Goal R.002 *Pulls hair out!* 5

[ModulusCT]

OK, so I was told to do the following by my supervisor today. I really hate having to tell him that he has no idea what he's doing and that he doesn't know the applicable drafting standard, so I thought that it would be better to simply offer a solution that does jive with the spec.

He wants to indicate R.002 +.003/-.000, but with an emphasis on trying to produce the radius at .002 rather than taking the total permissible error and shooting for the middle. His solution to this problem is to say .005 MAX; GOAL R.002, but I hate this personally because the word GOAL is not mentioned in the spec as an acceptable descriptor for tolerancing a part. In fact, I'm pretty sure there are parts of the spec that say not to do things like this.

Any suggestions? Is there a way that you can say, what my boss wants the drawing to say by using acceptable symbology and terminology?

 

[ewh]

I think we are starting to go into the weeds with this.
While this forum does address GD&T and tolerance analysis, this issue seems to be merely one of drafting standards and interpretation. It can be argued until the cows come home as to the best way to achieve what the OP's management is asking for, but that is beyond the purpose of the drawing.
Bottom line is that dimension = target (not suggested target). Of course, no (non-basic) dimension can be perfect. Tolerance = acceptable variation from target. As noted, fabricators often juggle the allowable tolerance to ease their task and increase productivity, and without further direction this is an efficient method and is acceptable to do so. Such a drawing as this IS NOT the place to provide that direction, as its purpose is to provide part definition, not process. Adding terms such as "GOAL" is redundant and only serves to clutter the drawing. The line is then crossed to start adding more terms that may be irrelevant to the part definition, such as REAM, BORE, TAP, etc which should only be used if it is critical to part function. The presence of the tolerance seems to negate the criticality of "GOAL". If the target dimension is not the goal, then then a dimension range should be used instead. Issues such as the OP's are production issues, not definition, and should be addressed in other production documentation, such as a work instruction, using statistical control or other methods, but not on the defining drawing. If there are still problems, then the tolerance is wrong, or some training needs to be given. Ignorance of simple drawing interpretation per whatever standard used needs to be addressed.

On the other hand, if management chooses not to follow a standard, you're on your own.

Technically, the glass is always full.

 

[ewh]

My apologies... a couple of posts snuck in before mine which make very similar points.

Technically, the glass is always full.

 

[MechNorth]

J-P,
You're assuming that the high-school football team has the ability to track the start point, kicker's capabilities, end point, calculate the trajectory, force and velocity. Unfortunately only a few colleges & the NFL have those abilities. The reality for everyone else is ... get it thru the uprights, or it doesn't count.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[Belanger]

Well, running a professional manufacturing company should involve some sort of consideration of capability, shouldn't it? Quite different than a high school team where all players are given playing time. (Obviously I was referring to a football team at a higher level.)

Bottom line in all this: Though all parts within a given range are acceptable, the designer is still trying to communicate something by assigning a unilateral or unequal bilateral tolerance.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

http://www.gdtseminars.com

 

[MechNorth]

I don't disagree with the second statement, J-P. The problem is effectiveness of communication. I would usually visit the Process Planner and the individual machinist to get the point across.

As for the first point, "should" is the operative word there. Reality is vastly different, and most companies of small-to-moderate size don't play on the same technology field as the big boys. Most companies just try to do their best (i.e. "tightest") processes and check the parts for acceptance. It's a gross waste of money because companies are wasting huge amounts of personnel and machine resources where they don't have to. Not terribly different from a company that bids on a project with GD&T on the drawings, but has never done it before; they crank up the price and just do their best without knowing what it all really means ... praying they get it right.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[ewh]

I don't think it is an effectiveness of communication issue so much as it is an inability to understand the different methods of tolerancing a dimension. You shouldn't need new procedures or processes to train someone in correct dimension interpretation.
Bi-lateral (targeted but variance allowed), unilateral (targeted but variance allowed), geometric (targeted but variance allowed) or range (general - don't care where the final dimension ends up as long as it is within the range); the OP's issue is covered once those four types of tolerancing are understood. Creating new procedures or processes to redefine such practices will only unnecessarily complicate things, and that will snowball as time goes by. Technology is great, but I still stand by Training and KISS first.
I truly appreciate the knowledge being shared in this forum, but more and more often it seems that theory is over-riding practicality. This is understandable when new principals are introduced or refined, but the OPs issue is cut and dried dimension interpretation. You can discuss theory all day and still not completely agree, but sometimes drawings actually need to get to the floor.

Technically, the glass is always full.

 

[ModulusCT]

"What matters to the machinist is not always what matters to the designer. Do you believe R.002 +.003/-.000 communicates exactly the same thing as R.0035 +/-.0015?"

@powerhound: That's the concern. I don't know if it's well founded. In your opinion it is not, but the problem is convincing my boss of that fact.

"I guess you have a way to tell the difference between a .002 radius and a .005 radius."

Nope! We don't inspect our parts /facepalm. We operate with a policy where we trust our manufacturers to give us good parts. We only do what's called a first article inspection, which means that we measure the assembled end product, not its constituent parts.


"They state that .002 +/- .001 is not achievable but that .002 to .005 is, which however you dress it up is +/- .0015 so that extra half a thou must make all the difference."

@ajack1: Well, that's just what my dept. head told me... He probably doesn't know himself. That's a good point though. It's unfortunate that I've got to deal with all of this silliness, but there are a lot of sensitive egos where I work I guess.

"On the other hand, if management chooses not to follow a standard, you're on your own."

@EWH: Yeah, we're 'soft ANSI', which means that our drawings do not say 'interpret per ASME Y14.5M'. We're supposed to make our drawings to the spec, but our (micro)managers reserve the right to do whatever the hell they want with it. Therefore, no one really knows or cares what the spec says at the end of the day. Hence, my grey hairs.

 

[ModulusCT]

"I truly appreciate the knowledge being shared in this forum, but more and more often it seems that theory is over-riding practicality."

@EWH: It's OK by me... I enjoy the conversation.

 

[KENAT]

Drawing says what is acceptable. So strictly yes R.002 +.003/-.000 communicates exactly the same thing as R.0035 +/-.0015.

You may get a machinist that believes that with unilateral tols you try and get as close to the nominal value as possible but can't guarantee it.

Like I said, if .005 isn't acceptable then adjust the tolerances, and if need be find a different process or change mating parts...

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[Belanger]

I still say be careful. Sure, Kenat, that works for onesies/twosies, but if the OP is talking about a mass-production environment, then that cavalier philosophy will goof up the tolerance stacks. This is why the Monte Carlo method is so useful -- it can assign different "weight" to those variables in a stack that are favoring more on the plus (or minus) side.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

http://www.gdtseminars.com

 

[powerhound]

KENAT,
I don't believe that the two callouts communicate the exact same thing. While the tolerance zone is the same width and disposed between the same dimensions, the message is different between a unilateral and bilateral callout..

Powerhound, GDTP T-0419
Engineering Technician
Inventor 2010
Mastercam X5
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[ajack1]

It still seems to me that you are debating fly poo.

The manufacturer has already stated that they cannot hold a +/- .001 limit so in order to be sure that all parts are over .002 radius they will have to aim for something over .003 radius.

Personally I very much doubt that an extra .0005 on either side of the limit will mean that they can consistently produce good parts on an etching process as the radius will never be “true”. Although how you are going to measure a radius that isn’t a true radius to within 0.0005 I have no idea.

 

[KENAT]

Powerhound, from an inspection pass/fail criteria they do communicate the same thing. If push comes to shove that is what an Engineering drawing to ASME stds does - define what will be accepted. Per 1.4 of the 94 std it generally doesn't relate to the production process.

Now you can argue that in this case the process materially affects the end performance ... but I think that's pushing it.

Out of interest Powerhound & Belanger, where in the standard are you interpreting it to say that unilateral tolerancing implies that you get as close to the nominal value stated as possible rather than aim for the middle of the range or other variations?

As ajack & others suggest it seems the real problem is that they are on the limit of their process capability. In which case .00X MAX again starts to sound like the way to go where X is less than the 5 which is apparently problematic to the OP's boss - I don't think we established that 0 radius was functionally unacceptable did we?

Or perhaps you address this with the process spec suggested above, or find a way to live with .005...

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[fcsuper]

I would also question the design intent on trying to hole such a tight tolerance for a radius in the first place. That can be extremely expensive, even for parts that engage each other.

There is no such thing as a goal in any standard. The very concept shows a radical misunderstanding of reality of tolerances. The best way to communicate the reference goal is as you alrady stated above. State the .002 as the nominal.

Matt Lorono, CSWP
Product Definition Specialist, DS SolidWorks Corp
Personal sites:
Lorono's SolidWorks Resources & SolidWorks Legion

 

[powerhound]

KENAT,

I'm not arguing the point from an inspection pass/fail criteria, I'm arguing it from a designer to machinist perspective. There are certainly cases where the nominal dimension of a unilaterally toleranced feature is more favorable and the favor declines as you move towards the other end of the tolerance band, for example, the slip fit of a shaft in a hole. A shaft may be toleranced +0/-.025 and the hole +.025/-0 with .025 of clearance built into the parts at MMC. If both parts are produced at MMC they will have .025 of clearance and less slop between them at assembly. While this condition is favorable, there still must be some tolerance allowed so .075 of clearance is allowed in the design but there will be more slop between the parts. You can aim for there always to be .050 of clearance between the parts and be okay but if you can consistently hit .050 of clearance then you can consistently hit .030 and since that is what the designer has preferred, why not just do it? If you process is not capable of consistently holding the .050 but staying within the tolerance zone, then that's when you should leave it alone.

Powerhound, GDTP T-0419
Engineering Technician
Inventor 2010
Mastercam X5
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[Belanger]

Kenat, you asked "where in the standard are you interpreting it to say that unilateral tolerancing implies that you get as close to the nominal value stated as possible..."

Wouldn't it be implied from paragraph 1.3.63, which says that the tolerance is permitted from the specified dimension? Also see paragraph 1.3.22, which states that a dimension defines the form, size, orientation, or location of a feature.

Put the two statements together: the dimension says "here's what I want," and the tolerance says, "here's what I'll accept." Sure, you are legal to make the part anywhere within range, but if all those parts please the customer equally, then everything on every print should be limit dimensions.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

http://www.gdtseminars.com

 

[axym]

I would generally agree with Kenat on this one. A designer might try to imply a certain target or preferred value by selecting a certain nominal, but it's not enforceable. At the end of the day, the tolerance defines goalposts.

J-P,

I'm not a designer, but I would hesitate to infer a desired target from the nominal dimension that a designer has chosen. I have heard of other reasons why a certain off-center nominal might be selected, such matching a common tool size and not wanting to modify the CAD data.

If the dimension says "here's what I want" and the tolerance says "here's what I'll accept", then the logic of unilateral tolerances could be questioned. A unilateral tolerance would say "I won't accept something that's on the wrong side of what I want, even by a millionth of an inch". I've even seen +/+ and -/- (unequal unilateral?) specifications, in which the specified value was outside of the tolerance range. This would say "I won't accept what I want".

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[KENAT]

Perhaps I'm thrown by having seen metric type fits where the nominal shaft/hole size or similar is given and then the tolerances based on that - sometimes the -/- or +/+ that axym mentions.

Sorry Powerhound & Belanger, you still haven't convinced me. It seems you're still trying to take advantage of aspects of the human psyche or something rather than what the standard explicitly says.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[Belanger]

Good point about the plus/plus tolerance, Evan.


John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

http://www.gdtseminars.com

 

[powerhound]

Evan,

"A designer might try to imply a certain target or preferred value by selecting a certain nominal, but it's not enforceable."

This statement makes it sound like you think I'm trying to say if the target dimension is not met then the part is not good. Again, if the part is in tolerance--regardless of how the tolerance is disposed from nominal--then it should be accepted. The only thing that should be enforced is whether or not it is in tolerance.

"I'm not a designer, but I would hesitate to infer a desired target from the nominal dimension that a designer has chosen."

Does this mean you would infer that even though a tolerance is shown as 0/-0.5 that what the designer really meant was +/-0.25 but just didn't know how else to express it?

"I've even seen +/+ and -/- (unequal unilateral?) specifications, in which the specified value was outside of the tolerance range. This would say 'I won't accept what I want'."

The +/+ and -/- are an exception to this for obvious reasons. It makes no sense to me why this even exists as it causes confusion and scrap parts. I used to work at a place that had JIS prints like this. There was one part in particular that we would scrap once or twice a year because a new machinist would automatically see +/- instead of +/+ and he would adjust his cutter comp to put the dimension right in the middle of the values which always resulted in a scrapped part. It seems like the next order of that part would come around right about the time everyone forgot about the last time it was scrapped and always with a new machinist and we would go through the drill all over again.

As I said in a previous post on this thread, where I live, most machinists will machine a part as close to the nominal dimension as they can because they interpret a unilateral tolerance as the designer telling them what they would prefer but also what they will accept if they don't get what they prefer.

Powerhound, GDTP T-0419
Engineering Technician
Inventor 2010
Mastercam X5
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II