Table of Contents
Advertisement
9 8640A Analog Input Interface
Why does the A-to-D card use a pipeline? Not, as you would expect, to make successive readings
happen faster.
It
turns out that single, non-pipelined readings could be made every 18 microseconds,
the same rate as for pipelined readings. The problem with non -pipelined readings is that they would
tie up the backplane for the full 18 microseconds, and the backplane simply doesn't allow that; two
microseconds is about all that is available at a time. By breaking the analog read process down into a
three-stage pipeline, the card can always be able to release a reading and take in a new address in that
2 -microsecond window.
Once you understand that the pipeline exists, the effect it has on the way you take readings is fairly
straightforward. For any series of readings, take two more than you need, and throwaway the data
returned by the first two.
3-26
Advertisement
Table of Contents
