Tester Variability

Guest

hi, im a new test engr. if someone said the customer return is due to 'tester variability', what does this mean? is this related to guardbands?

iluvher

Joined: 20 Jun 2026 Posts: 20

Iluvher,

I believe this is related to GB's. When there is large tester variability, a good device on one tester may become bad on another one. So its possible the customer return passed testing because its tester was not tight. If you make the guardbands large, then you ship only non-marginal parts, and prevent customer rteurns.

Test Engg 99

Guest

Customer returns can not be due to inhouse tester variability alone. If there’s a return, it means you are not correlated with the test result of your customer. If your customer’s setup is correct, that means the tester you used is not accurate. It’s possible your setup is correct so in that case your customer’s setup is tight. You must correlate with your customer’s test method asap. I suggest you agree on a common bench setup and agree on what the ‘true’ measurement is. The one with the wrong measurement (farther from the bench reading) must correct the problem with his testing.

SCUBA

Joined: 20 Jun 2026 Posts: 20

Scuba,

My explanation is assuming that the customer is correct, because their board is not working. So if a part is a reject, how did it get to the customer? Because the tester allowed it to pass. How? Because the tester used is not tight enough.

If the guardband is large, then the production program will screen out the marginal units before they reach oqa. Remember too, oqa is just sampling, so this unit may not be the one sampled.

Test Engg 99

Guest

You can put a very large guardband for a test, but that will not assure no customer returns. A large guardband will prevent oqa failures, but if all your testers are overaccepting, then you can still ship units that will fail with the customer. You really need two things to assure no customer returns due to testing: 1) tester accuracy and 2) adequate guardbanding.

SCUBA