Are your data pass/fail or do you have continuous measurements that are compared to a spec? If it's the former, you'll want to use the One Sample Proportion calculator, and you will need a lot of samples (hundreds). If your data are continuous, you can probably make the same quality claim with far fewer samples. There are quite a few ways to do with continuous data. One approach would be to calculate a 98% tolerance limit and show that it is contained within your spec(s).