<> At 08:51 AM 2/2/98 -0700, emanuel stiebler wrote:
<> >What makes this scsi so expensive ?
<> >Is there a patent on this MSCP, TMSCP protocols ?
<> >
<> >I'm thinking about developing one for the q-bus, should be cheaper tha
<> >.., and i could use my disks here.
<> >(its not my first one, i build such stuff on vme-bus before)

First off the controller does not have to do MSCP unless you wish to use 
DEC software.  So by eliminating that it means that NetBSD will provide
direct support.  Same applies to the VS2000.  There is no need to emulate
MSCP there is a need to present a logical interface to the kernel and
deal with bad sectors but that I presume is a driver problem.

Second, why scsi?  Why not IDE it's cheaper hardware, capable of better
performance than the RQDX3/mfm drives and requires a simpler controller
than SCSI.  IDE offers one thing that SCSI does not.  New surplus 1GB 
drives can be had for less than $100.  If IDE were used a fair 
amount of the basic logic on the DRV-11B could be done using a DRV-11P 
foundation module as a proto and then commit to circuit board layout.

What the board must have/do:

 DMA for transfer performance:
  SCSI/IDE would be slow with PIO, though IDE would be better.
  DMA for Q-bus only support 22bit(4Mb) transfers.

 IDE or SCSI bus
	IF SCSI likely local intelligence (z80,8751,8086, whatever) 
          will be needed to manage the protocal.  This adds cost and
	  design complexity.
	IF IDE only a hardware interface with DMA will be all that's 
	 needed.  

Drivers to actually make the board work for netbsd.

Both IDE and SCSI drives are smart enough to do part of the sector 
sparing work for a OS.  However if it is inadaquate there will have 
to be a protocal for that in the driver/kernel as well.  In either case 
the interface would be fully understood and there would be no excuse 
for a weak driver.

IDE offers a far lower cost interface.  It should be considered.


Allison