Minutes from the 19 September ibis-atm meeting are attached.
IBIS Macromodel Task Group
Meeting date: 19 September 2017
Members (asterisk for those attending):
ANSYS: * Dan Dvorscak
* Curtis Clark
Broadcom (Avago): Xingdong Dai
Bob Miller
Cadence Design Systems: Ambrish Varma
Brad Brim
Kumar Keshavan
* Ken Willis
eASIC: David Banas
Marc Kowalski
Ericsson: Anders Ekholm
GlobalFoundries: Steve Parker
IBM Luis Armenta
Trevor Timpane
Intel: Michael Mirmak
Keysight Technologies: Fangyi Rao
* Radek Biernacki
Ming Yan
Maxim Integrated Products: Hassan Rafat
Mentor, A Siemens Business: John Angulo
* Arpad Muranyi
Micron Technology: Randy Wolff
Justin Butterfield
SiSoft: * Walter Katz
Todd Westerhoff
* Mike LaBonte
Synopsys: Rita Horner
Kevin Li
Teraspeed Consulting Group: Scott McMorrow
Teraspeed Labs: * Bob Ross
TI: Alfred Chong
The meeting was led by Arpad Muranyi. Curtis Clark took the minutes.
--------------------------------------------------------------------------------
Opens:
- None.
-------------
Review of ARs:
- Walter to finish editing of BIRD158.6 Draft 5.
- Radek had issued a draft 5, so Walter produced draft 6.
- There is now also a draft 7.
--------------------------
Call for patent disclosure:
- None.
-------------------------
Review of Meeting Minutes:
- Arpad: Does anyone have any comments or corrections? [none]
- Walter: Motion to approve the minutes.
- Radek: Second.
- Arpad: Anyone opposed? [none]
-------------
New Discussion:
BIRD158.6:
- Walter: [sharing his draft 7]
- This is mostly my draft 6, but with comments from Arpad and Mike L.
- Background information section has not been maintained particularly well,
but I removed item #2, which was an entry for something we added and then
removed.
- It includes Radek's modified Tx picture (which shows Vp and Vn with no
DC supply symbols).
- I can't edit this figure, but I think we should change "pad terminals" to
"buffer terminals."
- Arpad: I agree.
- Radek: Moving forward, the right thing to do is to keep "pad terminals."
- Walter: This won't be a quick discussion. Let's defer until the end of the
presentation.
- Added "ideal" to the step stimulus description sentence. [per Arpad]
- Changed "Vp - Vn" to V(Vp, Vn) to adopt our new SPICE type voltage syntax.
[per Mike L.]
- Added a note that the triangle symbol is a "signal ground".
- Where should this definition of the triangle symbol go?
- Arpad: Right after the picture.
- I added the note that the triangle symbol represents node 0.
- I thought that's what Bob R. wanted.
- Bob: This is a 4 terminal definition of a 4 port file. It's a moot point.
The reference is node 0, but we can call it a local reference.
- Radek: The Ts4file contains 4 port data. There are 4 terminals and the
reference for the respective ports.
- That node will typically be grounded, that's fine.
- All those ref nodes in the figure can be global zero if the tool wants.
- But the tool could offer something else.
- All those reference nodes in the end-to-end picture have to be the same.
- Otherwise you're using the Touchstone data incorrectly.
- Walter: You're saying all those triangles represent (are connected to) a
single node, and that node could be node 0.
- Radek: Yes.
- Walter: We will iron out the exact wording later.
- We have three outstanding issues:
- Wording of the triangle symbol definition.
- "pad terminal" or "buffer terminal"?
- Need to add titles to the figures.
- Radek: I have some other issues to review:
- I think it's odd to say "ideal" step. A step is a step.
- I object to the V(Vp, Vn). Vp and Vn are not nodes or nodal voltages. They
are numbers. They are the delta between the + and - terminals in the
figure.
- Walter: Hearing Radek's argument, are you okay with changing that back Mike?
- Mike L.: Yes.
- Walter: [reverted to the Vp - Vn text].
- Radek: I understand the intent of the final sentence, "The slope of an ideal
step stimulus is infinite,..." But I think this is unnecessary and
outside the scope of the spec. to tell the EDA tool what to do.
- If it stays I'd rather we talk about 0 rise time than infinite slope.
- Arpad: I'm okay with removing it, but I would like to be clear that it's a
pure step and not some sort of trapezoidal waveform.
- Walter: I'll leave in your "ideal" in the earlier sentence, and remove this
final sentence.
- Bob R.: Please send your version to Mike and me, so we can do some additional
editing to restore the history from previous official revisions.
- Radek: Yes, we have to be careful to make the comparison to 158.5.
- Walter: I'll send my version to Bob R., Mike L., Arpad and Radek for final
editing prior to posting.
- Now we're down to resolving the final issues.
- Bob R.: I also prefer "buffer terminals" in the figure.
- We use "buffer terminals" in the text.
- This is supposed to be a replacement for [Model], so it should go to the
buffer terminals.
- Arpad: I prefer "buffer terminals", but I see Radek's point.
- If we have BIRD189 in place, are we going to exclude the possibility of
having a BIRD189 on-die interconnect model from buffer to pad and then
using it with BIRD158?
- Walter: If we have a BIRD189 interconnect model, and there was an on-die
interconnect between the pad and the buffer, should this BIRD 158
4-port connect to that?
- That is, is that on-die interconnect inserted between this Ts4file and
the pad?
- Arpad: Yes, would we be running the risk of double counting at that point?
- Walter: If this were an [External Model] that had terminals 2 and 4, and
in BIRD189 the interconnect model would be in series with it, then
it should be the same way here for BIRD158.
- Arpad: In that case it should be "buffer terminals".
- Radek: We have to worry about what people measure.
- Arpad: If we went with buffer terminal, Bob Miller's desire to measure at
the pad can still be accommodated. If we call it pad terminal, then
we could be boxed in and exclude other situations.
- Most of the time people make these models from simulation, and with
simulation you can "measure" anywhere you want.
- Walter: If Bob M. had measured his Ts4file data at the pad, then he simply
won't have a BIRD189 style interconnect model between the pad and
the buffer. Maybe that's a rule...
- Radek: You impose the restrictions on the BIRD189 models that can be used.
- Maybe that's the way to do it.
- Walter: We have that issue today. When you have a [Model] that model may or
may not include the capacitance of the on-die interconnect. We
already have that problem with BIRD189 model overlap.
- Bob R.: It's a questions of definition. Regardless of where it's measured,
by definition we can state that this ends at buffer terminals.
- Curtis: I'm on the fence. I understand Bob R.'s point that this is a [Model]
replacement, but I think Radek is correct if we say BIRD158 is for
providing on-die interconnect modeling.
- Radek: I won't fight changing it to "buffer terminals", but it doesn't make
sense.
- Walter: Motion to change "pad terminals" to "buffer terminals" in all figures.
- Arpad: Second.
- Bob R.: Let's have a roll call vote.
ANSYS : Abstain
Cadence : Not available
Keysight : Abstain
Mentor : Yes
SiSoft : Yes
Teraspeed Labs: Yes
Motion passed (3 Yes, 2 Abstain)
- Walter: How to define the triangle ground symbol?
- I believe Radek thinks since this is a differential signal they all have the
same reference node. They are all the same node, but not necessarily
node 0.
- Radek: Yes.
- Walter: In that case I think we should keep the definition of the triangle
symbol local to this section.
- Arpad: Is your "signal ground" term defined anywhere? We should define this
in the spec., not rely on outside terms from Wiki, etc.
- Walter: How about:
- "All triangle ground symbols are connected together, which may be node 0."
- [everyone agreed]
- Radek: I would remove the sentence, "For a Tx buffer, the transmitter circuit
defines the analog buffer model between the zero-impedance stimulus
input voltage source and the terminals."
- Walter: I think it's fine.
- Captions for the drawings?
- [captions similar to the headings were added to all three diagrams]
- Bob R.: Our AR (Bob, Radek, Arpad, Mike L.) is to work on it and send it to
full ATM group and upload it as draft8.
- Walter: Motion to adjourn.
- Curtis: Second.
- Arpad: Thank you all for joining.
AR: Walter to modify his BIRD158.6 draft 7 according to the meeting discussions
and send it to Bob R., Mike L., Arpad, and Radek for final review.
AR: Bob R., et. al., to perform final review and editing and post it as
BIRD158.6 draft 8.
-------------
Next meeting: 26 September 2017 12:00pm PT
-------------
IBIS Interconnect SPICE Wish List:
1) Simulator directives
