[SI-LIST] Re: Simulating frequency dependent s-parameters in HSPICE
- From: "Oh, Dan" <doh@xxxxxxxxxx>
- To: "ray.anderson@xxxxxxxxxx" <ray.anderson@xxxxxxxxxx>, Krishna Chaitanya <krishna1abc@xxxxxxxxx>
- Date: Tue, 13 Mar 2012 11:27:03 -0700
Hi Krishna,
One thing many people forget about S-parameter models in SPICE simulation is
that, in addition to the signal frequency content, you also need to consider
the frequency response of your model itself.
The maximum frequency range makes a perfect sense for frequency-domain analysis
but if you are interested in time-domain broadband simulation, you must
consider the side effect of IFFT. Basically, you have to make sure that your
frequency-domain model is roll off sufficiently in order to avoid aliasing. You
can either increase the frequency range of model or use a proper filter.
This problem often occurs for modeling of short interconnects such as connector
pins or vias. An alternative solution to this problem is to combine the overall
channel response in the frequency-domain as the overall S-parameter response
would have sufficient attenuation. This must be done by users as SPICE
simulators do not perform this for you.
If you need a reference material for this phenomena, please contact me directly.
Just my two cents,
-Dan Oh
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
Behalf Of Ray Anderson
Sent: Tuesday, March 13, 2012 11:01 AM
To: Krishna Chaitanya
Cc: si-list@xxxxxxxxxxxxx; Ray Anderson
Subject: [SI-LIST] Re: Simulating frequency dependent s-parameters in HSPICE
Krishna-
As with most SI issue, "it depends" when determining how high the interconnect
should be characterized to.
In general you should characterize your interconnect up to at least 3X the
frequency of maximum spectral content (which would be 3 X 10 GHz or
30 GHz for a signal with a NRZ data rate of20 GB/s) or 3 times the frequency
of the maximum spectral content as determined from the risetime (a commonly
used approximation is .35/Rt or .35/10ps = 35GHz for your case) which would be
3 X 35 GHz (105GHz). The factor of 3X is kind of arbitrary (some people suggest
5x) but the object is to have a bandwidth that most (say 90+ %) of the signal
significant spectral energy of your exciting signal fits in. If you wanted to
cover 100% of the spectral energy you would need "DC to Daylight" bandwidth.....
In this case the frequency determined by the risetime criteria will be the
determining factor (105GHz > 30 GHz).
10psec RT is quite fast, but seeing as you are concerned with on die signals,
it is possible. Accurately characterizing the interconnect up to the 100 GHz
region may be challenging. Even though it is easy to tell your extractor to run
up to 100GHz you need to make sure that your modeling assumptions are accurate
at those frequencies. If your modeling assumptions aren't good at very high
frequencies, the s-parameters generated at those frequencies may be highly
suspect.
-Ray
Xilinx Inc.
From: Krishna Chaitanya [mailto:krishna1abc@xxxxxxxxx]
Sent: Monday, March 12, 2012 4:50 PM
To: Sinha, Snehamay; Ray Anderson
Cc: Ihsan Erdin; si-list@xxxxxxxxxxxxx
Subject: Re: [SI-LIST] Simulating frequency dependent s-parameters in HSPICE
Thanks Snehamay, Ray and all others for helping me understand how s-parameters
are simulated in HSPICE.
Thanks for the set of guidelines Ray. I think Q3D has the adaptive sweep.
One quick question regarding the frequency range for s-parameter extraction
using Q3D. When you say the highest frequency of interest, should I consider
maximum data rate I am targeting or the bandwidth from the rise/fall time?
If I am looking for a 20 Gbps with rise/fall times of 10ps, should I consider
the frequency sweep from DC to 100GHz (corresponding to 10ps
rise/fall) or to 10 GHz (corresponding to 20Gbps)?
I am planning to use adaptive sweep so that the step size can be determined by
the solver as mentioned earlier.
Thanks,
Krishna
.
This email and any attachments are intended for the sole use of the named
recipient(s) and contain(s) confidential information that may be proprietary,
privileged or copyrighted under applicable law. If you are not the intended
recipient, do not read, copy, or forward this email message or any attachments.
Delete this email message and any attachments immediately.
------------------------------------------------------------------
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
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:
