Re: [802.3_ISAAC] [EXTERNAL] Re: [802.3_ISAAC] More information on “Complexity of ACT/GMSLE Implementations”

[Ragnar Jonsson <rjonsson@xxxxxxxxxxx>]

Hi Ahmad,

 

I agree that the DFE coefficients look different, but I would say that it is not entirely unexpected, given that we are using different cables, etc. in our simulations.

 

Regarding the S-parameter files, as is explained in jonsson_3dm_01_03_10_25.pdf these are simulated channels and not measured. It looks like your code for reading Touchstone files is hardcoded for including noise parameters, which is not a requirement of the Touchstone format. In fact, most S-parameter files for cables do not include the noise parameters. I recommend that you look for code that can read general Touchstone file format. I may be able to help you with code that can read the files I shared if you are not able to find a more general code for reading the Touchstone file format.

 

Another alternative is to load the .mat files I provided into MATLAB. These files have both the impulse response and the S-parameters.

 

BTW, just to be clear, the cable models only model the cables. You need to model the HP effect of the PoC separately with an IIR filter.

 

Ragnar

 

From: Ahmad Chini <chiniahmad.ieee@xxxxxxxxx>
Sent: Friday, June 27, 2025 3:47 PM
To: Ragnar Jonsson <rjonsson@xxxxxxxxxxx>
Cc: STDS-802-3-ISAAC@xxxxxxxxxxxxxxxxx
Subject: [EXTERNAL] Re: [802.3_ISAAC] More information on “Complexity of ACT/GMSLE Implementations”

 

Hi Ragnar Thanks for sharing the DFE taps. The DFE coefficients are very different from those in "https: //grouper. ieee. org/groups/802/3/dm/public/0325/Chini_3dm_02b_0325. pdf". I may be able to provide more info on the expected DFE

ZjQcmQRYFpfptBannerStart

Prioritize security for external emails:

Confirm sender and content safety before clicking links or opening attachments

    Report Suspicious    ‌

ZjQcmQRYFpfptBannerEnd

Hi Ragnar

 

Thanks for sharing the DFE taps. The DFE coefficients are very different from those in "https://grouper.ieee.org/groups/802/3/dm/public/0325/Chini_3dm_02b_0325.pdf". 

 

I may be able to provide more info on the expected DFE coefficients when HPF and CTLE are designed to reduce the channel.IR length. 

 

You can also see plots on page 8 of that presentation that shows how much of the low frequency noise is rejected by the internal PHY filtering (HPF).

 

I tried to open the cables S parameter with Matlab, I get the following error

 

Error using rfdata.data/read>readsnp
Noise parameters must have 5 columns.
Error in rfdata.data/read (line 51)
        refobj = readsnp(h, filename);

 

If you can share the actual VNA measurement of the cable's S2P, I will be able to check them on my platform.

 

Ahmad Chini

 

On Thu, Jun 26, 2025 at 4:34 PM Ragnar Jonsson <rjonsson@xxxxxxxxxxx> wrote:

All,

 

In today’s 802.3dm ad hoc meeting I gave presentation titled “Complexity of ACT/GMSLE Implementations” (see https://www.ieee802.org/3/dm/public/adhoc/062625/jonsson_3dm_02_06_26_25.pdf).

 

In discussions of the presentation, I was asked to provide more information about the channel I used in my simulations, and about the DFE coefficients.

 

The channels that I used are taken from the simulated coax cables that I shared in the March meeting in Atlanta: https://www.ieee802.org/3/dm/public/0325/jonsson_3dm_01_03_10_25.pdf

More specifically I used the 5^th and 7^th cables shown on slide 14 of that presentation. Note that the 7^th cable has particularly bad echo characteristics because it has the “ugly” connector with excessive impedance mismatch, and it has 6 Ohm cable impedance mismatch at each connector.

 

The S-parameters and impulse responses for these cables can be downloaded from links in the Atlanta 802.3dm home page: https://www.ieee802.org/3/dm/public/0325/index.html

 

The enclosed plots show the Decision Feedback Equalizer coefficients for the four different simulation runs shown in my presentation today.

 

Please let me know if you have any additional questions. 

 

Ragnar

---

To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1

---

To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1