Subject: [IBIS] BIRD79 - Non-linear buffer impedance (extension to C_comp)
From: Peters, Stephen (stephen.peters@intel.com)
Date: Wed Nov 06 2002 - 09:37:48 PST
Greetings:
Below is a new BIRD submitted by Luca Giacotto and Arpad Muranyi. BIRD79
greatly enhances IBIS's ability to describe the non-linear and frequency
dependant characteristics of buffer impedance.
BIRD79 will be discussed at the November 22 teleconference.
Regards,
Stephen Peters
Intel Corp.
Chair, EIA/IBIS Open Forum
========================================================================
Buffer Issue Resolution Document (BIRD)
BIRD ID#: 79
ISSUE TITLE: Non-linear buffer impedance (extension to C_comp)
REQUESTOR: Luca Giacotto (Alstom Transport) and Arpad Muranyi (Intel
Corp.)
DATE SUBMITTED: November 6, 2002
DATE REVISED:
DATE ACCEPTED BY IBIS OPEN FORUM:
****************************************************************************
**
****************************************************************************
**
STATEMENT OF THE ISSUE:
This BIRD aims to improve the modeling of the reactive component of the
buffer
impedance, currently represented by keywords C_comp and C_comp_*. (C_comp_*
was introduced in BIRD 65.2 and is now part of the IBIS 4.0 specification).
This extension models the non-linear (voltage dependent),
frequency-dependent
and state-dependent impedance behaviour of real buffers (as represented by
the
transistor-level model). These improvements were already foreseen in the
bottom comments of BIRD 65.2.
****************************************************************************
**
STATEMENT OF THE RESOLVED SPECIFICATIONS:
A new optional keyword [PZ Impedance] (abbreviation for "Pole/Zero
Impedance")
shall be introduced, with 24 possible subparameters. The new keyword should
be
placed in the Model section, after the [Model] keyword.
|===========================================================================
==
| Keyword: [PZ Impedance]
| Required: No
| Description: Used to enhance the modeling of the buffer impedance, beyond
| the possibilities given by the C_comp* subparameters in the
| [Model] keyword. It allows to account for the
| a) non-linear,
| b) frequency-dependent,
| c) state-dependent
| behavior of the buffer impedance.
| Sub-Params: Pole_high_pullup, Zero_high_pullup, Pole_high_pulldown,
| Zero_high_pulldown, Pole_high_gndclamp, Zero_high_gndclamp,
| Pole_high_pwrclamp, Zero_high_pwrclamp,
| Pole_low_pullup, Zero_low_pullup, Pole_low_pulldown,
| Zero_low_pulldown, Pole_low_gndclamp, Zero_low_gndclamp,
| Pole_low_pwrclamp, Zero_low_pwrclamp,
| Pole_in_pullup, Zero_in_pullup, Pole_in_pulldown,
| Zero_in_pulldown, Pole_in_gndclamp, Zero_in_gndclamp,
| Pole_in_pwrclamp, Zero_in_pwrclamp.
| Usage Rules: These subparameters define the pole/zero behavior of the
| buffer impedance at various die-node voltage values, and in
| each of the three basic states (whichever applicable to the
| specific buffer type): "output high", "output low" and
input".
| The four impedances in each state should be thought as a
| reactive network shunted to the corresponding I-V table.
| Moreover, the pole/zero data account for the presence of the
| I-V tables.
|
| Each of these subparameters, when present, is followed by a
| F-V table. The first column
| contains the voltage value at the die node, and the three
| remaining columns hold the typical, minimum, and maximum
| frequency values in [Hz]. The four entries, Voltage,
req(typ),
| Freq(min), and Freq(max) must be placed on a single line and
| must be separated by at least one white space.
|
| All four columns are required under these subparameters. If
| typical, minimum and/or maximum frequency values are not
| available, the reserved word "NA" must be used.
|
| Other Notes: A typical [PZ Impedance] description could contain two sets
of
| Pole* tables and one set of Zero* tables. Or, just one set
of
| Pole* tables.
|---------------------------------------------------------------------------
--
[PZ Impedance]
|"HIGH-state":
| *) a 2nd order non-linear impedance for the pullup element
| *) a 1st order non-linear impedance for the pulldown element
|pole_high_pullup |this is pole "1" in the high-state, related to the
pullup
|element
| [Volt] typ [Hz] min [Hz] max [Hz]
0V 6.37E7 NA NA
1V 5.36E7 NA NA
2V 1.15E8 NA NA
| ... ... ... ...
5V 2.53E8 NA NA
|
zero_high_pullup |this is the zero in the high-state, related to the
pullup
|element
0V 3.00E8 NA NA
1V 2.83E7 NA NA
| ... ... ... ...
5V 2.56E8 NA NA
|
pole_high_pullup |this is pole "2" in the high-state, related to the
pullup
|element
0V 9.28E8 NA NA
| ... ... ... ...
|
pole_high_pulldown |this is the pole in the high-state related to the
pulldown
|element
0V 5.37E7 NA NA
1V 5.22E7 NA NA
2V 9.05E7 NA NA
| ... ... ... ...
5V 2.03E8 NA NA
|
|"LOW-state":
|(1st order NL impedance for both the pullup and pulldown elements)
pole_low_pullup |this is the pole in the low-state related to the pullup
0V 2.00E8 NA NA
| ... ... ... ...
|
pole_low_pulldown |this is the pole in the high-state related to the
pulldown
0V 5.25E9 NA NA
| ... ... ... ...
5V 4.24E8 NA NA
|
|===========================================================================
==
****************************************************************************
**
ANALYSIS PATH/DATA THAT LED TO SPECIFICATION:
The fixed valued (constant) C_comp or C_comp* subparameters don't have the
ability to describe the voltage-dependent (i.e., non-linear) and
state-dependent behaviour of the die capacitance. Moreover, there are
buffers
showing an impedance that is also strongly frequency-dependent. This BIRD
intend to extend the concept of "die-capacitance" to "die-impedance",
making the necessary data available for a simulator to implement it.
The improvement in the impedance modeling that can be obtained with the
adoption of a second order non-linear "die-impedance" has been demonstrated
in
the presentation at the June'02 IBIS Summit by Luca Giacotto (the URL is
http://www.eda.org/pub/ibis/summits/jun02/giacotto.pdf). Formulas and a
modeling method can be found in the presentation too.
To meet the needs of power delivery or supply noise simulations, the
non-linear network from the above presentation has been splitted into four
networks, following the idea of BIRD 65.2.
How to arrange the data in IBIS
Three options were considered for arranging the new data in the
specification:
1) A new keyword with 24 subparameters (as in this version of the BIRD).
2) 6 new subparameters in each of the I-V table keywords (pole_high,
zero_high, pole_low, zero_low, pole_in, zero_in).
3) Adding the subparameters directly in the [Model] keyword, beside the
basic
C_comp* subparameters. Actually, this option has been avoided because of the
high number of subparameters (24) needed to cover all the types of buffers.
****************************************************************************
**
ANY OTHER BACKGROUND INFORMATION:
Documents related to this BIRD (available on the IBIS website):
- presentations at the Paris (March 2002) and New Orleans (June 2002) IBIS
summits;
- BIRD 65.2;
- Arpad Muranyi's presentation at the June 2001 IBIS summit, in which the
problem of the non-linear impedance modeling was first brought up (pp.
10-13).
****************************************************************************
**
-----------------------------------------------------------------
|For help or to subscribe/unsubscribe, email majordomo@eda.org
|with the appropriate command message(s) in the body:
|
| help
| subscribe ibis <optional e-mail address, if different>
| subscribe ibis-users <optional e-mail address, if different>
| unsubscribe ibis <optional e-mail address, if different>
| unsubscribe ibis-users <optional e-mail address, if different>
|
|or email a request to ibis-request@eda.org.
|
|IBIS reflector archives exist under:
|
| http://www.eda.org/pub/ibis/email_archive/ Recent
| http://www.eda.org/pub/ibis/users_archive/ Recent
| http://www.eda.org/pub/ibis/email/ E-mail since 1993
This archive was generated by hypermail 2b28 : Wed Nov 06 2002 - 10:07:48 PST