Device Modeling Is Critical for
Fast Time-to-Test

Seeking Opens

Following this first pass, the test generator automatically performs additional passes which seek to identify the operating modes that devices can be placed into so that opens can be tested. Successive passes then add test coverage for opens which was not gained during the test generator’s first pass. Once this information has been processed, additional test vectors for opens are generated and included in the test set.

Of course, during any test-generation pass, the test generator identifies any cluster device’s operating mode that simply must be taken into account. These types of modes are often hard-wired into the design. (See Figure 3 below for an example.) When the test generator encounters these conditions, it will constrain the test so that the hard-wired operating modes are maintained.

Figure 3. Hardwired Modes of a "245" Device

Iterative Passes

After the initial test generation, the test developer may continue to invoke the ScanWorks test generator to make additional passes to further maximize test coverage. At this point, the engineer or technician who is developing the test can decide to provide additional information to ScanWorks, such as cluster models for those devices that have not been modeled. This added information is used by the test generator in these successive passes to obtain higher and higher levels of test coverage.

This type of iterative process gives the test developer the option to invest his or her development time where it will have the biggest payoff in terms of test coverage or other test performance criteria. Quite often, ScanWorks will generate a test with the level of coverage required by the developer. Later, when the developer has time to devote to it, the test can be passed through the ScanWorks test-generation engine again with additional modeling information and the test coverage can be increased.

Throughout this process, ScanWorks continues to evaluate the operating modes selected for cluster devices in light of any possible bus conflicts which may ensue. Should the developer manually dictate an operating mode that might establish a hazardous bus contention, ScanWorks will notify the user of the possibility for board damage. At that point, the developer has the option of continuing with the operating mode originally selected or revising the mode to avoid the possibility of bus contention.

Optimizing Automatic Test Generation

There are many facets to automatic test generation, including time-to-test, test coverage, board safety and requirements for manual developer involvement. Of course, the most effective test-generation engine simultaneously optimizes all of these factors to provide the best balance for the inevitable tradeoffs that arise. For example, a test with high coverage might endanger the safety of the board. The algorithms of ScanWorks’ test-generation engine provide an efficient way to quickly generate high-coverage and safe tests while giving test developers the information they need to further refine the tests generated and decide how best to invest their time.

Previous : Next                    Page 3 of 3          Back to Page 1