Re: [IBIS-Users] Rref_diff keyword

From: Bob Ross <bob_at_.....>
Date: Wed Jul 08 2009 - 16:27:58 PDT
Hi Cynthia:

To add to the response, Rref_diff and Cref_diff were originally
proposed when the multilingual extensions for IBIS 4.1 were
implemented to deal with the true differential buffer case.  The
I/O_diff, Output_diff and 3-state_diff model types were also
added at that time, and Rref_diff and Cref_diff could be the
only timing subparameters when [External Model] existed.  So,
it is documented in Section 6b.

However, at that time we also wanted to use it for the psuedo
differential structures created by two single-ended buffers, as
already supported by IBIS Version 3.2.  However, we stabilize
the set of keywords and subparameters in older versions, so,
we would require the file to be designated as [IBIS_Ver] 4.0
or above.

Some specific responses are in your text.

Best Regards,
Bob

Timothy Coyle wrote:
> Hi Cynthia,
> 
> I think the best approach is to contact your tool vendor support and see 
> if they support the keyword and how it is used in their toolset. This 
> way you can get concrete data on how you can and cannot use the keyword. 
> As with any specification there are a lot of things that can be done but 
> whether and how they are supported by the industry is always a different 
> story.
> 
> Best,
> 
> Timothy Coyle
> 
> President
> Signal Consulting Group LLC
> 405 Western Ave #430
> South Portland, ME  04106
> Tel:  617.297.2566
> Email: tim.coyle@siconsultant.com <mailto:%20tim.coyle@siconsultant.com>
> Web:   http://www.siconsultant.com <http://www.siconsultant.com/>
> SharkSim <http://www.sharksim.com/> - IBIS Modeling Made Easy
> 

> From: Cynthia Low [mailto:ibis_user@yahoo.com]
> Sent: Wednesday, July 08, 2009 7:07 AM
> To: Bob Ross; Timothy Coyle
> Cc: ibis-users@server.eda.org; bob@teraspeed.com
> Subject: Re: [IBIS-Users] Rref_diff keyword
> 
> 
> Thanks Bob and Tim.
> 
> I have a follow up questions, do you know why this Rref_diff keyword is 
> noy create back in v3.2 IBIS specification when it is use to represent 
> just the differential buffer test load parameters, just like the 
> other single ended test load (Cref, Vref and Rref).

We do not allow new information (keywords and subparameters) in
older versions so that older versions always define the same maximum
set of information.

> 
> Can you confirm on the definition of using this subparameters of 
> Rref_diff as the differential buffer test load parameters in the IBIS 
> models?

It is in the Specification, and ibischk4 accepts it.  But the file has
to be version 4.0 or above.  Tim's tool comment applies, and some tools
do support Rref_diff.

> 
> Is the usage to let the IBIS user know how the relevant software will 
> generate the clock to out time based on the software default test load 
> setting, and with this test load information (including Rref_diff, Cref, 
> Rref and Vref) the user will know how the software will interprete this 
> set of default test load and obtain the TCO value in teh 
> simulation Hyperkynx (for example), then we can then calculate the 
> flight time?
> 

Contact vendor to get the answer.  We do not use the IBIS reflectors
for vendor-specific questions.


> 
> Some question that i am still unclear:
> 
> 1. In the IBIS spec example, the differential buffer use is with Model 
> Type I/O_diff. (Please see below). But for my case, it is just an output 
> buffer and not a 'diff' type. will that matter?

This subparameter requires and [External Model], and that model should
be a true differential buffer and have two output terminals (for outp
and outn).  Otherwise it should be psuedo differential

> 
> 2. Also, eventhough the parser actually allow this Rref_diff keyword to 
> pass the warning and error check but can we conclude that the 
> correctness of the usage on this Rref_diff subparameters?

At this point I would assume it is correct.  I would also verify
with the EDA vendors of interest if they support this.  They might
still use Only the single-ended loads rather than a differential
simulation to determine the timing reference point, or they might
not have implemented support of Rref_diff for timing or require
vendor-specific work-arounds.

However, from the IBIS specification perspective, Rref_diff is
correct and should be supported with true differential model types
or with psuedo differential model types.

> 
> 3. Can we conclude that the Rref_diff subparameters are just like the 
> Cref, rref, and vref subparameters that is use to specified the timing 
> test load? Can you please comment?

Yes for Version 4.0 and above.  In fact, the intent is to allow for
psuedo and true differential buffers any combination of Rref_diff,
Cref_diff, Rref, Cref, Vref.  So the timing test load can be only
differential, only a single-ended set, or both.


> 
> Lastly, can someone comment on the EDA usgae on this keyword? Or i need 
> to file specific request to Menthor to requst for this information? Many 
> many thanks.
> 
>  
> 
> [Model]         External_Model_Diff
> Model_type      I/O_diff               | Requires [External Model]
> Polarity        Non-Inverting
> Enable          Active-High
> | The [Diff Pin] vdiff value overrides the thresholds below
> Vinl = 0.8V                            | Input logic "low" DC voltage, 
> if any
> Vinh = 2.0V                            | Input logic "high" DC voltage, 
> if any
> |                         | The true differential measurement point is at
> |                         | the crossover voltage
> |                         | The Vmeas value is overridden
> Vmeas = 1.5V              | Reference voltage for timing measurements
> |                         | Single-ended timing test load is still permitted
> Cref = 5pF                | Timing specification test load capacitance value
> Rref = 500                | Timing specification test load resistance value
> Vref = 0                  | Timing specification test load voltage
> |                         | These new subparameters are permitted for
> |                         | single-ended differential operation based on the
> |                         | [Diff Pin] keyword
> Rref_diff = 100           | Timing specification differential resistance 
> value
> 


Contact Mentor about this.  With [External Model], this looks like correct
IBIS syntax; without [External Model], I/O_diff is technically an Error.
There is more detail, but you need to contact the vendor.  There
might be vendor-specific extensions.


> 
> Regards
> 
> Cynthia
> 
> ------------------------------------------------------------------------
> 
> From: Bob Ross <bob@teraspeed.com>
> To: Timothy Coyle <tim.coyle@siconsultant.com>
> Cc: Cynthia Low <ibis_user@yahoo.com>; ibis-users@server.eda.org; bob@teraspeed.com
> Sent: Wednesday, July 8, 2009 8:02:47 AM
> Subject: Re: [IBIS-Users] Rref_diff keyword
> 
> Hi Cynthia:
> 
> Tim is correct that you need to check with EDA vendors if and how
> they might support Rref_diff and Cref_diff.  Also, some tools
> may not support many features in the latest versions of the
> Specification or use the latest version of the parser in its
> internal checking.
> 
> The latest parser is ibischk4 V4.2.2, which corresponds to the
> latest version of IBIS (Version 4.2) for which a parser exists.
> I am glad you found that the latest parser gives fewer Warnings
> for legal cases than I suspected.  Therefore you do not need
> the extra parameters that I suggested.
> 
> I have more responses in your text.
> 
> Bob
> 
> 
> Timothy Coyle wrote:
>>  Hi Cynthia,
>>
>>  It will probably be best if you contact Mentor about their keyword 
> support for Rref_diff. Mentor has a couple of different tools and 
> parsers (ICX Pro, ICX IS, HyperLynx) and they all work a little 
> different depending on the version you have. And while the IBIS 
> Specification allows you to specify keywords like Rref_diff there is no 
> guarantee that a specific tool vendor is supporting that keyword. In 
> general the Rref_diff keyword is used as part of a reference load for 
> the manufacturer?s AC timing. Depending on the differential technology 
> the manufacturer is using it may or may not be applicable.
>>
>>  Best,
> 
>>
>>  Timothy Coyle
>>
>>  President
>>  Signal Consulting Group LLC
>>  405 Western Ave #430
>>  South Portland, ME  04106
>>  Tel:  617.297.2566
>>  Email: tim.coyle@siconsultant.com <mailto:tim.coyle@siconsultant..com> 
> <mailto:%20tim.coyle@siconsultant.com <mailto:20tim.coyle@siconsultant.com>>
>>  Web:  http://www.siconsultant.com <http://www.siconsultant.com/>
>>  SharkSim <http://www.sharksim.com/>  - IBIS Modeling Made Easy
>>
>>  From: owner-ibis-users@server.eda.org 
> <mailto:owner-ibis-users@server.eda.org> 
> [mailto:owner-ibis-users@server.eda.org 
> <mailto:owner-ibis-users@server.eda.org>] On Behalf Of Cynthia Low
>>  Sent: Tuesday, July 07, 2009 6:55 AM
>>  To: Bob Ross
>>  Cc: ibis-users@server.eda.org <mailto:ibis-users@server.eda.org>; 
> bob@teraspeed.com <mailto:bob@teraspeed.com>
>>  Subject: Re: [IBIS-Users] Rref_diff keyword
>>
>>  Hi Bob
>>
>>  Thank you so much for your reply. It was indeed helpful. However, 
> there is something that I was not certain off and hope you can help to 
> clarify.
>>
>>  In your reply, you mentioned that Rref_diff can be used but to prevent 
> the parser warning message by adding in the Cref,Vref and Rref 
> subparameters.
>>
>>  Which parser checker are you refering to? IBISchecker v4.x..x? Can you 
> please specify?I basically use the latest IBIS checker v4.2.2 to check 
> and there is no warning on the usage.
> 
> Ibischk4, Version 4.2.2.  I am glad you found fewer messages.  I believe
> the intent was to allow Rref_diff and Cref_diff to exist without the
> the Rref, Cref, and Vref that I suggested.  So if the latest parser
> does not complain, the model is ok.
> 
> However, if there are cases where the parser complains, you can
> add the extra subparameters to suppress the messages without
> changing the test conditions.  We may have already tested
> for the case that the buffer is never used in single-ended
> mode, and therefor the extra subparameters are not needed.
> 
>>
>>  Also, I will be using this Rref_diff keyword for a dedicated 
> differential output buffer and Cref, Vref and Rref will not exist, 
> (which you suggest to state Cref = 0F ,Vref=1.163 (to follow Vmeas?) and 
> Rref=1E9), this can be applied? It is a dedicated output buffer which do 
> not share any single ended timing test-load information, it could still 
> be use in that way? You can refer back to the previous attachement for 
> the timing test load that I wish to specified (resistor).
> 
> Since the parser does not issue any Warnings, and all [Model]s in the
> [Pin] list where Rref_diff and/or Cref_diff exist are described as
> differential (by the [Diff Pin] association), then you do not need
> the Cref, Rref, Vref (which can be with the other subparameters
> such as Vmeas in any order) information.
> 
> So if the timing test load for the LVDS example passes the parser,
> then it is ok as is.
> 
>>
>> 
>>  Lastly and most importantly, I noticed that this Rref_diff keyword, 
> which is used in v4.1 IBIS specification is not integreted into many 
> simulation tool yet, and also the ICX checker may not allow this keyword 
> to pass, if so, can this subparameters still be use? How will the 
> simulator tools interprete the Rref_diff keyword. Also, if the simulator 
> is using a v3.2 ibis checker, this keyword will be redundant as it will 
> not pass, am I rite? Could you please kindly comment on this?
> 
> Tim responded to this question.  To add, the tool should report an
> error if it is using ibischk3 as the basis for checking and reject
> the model.  You need a tool that has at least IBIS Verison 4.1 support,
> although with ibischk4, Version 4.2.2, a file that is designated
> as 4.0 will pass since in the final versions, we are interested
> in maintaining only major version distinctions (such as between
> 3.2 and 4..2 and not sub-version destinctions such as between 4.0, 4.1
> and 4.2.
> 
>>
>>
>>  Rref_diff and Cref_diff can also be used for [Model]s when
>>  they are connected by the [Diff Pin] keyword without the [External
>>  Model] extension.  However, to prevent parser warning messages,
>>  it is strongly encouraged to still add the Cref and/or Vref and Rref
>>  subparameters.  Cref can be 0 F and/or Rref can be very large.
>>  The parser check was added to be sure that the model maker intentially
>>  meant to have a negligible single-ended timing test load.
> 
> I was wrong with this advice.  Your model is ok if it passes the
> ibischk4.2.2 without Warnings or Error.
> 
>>
>>  Thanks and best regards
>>
>>  Cynthia
>>
>> 
>>  ------------------------------------------------------------------------
>>
>>  From: Bob Ross <bob@teraspeed.com <mailto:bob@teraspeed.com>>
>>  To: Cynthia Low <ibis_user@yahoo.com <mailto:ibis_user@yahoo.com>>
>>  Cc: ibis-users@server.eda.org <mailto:ibis-users@server.eda.org>; 
> bob@teraspeed.com <mailto:bob@teraspeed.com>
>>  Sent: Tuesday, July 7, 2009 1:45:08 PM
>>  Subject: Re: [IBIS-Users] Rref_diff keyword
>>
>>  Hi Cynthia:
>>
>>  My answers are in your text.
>>
>>  Best Regards,
>>  Bob
>>
>>  Cynthia Low wrote:
>>>  Dear IBIS experts
>>>  I am currently looking at the Rref_diff subparameters under the
>>  [model] keyword and the usage of this subparameters. This 
> subparameters is only available in IBIS specification of v4.1 but not in 
> older version. May I know how this subparameters keyword is being used? 
> the detail of the question is as below.
>>>  IBIS Spec:
>>>  Rref_diff is used for true differential buffer timing test loads.
>>  where the buffer hv to be specified only when the [Diff Pin]
>>>  keyword connects two models, and each buffer references the same
>>  model.  The Rref_diff and Cref_diff subparameters can be used with the 
> Rref, Cref, and Vref subparameters for a combined differential and 
> signal-ended timing test load.
>>>  1. My questions will be, can this Rref_diff sub-parameters be used
>>  ONLY when we engaged using the true differential [External Model] with 
> either spice or vhdl-ams? As shown in the IBIS spec.
>>
>>  Yes, Rref_diff can be the only timing test load parameter for true
>>  differential buffers, as illustrated in Section 6b.
>>
>>>  2. Can this Rref_diff keyword be used in the context of a normal true
>>  differential buffer under the [Model] keyword, and at the same time 
> the model is define under [Diff Pin] section to describe the timing test 
> loads just like how the rest of the Cref, Vref keyword are used to 
> facilitate for board-level timing simulation, with the setup as 
> attached. Can the Rref_diff keyword used under the example below?
>>
>>  Rref_diff and Cref_diff can also be used for [Model]s when
>>  they are connected by the [Diff Pin] keyword without the [External
>>  Model] extension.  However, to prevent parser warning messages,
>>  it is strongly encouraged to still add the Cref and/or Vref and Rref
>>  subparameters.  Cref can be 0 F and/or Rref can be very large.
>>  The parser check was added to be sure that the model maker intentially
>>  meant to have a negligible single-ended timing test load.
>>
>>  So the example below may produce parser Warning messages.  If that is
>>  the case, I would add
>>
>>  Cref = 0F
>>
>>  and/or possibly
>>
>>  Vref=1.163
>>  Rref=1E9
>>
>>  as needed to remove any Warning messages.
>>
>>>  [Model] 25_lvds
>>>  Model_type Output
>>>  Polarity Non-Inverting
>>>  Enable Active-Low
>>>  Rref_diff = 100
>>>  Vmeas = 1.163
>>>  C_comp  1.8pF  1.8pF  1.8pF  Please kindly comment. Thanks.
>>>  Regards
>>>  Cynthia
>>>
>>>
>>>  --


-- 
Bob Ross
Teraspeed Consulting Group LLC     Teraspeed Labs
121 North River Drive              13610 SW Harness Lane
Narragansett, RI 02882             Beaverton, OR 97008
401-284-1827                       503-430-1065
http://www.teraspeed.com           503-246-8048 Direct
bob@teraspeed.com

Teraspeed is a registered service mark of Teraspeed Consulting Group LLC


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

--------------------------------------------------------------------
|For help or to subscribe/unsubscribe, e-mail majordomo@eda-stds.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 e-mail a request to ibis-request@eda-stds.org.
|
|IBIS reflector archives exist under:
|
|  http://www.eda-stds.org/pub/ibis/email_archive/ Recent
|  http://www.eda-stds.org/pub/ibis/users_archive/ Recent
|  http://www.eda-stds.org/pub/ibis/email/         E-mail since 1993
Received on Wed Jul 8 16:28:26 2009

This archive was generated by hypermail 2.1.8 : Wed Jul 08 2009 - 16:29:14 PDT