simulating datum target area on cmm 1

[acnme]

I have a customer ask us how do we established datum target area of .250 dia, I take 1 point in this area s.b enough to define datum target area. The customer did not like my answer? so I need your advice on how to simulate datum target on the cmm machine.Thank

 

[DeanD3W]

Dave,
To follow the drawing specifications, it could be said that when a datum target area is specified, then you must find the high point of each area, no matter how small the area is. Of course you can then bring up the fact that there is uncertainty in all measurement and the single point per area just introduces more uncertainty (which may still be acceptable, if associated tolerances aren't too tight), so that may be the practical way to follow the datum target area specifications.

If it was not the desire to have the inspection cost include the work of finding the high point in each area then datum target points should have been specified rather than areas. Whether to specify, and inspect in accordance with, datum target areas depends upon how imperfect the datum feature surface is and how tight the tolerances are that include reference to that datum feature. Design should make that determination and inspection should follow what is specified (in a perfect world anyway).

I hope the datum targets were specified based upon functional contact with the mating assembly... If the mating feature has three small circular areas then to base the datum from the high point of each circular area is doing what simulators normally do, which is to model an ideal/perfect mating feature. If the mating features for the specified datum targets are spherical bumps that are located accurately enough then maybe datum target points model the mating relationship. I generally don't think it's a good idea to specify targets just to improve measurement repeatability... The determination of whether to specify targets, and what type of targets to specify should be based upon how the mating assembly contacts the datum feature. I'm taking this position because the point of the specifications is to capture design intent, which means producing data that relates to function, which can only be done if the datum feature/target specs model the way the mating assembly contacts the datum feature.

Please pardon my wordiness (was that my "outside wordiness" ).

Dean Watts

http://www.d3w-engineering.com

 

[dingy2]

Dean:

Most Designers have trouble with datum selection. Which should be the primary, secondary and tertiary. Should one use a hole or a surface for a secondary? When it comes to datum targets, it usually is requested by Quality or Manufacturing Engineering and, in my experience, not by the Design group. As a matter of fact, some Designers aren't readily agreeable to placing them on the drawing and really want process instructions to include datum targets rather than the part drawing.

Quality personnel need datum targets to set up on the datums to measure the part consistently. If the part does not have datum targets, someone usually hand bombs the datum targets in pencil on the marked up drawing just as they number all dimensions.

If the part really touched 3 points of a diameter of.250 on the mating surface rather than the full surface, then those 3 areas should be shown on the drawing using a chain line boundary and then reflected as the datum - not the datum target. One would then create a plane using the three (3) datum areas shown in the chain line boundaries.

Just some food for thought.

Dave D.

http://www.qmsi.ca

 

[DeanD3W]

Sorry Dave, but that food doesn't taste good at all.

1) Design should agree that the datum targets, if applied, adequately model the function of the part, or else the inspector should be after the points that would contact a proper datum feature simulator. Arbitrarily choosing their own datum target points will still produce data but, if the datum feature's form has significant imperfections, that data will not be of adequate quality.

2) Partial datum features (designated by chain lines) would not be used as you describe, datum targets would be used.

Best Regards,
Dean Watts

http://www.d3w-engineering.com

 

[axym]

Dave,

I can understand why some designers don't want datum targets put on the drawing. It's probably because they want their drawing to communicate the intended requirement! If quality or manufacturing wants to introduce error by setting up on non-functional datum targets, that's their business.

I agree with Dean that datum target areas are the proper tool to use for the three .250 diameter areas. I don't see the point of annotating chain line boundaries in this case - the datum target areas convey exactly the right intent.


Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[dingy2]

Dean:

I enjoy discussions like this since the standard is vague in some areas and other areas, in my opinion, not really practical. From these online discussions, I have created web pages covering the controversial topics so bear with me and don't be offended.

I am going to use your description of the mating part having three (3) flat round features that contact on our part. I don't think this example is practical but I will use it anyway. The tops of the three (3) flat round surfaces on the mating part would be its primary datum.

The first question would be - is the full surface on our part the primary datum? If so, the geometric counterpart must contact on the full surface. You stated in your hypothetical example that the mating part has three (3) .250 flat round features that contact on the part rather than a full surface contact.

I could not find anything in the ASME Y14.5-2009 with this example but I did find a situation in ASME Y14.43-2003 (gauges and tolerances). There is a drawing where on the workpiece datum A is 1/2 the full surface as shown by a chain line boundary. To me this means that the mating part only contacted on the surface bounded by the chain line, not the full surface. Using this same concept, should have 3 areas on our surface shown in chain line boundaries (not the full surface) designated as datum? We could still use the datum targets on those 3 areas but the datum is not the full surface but only the areas where the mating part contacted.

Flat round contacts are usually shown on castings since surface is rather rough. Sometimes the flat round datum targets are the same size as the machine set up points for broaching or milling a surface. I have also seem flat round datum targets on a plastic automobile dash with 24 points shown on 24 tabs. The part was rather flexible and one needed it to be stable for measurement.

I have found datum targets are process rather than design generated. We need them in Quality to set up on the product consistently to achieve a good repeatability and reproducibility. If the drawing does not have the datum targets shown, how should a CMM Operator set up the part? Scan the surface looking for high points . That just doesn't give a stable set up.

We use a 3 point set up on a rigid part, hopefully, about 120 degrees apart towards the perimeter but not on the edge of the part. It is best to have the datum target shown on the drawing to make it official or valid but I would never suggest to a CMM Operator to send out a marked up drawing without some sort of set up criteria.

Dave D.

http://www.qmsi.ca

 

[axym]

Dave,

Have a look at Fig. 4-48 in ASME Y14.5-2009, with the three datum target areas A1, A2, and A3. If the mating part had three 6mm flat round features that contact the part, then this datum target specification would make perfect sense. I still don't see what your objection is, and why you would use the chain line boundary method to try to depict the same thing.

Part of the problem here might be that we're not all using the terms in the same way. I hate to be the terminology police, but we have to be very careful here.

A datum feature is a surface (or group of surfaces) on the part. In 4-48, the datum feature is the group of 2 roughly rectangular surfaces, that are nominally coplanar.

A datum feature simulator is a perfect inverse feature that contacts the datum feature. The default datum feature simulator covers the entire area of the datum feature. In 4-48, this would be two perfectly coplanar plates that cover the entire datum feature.

Datum target specifications specify different (non-default) geometry for the datum feature simulators. In 4-48, the datum target areas specify that the datum feature simulators are to be three circular areas 6mm in diameter. See the three flat-tipped pins in the "means this" diagram.

A datum is a theoretically exact entity defined in the datum feature simulators. In 4-48, Datum A is a plane that passes through the three circular areas.

So datum target specifications do not change what the datum feature is. They change the geometry of the datum feature simulators, which means that they change the geometry of the fixture that the part will be held in to establish the datum reference frame.

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[dingy2]

Evan:

Fig. 4-48 makes perfect sense but please note that the datum surface is the 2 coplanar top surfaces of the part - datum A.

Here is where things seem to go wrong but maybe I misinterpreted Dean's thoughts. He stated "If the mating feature has three small circular areas then to base the datum from the high point of each circular area".

I think that he is talking about the mating part having 3 small circular areas that actually contact onto our part on the datum surface. The full surface is now not the datum but just the areas of contact from the mating part. Am I wrong to conclude that the mating part does not fully contact on the datum surface but only in the three (3) cylindrical areas noted??

As I stated, fig. 4-48 makes perfect sense using datum targets to set up on a plane and I have actually performed this many times in the past. Both Processing and Quality really need datum targets on a drawing but generally have to request them.



Dave D.

http://www.qmsi.ca

 

[DeanD3W]

Dave,
You were correct to take my meaning to be that the only contact with the datum feature in question would be within the three circular areas.

Datum targets should only be specified if they model the functional contact with the mating assembly or possibly if you're dealing with an as-cast or as-forged part. If someone in inspection is choosing "their own" datum target points then their data will likely not model function as well as it should, since those points may not be the high points that would contact a set of datum feature simulators.

For the case we're discussing here, unless uncertainty is deemed insignificant, or the additional work too expensive (almost the same case said two different ways) the inspector should be looking for the set of three high points, one from each of the the three circular datum target areas. They need to do this since those would be the points that make contact with a datum target simulator. To build a fixture with 3 cylindrical pins is one approach, but to properly locate that set of 3 pins relative to the part is not so easy... Including 2 "B" targets and 1 "C" target as part of the fixture was mentioned in an earlier post. That may be OK, but there aren't many parts in the world that really function as a simple rectangular block with three mutually perpendicular planar datum features. Most/many parts will end up with a cylindrical datum feature if a proper, function based, approach is used to select datum features... This makes building that "simple fixture" not so simple. Maybe a lot of parts are handled like a simple rectangular block, but that doesn't mean that they should be.

Processing and quality do not need datum targets. They may want datum targets, since that makes their job a bit easier, but if the designer says "no, there are no datum target points, lines, or areas on the datum features because the mating assembly does not contact the datum feature in that way... The mating features have the full footprint of the datum features or more" then the downstream processes need to build or model a set of full datum feature simulators. That perceived/invented need you describe is part of the reason poor quality data is gathered every day and part of why cmms get an unjustified bad rep... Just sampling a few points and thinking/using a "plane/line/point" approach within the software does not produce a proper datum reference frame. All the inspectors using CMMs should be 1) Sampling an adequate quantity of points on each datum feature to do an adequate job of finding the points that would contact a physical datum feature simulator, and 2) iterating when establishing a datum reference frame, so they can repeat the process with the very same points each time (an important thing whether datum targets are specified or not), and 3) examining additional points on the surface to verify that they've adequately modeled what a physical datum feature simulator set would produce (meaning no "higher" points are found that would have contacted a physical set of simulators)(not necessary if datum target points were specified). Of course the amount of rigor depends upon whether one is building toy trucks or Trident submarines, that's why the word "adequate" is used a couple of times above... It depends upon how imperfect the parts are relative to how tight the specified tolerances are.

Have you ever done a data correlation between a good, careful knowledgeable metrology lab and a typical supplier's metrology lab? If so, then you know that there will be very significant differences in the data. Those differences almost always include a less than good enough process for establishing the datum reference frame(s).

Dean

http://www.d3w-engineering.com

 

[dingy2]

Dean:

Thanks for the clarification although I certainly do not agree with you on this one - not even close.

Dave D.

http://www.qmsi.ca

 

[DeanD3W]

OK Dave, We don't agree for sure.

The reason GD&T was developed was because it was found that the pre-GD&T method of directly toleranced dimensions was not enabling adequate data to predict whether a part will function or not. The "shop floor assignment of datum target points" that you describe as method will absolutely tend to make data of lower quality with regard to predicting whether the part will function or not. Why bother with GD&T if you're going to implement practices that significantly degrade its benefit?

What if those shop-floor-selected datum target points happen to fall into low points on the surface that, relative to the level of precision of the tolerance specified, cause a significant error in the establishment of the datum reference frame?

Dean

http://www.d3w-engineering.com