Figure 4-16 in the '2009 version vs. figure 7-22, 7-23, and 7-24 in the '2018 version

[3DDave]

How fun. I hadn't looked until now but the figure that was 4-16 in the '2009 version and now 7-24 in the '2018 version no longer uses the virtual condition. I wonder why.

It still fails to mention that for the 3rd condition the maximum dimension only applies in one direction, but still, that's quite a change. Converting it to RFS means that there's less point in making that calculation as whatever the size the mating datum simulator has to collapse to meet it. Since D is referenced to arrest rotation the radial component of the RFS feature reference doesn't matter; the tangential one does and that is limited by the perpendicularity. Starting from a cylinder when only the width centered on datum feature B is important is not useful.

As I recall, the combined effects in the first example produced a dimension in a slightly diagonal dimension that was greater than the simple calculation showed so it is still wrong, but just in a different way. The addition of the straightness tolerance is a nice touch, but I don't understand why most values are changed. Change the text, change the test perhaps.


Edit: Title fix.

 

[pmarc]

The figure that was 4-16 in the 2009 is now 7-22 in the 2018.

 

[3DDave]

Thinking more on the RFS version - the datum feature simulator for D will be perfectly centered at the 29mm distance from datum feature simulator for B. When it starts to contract around datum feature D, it may only contact one point that is on a plane through the axis of the datum feature B simulator and the axis of the datum feature D simulator, but since that limit is on an RFS basis, it could be that this large diameter of the simulator will contact the that single point at the lowest diameter of the feature. Determining exactly where that one point is won't be easy and small errors in measurement will allow large movement of the part relative to the datum reference frame.

Had they used the translation modifier then the position tolerance would no longer apply and the feature would better control clocking. This case is what the translation modifier was created for.

 

[3DDave]

pmarc - good catch. I should have known that they would add additional cases. So many parallel construction paragraphs that bulk up the document. Serves me right for skimming.

Now there are two of them that are certainly not correctly describing the situation.

I'm not sure, having not worked though it, but I expect 7-23 also is flawed for the same reason. I expect that the LMB situation leaves a thicker wall tangentially relative to [A|B(L)|D(L)] than the calculation would suggest.

 

[pmarc]

3DDave,

I am glad that you finally saw the reason for introducing the translation modifier to the standard ;-)

 

[3DDave]

I don't see a reason for introducing it, but if it's in there might as well use it. I prefer controls to be explicit rather than forcing the drawing reader to have to figure out what degrees of freedom might be left unconstrained. Using parallel extensions on the width of the feature shows the exact direction the control is to be applied. They could have used the customized datum reference frame instead of creating the translation modifier as well; that would also make the direction of the control explicit.

 

[3DDave]

<off topic> One concept that seems to be skipped is that datum feature references are best used as idealized versions of the mating features on the mating part(s.) If a planar mating surface is identified as a datum feature, the pretend version of the mating feature on the mating part is a planar datum feature simulator - a stand in for the planar imperfect mating surface of the actual part.

With that in mind, when there is a case that something like a protruding pin on a mating surface is going to mate with a slot, then the depiction of how that slot is oriented is important. It isn't always the case that the slot will intersect with a locating feature/axis of rotation to simply control rotation. It's possible the slot will not intersect. The problem with the translation modifier is that there is only one sort of slot that can be represented and it takes anyone else depending on the drawing time to figure out what can be explicit - in the case of the customized datum reference frame that direction is easily machine readable without having to depend on some AI to sort through the options that are unclear from the translation modifier. With extension lines, it is discernible at a glance.

Done graphically with extension lines and more, one could have a datum feature defined to account for a curved slot in the mating part.
</off topic>

 

[Burunduk]

Regarding the RMB case:
Contrary to what you allude to,
The maximum external boundary of datum feature D at case (c) is an 8.6 diameter. This results from the combination of size+form error plus the position tolerance creating an outer boundary of datum feature D, and it would be so even if the position tolerance applied to it was referencing only A and B rather than A, B, C. It is true that for the position control of case (c) datum feature D is free to rotate about datum axis B before being constrained to its simulator, but it doesn't alter the boundary dimensions in any direction because in that stage the pin feature is still potentially able to contact any point on an 8.6 diameter boundary perpendicular to datum A and centered to any point located at exactly 29 mm from datum B. This is true with or without the perpendicularity within 0.3 control applied to datum feature D.

Edited: the bolded portion for clearer description.

 

[3DDave]

Datum feature D is allowed to rotate 360 degrees about [A|B] so the width is a 360 ring. It can contact some point(s) on the inner and/or outer surface of that ring. When D is added as a constraint it is only the tangential tilt that sets the width. The width due to that tangential tilt is less than the radial width of the ring.

 

[pmarc]

3DDave,

Customized DRF concept and the translation modifier are two different animals.

In the scenario that you described in the off topic reply, customization of DOFs constrained by the tertiary slot would make no sense because there would be no DOF to customize (as there would be only 1 rotational DOF left to constrain by the tertiary simulator). It would makes sense, however, to specify that the location of the tertiary simulator with respect to the secondary simulator does not need to be basic.

 

[3DDave]

The problem at hand was the origin of the translation modifier suggestion - that is: while there is one remaining degree of freedom the tertiary feature controls two degrees of freedom and, in a practical application it may be the feature that forces the part into a particular X location that secondary one does limit. Example: if the tertiary is a tighter fit than the secondary.

By specifying a customized datum reference frame it becomes explicit that only one degree of freedom from that feature can be used, so force-fit away.

If it was interpreted the other way then there would never be a push for the translation modifier.

edit:

Another alternate is to specify something like [A|B|B-D]

For the customized datum reference frame the modifiers could be [A[z,u,v)|B(x,y)|D(x,y)]** for the non-translation mode and then the translation mode is the default and requires no symbol.

** Yeah - they should be in square brackets, but I'm already using square brackets for the feature control frame delimiting and this editor doesn't have text size control to make them little like they did in the standard.

As I indicated, there should be a way to define the direction that the datum feature (or any feature) is allowed to move tangentially when there is a radial change.

 

[pmarc]

I don't think the true origin of the translation modifier was to prevent the tertiary from constraining more degrees of freedom than it normally should, although I agree that it could be a potential application.

The origin is relatively decently explained in fig. 4-32 in Y14.5-2009. Notice that both Means this illustrations don't differ when it comes to which and how many DOFs are constrained by B secondary.

 

[3DDave]

The competition in 4-32 is whether the slot location and slot width are both used to set the orientation or whether only the slot width is used - the translation is eliminating an over-constraint by allowing the centerplane of the slot to be at a different radius from the axis. Removing that constraint allows a wider range of parts to be acceptable, (whether or not that wider range is also useful is another task beyond Y14.5)

AFAIK the first area was in 4-9 vs the added 4-19 example where the desire was to use RFS callouts for the holes used as datum features and the complaint was this might require a mallet to get the part onto the gauge. I don't recall if it was this forum in the Yahoo! Y14.5 group but that distance between pins problem was the first I came across it and that was before the '2009 was a twinkle in anyone's eye.

4-9 was kind of like figure 4-8 from '1994, but they removed all the maximum material references in the '1994 version so it could use the fictional uniform expanding pin and then 4-19 that still requires the fictional uniform expanding pin plus a precision made gauge if real gauges are to be used as was once the case for mass production in-process inspection. I guess they felt they covered the maximum material case in other figures to not double onto it in '2009.

 

[Burunduk]

Datum feature D is allowed to rotate 360 degrees about [A|B] so the width is a 360 ring. It can contact some point(s) on the inner and/or outer surface of that ring. When D is added as a constraint it is only the tangential tilt that sets the width. The width due to that tangential tilt is less than the radial width of the ring

In the RMB case option (b), had the datum references been |B|D|, would you say the OB (Outer Boundary) for datum feature D is a 360° ring?

 

[3DDave]

D has a datum reference directly to A. That's where the perpendicularity fits in.

Another analogy since the Monte Hall problem didn't work.

You are standing beside a highway. Cars go by at 100kph. If you reach out to touch one you'll be injured.
You then get into a car and you travel with the stream of cars at 100 kph. You can reach out and touch another car next to you. You aren't injured.

Changing the frame of reference changes the outcome.

The tangential component for the location of the feature used as datum feature D is only fixed by the initial reference to datum feature C. Once that is gone all that remains in the tangential direction is the orientation limitation and that is refined to a smaller value by the perpendicularity tolerance. The radial variation remains only as long as there is a reference to datum feature B. With that gone, the orientation component of the position tolerance remains, and the position tolerance control of orientation is again refined away by the perpendicularity tolerance when the only reference is datum feature A.

The explanation in 7.11.9.1 (b) is dissatisfying for not telling why the position tolerance isn't part of the calculation.
"to ensure that datum precedence is not violated" doesn't say how it could be violated and why this selection avoids violating it. I think too much discussion happens at the meetings until everyone is familiar with the supporting analysis and then they forget they only got to that understanding by seeing that supporting analysis which they don't put in the document.

The cynic suggests that's held back to sell training, but the group-thought experience says the more likely explanation is they are so used to it they don't notice. Like the use of UAME in some posts - an abbreviation that isn't in the standard.

 

[Burunduk]

Since you posted your reply at the same time I edited my post (very short time after posting it), let me ask you one more time:

In the RMB case option (b), had the datum references been |B|D|, would you say the OB (Outer Boundary) for datum feature D is a 360° ring?

 

[3DDave]

It would not be 360 because D is what gets you going 100 kph to align with D. If it is [A|B] only then datum feature D is not constrained and the part is free to rotate resulting in that ring.

If it is B only then it's too complicated as predicting the orientation of D relative to B is not trivial, but the pin feature will be able to occupy anywhere in some ring shape. A fixture to accept all locations/orientations of the pin will require a circular trough for that side of the part.

 

[Burunduk]

It would not be 360 because D is what gets you going 100 kph to align with D.

Then what would be the shape of the OB, and what would be the shape of the datum feature simulator/true geometric counterpart for datum feature D, in a control referencing B primary D secondary?

 

[pmarc]

3DDave,

In my mind, figs. 4-9 vs. 4-19 and 4-32(a) vs. 4-32(b) are about the same thing - the translation modifier allows the simulator to fully engage with the feature designed to constrain the rotational DOF.

My main message still remains - in principle, the translation modifier does not work the same way as the customization of DRF, as one might conclude by reading your previous comments in this thread; in 4-19 and 4-32(b) no DRF customization really takes place.

Side note: Although the gage pins B and C in fig. 4-8 in '94 would indeed be of fixed size, according to para. 4.5 3(d) the distance between them should be variable, and so the whole gage would not be that trivial anyway.

Side note 2: A mallet might be needed to install the part from fig. 4-9 in '09 onto the gage, because generally in the gage design (assuming someone would even like to design a gage to simulate B and C at RMB), the sequence of DOF constraint, as defined by the datum portion of the FCF, is ignored for obvious economical reasons. Even Y14.43 standard for gages doesn't pay too much attention to the topic - there is just one figure, as far as I remember, that talks about it in greater detail.

 

[3DDave]

4.5.3(d) in '94 was about RFS gages, so for 4-8 would not apply and a moveable pin would not ever be required.

All that the translation modifier does is relieve the constraining feature of one degree of freedom, which is explicit in a customized datum reference frame. How do you suppose a customized datum reference frame is able to shift constraints from a feature they ordinarily control to one later in the sequence?

As I related earlier - indicating the width of a hole or pin in the direction of desired restriction on the datum feature handles these cases.

As for the orientation by the slot - I haven't seen any real mechanism that both required a fixed offset to define the location of such a feature that then said - doesn't matter where it is. However - on an RFS basis the slot competes for control of the Y location and excluding that control explicitly would indicate that the datum simulator could move.

For '2009, "in 4-19 and 4-32(b) no DRF customization really takes place." That's true - both are MMC/MMB cases.