# % STUDY VARIATION VS. % TOLERANCE

#### abichou

##### Starting to get Involved
Hi everyone,

So i'm trying a new product with a tolerance from other products ,because we still don't know the tolerance for this new product.

Exactly we want to see if our measurement system which is used for other products is capable of measuring this new Product or no.

After performing the MSA i got a% Study Variation of 2.81% and a % Tolerance of 18.26% as you can see the difference is high what does it mean?

Any help would be appereciated, Thank you in advance.

I will add also i got a number of distinct categories of 50 and a number of distinct categories of 0.08.

#### pziemlewicz

##### Involved In Discussions
What was your setup? Maybe more parts and/or operators would help...

#### abichou

##### Starting to get Involved
What was your setup? Maybe more parts and/or operators would help...
Hi the setup was 3 operators 10 samples each measures 2 times the same sample.

#### pziemlewicz

##### Involved In Discussions
Are you using MiniTab? Can you share the below table to see if something stands out?

#### Miner

##### Forum Moderator
Hi everyone,

So i'm trying a new product with a tolerance from other products ,because we still don't know the tolerance for this new product.

Exactly we want to see if our measurement system which is used for other products is capable of measuring this new Product or no.

After performing the MSA i got a% Study Variation of 2.81% and a % Tolerance of 18.26% as you can see the difference is high what does it mean?

Any help would be appereciated, Thank you in advance.

I will add also i got a number of distinct categories of 50 and a number of distinct categories of 0.08.

This means that your part variation is very wide compared to the tolerance. Either your process is not capable, or the parts do not represent the actual process variation.

#### abichou

##### Starting to get Involved
This means that your part variation is very wide compared to the tolerance. Either your process is not capable, or the parts do not represent the actual process variation.
So what should i do in this case go with higher tolerance or reduce the variation between the samples?

#### pziemlewicz

##### Involved In Discussions
Your biggest variation is part-to-part, with sample outside control limit. This suggests your tolerance is not wide enough.

I would look at 30 samples, measured one time each, and evaluate CPK for current specification. What does the specification need to become in order to make the CPK > 1.66? Can your application tolerate that variation?

#### abichou

##### Starting to get Involved
Your biggest variation is part-to-part, with sample outside control limit. This suggests your tolerance is not wide enough.

I would look at 30 samples, measured one time each, and evaluate CPK for current specification. What does the specification need to become in order to make the CPK > 1.66? Can your application tolerate that variation?
So you are saying that the main problem is coming from the tolerance right.
One more thing like i said as a new product that has no specified tolerance can i ignore the P/T value and only present the gage R&R value which is perfect 2.81.

#### pziemlewicz

##### Involved In Discussions
You really can't do a GRR without tolerance. The gauge you use must be capable of measuring to the tolerance band specified. For argument's sake, assume you have a mechanical part with one inch nominal: depending on tolerance, you may select to measure with ruler, caliper, vision system, or micrometer.