All, Two of my kids are competing in this years BEST robotics design competition. Rules are here, FYI: http://www.bestinc.org/b_game_rules.php The competition is to build a robot which will assemble logic element stand-ins into logic gates and CPU elements (not actual ones, everything is sylized so that the robots can be made out of easily available macroscopic materials). However, an important element of the competition, on which the older child is working, is to write a "business case" for such a robot - do market research, in essence, and convince investors why such a device is useful. When I heard "market research" I immediately thought of this group as probably the best concentration of knowledge anywhere on why anyone would want to build custom-configured logic circuits as opposed to using a generic general-purpose microcontroller for the same task. I *think* I know part of the answer. My thoughts are generally: 1) Anything custom-configured to the task at hand can have number of logic gates, number of transistors, etc. optimized to that task. Register length can match precision of sensors, etc. Thus, *all else being equal*[1], minimum resources (power, heat generated, etc.) will be required by the resulting system. 2) Anything custom-configured can have performance optimized - state machines or flip-flops (vs. combinatorial logic) implemented only when necessary, path lengths shortened where possible, etc. to provide minimum feedback-control times. 3) Anything custom-configured can have minimum complexity. For mission-critical applications, this means it becomes exponentially easier to "prove" correct control system functionality and to predict performance even in the case of SEU or other problems; general-purpose microcontrollers are much more difficult to validate in this sense. I would love comments, and particularly references (places to look on internet, conference proceedings, etc.) from anybody who would care to provide them. Replies on the list to the extent you think they are on-topic, or direct to me and my cc'd daughter if off topic. Thanks in advance! [1] I know microcontrollers have the advantage of gate size *far* below what any discrete components currently can achieve; that's the caveat I'm trying to leave with the "all else" comment. I assume part of the business case has to be that a robot assembling logic designs can be made arbitrarily small so that gate feature size for the discrete logic can compete with gate feature size for large-scale IC's. -- - Mark 210-379-4635 ----------------------------------------------------------------------- Large Asteroids headed toward planets inhabited by beings that don't have technology adequate to stop them: Think of it as Evolution in Fast-Forward.