Single-ended termination current
aloisi at unitrode.com
Wed Oct 14 09:20:11 PDT 1998
* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* "Aloisi, Paul" <aloisi at unitrode.com>
The actual document that set the SPI-2 standard was 97-236R1 which was =
on the the Plenary Sept 97.
96-245R2 was not accepted by the working group because of the =
Coefficent that was documented in the proposal.
The data presented showed reflections that could cause signals in the =
of the cable to reverse through the full threshold range causing a =
pulse. Bill Ham commented about the data that you presented for Dean, =
never came back with the answers to Bill comments.
Left out of your notes is about a years worth of meetings; 96-251, =
97-180 reviewed the termination issues.
Reflection was a major consideration of the 45 ohm impedance proposed =
Driver current was a major concern of the controller manufacturers, it =
not to exceed 48 mA at 0.5 volts. That was a qualification and =
SPI allows 24 mA at 0.2 volts, when this was questioned in November 93
document 93-149=A0 (point 6)- the standards group at that time would =
more than 24 mA at 0.2 volts. It was pointed out at that time a Linear
terminator that worked for SCSI-2 22.4 mA at 0.5 volts would be 25 mA =
volts requiring a special series of termination for everything after =
The difference between SCSI-2 and SPI was expand as Bipolar technology
drivers don't sink the signal lines a low as CMOS drivers. The =
voltage point was moved to 0.2 volts for the CMOS drivers. The input =
current was dropped from 0.4 mA at 0.5 volts to 20 microAmps at 0.5 =
SPI-2 working group allowed Fast-5 devices to have the higher leakage
current of SCSI-2, allowing for older bipolar devices to be used on the =
=A0Gene Milligan opened the terminator current issues with 96-222, =
only the 0.2 volt current limit be increased for the same performance =
A SPI and Fast-20 terminator actually is 1 mA less current pull up than =
SCSI-2 terminator, as pointed out in November 93.=20
=A0Linear terminators that meet the 22.4 mA at 0.5 volts and 25.4 mA at =
volts where considered the correct levels in the SPI-2 meeting. 24 mA =
volts was considered a problem for the drivers because no margin is =
system Leakage. Drivers would need to be specified at 49 mA at 0.5 =
mA at 0.5 volts was not acceptable to the working group.
SPI actually specifies 1 mA less current at 0.5 volts than SCSI-2 =
when this was discussed in the meeting.
SPI-3 should call out the same termination limits as SPI-2.
From: Louis Grantham [mailto:lgdatcom at ix.netcom.com]
Sent: Wednesday, October 14, 1998 9:53 AM
To: Aloisi, Paul
Subject: Single-ended termination current
At the SPI-3 working group on 9/15/98, I brought up the following
(X3T10/96-222r1) identified the single ended termination =
between SCSI-2 and SPI. The current each terminator can source =
SCSI-2 is 22.4 mA at 0.5 V and above and SPI is 24 mA at 0.2 V and =
This can make SCSI-2 terminators not SPI compliant. The notch in the
terminator IV curve document was intended to reconcile this difference. =
original proposal (X3T10/96-245r2) accomplished this with 25.4 mA at =
and 24 mA at 0.5 V. This was subsequently changed to 22.4 mA at 0.5 V =
25.4 mA at 0.2 V. This change meant that one could still have an
incompatible terminator if at 0.5V the terminator sourced between 22.4 =
and 24 mA which defeats the original intent.
I want the spec changed back to 24 mA at 0.5 V in 7.1.1 b).=A0=20
The only technical reason that I can find of why the current at 0.5 V =
spec'd at 22.4 mA is for FAST-5 leakage current (Table 25 FAST-5 c))=A0
ASSUMING the potential leakage from receivers or open drain drivers to =
0.4 mA at 0.5 V and given the driver sink current of 48 mA, then the
terminator source current should be limited to 22.4 mA (max) at 0.5 V. =
number is driven entirely by assuming the 0.4 mA worst case leakage of
legacy FAST-5 SCSI drivers and receivers.
i.e. (22.4 mA)(2 terminators) + (0.4 mA leakage)(8 devices) =3D 44.8 mA =
mA =3D 48 mA.
Millions of terminators have been shipped to the SPI spec. of 24 mA at =
(and above) with no adverse consequences for end users. This tells me =
in the real world:
1. leakage currents will not approach the 0.4 mA @ 0.5 V
number...especially with today's receivers and drivers=20
2. drivers have excess margin beyond 48 mA
QUESTIONS FOR REFLECTOR:
1. Was SPI wrong?=20
2. Is there really 0.4 mA leakage with FAST-5 devices?
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com
More information about the T10