[previous] [next] [contents]    Fault detection in the BS infrastructure

Two main types of faults may affect the integrity of the BS infrastructure: Faulty TAP interconnects (open or short faults in the TDI, TDO, TMS, TCK or /TRST pins) or faulty / misplaced components. The sequence of operations usually carried out to detect these faults starts with the initialisation of the BS logic in each component (the TLR TAP controller state can be reached from any other state by the application of a 1 in TMS during not more than 5 TCK cycles), which is followed by:
  • A capture-and-scan operation through the instruction registers (IRs) in all BS components.
  • A capture-and-scan operation through the identification registers of those BS components supporting this optional instruction.

The IEEE 1149.1 std states that the result of a capture operation in the IR (taking place in the Capture-IR TAP controller state) must be XX01, where X represents a "dont care" condition and the rightmost bit is closest to TDO. This feature was defined with the objective of enabling the simple integrity check operation shown in figure 1, that consists of bringing the TAP controller FSM to the Capture-IR state and then on to Shift-IR and to scan out the IR contents in all BS ICs.

Figure 1: Integrity checking of the BS chain by capture-and-scan through the IRs.

Since the two least significant bits in each IR are 01, every TDO-TDI connection will be forced to 0 and to 1, which enables the detection of a large spectrum of faults in the TAP pins:

  • Any stuck-at TAP pin will disrupt the 01 sequence shifted out and be detectable at TDO of the last BS IC (the "board-level TDO").
  • Most short-circuit faults involving TAP pins will have a similar effect and therefore be detectable in the same way.
  • Open interconnects among TAP pins are generally easy to detect, since TDI and TMS are defined in the standard as capturing a logic 1 when floating, which again disrupts the sequence and is easily detectable.

Some specific faulty conditions might remain undetected by this simple test, but these situations would normally prevent correct operation in the following steps of the test protocol and as such be rapidly detected (alternatively, a more sophisticated infrastructure test may be specified). An example of one of these specific faults might be a short circuit between TDO and TDI of the same TAP, in a chain where all ICs have the same specification for the IRs. The reader is invited to explain why this particular fault would not be detected by this test and to suggest a way to detect it.

Boards with more than one BS chain may have their BS infrastructures checked independently, but again there are a few specific faults which will only be detected if the integrity check is co-ordinated in all chains. An example of one of these faults might be a short circuit between two TDO pins in different chains, which will be in high-impedance except when scanning is in progress. If one BS chain is in TLR state while the other is being checked, any short-circuit between two TDO pins in different chains has no effect, since only one of them will be active. Again the reader is invited to specify a co-ordinated integrity check procedure that is able to detect the situation referred above for the case of multiple BS chains.

If the capture-and-scan test runs without problems there is a high probability that no faults are present in the BS chain. However, and if there are components supporting the optional IDCODE or USERCODE instructions, it is recommended that these instructions are used to check that the proper components were placed where they are expected to be.