Surface roughness comparisons (profileometer)

[QAFitz]

I was recently told that when using a surface roughness tester that as long as the results are within +/- 4, the results are acceptable. For instance, on the same sample one machine gets a reading of 32 and another gets a reading of 29 both are accurate/acceptable, or, when checking a "known" if the machine is within +/- 4 of the "known" the machine is giving accurate/acceptable readings. Will someone please:
1) validate this or
2) list the proper +/- variation

ps: We know that when a drawing requires a certain finish (i.e., 32) the finish has to be at least that good (i.e., 32 or lower).

Thanks in advance for your time.

 

[IRstuff]

And just how are total strangers supposed to know what your company's acceptance criteria are? Why aren't you asking the person who gave you this information?

On the face of it, your test procedure is non-compliant, but that is really dependent on the exact wording of the requirement, and other information that only your company has.

TTFN

FAQ731-376

 

[CoryPad]

Surface roughness testing isn't different from other testing types. You have product tolerance requirements, you have testing accuracy/precision/repeatability, etc. There is no blanket statement regarding a +/- value that is acceptable. Also, surface finish can be defined in so many ways (Ra, Rz, length of measurement, etc.), it is impossible to generalize.

Regards,

Cory

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

 

[IRstuff]

From a strict constructionist perspective, your testing procedures are noncompliant to the specification. Since your instruments have a +/-4 tolerance, you could have an actual roughness of 40, which reads as 36, which is allowed by your test limits. But the part would be noncompliant, being actually +8 over the specification.

From a strict constructionist perspective, your upper test limit should actually be 28, but what that would mean is that you potentially throw out gobs of product, which is bad for the bottomline.

The correct answer to your specific problem is that your processing limits need improvement. Your process should be designed to result in no more than 28, which allows for the +4 error in the testing.

Alternately, you need a tester with better accuracy.

TTFN

FAQ731-376

 

[EdStainless]

Get a copy of ANSI standard on surface texture and lay. These methods are known for being in-precise, but there are rules to follow.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[QAFitz]

Thanks.