Low Power Differential

ham at subsys.enet.dec.com ham at subsys.enet.dec.com
Tue May 23 08:50:27 PDT 1995

With all the traffic on the pin count for Low Power Differential SCSI 
(LPDS) I would like to call for generating some common vision for the 
application spaces that LPDS would satisfy.  Without this common vision 
I see no way to close on the technical solutions.

Lowest cost consistent with working in all of these chosen application
spaces will drive the answer.

I certainly hope we will drive for a single solution in the interest of 
simplicity and higher volumes.

If we intend to support say 10 meters and FAST 40 in a multicab 
application my initial guess is that we will need differential on every 
line.  For longer lengths and FAST 80 the ability of differential to 
reject cross talk may be key in successful use on low speed as well as 
high speed lines.  As the length is reduced and the grounding 
environments become better controlled, the need for differential is 
reduced.  We sort of know (from FAST 20) that 3 meters with good
grounding control will probably work with a single ended scheme on low 
speed lines. 

Some of the features that will be critical in these discussions will be:

	.Common mode voltage levels supported (The DIFFSENS line limits
         this to around 0.8 volts as it now stands and using single
         ended drivers for any signal further erodes this common mode
         level by the amount off local ground of the driver) 

	.Single segment bus lengths needed (The longer these segments 
         the more important the differential features are to all the

	.Maintenance of the present timings in SCSI-3 for all except 
         data phase (it may require changes elsewhere if we abandon a
         homogeneous transmission scheme for all the dynamic SCSI lines) 

	.Real cost effects of pin/driver counts (Eliminating the 
         external differential drivers is a huge cost reduction -- this
         needs to be quantitatively played against the cost differences
         in silicon/packaging) 

	.Importance of a common SCSI interface chip footprint (uniform 
         pincounts etc.)

	.There are probably other important points I have not listed here

	Until these and related points are agreed it may be difficult to 
close on the technical answers.


					(Digital Equipment)

More information about the T10 mailing list