Fractional dimensions 5

[paulyg]

My company has a long history of designing all in fractional or architectural dimensions, except the rare case where you need a tolerance to the thousandth. We deal with mostly castings (sand castings) and fabricated (welded) items. Being a younger engineer I was only exposed to working in decimal until I started here. My personal rule of thumb up to this point has been to design anything machined all in decimal and anything fabricated, especially big stuff (some of the fabrication you could fit inside) in feet & inches (architectural). I am dealing with more and more castings now and am note sure if I should continue to use fractional measurements or decimal. Because of the loose nature of sand casting only two place decimal is really needed. The head of our department doesn't care which method I use. I am curious what the collective experience here has to say about this choice. What is your personal or your company's position? Pros & cons of either choice?

 

[CoryPad]

Fractions don't lend themselves well to inspection, tolerances, and significant digits. This is one of the reasons why they are not used in complex mechanical systems like automobiles.

Regards,

Cory

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[KENAT]

ASME Y14.5M-1994 (1.6) only deals with decimal dimensioning, doesn't mention inch fractional that I'm aware of.

You talk about only holding to 2 decimal points but this is meaningless withouth the title block tolerances. I've sean tol blocks where 2 dp was +-.005, more typically +-.010.

The number of decimal points only indicates tolerance if your title block is set up that way. In my last company in the UK the number of decimals had no significance. We assigned a singe general tolerance, typically +-.25mm and every dimension that required a different tolerance had it indicated on the dimension.

You can have a 1.375 dimension and apply +-.050 to it if it makes sense.

KENAT, probably the least qualified checker you'll ever meet...

 

[MintJulep]

I continue to be confused at why people believe that there is some association with the format or number of significant digits needed to describe a nominal dimension and the necessary tolerance of that dimension.

Tolerance is the amount of part-to-part variation that is acceptable without preventing proper function of an individual part, or its parent assembly. It is a design function, and is separate from and unconnected to the nominal dimension.

If I have a part that must have a feature within the range of 20.200 and 20.206 to function the requirement does not change if I call the nominal dimension 20.203 or 20-13/64.

Or maybe the part must be between 19.997 and 20.003 to properly function. The nominal is 20.

The fixation on number of decimal places "defining" the required tolerance - as associated in a typical "standard title block" - is simply laziness, indicating that the engineer or designer could not be bothered to actually engineer the tolerances, and has simply made a guess, or more likely given no thought to tolerance at all and has simply let the CAD program variable the sets the number of significant digits decide the tolerance.

 

[ctopher]

I would never use fractions. They can be rounded either way to the nearest .XX, .XXX or .XXXX, depending on who is working with them. Some machinist's don't know how to convert them to decimal.
I agree with others comments above.

Chris
SolidWorks 07 4.0/PDMWorks 07
AutoCAD 06
ctopher's home (updated 04-21-07)

 

[ewh]

"The fixation on number of decimal places "defining" the required tolerance - as associated in a typical "standard title block" - is simply laziness"

Give us a break...
Block tolerances are usually found to be reasonable manufacturing tolerances, and are no different than if every dimension had its own tolerance. The design should allow for this. It serves no purpose to put a tolerance behind every dimension if the part was designed with the noted tolerances in mind. Drawings are supposed to be CONCISE, not verbose.
When and where a block tolerance is not appropriate, then call out something different.
YOU did not write the Y14.5 standard. If you had, I'm sure that it would be a very different animal.

 

[KENAT]

The laziness is if you don't verify the block tolerances make sense/support/are driven by function, which sadly is very common.

Lots of parts here either have potential interferences or unnecessarily tight tolerances because people don't think it through and just use the default 3dp our CAD system places which invokes +-.005. On an older format 3dp was +-.002, we had large invar pieces which could easily have been +-.01 or more on overalls held at .002 because the person doing the drawing didn't think about it!

Using block tolerance and checking to see if any dimensions can/need to be different is not laziness, just trying to keep a tidy drawing.

You think standard tolerance block is bad look at ISO 2768, especially the last paragraph in the notes.


KENAT, probably the least qualified checker you'll ever meet...

 

[ringman]

If my memory serves me correctly, we were directed by Mil-Std-100, and some other standards, some 50 years ago to use decimal dimensioning without making it an equivalent to a fraction. That is: a new designed feature would be .30 or .32 perhaps rather than .312 which is the dec equivalent for 5/16. The exception being that if the feature were driven by a fractional tool size.

This was not fully applied then and most probably not applied today by CAD operators.

Anyone else recall this?

 

[KENAT]

ASME Y14.5M-1994 only talks about metric and decimal inch, it doesn't include fractions, as I posted before. Is this kind of what you mean?

KENAT, probably the least qualified checker you'll ever meet...

 

[ringman]

I should have mentioned that decimal dimensioning was not DIRECTED TO APPLY to Architectural nor Civil engineering drawings.

 

[ringman]

Kenat,

Not quite. The Standard is directed towards metric dimensioning for the most part, I believe.

BRIEFLY STATED: THROW AWAY THE FRACTIONAL SCALES AND USE ONLY DECIMAL-INCH SCALES.

 

[KENAT]

Most of the examples are metric in 14.5M-1994 but it makes it clear in 1.6 that it applies to decimal inch as well. I don't see any mention of fractional, be it to use or not use.

I agree that 14.5 doesn't apply to architectural etc. However I'd expect it does generally apply to castings which the OP is asking about. Don't know what ASME Y14.8M-1989 Castings and Forgings might say.

KENAT, probably the least qualified checker you'll ever meet...

 

[ewh]

My apologies for straying, but here we go again...
KENAT is correct that the laziness is in the designer/drafter not paying attention to the accuracy for which he is asking. Having a tolerance on each dimension is no guarantee against this. A checker would (should) catch these mistakes. But who needs checkers....

 

[KENAT]

On topic:

I had some foam drawings the other day that didn't call for more than +-.03 (which we invoke with 1 dp), in fact this may have been a bit tight. The designer tried to get around it by using fractions for any dimensions that would otherwise be more dp. My preference in this case is to individiually tolerance the dims. Another option is to change the block tol but this can cause mistakes when people are used to the stanard block tol and don't look at differences.

Getting way off topic but...

In my opinion the checker should spot it the first few times a designer does it.

In future the designer should learn from this and catch most 'mistakes' themselves with the checker only catching any they miss, or the designer should be disciplined and eventually fired if they continue to produce sloppy work.

Checkers shouldn't just be catching mistakes/problems, they should be educating designers through their mark ups so that designers make less problems in future.



KENAT, probably the least qualified checker you'll ever meet...

 

[MintJulep]

Give us a break...
Block tolerances are usually found to be reasonable manufacturing tolerances

Sorry, but tolerances have nothing to do with manufacturing.

Tolerances are a design requirement. They must be defined by the engineer.

Manufacturing processes result in variation from part to part.

The purpose of tolerances is to define how much variation from the nominal dimension can be tolerated before the part is no good.

 

[MikeHalloran]

If you ever need to make a drawing for an American carpenter, you will have to use fractions. They just refuse to deal with decimals. Mine did, anyway.

In AutoCAD, you could design in decimal and then set your dimensions, rounding, etc. to fractional values and update the dimensions. There are probably similar capabilities in other packages. I.e., you could work in one system and convert.

Beware of mixing systems, e.g. using .31 or .32 when the real dimension is 5/16. It worked okay in pencil, but in CAD systems it causes awful stackup problems, where lines that should meet, don't, or overlap. You can waste a lot of time finding a .0125" error in a CAD model.

My advice: Get used to fractions, use them where appropriate (as has already been established by local custom)... and get used to the idea that you'll have to adopt a new set of local customs in your next job.

Speaking of your next job, you might as well memorize this now: 5/16" is pretty close to 8mm.



Mike Halloran
Pembroke Pines, FL, USA

 

[MintJulep]

Speaking of carpenters....

Japanese carpenters use tapes that are divided into 33rds of a meter.

 

[btrueblood]

"I would never use fractions"

Remind me never to order you a half pint.

 

[KENAT]

MintJulep you clearly haven't looked at ISO 2768. It's all about manufacturing tolerances and very little about being driven by function. I think it sucks but what would I know.

I basically agree with you Mint that tolerances should be driven first by function. However, manufacturabily should be kept in mind. If function appears to demand very tight tols you should start thinking if there's a way to modify the design so that looser tols are ok.

So I have no problem with block tols for different dp as long as they are used correctly.

Mike, I'm with you on the decimal/fractional modelling, I've seen it cause problems so many times. If you want you use .38 on drawing as 3/8 fine, but model .380 not .375.

KENAT, probably the least qualified checker you'll ever meet...

 

[ctopher]

Kenat,
I agree 100 +/-.001%.

Chris
SolidWorks 07 4.0/PDMWorks 07
AutoCAD 06
ctopher's home (updated 09-17-07)