Not All Xs are Bad for Scan Compression

Scan compression technology combines the expected responses from multiple scan chains to be observed at fewer scan outputs. As a result unknowns (Xs) in the test response interfere with the good values that could be observed. Prior to this paper, Xs in the test response were treated as bad for compression and solutions either removed, bypassed, or blocked the Xs from interfering with the other responses. In this paper we show that some X scan be added to improve test compression quality of results. The trade-off between improved observability due to simultaneous clocking of interacting clock domains is played against the reduced observability caused by the Xs in the response due to race conditions. In this paper we show that when the inter clock domain Xs are added but limited, the gains achieved by adding the Xs far exceeds the losses in bringing together the Xs with other observes in scan compression.