[ibis-quality] Minutes from 2 Jun to 16 Jun 2009 ibis-quality meetings

Minutes from the 2 Jun to 16 Jun 2009 ibis-quality meetings are
attached.

Mike
Minutes, IBIS Quality Committee

16 June 2009

11-12 AM EST (8-9 AM PST)

ROLL CALL
  Adam Tambone
* Anders Ekholm, Ericsson
  Barry Katz, SiSoft
  Benny Lazer
  Benjamin P Silva
  Bob Cox, Micron
* Bob Ross, Teraspeed Consulting Group
  Brian Arsenault
  David Banas, Xilinx
  Eckhard Lenski, Nokia Siemens Networks
  Eric Brock
* Guan Tao, Huawei Technologies
  Gregory R Edlund
  Hazem Hegazy
  Huang Chunxing, Huawei Technologies
  John Figueroa
  John Angulo, Mentor Graphics
  Katja Koller, Nokia Siemens Networks
  Kevin Fisher
  Kim Helliwell, LSI Logic
  Lance Wang, IOMethodology
  Lynne Green
* Mike LaBonte, Cisco Systems
  Mike Mayer, SiSoft
* Moshiul Haque, Micron Technology
  Muniswarareddy Vorugu, ARM Ltd
* Pavani Jella, TI
  Peter LaFlamme
  Randy Wolff, Micron Technology
  Radovan Vuletic, Qimonda
  Robert Haller, Enterasys
  Roy Leventhal, Leventhal Design & Communications
  Sherif Hammad, Mentor Graphics
  Todd Westerhoff, SiSoft
  Tom Dagostino, Teraspeed Consulting Group
  Kazuyoshi Shoji, Hitachi
  Sadahiro Nonoyama
  Liqun, Huawei

Everyone in attendance marked by *

NOTE: "AR" = Action Required.

-----------------------MINUTES ---------------------------
Mike LaBonte conducted the meeting.

Call for patent disclosure:

- No one declared a patent.

AR Review:

- Mike post new draft
  - 1.1as was posted

- Mike fix font and reference issues in IQ specification identified by Anders
  - TBD

New items:

A Motorola group has contacted Mike asking for a presentation on IBIS Quality
- Bob: They could join our meetings too

Anders Eckholm review of the IQ 1.1 specification (2nd half):

3.1.1.  {LEVEL 2}  [Package] must have typ/min/max values
- We had changed this in the last meeting
- Moshiul: The [Package] min/max L and C do not necessarily correspond
  to any single pin
- Bob: Worst case time delay is max L max C
  - Worst case impedance is max L over min C or min L over max C
  - May use max/max and min/min, but EDA tool can do whatever
  - A good practice is to find min & max of the [Pin] RLC list
- Mike: We use [Package] calculated overall skew to decide when RLC is needed
- Bob: We call for pin RLC in 3.3.2
  - That is a level 3 check

From Anders' email:
    3.1.2 
    here we state pin parasitics should represent the range spanned by the
    actual pin parasitics for signal pins..  Which raises the question which
    return pin these parasitics are suppose to be related to ?  At the end
    of the paragraph it clearly states that power and Gnd pins should not be
    included in the determination of the [Package] parasitics. Related to a
    discussion we had concerning defining pin parasitics for power & Gnd pins.
- Anders: The return path assumption is the problem here
- Bob: This has to be figured out from [Pin Mapping]
- Mike: [Pin Mapping] gives just a loose pin association, is that enough?
- Bob: There can be multiple loops using that same parasitic
  - For example, 5 returns using the same ground
  - A TDR measurement might be measuring the whole loop
- Mike: The ATM group is working on better interconnect modeling now
- Bob: Modeling is always done piecewise anyway
  - IBIS is extraction-centric
- Mike: Should we change 3.1.2?
  - Anders: It is not clear what we could do
- Mike: Is consideration of return paths too much detail for [Package]?
- Bob: If the power and ground pins have zero LC that might have been
  factored into the signal pin RLC
- Anders: There should be a comment about extraction
- Moshiul: With some extraction tools you can not specify return paths.
  - They make assumptions
  - The mounting makes a difference
- Mike: We could delete: "typically it is the average of the signal pin
  parasitics, but it need not be"
  - Also delete: "The power and ground pins should NOT be included in the
    determination of the [Package] parasitics"
  - We could add a comment calling for documentation of the method used
- Moshiul: The limits L < 100nH, C < 100pF, R < 10 ohm are not reasonable
  - for power and ground pins
  - We changed it to apply only to signal pins
- Mike: This IQ check calls for limits lower than IBISCHK
- Bob: It would be better to not have numeric limits
  - But it is OK as is
  - The lower limits might be a CAUTION in future IBISCHK

- No new draft will sent this time as there are few changes

Meeting ended at 12:02 PM Eastern Time.
Minutes, IBIS Quality Committee

02 June 2009

11-12 AM EST (8-9 AM PST)

ROLL CALL
  Adam Tambone
* Anders Ekholm, Ericsson
  Barry Katz, SiSoft
  Benny Lazer
  Benjamin P Silva
  Bob Cox, Micron
* Bob Ross, Teraspeed Consulting Group
  Brian Arsenault
  David Banas, Xilinx
* Eckhard Lenski, Nokia Siemens Networks
  Eric Brock
  Guan Tao, Huawei Technologies
  Gregory R Edlund
  Hazem Hegazy
  Huang Chunxing, Huawei Technologies
  John Figueroa
  John Angulo, Mentor Graphics
  Katja Koller, Nokia Siemens Networks
  Kevin Fisher
  Kim Helliwell, LSI Logic
  Lance Wang, IOMethodology
  Lynne Green
* Mike LaBonte, Cisco Systems
  Mike Mayer, SiSoft
* Moshiul Haque, Micron Technology
  Muniswarareddy Vorugu, ARM Ltd
  Pavani Jella, TI
  Peter LaFlamme
  Randy Wolff, Micron Technology
  Radovan Vuletic, Qimonda
  Robert Haller, Enterasys
  Roy Leventhal, Leventhal Design & Communications
  Sherif Hammad, Mentor Graphics
  Todd Westerhoff, SiSoft
  Tom Dagostino, Teraspeed Consulting Group
  Kazuyoshi Shoji, Hitachi
  Sadahiro Nonoyama
  Liqun, Huawei

Everyone in attendance marked by *

NOTE: "AR" = Action Required.

-----------------------MINUTES ---------------------------
Mike LaBonte conducted the meeting.

Call for patent disclosure:

- No one declared a patent.

AR Review:

- Mike post new draft
  - Not done

New items:

Editorial review of the IQ 1.1 specification:

- Mike: Should we use V-T or V-t?
- Bob: A quality document should conform to some convention
  - Uppercase T often refers to a time interval
  - But we comply with IEEE standards in using V-T and I-V
- Bob: There should be a space in "10 V"
  - But in IBIS syntax examples that call for no space that is different
  - Spaces can be omitted on whitespace-parsed lines

- Bob: We do not have to be editorially perfect
  - But there should be no obvious problems
  - If we are inconsistent people will have trouble finding things
  - We would have to follow IEEE guidelines if this was to be an ANSI standard

- We changed all vcc and VCC to Vcc in the IQ specification
- Bob: We do not have to change VDDQ because it is that way in datasheets
- Moshiul: Often the trailing letters are small caps
- We will leave VDDQ uppercase

- Bob: Keyword and parameter references have to be precise
  - For example, pullup
- We reviewed all use of pullup and pulldown

5.4.4.  {LEVEL 2}  V-T table endpoints match fixture voltages
- Bob: There is a difference between:
  - Pullup transistor structures
  - Pullup resistors on-die
- Moshiul: Internal pullups usually does not mean active structures
  - It usually is in the clamps
- Mike: Should we remove "where internal pullups/pulldowns"?
  - Bob: No that is correct here
- Moshiul: Should we say where internal pullup/pulldown terminations"?
  - Mike: This checks is about output buffers
- We left it as is

Bob: We should not use global search and replace to make these changes
- We should search for consistent use of typ/min/max next

Next meeting: 09 Jun 2009 11-12 AM EST (8-9 AM PST)

Meeting ended at 12:05 PM Eastern Time.
Minutes, IBIS Quality Committee

09 June 2009

11-12 AM EST (8-9 AM PST)

ROLL CALL
  Adam Tambone
  Anders Ekholm, Ericsson
  Barry Katz, SiSoft
  Benny Lazer
  Benjamin P Silva
  Bob Cox, Micron
* Bob Ross, Teraspeed Consulting Group
  Brian Arsenault
  David Banas, Xilinx
* Eckhard Lenski, Nokia Siemens Networks
  Eric Brock
* Guan Tao, Huawei Technologies
  Gregory R Edlund
  Hazem Hegazy
  Huang Chunxing, Huawei Technologies
  John Figueroa
  John Angulo, Mentor Graphics
  Katja Koller, Nokia Siemens Networks
  Kevin Fisher
  Kim Helliwell, LSI Logic
  Lance Wang, IOMethodology
  Lynne Green
* Mike LaBonte, Cisco Systems
  Mike Mayer, SiSoft
  Moshiul Haque, Micron Technology
  Muniswarareddy Vorugu, ARM Ltd
* Pavani Jella, TI
  Peter LaFlamme
  Randy Wolff, Micron Technology
  Radovan Vuletic, Qimonda
  Robert Haller, Enterasys
  Roy Leventhal, Leventhal Design & Communications
  Sherif Hammad, Mentor Graphics
  Todd Westerhoff, SiSoft
  Tom Dagostino, Teraspeed Consulting Group
  Kazuyoshi Shoji, Hitachi
  Sadahiro Nonoyama
  Liqun, Huawei

Everyone in attendance marked by *

NOTE: "AR" = Action Required.

-----------------------MINUTES ---------------------------
Mike LaBonte conducted the meeting.

Call for patent disclosure:

- No one declared a patent.

AR Review:

- Mike post new draft
  - 1.1ar was posted
  - The PDF file had a strange right side alignment

New items:

Anders Eckholm review of the IQ 1.1 specification:

- Mike: Anders reports several font consistency issues
  - Mike will fix these
- Also there are some dangling reference problems

AR: Mike fix font and reference issues in IQ specification identified by Anders

- From Anders' email:
    1.2.2 and 1.2.3 
    Comparing these two paragraphs to 1.2.1, the G designator can be used if
    only a subset of buffers models have golden waveforms
    in 1.2.2 and 1.2.3 no similar statements exists ? 
- We decided M + S can be used if a reasonable set of models are correlated.
- We added the 1.2.1 statement about this to 1.2.2 and 1.2.3

- From Anders' email:
    1.4 In the first bullet, question for any buffer for every buffer ? 
    Bullet number 2, How many buffers ? 
- We inserted "a reasonable set of"

- From Anders' email:
    1.4 The summary IQ score is stated to be placed either in a comment or
    in the [Notes] section, could we define which or do we want it loosely
    specified?
- Bob: Maybe the IQ score should be in the [Notes] section 
- Mike: Since this is not per [Model] it only needs to be at the top of the file
  - [Notes] can have lines with and without the | character
  - Bob: Tools that display [Notes] may not show the | lines
  - Mike: Most people probably look at the IBIS file directly
    - Tools might read both comment and non-comment lines anyways
- Bob: The IBIS Accuracy Handbook called for a trailer
- This remained undecided

- From Anders' email:
    2.1 middle of page.  The [IBIS ver] value can be set to at least.......
    Is this text necessary, it refers to tools and does not seem to be
    focused on model quality.
- Bob: This is about using newer versions of IBISCHK to check more things
- Mike: This paragraph is a commentary asking people to use higher [IBIS Ver]
  - It is a little off topic, written into a "Passes IBISCHK" IQ check
  - But it is a good recommendation
- Bob: Some tools support only up to IBIS 3.2

Bob:
- We should use uppercase IBISCHK throughout.
  - We first mention IBISCHK in 1.1.2 but do not explain what it is
- Need to use typ/min/max consistently
- We do not explain NCSU
- We have to use uppercase POWER as IBIS uses it

3.1.1 is unclear, the paragraph states "must have typ/min/max...." 
    And than shortly after states allows NA if not available" ? 
    And  than again says "must have ..." 
- It was unclear to us what this meant. 
- Mike: Min and max are the most important for timing checks
- Bob: This is just [Package Model], but for quality we are going beyond IBIS
  - Min and max have to each be numerically least and greatest
    - Even if that combination does not make sense
  - Only [Model] requires fast/slow correlation
  - Max L and min C might be worst case for impedance
- Mike: "NA if not available" refers to what the IBIS specification says
  - IQ clearly is going beyond IBIS requirements
- We changed the text

Meeting ended at 12:04 PM Eastern Time.

Next meeting: 16 Jun 2009 11-12 AM EST (8-9 AM PST)

Other related posts:

  • » [ibis-quality] Minutes from 2 Jun to 16 Jun 2009 ibis-quality meetings - Mike LaBonte (milabont)