jmm@skivs.UUCP (Joel M. Miller) (07/19/89)
There is a large class of research laboratory data acquisition and control
(DAC) applications that are not well-supported by any commercial system I
know of. Mainly, these applications require "realtime response" (closed-loop
control) at kilohertz rates, the ability to monitor the progress of an
ongoing experiment, and support for complex and easily modified experimental
protocols. An example of a realtime response task is a visual perception
study in which a subject's eye movements are monitored as he or she views an
eye position-contingent visual display. Such experiments are usually
custom-programmed in a low-level language, resulting in user-hostile systems
that are difficult to modify on a day-to-day basis as an experiment evolves.
Q1: Has anyone thought about the "realtime response" problem and developed
an approach?
I am aware of (at least some of) the problems in maintaining a "live" visual
display during time-critical processing on the Mac II (the video card can
grab the NuBus when the CPU and DAC cards want to talk). The solutions seem
to be: (1) collect data in epochs, and display results only at the end of an
epoch; (2) produce a fast, impoverished, custom visual display; (3) Let the
host CPU and NuBus handle the display, user interaction, and other
"executive" functions (roughly: events that occur at the user's time scale),
and provide a coprocessor with its own bus to handle the experiment's realtime
events. (Note that DMA does not help with the realtime response problem,
since the CPU must be able to compute a timely response to each set of data
samples).
Q2: Has anyone found another way around this problem?
The coprocessor solution is, clearly the solution of choice. One way to
implement it would be to simply connect two Mac II boxes, dedicating one
as the "executive" and the other as the "DAC processor".
Q3: Any ideas on how to best connect two Mac II's for transmission of data
(for display on the executive), commands (to control the DAC proc), etc.
SCSSI? Serial? Digital IO?
Another coprocessor solution would be a (single Mac II with a) commercial
coprocessor, either with on-board DAC devices, or with its own bus to connect
it with DAC cards plugged into the NuBus. (I had expected National Instruments
to use its RTSI bus for this).
Q4: Any info on such coprocessor boards?