[SI-LIST] Re: Touchstone for transmission lines (was Re: touchstone models)
- From: Conrad Herse <conrad.herse@xxxxxxxxxxxxxxxxxx>
- To: si-list@xxxxxxxxxxxxx
- Date: Mon, 31 Oct 2011 08:54:11 -0500
Thanks Aubrey, and to everyone else who replied. Apparently the issue
was not with the s-parameter models themselves but rather with the
process/tools I was using to generate them (I've since moved to using
scalable transmission line models for the PCB routes themselves). I find
it a bit counter intuitive that I need to decrease the frequency step
interval as the transmission line length increases, I'll need to ponder
that a bit more.
Thanks,
Conrad
Aubrey K Sparkman wrote:
> Conrad,
>
> Just to pick on Scott... :-) Hi, Scott! The "inherent problem with
> using S-parameter models for long transmission line segments" is that
> people forget to use sufficiently small frequency step intervals.
>
> The thing that can get you is file/model size if you remember to include
> fine enough resolution or if you don't remember, you will have accuracy
> issues. As the transmission line length (or end to end delay) gets
> longer, you need smaller and smaller frequency step intervals to get the
> correct response.
>
> Easy to verify. Just setup a simple transmission line, pick a line length
> (ex: 1" long) sweep frequency (1GHz to 10GHz) and frequency step (0.1GHz).
> S11 looks like you might expect, right? Then increase the line length
> to 10"and take another look at S11. Looks weird, right Now change the
> frequency step to 0.01 GHz and take another look at S11. Back to what you
> might expect? As the length gets longer the file size explodes. A good
> "per unit length" (accurate, passive, causal) model will save time and
> disk space.
>
> Good Luck,
> Aubrey Sparkman
> SparkRight Solutions
>
> On 10/27/11 11:33 AM, "Scott McMorrow" <scott@xxxxxxxxxxxxx> wrote:
>
>> Conrad
>>
>> There is no inherent problem with using S-parameter models for long
>> transmission line segments. The trick is using a solver that can create
>> accurate causal models (Ansoft 2D, HFSS, Simberian Simbeor, ... etc),
>> with resolution that is appropriate for the maximum length of entire
>> interconnect system. The disadvantage in using them is in the lack of
>> scalability. However this can be resolved by some solvers, such as
>> Simbeor, with length scaling at the time of touchstone file generation,
>> or by creating shorter length touchstone files that can be concatenated,
>> either in a network simulator, or with additional custom code. I
>> routinely create sweeps with touchstone interconnect models swept from 1
>> mm to multiple meters in length, in increments of 1 mm, with no loss of
>> accuracy. I've published multiple papers that show our modeling match
>> to measurements with extremely low error from DC to 50 GHz.
>>
>> Other transmission line models, such as the W-element, can be
>> attractive, since they are inherently scalable. However,
>> implementations from tool to tool may be different. If you use them, I
>> would definitely advise correlation of your solver and your simulator
>> with physical measurements.
>>
>>
>> best regards,
>>
>> Scott
>>
>> Scott McMorrow
>> Teraspeed Consulting Group LLC
>> 121 North River Drive
>> Narragansett, RI 02882
>> (401) 284-1827 Business
>> (401) 284-1840 Fax
>>
>> http://www.teraspeed.com
>>
>> Teraspeed® is the registered service mark of
>> Teraspeed Consulting Group LLC
>>
>>
>> On 10/27/2011 11:42 AM, Conrad Herse wrote:
>>> This thread reminded me of a discussion we had awhile back with a CAD
>>> vendor (who shall remain nameless). They discouraged us from using
>>> s-parameter models for long segments of traces, stating that s-parameter
>>> models were better suited for small geometry discontinuities such as
>>> vias, connectors, and packages. For long PCB traces they recommended use
>>> of transmission line models rather than an s-parameter model.
>>>
>>> We had been using s-parameter models for sections of a channel when
>>> modeling a full system, the s-parameter model included sections of PCB
>>> routing along with vias, connectors, etc. Their recommendation was to
>>> model the trace portions of the channel with transmission line models,
>>> and use s-parameter models for just the vias, connectors, etc. I can't
>>> remember the issue which triggered our original discussion with them but
>>> I believe it was related to flight time delays (or lack there of) when
>>> including a long transmission line in an s-parameter model.
>>>
>>> I'd be interested in hearing other's thoughts on this topic, is it bad
>>> practice in general to include long transmission line segments within an
>>> s-parameter model?
>>>
>>> Thanks,
>>>
>>> Conrad Herse
>>>
>>>
>>> Carrier, Patrick wrote:
>>>> Hi Megan--
>>>> Others have forwarded links to some other great resources, but I
>>>> wanted to let you know that there are a number of videos which discuss
>>>> S-parameters up at mentor.com, the most relevant being:
>>>>
>>>> http://www.mentor.com/products/pcb-system-design/multimedia/s-parameters
>>>>
>>>> Many connector models are delivered in S-parameter format.
>>>> S-parameters are also useful for comparing your total interconnect
>>>> loss (including traces, vias, and connectors) against a specified loss
>>>> budget.
>>>> For AC power analysis, Z-parameters are most often used, which are
>>>> similar. The HyperLynx Touchstone Viewer will let you convert between
>>>> S-, Z-, and Y-parameters.
>>>>
>>>> Enjoy!
>>>> --Pat
>>>>
>>> ------------------------------------------------------------------
>>> To unsubscribe from si-list:
>>> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>>>
>>> or to administer your membership from a web page, go to:
>>> //www.freelists.org/webpage/si-list
>>>
>>> For help:
>>> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>>>
>>>
>>> List technical documents are available at:
>>> http://www.si-list.net
>>>
>>> List archives are viewable at:
>>> //www.freelists.org/archives/si-list
>>>
>>> Old (prior to June 6, 2001) list archives are viewable at:
>>> http://www.qsl.net/wb6tpu
>>>
>>>
>> ------------------------------------------------------------------
>> To unsubscribe from si-list:
>> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>>
>> or to administer your membership from a web page, go to:
>> //www.freelists.org/webpage/si-list
>>
>> For help:
>> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>>
>>
>> List technical documents are available at:
>> http://www.si-list.net
>>
>> List archives are viewable at:
>> //www.freelists.org/archives/si-list
>>
>> Old (prior to June 6, 2001) list archives are viewable at:
>> http://www.qsl.net/wb6tpu
>>
>>
>>
>
>
> ------------------------------------------------------------------
> To unsubscribe from si-list:
> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>
> or to administer your membership from a web page, go to:
> //www.freelists.org/webpage/si-list
>
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
>
> List technical documents are available at:
> http://www.si-list.net
>
> List archives are viewable at:
> //www.freelists.org/archives/si-list
>
> Old (prior to June 6, 2001) list archives are viewable at:
> http://www.qsl.net/wb6tpu
>
>
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
