Document 99-111r1 on Cable Imdedance

Vince Bastiani vince_bastiani at
Wed Jun 2 09:07:14 PDT 1999

* From the T10 Reflector (t10 at, posted by:
* vince_bastiani at (Vince Bastiani)
Revision 1 of document 99-111 is a proposal to specify the impedance of
SCSI cables for SPI-3. The intent is to take the two tables (Tables 10
and 11) in the current SPI-3 proposal and replace them with a single
table detailing the impedance requirements for both single ended and
differential cables both flat and round.

The problem with the new table is it raises the minimum differential
impedance for twisted flat and flat cables to 115 ohms from the current
110 ohms. This will make some Ultra 2 cable plants not compliant with
Ultra 160 (SPI-3) specification. The reason given is the higher
impedance cable will suffer less loss and be better able to match the
terminator when drive loads are added (which decrease the apparent cable

This is not needed for the following reasons. The difference between
Ultra 2 and Ultra 160 (SPI-3) is the use of dual edge clocking which
does not increase the maximum frequency on the wire. Hence, attenuation
should be the same since the frequency is the same. Secondly, the
maximum load capacitance of the Ultra 160 (SPI-3) loads was reduced from
20 pf to 15 pf. Ultra 160 (SPI-3) loads represent less of a load to the
cable and will not degrade the impedance as much as Ultra 2 loads. The
new table should be changed to have the minimum value for twisted flat
and flat cables be 110 ohms consistent with Ultra 2 cable plant
specifications. When marketing Ultra 160 we can state it requires no
change to the cable plant.

One other reason given for the change was to prepare for SPI-4 where the
frequency will increase and cables characteristic may have to change.
But SPI-4 may require many more changes than a impedance change. We
should wait till SPI-4 is defined before deciding cable requirements for

* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at

More information about the T10 mailing list