# Best Practice for setting tolerances on a Drawing

#### marcusja2002

##### Involved In Discussions
I looked through the forum and couldn't find the information I was looking for, so hopefully the forum can help me out.

I'm looking for information to help guide my engineering department toward a better way to control and standardize our drawings. Specifically tolerances.

They seem to be stuck on the methodology of using decimal places of a dimension sets the tolerances.

I feel this method limits accuracy, for example.

If I need to convert a dimension there is a big difference between .125" and .13". (3.175mm vs 3.302mm). I also have a tough time believing 1-1/8"(1.125) needs to be held tighter than 1-1/4" (1.25).

GD&T is the best, but its \$\$\$\$ to train. I'm trying to suggest an improved methodology without the need for a whole new drafting language.

#### Michael_M

Trusted Information Resource
On the mechanical drawings I have seen throughout my career, for dimensions that do not specifically state a tolerance (i.e. Unspecified tolerance), it uses the number of decimals to determine the tolerance:

.X = ±.030
.XX = ±.010
.XXX= ±.005
Fractions = ±1/16

This is put into the title block of the drawing. When the print puts the number of decimal place needed: .250 would be ±.005 while .25 would be ±.010

This works for non-specific tolerances. If a tolerance is listed next to the dimension it overrides the unspecified: For example .250 ±.001. Other things can be put into the title block such as non-specific surface finish requirements, edge breaks, and fillet radii.

NOTE: I don't know what type of drawings you are dealing with, my career has only dealt with mechanical drawings following ASME Y14.5 standards.

.125" and .13". (3.175mm vs 3.302mm). I also have a tough time believing 1-1/8"(1.125) needs to be held tighter than 1-1/4" (1.25).

From an manufacturing point of view, .125 and .13 are different. Using the listing above, there is a difference between the two dimensions. .125 would need to run between .120/.130 with .125 being the target. .13 would need to be held between .120/.140 with .130 being the target.

Last edited by a moderator:

#### Golfman25

Trusted Information Resource
It should not be arbitrary. It should be specific to the needs of your product and the capabilities of the process your using.

We have more trouble with the arbitrary "box" tolerances because in our process, we can hold some features better than others.

#### SpinDr99

##### Involved In Discussions
I agree with Michael M, and will add: +/- 1 degree for angles.

#### marcusja2002

##### Involved In Discussions
Michael

that bit I understand. I'm trying to figure out if there is a supperiour method out there I could present that will remove the shortcomings I indicated while not needing to move to GD&T.

Super Moderator
GD&T......

#### Golfman25

Trusted Information Resource
I agree with Michael M, and will add: +/- 1 degree for angles.

That's funny right there.

Which is my point. To expect our process to hold +/- 1 degree is a pipe dream.

#### Michael_M

Trusted Information Resource
Michael

that bit I understand. I'm trying to figure out if there is a supperiour method out there I could present that will remove the shortcomings I indicated while not needing to move to GD&T.

I think GD&T will be the best option, when it makes sense to use it. Depending on the part requirements, you can still use the basic design practices with little or no GD&T.

That's funny right there.

Which is my point. To expect our process to hold +/- 1 degree is a pipe dream.

We are a machine shop and holding +/-.5 (1/2) degree is standard across most drawings for us. Our angles are machine controlled so it is not difficult for us to hold.

#### Proud Liberal

##### Quite Involved in Discussions
GD&T is almost always indicated if you're making parts that need to assemble with other parts (especially when things are fastened together) and you don't want to reject functional parts.

#### clark1

##### Registered
I totally agree with the OP - title block tolerances that are based on number of decimal places affects accuracy. It also often conflicts with a CAD principal of always modeling features to the exact nominal size. There must be better systems out there, right?

At my prior company I implemented an alternative - the tolerance was expressed by the color of the dimension text. In my case, we used red for +/-.001, blue for +/-.005 and green for +/-.015. The only downside was that any printouts or copies must be in color.

Has anyone seen any other alternatives?