Joined: 20 Jun 2026
Posts: 17
 Posted: Mon Sep 08, 2026 1:24 pm Post subject: Guardbands for 'Go-No Go' Parameters |
 |
| Pipol,
I read somewhere here that 'go-no go' parameters don't need a GB. Now my counterpart is telling me that 'go-no go' parameters must also have a GB. How can you calculate this if the parameter is 'go-no go'? Test Engg 99 |
|
  |
Joined: 16 Jun 2026
Posts: 108
Location: Paranaque City, Phils.
| Posted: Tue Sep 16, 2026 10:30 am Post subject: |
|
| Hi Test Engg 99,
At first glance, it seems that ‘Go-No Go’ parameters can't have guardbands because the tester classifies them into one of only two states, ‘1’ (or ‘Pass’) and ‘0’ (or ‘Fail’). Thus, the result is only one or the other – there’s nothing in between. Nonetheless, your counterpart is correct in the sense that these parameters can be guardbanded within the measurement part of the test block prior to Pass/Fail classification (internal guardbanding). Note that when testing any ‘Pass/Fail’ parameter, the tester still has to measure ‘something’, so that it can classify the parameter as either ‘Pass’ or ‘Fail.’ This ‘something’ is a ‘real’ measurement that you can guardband against, albeit at a level that’s invisible in the datalog print-out. As an example, consider a test block that checks if a digital output pin is functional or not. To do so, it excites the DUT such that a ‘1’ should appear at this pin if the DUT is good. How do you suppose the output gets classified as a ‘1’ or ‘0’? The tester has to measure the voltage level at the output and compare this to Voh, or 2.4V. If it’s equal or greater than 2.4V, then it gets classified to a ‘1’. By setting Voh to 2.6V in your production program and 2.4V in your OQA program, you can actually guardband this ‘Pass/Fail’ parameter internally, with the guardband = 0.2 V. Thus, in this sense, you can indeed guardband a ‘Go/No Go’ parameter. EE |
|
| |