inspection significant figures 1

[drdherl]

When a dwg specfies the locational tolerance of part soldered to a board as example +/-.005, is a part that measures .0051 off location out of tolerance? Do typical quality/inspection guidelines mandate that non specified significant figures are zeros as opposed to +/-.0054999.

I find it hard to believe that common sense, especiallly for run of the mill, non safety critical parts that rounding off/rounding down isn't practiced.

 

[NomLaser]

Round the non-significant digits. Use the rules you learned in school about sig. digits.

 

[IRstuff]

Well, but why spec something if you aren't going to adhere to it anyway?

For something like that, the only technical out is if your measurement accuracy can cause a conforming part to exceed the spec.



TTFN

FAQ731-376

 

[handleman]

So where do you stop your rounding? Does ±0.005 really mean ±0.0054?


-handleman, CSWP (The new, easy test)

 

[drdherl]

I would have thought that if I didn't want to permit rounding, I would specify tolerance as +/-.0050. If I would accept rounding, specing +/-.005 would imply (to me) that rounding off +/-.0054 to +/-.005 is acceptable.

Splitting hairs...but I have 3 parts hung up bc of this hair splitting....

 

[NomLaser]

Then the part isn't toleranced correctly. Rounding non-significant digits is what happens when you have a measuring device more accurate than your tolerance. What would happen if your caliper only had a .001" resolution (many still do). They round and .0054" should round down to .005".

 

[CoryPad]

Rounding digits is commonplace since most measuring instruments are more accurate than most production equipment. ASTM E 29 Standard Practice for Using Significant Digits in Test Data to Determine Conformance with Specifications is an example of a consensus standard created to address this issue.

 

[KENAT]

Assuming the drawing is to a drawing standard like ASME Y14.5M-1994 then +-.005 actually means +-.00500000000000000000000 ad infinitum. See section 2.4 os ASME Y14.5M-1994.

So +.0051 is out of specification.

If the part doesn't need to be accurate to +-.0050000000 then the drawing should be changed to show a larger tolerance. The tolerance should be chosen by function while meeting manufacturing capabilities.

As to the inspection specific aspect of this, I'd look in forum286 as I think similar has been asked before.

A problem with rounding is there are different techniques, you'd have to formally standardize a technique.

KENAT,

Have you reminded yourself of faq731-376 recently, or taken a look at posting policies:

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

What is Engineering anyway: faq1088-1484

 

[IRstuff]

I think that you have a separate problem of process control. If you have 3 units hung up because of this problem, then some fundamental assumptions about the ease of fabrication of this step were incorrect. Either that, or you got a production line that doesn't give a hoot, because it knows that you'll knuckle under to ship parts.

TTFN

FAQ731-376

 

[hydtools]

+.0051 is out of spec. +.005 is the upper limit. It does not mean +.005 plus a little more. Be advised, that if you continually accept +.0051 then you should increase your tolerance to indicate what you will accept and admit that the original spec was really not what you wanted.

Ted

 

[NomLaser]

Kenat's right. I stand corrected...good thing that I'm not in quality control I had to pull out my copy of ASME 14.5M-1994 and it's in section 2.4.

"All limits are absolute. Dimensional limits regardless of the number of decimal places, are used as if they were continued with zeros. To determine conformance within limits, the measured value is compared directly with the specified value and any deviation outside the specified limiting value signifies nonconformance with the limits."

 

[IRstuff]

One final comment. Your company obviously ignored the need to create a Test Requirements Document, in which you determine the requirement, the test method, its measurement accuracy, and the acceptance limits.

Once you've done that, you're done with niggling with test results; they are either in spec or not, no exceptions.

Likewise, you are probably violating your ISO-9000 certification with this line of investigation, since your procedures should have already been formalized and approved for production.

TTFN

FAQ731-376

 

[drdherl]

Thanks folks.