A considerable part of the 11/17 meeting was spent discussing data related to non-BScan devices, and much of that was debate related to the possible use of model formats such as STIL.
Looking back on that discussion, I feel we maybe missed the point which was to try to understand what we needed to know about the devices to generate a test (or other operation on the system) - the precise format of that data is, I think, secondary and at this stage I'd suggest that it's up to the tool vendors to determine how best they get that data into their own tooling. It may well be that more than one source is required to provide all the necessary data for any given device.
From the discussion, I will summarise the data items that we did identify (these are perhaps a little vague, and will require refinement):
- A functional model to support cluster test generation
- Control mechanisms for tri-state or output enable
- Conditions that cause a device to be inactive within the system (ie powered off, redundant, etc.)
There will need to be a similar set of attributes derived for use at board and higher level assemblies.