Good Dimensioning Practice - 3D to Drawings

[D.Wu]

I am new in the workforce & I am being put in a draftsman position despite me being trained for manufacturing.

Simple but important question here is what is good practice when dimensioning/designing 3D parts. specifically the amount of sig. figs to use in the 3D & drawings.
I work 99% with sheet metal.

For example, with a 3D Sheet Metal part has a flange 2.125" but in the drawing it is indicated 2.13" because of the demands of tolerancing.

Should the 3D sheet Metal part be only dimensioned to 2.13" also instead of 2.125"? Should the sig. figs match everywhere?

Thanks

 

[TheTick]

Make 3D models as precise as possible. You can always reduce precision or add tolerance to the drawing.

 

[3DDave]

If the model will drive downstream processes it's generally best if the model and dimensions match; otherwise you start off with an error.

The exceptions are where the process favors a shift during the process. For example - a waterjet kerf will always remove material to each side of the cutting path. Therefore making the part towards the MMC values will allow the kerf to work towards making the part to the nominal value.

It also depends on any engineering use of the models - changing the dimensions away from nominal may produce misleading results at the assembly level.

 

[ewh]

For the sake of accuracy and to avoid the introduction of added errors (as mentioned above) you can still place dimensions with three decimals (as modeled) and directly tolerance to reflect the (I assume) default two decimal place tolerance, i.e. 2.125 ±.030. This will reflect the "as modeled" part while allowing for a looser tolerance.

"Know the rules well, so you can break them effectively."
-Dalai Lama XIV

 

[D.Wu]

Thanks for all the answers!

 

[djhurayt]

and always remember that zeros cost money
i.e. a shaft dimensioned with a diameter of 1.0 will be less expensive than the same shaft dimensioned with the diameter called out as 1.000

 

[ewh]

djhurayt brings up an important caveat to my earlier suggestion... added decimal places mean more precise inspection methods will be required. For this reason it is best to model to the intended dimension when possible. Sometimes the modeling method does not lend itself to this and rounding of dimensions will be necessary for ease of inspection.

"Know the rules well, so you can break them effectively."
-Dalai Lama XIV

 

[ctopher]

Get copies of standard, (MIL, ASME) or whatever you use. They cover the practices you need.

Chris, CSWP
SolidWorks '17
ctophers home
SolidWorks Legion

 

[TheTick]

Decimal places mean what you (or the title block) decide they mean. With GDT, the mean nothing, and 1.05 means the same as 1.0500000.

When in doubt, use explicit tolerances.

When in doubt, go with a tighter tolerance. It’s far easier to loosen a tolerance than it is to fight for a tighter tolerance.

 

[pylfrm]

D.Wu,

I agree that it's generally best for the model and drawing dimensions to match exactly.


djhurayt, ewh,

Are you saying that 1.0 +/- 0.2 will be less expensive than 1.000 +/- 0.200? If so, could you please explain why?


pylfrm

 

[Sem_D220]

The model should match accurately the nominal dimension, so model it 2.125", and follow ewh's advice if this is the intended dimension.

In other cases, it is important to make the differentiation between nominal size and "middle of tolerance" size. Especially when dealing with standardized engineering fits. Always set the nominal size at the CAD model, even when the tolerance zone is unilateral.

 

[ewh]

pylfrm,
Consider the inspection process. I am not in inspection, but my understanding is that to allow for tool resolution, a two decimal dimension has to be inspected to three decimal places and three decimal dimension to four decimals.
I'm sure that someone will correct me if I got that wrong.

"Know the rules well, so you can break them effectively."
-Dalai Lama XIV

 

[3DDave]

ewh - the assumption in Y14.5 is that every tolerance has an infinite number of trailing zeros.

2.4 "Dimensional limits, regardless of the number of decimal places, are used as if they were continued with zeros."

The precision of the measurement should be proportionally better based on how close to the limit one expects to be, but that's tougher than a simpler rule of thumb.

 

[ewh]

So a three decimal place dimension is not inspected to four decimal places?

"Know the rules well, so you can break them effectively."
-Dalai Lama XIV

 

[SeasonLee]

but my understanding is that to allow for tool resolution, a two decimal dimension has to be inspected to three decimal places and three decimal dimension to four decimals.

This is the 10% rule we followed on our QA lab.

Season

 

[djhurayt]

pylfrm (Mechanical) 27 Jul 18 22:55
djhurayt, ewh,
Are you saying that 1.0 +/- 0.2 will be less expensive than 1.000 +/- 0.200? If so, could you please explain why?

Yes, I am. Mostly in the QC costs. I can measure a 1.0 ±0.2 with an elementary school wooden ruler. In order to measure thousands of an inch I need vernier calipers at minimum. And if I am trying to measure a shaft down to microns we are talking even more expensive gauging.

 

[3DDave]

With a wooden ruler you can tell if the part is at 1.1999999999999999 and not 1.200000000001?

And if the 1.000+/-0.2000 part measures 1.03 with a calibrated wooden ruler are you not confident it is within tolerance?

The costs should go up on two fronts: 1) the range of values is smaller. 2) when acceptance is expected to be very close to the limits.

If the error budget of the inspection is not being considered as part of the rejection criteria then either excess money is spent for no reason or out-of-tolerance parts are likely being accepted.

 

[looslib]

A lot of the 'cost' in producing a part to 1.0200 +/- .0100 instead of 1.02 +/- .01 is in perception. Since the callout of the tolerance has been listed there should be no difference. However, you give 2 prints to different shops, one with 1.0200, the other with 1.02, you will probably get back 2 different prices to make your parts.
If you depend on the title block for default tolerances, then using trailing zeros will change the inspection criteria and the extra cost is more easily justified as the precision has gone up.


"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli

 

[drawoh]

D.Wu,

Learn what tolerance your fabricator can achieve. A sheet metal vendor I have worked with recently claims he can punch repeatably to [±].004", and locate bends to [±].015". That works out to [±]0.1mm and [±]0.4mm, respectively. If you call up tighter tolerances than this, expect your part to be expensive, and/or frequently non-compliant.

--
JHG