On Tuesday 11 January 2005 16:40, Jon LaBadie wrote:
>On Tue, Jan 11, 2005 at 03:40:47PM -0500, Brian Cuttler wrote:
>> Amanda user list,
>>
>> I've spoken to Harry at Condre systems and he confirmed what we
>> where seeing. The jukebox robot is a narrow bus so that are two
>> changes we need to make.
>>
>> 1) We want to move the jukebox/SDLT to be the last device on
>>    the daisy chain.
>>
>> 2) We want to terminate the jukebox-robot, so we need to make
>>    certain that the live port on the box hits the SDLT before
>>    we reach (electically) the robot.
>>
>> Hopefully this will have a positive effect on both SDLT drives.
>
>Not a scsi expert here, but if you terminate the jukebox, wouldn't
>that terminate only the narrow portion of the bus leaving the extra
>lines of the wide bus unterminated.  Might there be some kind of
>device/cable to go from wide -> narrow terminating the unused lines?
>
>Also, I think that if both types of devices exist on the same bus,
>the lower performance one determines the performance of the entire
>bus.  That narrow jukebox may hurt your sdlt performance just by
>being on the bus.  Some of my scsi adapters have multiple channels
>and buses.  Perhaps you could run the sdlt's on one channel, wide,
>and the jukebox on another, narrow or wide.  I added a cheap second
>controller from a dead system for a very similar purpose, external
>devices that were narrow when the main controller was driving only
>wide devices.  Its two channels were used for different speed
> devices.
>
>jl

Generally speaking Jon, thats very good advice. In any event, where 
there is a transition to narrow devices, the unused lines must be 
properly terminated. To many folks forget that a a scsi bus is indeed 
an rf transmission line, subject to the usual rules about vswr.  And 
double handicapped because the cable is, compared to a piece of well 
built coax, pretty much a guestimate as to its operating impedance, 
usually quoted as being in the 120 to 130 ohm territory, hence the 
commonly used resistive termination of a pair of resistors attached 
to the data line, a 220 ohm connected to the +5 volt line, and a 330 
ohm connected to ground.  It works quite well provided the resting 
voltage on the data lines can be held well above the 2.6 volt region, 
preferably at nearly 3.0 volts, giving an equal noise margin for both 
a logic zero and a logic one.  Having that isolation diode in series 
with this feed often drops that voltage into the very low noise 
margin region below 2.6 volts.  At that point you start sacrificing 
virgins etc to make it work.

As to the whole chain being restricted to the speed of the slowest 
device, I've heard that, but it doesn't, on the face of it, make a 
lot of sense to me given that the drivers are properly written to 
auto-detect whats narrow and slow, and whats wide and fast.  But 
thats a given I wouldn't swear about, at maybe...  Driver quality 
does vary unfortunately depending on the authors whims and expertise 
at the time.

If Brian has an old card he can stick in just to run the robot, it 
sure would be worth the try.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.31% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attorneys please note, additions to this message
by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.