[mod.computers.vax] DRV11 on the uVAXII

A0061@DK0RRZK0.BITNET (12/05/86)

Yesterday I worked on some data acquisition program for a uVAXII. It
uses a DRV11 16 Bit paralel interface to poll for the status of some
digital input lines. The same type of program on a PDP 11/73 or so
works fine, but on the uVAX I often read some trash instead of the
correct data word out of the DRV11 (using $CRMPSC with page frame number
mapping). Note that no interrupts are involved, just normal device-
register to memory operations. A work around was found by reading the
register twice and if good compare taking the data as valid.

How come ? Any ideas ?

Claus Kalle, Regional Computing Center, Cologne University, West Germany

LEICHTER-JERRY@YALE.ARPA (12/09/86)

    Yesterday I worked on some data acquisition program for a uVAXII. It
    uses a DRV11 16 Bit paralel interface to poll for the status of some
    digital input lines. The same type of program on a PDP 11/73 or so
    works fine, but on the uVAX I often read some trash instead of the
    correct data word out of the DRV11 (using $CRMPSC with page frame number
    mapping). Note that no interrupts are involved, just normal device-
    register to memory operations. A work around was found by reading the
    register twice and if good compare taking the data as valid.
    
    How come ? Any ideas ?
There are several different flavors of the DRV11 board.  At least one - I
think it was the DRV11-W - does not work properly on the uVAX (I or II?  I
don't recall, and you mention both).  I don't remember the details; it had to
do with the designer of the board making an assumption about the timing of
some pair of signals on the Qbus that was not in fact required by the Qbus
spec, but which all previous Qbus masters had happened to maintain - but the
uVAX does not.  There's a very simple ECO for it - something like cutting
one etch and adding a wire to ground one pin - that your Field Service rep
should be able to run down for you.

(Note:  I'm assuming you've checked for obvious problems, like using longword
access to a 16-bit register, or any of the opcodes or addressing modes that
are not legal in I/O space.)
							-- Jerry 
-------