Qualifyinig Datums

[gb2835]

Hi Everyone,

In the basic example where Datums A,B, and C are perpendicular planes and are in order from primary to tertiary, I have really only seen (in inches) Datum Feature A with .005 Flatness, Datum Feature B with .005 Perp [A], and Datum Feature C with .005 Perp [A|B]. I understand doing the qualification allows for repeatable inspection measurements, but is there a more formal approach for determining the tolerance values? It's hard for me to picture an approach other than always assuming a convex surface and finding a value for the allowable "rock," but I know that is a topic in itself (I believe Y14.5-2018 addresses this now). Any help or references on this would be greatly appreciated as I'm having trouble wrapping my head around it.

 

[Burunduk]

There will be features controlled by geometric tolerances with ref. to those datum features. The ability of holding the tolerances of the controlled features will be influenced by the form and interrelationship deviations between these datum features. You should find a balance between not specifying too difficult to hold tolerances for the datum features and allowing reasonable rates of conformance for the controlled features. The datum features should anyway be more precise than the features controlled relative to them. How much more precise? It is a matter of experience with similar products, or a tolerance stack up analysis.

 

[3DDave]

Since True Geometric Counterparts are considered perfect, they are always more precise than any feature inspected relative to them. If any features are to be more precise it is the features on the mating parts.

Small tolerances on datum features are to benefit the inspectors and those making quality charts.

 

[Burunduk]

True Geometric Counterparts exist only in the standard. Their practical application is Datum Feature Simulators. These are made to gages and fixtures tolerances if they are physical, or programmed to attempt to mimic the behavior required from True Geometric Counterparts if they are digital. If the tolerances on the datum features are looser than on the features controlled relative to them, it increases the chances for high scrap rates due to greater variation in the way the part is positioned relative to the Datum Feature Simulators, and subsequently relative to the datum reference frame and the tolerance zones in that datum reference frame.

 

[3DDave]

Physical datum feature simulators, if you like being a pedant and, as you point out, True Geometric Counterparts, if they are digital, exist outside of the standard.

Also, you must not believe "The principles in this Standard are based on true geometric counterparts"

Nevertheless, it is the mating part that is the source of precision in placing the item in the assembly and the inspection equipment is a stand in for that mating part.

As a term, "physical datum feature simulators" also only exist in the standard. You cannot order one in a catalog.

"If the tolerances on the datum features are looser than on the features controlled relative to them, it increases the chances for high scrap rates due to greater variation in the way the part is positioned relative to the Datum Feature Simulators, and subsequently relative to the datum reference frame and the tolerance zones in that datum reference frame."

This is simply false. Bad inspectors increase those chances - and it's not high reject rates that are the worst problem. You know that, or should.

 

[Burunduk]

According to the latest version of the standard, there are no longer "Physical" and "Theoretical" Datum Feature Simulators. There are only Datum Feature Simulators (physical or digital/mathematical) and True Geometric Counterparts which are only conceptual. Any imperfection in the datum simulation procedure, resulting either from form/mutual orientation/location of the physical devices or insufficient gathered data/mathematical precision for digital datum simulation, is related to Datum Feature Simulators and not True Geometric Counterparts.

If the datum features are sloppy and this causes rejection of controlled features, it may not be only the inspector's fault. Datum features are often used for fixturing the part during machining (although not obligatory). You can't expect the fixturing for manufacturing to always follow the datum precedence rules, therefore greater variation in the relationship between datum features means greater difference between the manufacturing process precision and the tolerance conformance outcome.

 

[3DDave]

7.6 TRUE GEOMETRIC COUNTERPARTS AND PHYSICAL DATUM FEATURE SIMULATORS
7.20.4.5 Contacting Physical Datum Feature Simulators.

 

[3DDave]

It's a QC problem if the parts are incorrectly fixtured when being machined and it's up to inspection to reject the ones that are not compliant and accept the ones that are.

 

[Burunduk]

If you look at section 3, there is no longer a definition for a Physical (or Theoretical, for that matter) datum feature simulator. There is only "Datum Feature Simulator" (3.18). The examples in the note under that definition include "a mathematical simulation". So even if they call simulators "physical" in other portions of the standard, the concept is the same - Datum Feature Simulators are the practical embodiment of the conceptual True Geometric Counterparts.

All this is beside the fact that it is generally useful to have the datum features controlled more tightly by their qualifying tolerances than the features controlled with ref. to these datum features. Datum features that are too sloppy for the tolerances applied relative to them may increase rejection, regardless whose fault it is: whether it's the machinist, the machine, the process, the inspector, or the measurement equipment to blame.

 

[3DDave]

It hardly requires a definition. There is a plain statement using ordinary words. I note that you can clearly make out the meaning or you would have been posting "what is a physical datum feature simulator" years ago.

Not every phrase is defined, nor does it have to be. Definitions should be used for unexpected words or phrases. The fact that the definitions subgroup is separate from the group that actually put the phrase in context puts their efforts at the bottom of the list for caring what they wrote. They defined something else and there is no conflict.

Lots of things "may increase rejection." Wasting time and money on precision where it isn't necessary is one of them.

 

[greenimi]

.........but is there a more formal approach for determining the tolerance values?

I am not sure what the OP is asking?

gb2835,
Are you asking where tolerances (.005 in your exemplified case) are coming from?
Or what exactly you are asking?

I am saying that because the current discussion went already in the theoretical weeds.

 

[drawoh]

gb2835,

Read the standard. In your three face example, datum[ ]A is the three points on the face that make contact with the inspection fixture. Datum[ ]B is two points on the face that make contact with your fixture, and datum[ ]C is one point. Your flatnesses and perpendicularies do not affect any this. They just make the results more predictable.

--
JHG

 

[Burunduk]

3DDave,
I was not implying that a separate definition for any subtype of a Datum Feature Simulator is required.
I was pointing out that, contrary to your assertion, True Geometric Counterparts are only a theoretical concept that doesn't exist outside the standard.
Any practical implementation is carried out by Datum Feature Simulators, whether physical or digital.

I was also not implying that "wasting time and money on precision where it isn't necessary" is a good idea. If you read my posts in this thread, the more reasonable message you could get is that investing some time and money on adequate precision where it's required may spare you waste, but overtightening the tolerances on the datum feature is not a good way to go about it.

 

[gb2835]

Thanks everyone, I sort of walked myself into what I was looking for. In terms of stack-ups I am capable of applying datum feature controls accordingly. I was more looking for simpler cases of plates with mounting holes if there is a "proper way" to tolerance the primary, secondary, and tertiary datums; outside of the continual examples I've seen of just applying .005". There was a lot of information provided by Burunduk and 3DDave that I will have to sit and comb through later to better understand both sides. Again, thanks for the input.