What are electrical test guardbands and what are they for?

EESemi.com FAQ

- questions most frequently asked by visitors of www.EESemi.com

Electrical Test Guard Bands

What are electrical test guard bands and what are they for?

One may think of electrical test parameter guard bands as built-in allowances or tolerances for whatever errors will be accumulated during the entire parameter testing process. Guard bands are implemented to make production electrical testing more conservative (i.e., slightly over-rejecting) relative to published datasheet specs, and decrease the probability that the customer will get products that do not meet datasheet specifications.

In most semiconductor testing operations, two versions of test programs are used - one is the production test program and the other is the quality assurance (QA) test program. The production test program is used for the electrical testing of all the units on the production line, while the QA test program is used to test a set of samples of electrically tested good units just before their batch is shipped to the customer.

The QA test is done to ensure 'for one last time' that the electrically good units are indeed good. Since the QA test is done on units that already passed production testing, devices are always expected to pass QA testing. An outgoing QA test failure should therefore put a lot on hold until it is cleared for shipment (if at all).

The QA test program is designed to test the device to its published datasheet limits. The production test program, on the other hand, employs a more stringent set of test limits, and is therefore tighter than the QA test program. The amount by which each production test parameter limit is tighter than the QA test limit is the guard band for that parameter. Note that many test parameters have two-sided limits, having both a lower and upper limit. In such cases, both the lower and upper limit must be guard-banded.

Now, why is there a need to put a guard band between each production and QA test limit? The answer lies in the fact that no two electrical test systems are perfectly identical. One will always be tighter than another. As such, two test systems testing the same unit can give different quantitative results. The difference in quantitative test results can swing a 'pass' into a 'fail' or vice versa if the device tested is marginal. In fact, even the same test system can give a different result every time it tests the same unit.

The differing test results from test systems are due to errors contributed by many, many factors. The fact that the measured characteristics of a device can vary from one tester to another or from one point in time to another necessitates that a tolerance for total testing error is introduced into the system.

This is the purpose of the guard bands - to make production testing stricter than QA testing, in such a way that all units passing production testing are good enough to 'always' pass QA testing, whether or not the tester used in production testing and that used in QA testing are the same. This ensures that the customer will only get products that conform to the device's published electrical specs.