AHA-2940 Ultra/Ultra Wide
Taras M. Dowhaluk
tarasd at visiondb.com.au
Wed Sep 17 05:19:41 PDT 1997
> Date: Wed, 17 Sep 1997 13:58:44 +0200
> From: Klaus Steinberger <Klaus.Steinberger at Physik.Uni-Muenchen.DE>
> Organization: Accelerator Lab LMU and TU Munich
> To: "Taras M. Dowhaluk" <tarasd at visiondb.com.au>
> Cc: aic7xxx at FreeBSD.ORG
> Subject: Re: AHA-2940 Ultra/Ultra Wide
> Taras M. Dowhaluk wrote:
>
> > On advice from an Adaptec FAQ, I disabled "Initiate Sync
> > Negotiation" (what ever the hell that is) and set the "Max Transfer
> > rate" down to 10mbps on all devices, and rebooted.
>
> On SCSI bus, transfers could be asynchronous and synchronous.
> Async transfers are slower, maximum is 5 Megatransfers per second
> (that's 5 Mbyte/sec on narrow SCSI).
>
> The "Max transfer rate" setting is only useful and active for
> synchronous transfers.
The thing is that at one point when I had both "Initiate Sync
Negotiation" and 20Mbps xfer rate enabled I was getting errors both
at boot and later. When I disabled Initiate Sync/rebooted/re-enabled
Initiate Sync/rebooted, why didn't I get my boot-time and later
errors again, ie. I have returned the scsi card settings to how they
were when I got the timeout problems but no timeout problems. There
has been no change to the scsi chain.
Could we just put this down to the AHA-2940UW 'getting confused' and
needed to be reset ? Firmware is software after all.
>
> > Errors went away, beaudy !
> You should check for cabling and or terminator problems. Async
> transfers are more forgiving against such problems (like cable
> length, too much or too less termination, bad cables).
>
> Also you need definitly active terminators for 10 Mbytes/sec and
> more.
>
> > To prove a point I then re-enabled "Initiate Sync Negotiation" and
> > they did NOT come back ! What the ???
> > I then bumped the speed back up to 20mbps, and lo ! still no errors
> > at boot.
> As I said above, "Max transfer rate" has no effect with asynchronous
> transfers.
>
> > If anyone has got some 'real' scsi bus exercising (exorcising ?)
> > software, pls email the location.
> Cable problem could not be exercised with software. You should check
> for the following things:
>
> 1. Cable length: A maximum of 3 Meters for up to 10 Mbyte/sec, and
> better much less on higher speeds.
1 metre ribbon-cable, no twists.
>
> 2. Check that the bus is __only__ terminated on its far ends, so
> that you have only 2 Terminators on the bus, and they must be
> on the end!
AHA2940UW at one end, Iomega supplied terminator at the other. Now
I'm assuming that passive means no power to it, eg. a plug, and
active means powered, eg. scsi-card, disk drive et al.
>
> 3. Check that you don't have "T" structures on the cabling.
None.
>
> 4. Check that you have active terminators on both ends. Some drives
> have passive terminators built in, even if they could transfer
> at Fast rates! Your adaptec controller has active termination,
> and does the right things even if its in the middle of the bus,
> as long as you let the default (automatic termination) enabled.
AHA-2940UW has auto terminate. Other end is the Iomega terminator.
Disk drives have jumpers for scsi-id only, other jumpers removed.
>
> 5. If you have a mixed wide/narrow cable you have to be very careful
> about termination of the upper 8 bits, as they must be terminated
> at their ends too, but not all adapter connectors or cables do
> this for you!
narrow ribbon cable.
>
> Sincerly,
> Klaus
>
> --
> Klaus Steinberger Beschleunigerlabor der TU und LMU
> Muenchen
> Phone: (+49 89)289 14287 Hochschulgelaende, D-85748 Garching,
> Germany
> FAX: (+49 89)289 14280 EMail:
> Klaus.Steinberger at Physik.Uni-Muenchen.DE
> URL: http://www.bl.physik.tu-muenchen.de/~k2/
>
regards, taras
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Taras M. Dowhaluk
Director - Technical Operations
VisionDB Pty Ltd
Sydney, Australia
email: tarasd at visiondb.com.au
www: http://www.visiondb.com.au
www: http://www.biz.com.au
voice: +61 2 99226615
facsimilie: +61 2 99568452
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
More information about the aic7xxx
mailing list