RE: [10GBASE-T] RE: CIR estimation
Victor
The Fext response is the coupling into the victim line shown in chart 25. All four pairs in the cable are of the same length. But, there usually is a different nonzero pair-to-pair skew. The 1000BASE-T standard specifies the maximum to be around 50 ns. Using this number for the 833 Msymbol/sec system, the skew can be in excess of 40 symbols. So, depending on the skew and on how the Fext couples into the victim pair, the main cursor of the Fext can lead or lag the main cursor of the victim line response by a similar number of symbols. I do think you have identified a formatting problem with the horizontal axis of the two plots. I believe the origin in the Fext response does not correspond to zero time.
regards
Bill
-----Original Message-----
From: Ermolayev, Viktor [mailto:viktor.ermolayev@intel.com]
Sent: Tuesday, January 28, 2003 1:18 AM
To: William Jones; Ermolayev, Viktor; stds-802-3-10gbt@ieee.org
Cc: George Zimmerman
Subject: RE: [10GBASE-T] RE: CIR estimation
Bill,
Thanks for quick response. Your answers are quite clear. Could you please
introduce full clarity into my second question.
snip"1. For computer simulations, try a long FIR filter."snip - O.K. We
make impulse response models in a similar manner.
snip "2. Different twisted pairs in the cable will have different delay
characteristics."snip - It means that responses of Wire Line on slide 25 and
FEXT on slide 28 were given for different twisted pairs (say for different
lengths) as examples only.
Is this correct?
Thank you in advance, Victor.
-----Original Message-----
From: William Jones [mailto:wjones@solarflare.com]
Sent: Tuesday, January 28, 2003 12:10 AM
To: Ermolayev, Viktor; stds-802-3-10gbt@ieee.org
Cc: George Zimmerman
Subject: [10GBASE-T] RE: CIR estimation
Viktor
Your questions were not completely clear. Here are the answers based on my
interpretation. Feel free to follow up if I answered the wrong question.
1. For computer simulations, try a long FIR filter.
2. Different twisted pairs in the cable will have different delay
characteristics.
Slide 10 of the tutorial at the November meeting is the measured frequency
response of the coupling functions.
Bill
William W. Jones, Ph.D.
Director of Systems Engineering
SolarFlare Communications, Inc.
949-581-6830, ext. 2550
mobile: 619-405-2445
fax: 949-581-4695
wjones@solarflare.com
-----Original Message-----
From: Ermolayev, Viktor [mailto:viktor.ermolayev@intel.com]
Sent: Monday, January 27, 2003 5:14 AM
To: stds-802-3-10gbt@ieee.org
Cc: George Zimmerman; William Jones
Subject: CIR estimation
I wish great successes to all participants of Reflector.
The team from Intel (N.Novgorod, Russia) was subscribed
stds-802-3-10GBT-Modeling. We are going to make MatLab simulation of 10Gb/s
system as full as possible. We think this system should be provided the
channel estimates more widely than previous one. We have to estimate all
channel impulse responses (CIR). We have studied the presentation by George
Zimmerman and Bill Jones and found the examples of Wire line, Echo, Next and
Fext CIRs. Arising questions are:
1.How can we make Next model, which has such a large delay spread ?
2.Why is the Fext front delay less than Wire line delay?
We have also the question regarding slide 10. Is frequency response or
interference signal spectrum shown on this slide in form of fast oscillating
curve?
Thank you in advance.
Victor on behalf of team.