of any number of features on the chips.

Is there any chance you could provide a simple example showing what you mean? As you said before the actual orientation of the outside profile of the chip may not be critical as long as the nominal center is in the correct position - utilizing faces/widths would put an orientation requirement/constraint on the outside profile would it not?

I can visualize what you might intend by "separate radial and angular position tolerances" but I just can't see how to actually apply that - the radial tolerance that I think you are describing requires the simultaneous "dilation" of the pattern that I described earlier which would essentially require simultaneous change in the basic distance between the tolerance zones as well as the basic distance to the center axis of the board. I don't know of a way to do this - as far as I can tell multiple single segment/composite does not allow this. If instead you apply something like Fig 7-29 with bidrectional polar tolerancing I don't think this allows this simultaneous dilation either - it just describes the shape of the tolerance zone.

If free rotation of the package about the invisible hall-effect element is to be allowed, and estimating the center isn't, then an electro-magnetic test is required as no FCF will be suitable to describe the required performance.

I didn't say estimation of the center wasn't allowed, I just said that I believed that in its current form on the drawing shown was not sufficient - though admittedly I did not provide a solution (I do below in my response to mkcski). I take your point about the actual magnetic performance of the sensor being the ultimate goal, but I was thinking that a robust dimensioning scheme would be a more economical way to try and ensure performance than 100% magnetic testing. I may be wrong though.

At this point I go with what I have hinted at before, that there is an electrical performance specification without any FCFs at all.

I also just saw this response as I was typing the above. Perhaps I am continuing to beat a dead horse or trying to reopen a closed discussion, if so I apologize.

mkcski,
I think the entire discussion was sort of predicated on the fact that the center was somehow defined better than it is currently in the drawing - as I said before I may be part of the problem because I have not myself presented a solution. Allowing it to be defined with a clearly defined note that it is the intersection of the centerplanes of the length/width dimensions is perhaps a sub-optimal solution which does not strictly follow the letter of the law in Y14.5 applying position to something other than a FOS. Another could be the center of the circumscribed circle/cylinder around the points of the rectangular chip - assuming the sensing element is directly in the middle.

 

[mkcski]

chez311: Thanks. When getting into non-standard applications, notes are critical to communicating design intent.

Certified Sr. GD&T Professional

 

[greenimi]

Another could be the center of the circumscribed circle/cylinder around the points of the rectangular chip - assuming the sensing element is directly in the middle.

I was thinking about the same thing, but I could not solve it/ find an acceptable or standardized solution. Could you? How to do it? All the phantom lines for irregular FOS in ASME Y14.5 show a circle. How to drive an actual mating envelope from a non-circular feature?
Irregular FOS :..."or collection of features that may contain or be contained by an actual mating envelope other than a sphere, cylinder, or pair of parallel planes"