REVIEW - INFORMATION MODEL OF IBIS

Alan Williams and Hilary Kahn
Department of Computer Science
University of Manchester

Subject:
Comments on Version 0.1 Information Model of I/O Buffer Specification
(IBIS) Version 2.0

Alan and Hilary:

Attached are my review comments of the above document. In general, you
have provided an excellent document which can be used for an ANSI/EIA-656
(IBIS Version 2.1) information model. Several comments relate to the
Version 2.1 additions.

Feel free to contact me. We may need to clarify some other points which
I may have missed.

Bob Ross
Interconnectix, Inc.
(Secretary, EIA/IBIS Subcommittee)

Title Page: Change Version 2.0 to 2.1 and make reference to ANSI/EIA-656.

Page 5: Change Version 2.0 to 2.1.

1.0 Introduction: Rewrite sentence to read:
"This document contains an information model of the I/O Buffer Specification
(IBIS) Version 2.1, also known as ANSI/EIA-656."

Page 6:
2.0 Component UoF
What does "UoF" mean? I am not familiar with the Information model conventions
and I did not see it defined.

Page 9, Lines 195-206:
Line 198: revision : STRING (This should follow Line 195 since it
                                      is the first required keyword in an
                                      IBIS file. While 1.0, 1.1, 2.0, 2.1
                                      are current legal arguments, the argument
                                      may be a string such as 2.1.1 or 2.1a.)
                                      
Line 199: date : OPTIONAL STRING (40)

Add: After Line 203:
copyright : STRING

Note, source, notes, disclaimer, and copyright can be multi-lined strings.
All strings must not exceed either the stated maximum length or must not
extend beyond the 80th character for each line.

Delete lines 204-206: revision is a STRING

Line 197:
status is also a STRING. While an enumeration is recommended, the
actual argment syntax is arbitrary and for information only.

Page 10, Lines 228-238:
After line 227, the [Model Data] and [End Model Data] Syntax is not
included. I presume that because this is used for syntactical bracketing,
it is not needed for the information model. Question: Do all keywords
and subparameters within the IBIS Specification have to have some informational
representation?

Line 230: The resistance_matrix is optional.

Lines 237-238:
Perhaps matrix_size should be max_matrix_size since several formats allow
storing the data in more concise forms.

Page12, Line 310:
What does not_yet_defined refer to?

Page 14,
Lines 485 and 486 should be OPTIONAL

After Line 486 Add for Vref:
reference_voltage: OPTIONAL voltage_value

Page 17,
Line 563: capacitance_value should be REAL. Its value within a Capacitance
matrix on Page 10, Lines 232, and 257 can be mostly negative.

Line 599: inductance_value should be REAL for a similar reason.

 
Page 19,
Line 715-716: Non-monotonic tables are permitted. A parser warning will be
given. I do not know what not_yet_defined means.

Page 20, Add after Line 763 (A Version 2.1 addition):
fixture_voltage_min : voltage_value
fixture_voltage_max : voltage_value

Page 21, Glossary
8.0: Change Version 2.0 to Version 2.1

8.1: Change sub-parameter to subparameter throughout since this was changed
in the Version 2.1 specification.

8.2: Regarding the question concerning sub-typing the "bus" entity, the
bus implies that all voltage sources on the same bus are physically CONNECTED
and all other models that have models ASSOCIATED with that bus use that
particular connected voltage source. I do not know whether this implies
a sub-type relationship, but I think it should.

8.5: Capatilize c in [Component]

8.6: The current_range does not always require numerical values in the
typical column. The typical table itself must contain a minimal amount
of numerical data, whereas the min and max columns may contain no
numerical data. So the nomenclature might be changed to table_current_range.

Page 25,
8.33: The pin_differential ASSOCIATES two pins and adds differential
voltage and relative time delay specifications. Should this also be a
sub-type relationship?

8.37-8.39 pin_to_pin_capacitance and pin_to_pin_inductance can also
include capacitance to global ground and self inductance. The definition
may need to be expanded.

8.39: pin_to_pin resistance is the resistance associated with each
pin of a circuit with two package_pins.

Page 28,
8.60 A waveform consists of "table_voltage_ranges" because the typical
value is not always required. However, similar to 8.6, the typical column
must contain a minimal amount of numerical data, whereas the min and max
columns may contain no numerical data. This is similar to 8.6.

Pages 31-42 Attribute Summary
As a general comment, this table is technically correct because attributes
may be associated with names even if they provide no functionality. For
example, the ac_terminator_pin_model and all input models will disregard
the pullup, pulldown, falling_waveforms, rising_waveforms, and has_ramp
attributes and several others.
Received on Thu May 30 14:35:32 1996