A list server problem prevented last week's meeting minutes from reaching
the IBIS Interconnect list. Here you go.
From: Randy Wolff (rrwolff)
Sent: Wednesday, May 27, 2015 10:00 AM
To: 'ibis-interconn@xxxxxxxxxxxxx'
Subject: IBIS Interconnect Meeting Minutes, 5/27/2015
Interconnect Task Group Meeting Minutes, 5/27/2015
Attendees
ANSYS Curtis Clark
Mentor Arpad Muranyi
Micron Justin Butterfield, Randy Wolff
Teraspeed Bob Ross
No patents were declared.
Bob Ross led the meeting.
Opens
Randy requested a brief overview of last week's discussion on the GND
topic.
Discussion
Randy shared Walter's presentation TerminalNamesInIBIS.pdf.
Arpad noted on slide 5, it shows proposed changes. Voltage definitions in
IBIS such as [Pullup Reference] should be described by V(A_puref, GND).
Is the GND on the bottom terminal a local ground or A_gcref or A_pdref?
The CAC capacitance might only connect to a voltage other than A_pdref or
A_gcref in a buffer such as RS232 with a -5V connection. We're ok if we
consider the GND to be die ground. Bob added that we're covered with
defining A_gcref with respect to die ground. We don't show split C_comp.
Arpad noted that split C_comp is only needed with non-ideal power and
ground supplies.
Arpad asked if when measuring I-V curves, do we have ideal voltages or
not. Randy said we measure them in a DC sense, so power supplies are
ideal. Bob noted this is even the case for V-t curves, where the package
is not included in Spice.
Arpad noted he looked up a databook, and they are sloppy in labeling VCC,
-VCC, and GND. GND in the databook for a pin does not mean universal
ground.
Arpad said he responded to Walter by email that we should be consistent
with use of VCC, VDD, VSS, etc. in the specification. It would be good to
have a universal term in the specification to refer to power and ground.
He thinks we picked the model_name in IBIS of POWER and GND to be neutral
on naming. The signal_name meaning can be unique. Bob agreed that we
don't need to change any examples in the specification that have GND in
the signal_name to Vss. Randy also agreed, as Vss vs. GND makes no
difference. It refers to a databook name for a pin. The databook could
use GND.
Bob noted that it was discussed previously to move discussion of GND
changes in IBIS to the ATM task group. He felt that the discussion should
be deferred until other topics are exhausted. It could be discussed by
the Editorial task group further once IBIS 6.1 work is done, then passed
back to ATM.
