SCSI Termination

Paul D. Aloisi 603-429-8687 ALOISI at uicc.com
Wed Aug 14 05:21:06 PDT 1996


* From the SCSI Reflector, posted by:
* "Paul D. Aloisi 603-429-8687" <ALOISI at UICC.COM>
*
Howard,

This has been a topic in the last few meetings and actually came up
several years ago. The SCSI-2 terminator is just out of spec for
SCSI-3 SPI. We actually have two full families of terminators, I brought
the point up when I joined the committee Fall 93 because we didn't want
to have two families of terminators. SPI had just finished the letter
ballot for approval. The committee did not think is was an issue. The SCSI-2
terminator meets 22.4 mA Maximum at 0.5 Volts, when a linear terminator
is measured at 0.2 Volts the maximum current is 25 mA exceeding the 24 mA
maximum in SCSI-3 SPI.

A Linear terminator designed for SCSI-3 SPI actually provides 1+ mA less current
than a SCSI-2 terminator. Even with the reduction it is a much better 
terminator than the passive terminator. SCSI-3 SPI does not spec the
impedance only the current and voltage points, one of the main developers
of SPI had designed a current mode terminator, SPI made it legal. It
has been designed out by about everyone who tried it because it is only
a current source it does not match impedance leaving the bus open to
reflections. We designed a current mode terminator, after testing scrapped
it. 

I have several customers that use the SCSI-2 terminator for SCSI-3 because
of the better performance. 

The outcome of yesterday's SPI-2 meeting a group has been tasked to write
a proposal which will make the SCSI-2 terminator legal for SCSI-3, but it
may not show up until the SPI-2 document is published replacing SPI.

Thec current reading of the standard the SCSI-2 terminator is over the current
limit and can not be used for SCSI-3 SPI or Fast-20. The way we changed the
terminator to meet SPI and Fast-20 was to reduce the regulator voltage to
a Nominal 2.7 Volts instead of the nominal 2.85 Volts. This actually helps
some of the high speed systems get to the low level in the first step 
allowing some drivers to work better for Fast-20. Sinking the signal
is just as important as pulling up the signal, it the signal is too high
the driver can not sink the line to below 1 volt for the Fast-20 spec.
Active negation drivers are required for Fast-20 which actually balances
the system pull up current to a typical 60 mA to the typical
driver sinking ability 60 mA. 

SPI allows a terminator to be designed to run off of 3.3 Volt supplies, 
5 Volts is not available on some systems.

Paul Aloisi
Unitrode 




More information about the T10 mailing list