Advanced Simulation Technology inc.
ORTT Decision to Centralize Panel Logic Software
ASTi examined two options: placing the state machine on-board the microcontroller PROM within each panel, or placing each state machine at a central location with networked communications providing final command and control actions.
The use of custom state machines, embedded within each of the panels would yield:
  1. High risk due to still unknown software tasks and limited microcontroller and memory resources.

  2. Changes to the state machine software would require the PROM in each of the panels (of each unique type) to be replaced.

  3. Changes to the network communications software would require the PROM in each of the panels to be replaced.

  4. Panels could not achieve final release and testing until each of the individual state machines were completed.

  5. Unique, on-board state machine code would result in 12 separate interface board assemblies, one for each unique panel type.

  6. Low latency measured from key or switch closure to the resultant action was nearly ensured by this architecture.
The second option, the use of a centralized processing node which is used to execute each of the panels state machines offered:
  1. Relatively unlimited processing and memory resources.

  2. Changes to the state machine software would only require that the code resident within the processing node be changed.

  3. Changes to the network communications software would require the PROM in each of the panels to be replaced.

  4. Panels could achieve final release and testing prior to completion of each individual state machines by using a multi-purpose Panel Interface Unit (PIU) and test state objects.

  5. Generic PIU code would result in only a single interface board assembly.

  6. Latency measured from key or switch closure to the resultant action was unknown.

  7. A new switch state data transport protocol would be required to support off-board processing.
In consideration of the risks and issues noted above and of resources and schedule, we decided to press ahead with development of the centralized processing node architecture, thus, the Model Server was born.