[ibis-macro] Re: Where are IBIS VT curves measured, and should we have a C_comp_KT?
- From: "Muranyi, Arpad" <Arpad_Muranyi@xxxxxxxxxx>
- To: IBIS-ATM <ibis-macro@xxxxxxxxxxxxx>
- Date: Tue, 4 Nov 2014 16:49:02 +0000
Walter,
We need to ask a more fundamental question first. What does
C_comp really represent?
In the days when IBIS started, we didn't imagine that the
on-die interconnect between the buffer's transistor and the
pad would be modeled separately, so C_comp included the
parastitics of the transistors and the metal connecting them
to the pad. I would even say that the pad's capacitance was
also included.
But we see things differently today. We would probably think
that the pad and metal would be in a separate interconnect
model, and C_comp is only the parasitics of the transistors
inside the buffer. But in order to be able to "measure" the
waveform at the transistors we would need very special probes
which are often not available. Even probing the die pad is
challenging most of the time. (At least in the lab).
We could generate these waveforms easily in simulations, and
then do the correlation work for the pin or pad waveforms...
But we will need to clarify in the IBIS spec what we really
mean by C_comp to get all this right...
Thanks,
Arpad
==================================================================
From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx]
On Behalf Of Walter Katz
Sent: Tuesday, November 04, 2014 7:21 AM
To: IBIS-ATM
Subject: [ibis-macro] Where are IBIS VT curves measured, and should we have a
C_comp_KT?
All,
IBIS is (was) a measurement based standard. One assumption is that the VT
curves should be measured at an accessible probe point (e.g. die pad or
component pin). Since IBIS models are now often generated from simulation, the
VT curves can be generated wherever the SPICE deck allows a probe which can be
at or even inside the buffer on the silicon. If we implement a Final Stage
circuit to be inserted between the "B" element and the on-die/package
interconnect, what C_comp (C_comp_KT) should the EDA tool use to generate the
KT functions that turn on and off the Driver Pullup and Pulldown circuitry?
The idea suggested here and in the enclosed slide is to allow IBIS ISS subckts
to be defined between the legacy IBIS B element, and the buffer terminal, and
to add a new IBIS parameter C_Comp_KT. The EDA tool uses C_Comp_KT to generate
the KT functions, but uses C_comp (and the Final Stage subckt) during
simulations.
Walter
Walter Katz
wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>
Phone 303.449-2308
Mobile 303.335-6156
Other related posts:
