Ultra SCSI capacitance - ag

Kevin Gingerich SN=Kevin_Gingerich%S=Gingerich%G=Kevin%TI at mcimail.com
Tue Jun 14 05:54:00 PDT 1994

Application message id:  TI.msg.432206
Grade of Delivery:  Normal


  To  SCSI Reflector   X400
>From  Kevin Gingerich  GING
Subj  Re: Ultra SCSI capacitance - ag
Basic transmission line physics require that not only the amount of capacitance
but, the distance between them be considered in the circuit response. In an
ideal model, adding 25 pF to a transmission line every 2.3 ft is the same as
adding 20 pF every 1.8 ft. The sin is not the extra 5 pF/node but, placing
devices too close together.
-********** ORIGINAL MSG RECEIVED ON 06/13/94 AT 16:29 FOLLOWS **********-
-MSG M#=   416899 FR=X400 TO=GING   SENT=06/13/94  04:29 PM
     R#=095 ST=C DIV=0003 CC=00618 BY=X400 AT=06/13/94 04:29 PM
To: /PN=Dal_Allan/ID=0002501752/ADMD=MCI/C=US/
    /PN=SCSI/MBX1=SCSI at wichitaks.ncr.com/EMS=INTERNET/ADMD=MCI/C=US/
    /PN=Jim_McGrath/MBX1=jmcgrath at qntm.com/EMS=INTERNET/ADMD=MCI/C=US/
From: /PN=dsteele/MBX1=dsteele at ccmgate1bb.ftcollinsco.ncr.com/EMS=INTERNET/ADMD
Subject: Re: Ultra SCSI capacitance - ag
I have to take issue with the statement that there is no data to support
reducing device capacitance.  Data on this has presented more than a year ago.
We have seen over and over again that device capacitance is THE MAJOR
contributor to the the bus reflections that cause the REQ/ACK double clocking
problem.  This is explainable with basic transmission line physics.  I am not
sure how practical it is to require 20 pF.  However, it is clear that it
desirable if we can achieve it.
Dave Steele
                      Subject:                              Time:  5:39 PM
  OFFICE MEMO         Ultra SCSI capacitance - again        Date:  6/9/94
I read Larry's minutes of our last meeting, and was shocked to see the device
capacitance once again changed to 20 pF from 25 pF.  I realize that I was
stepping in and out of the room, but I did not notice that change.
Once again, Quantum's official position is that no data we have seen compels
us to go with 20 pF - 1.5m cables and 8 devices work fine, 3 m and 4 devices
also work fine.  In practice, desktop systems using Ultra SCSI drives will
probably have old SCSI devices at 25 pF on them anyway, and we are convinced
that 20 pF will cost our desktop customers money that they are not willing to
spend for no reason.
I am getting really pissed off at revisiting this issue, especially since we
see this urge to 20 pF as not being supported by any technical data we have

More information about the T10 mailing list