Re: OMS VME-8 motor controllers: PAL upgrades

[RIVERS@BNLX26.NSLS.BNL.GOV (Mark Rivers)]

Joe Sullivan writes:
>Currently the OMS boards must be polled in order to provide dynamic motor 
>position readback. This would make using the 'operation complete' 
>interrupt redundant. The (as yet unavailable) OMS VME58 controller places
>the current motor position in dual ported RAM. The driver for this board
>could make effective use of the interrupt capability.

The interrupt capability can be useful even with the current board. For
example, I believe the motor device support routine polls the OMS board at 10
Hz. This means that if I tell the motor to move, and then want to wait for it
to get done before doing something else, I will suffer an average latency of
50ms and a worst-case latency of 100ms. If interupts were used on 'operation
complete' it would reduce the latency dramatically. The interrupt would cause
a poll and record processing.

Note also that the motor record device support does not use interrupts to
determine when the board is ready to accept another character in its input
buffer, or when another character is available in the output buffer. Device
support sits in a tight loop waiting in both of these cases. Has anyone done
any measurements to determine the average and worst case times for the OMS
board to accept/output another character? If they are longer than a few
hundred microseconds, then the device support should be rewritten to use
interrupts for passing the strings to/from the board.
 
____________________________________________________________
Mark Rivers                             (312) 702-2279 (office)
CARS                                    (312) 702-9951 (secretary)
Univ. of Chicago                        (312) 702-5454 (FAX)
5640 S. Ellis Ave.                      (708) 922-0499 (home)
Chicago, IL 60637                       rivers@bnlx26.nsls.bnl.gov (Internet)