Have not investigated common mode behaviour.
Didn’t see a “large” difference in differential behaviour, just enough to see that ADS and com didn’t’ agree under certain conditions (but not a large difference)
I just find that the conversion from 4 port to SDD has certain implicit assumptions tied to it.
I believe Sameh Elnaggar had a paper in Design Com about this.
bk
From: Richard Mellitz <Richard.Mellitz@xxxxxxxxxx>
Sent: Wednesday, June 18, 2025 12:17 PM
To: Bill Kirkland <wkirkland@xxxxxxxxxxx>; STDS-802-3-B400G@xxxxxxxxxxxxxxxxx
Subject: RE: [802.3_B400G] Proposal for Impedance change for IEEE P802.3dj d2.0
|
Caution: This email originated outside of Semtech.
|
We tried this. Unless the reference package common mode there is no difference. The IP is embedded in reference die load and package common mode. Another discussion.
Right now, all COM computations are normalized to Z0 for correct network concatenation. You should get same COM if you supply s4p’s and any impedance or use any Zo reference. It’s a 4-port renormalization
of passed s-parameters in COM right now.
… Rich
|
From: Bill Kirkland <00001cd3c07be0a7-dmarc-request@xxxxxxxxxxxxxxxxx>
Sent: Wednesday, June 18, 2025 11:35 AM
To: STDS-802-3-B400G@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_B400G] Proposal for Impedance change for IEEE P802.3dj d2.0
|
Caution: This email originated from outside the organization. Do not click links or open attachments unless you recognize the sender and know the content
is safe. |
Rich, shouldn’t be any “IP” issues with my suggests. One can easily build up a 4 port pkg/die by simply adding the com pkg/die to each leg and setting the cross terms to 0. Renormalize sparameter impedance
is a standard technique as well.
e.g. 2 take 2 port die/pkg to 4 port
Sobj_Tx_pkg = Create_Pkg (Freq, Pkg_param, Pkg_Length_Tx(:,z_p_select), Pkg_Z(:,z_p_select), C_diepad(Tx_Rx,:), L_comp(Tx_Rx,:), C_bump(Tx_Rx), C_pkg_board(Tx_Rx), C_v(Tx_Rx) );
Data = "">
Data(1:2:3,1:2:3,:) = Sobj_Tx_pkg.Parameters;
Data(2:2:4,2:2:4,:) = Sobj_Tx_pkg.Parameters;
Stx_s4p_pkg = sparameters(Data,Freq,50);
From:
richard.mellitz@xxxxxxxxx <richard.mellitz@xxxxxxxxx>
Sent: Wednesday, June 18, 2025 11:07 AM
To: Bill Kirkland <wkirkland@xxxxxxxxxxx>;
STDS-802-3-B400G@xxxxxxxxxxxxxxxxx
Subject: RE: [802.3_B400G] Proposal for Impedance change for IEEE P802.3dj d2.0
|
Caution: This email originated outside of Semtech.
|
Hi Bill,
Good points. I agree the multi-port s-parameters are the way to go but much of the information needed may be proprietary. COM and commercial simulators/instruments are compliments.
We can improve COM by improving the complements of multi-mode specifications which tends avoid some sensitive IP issues.
To paraphrase John Calvin:
COM is the performance you need to meet.
Measurement (and full simulation) is the performance you expect to get.
Rich Mellitz, Samtec
From: Bill Kirkland <00001cd3c07be0a7-dmarc-request@xxxxxxxxxxxxxxxxx>
Sent: Friday, June 13, 2025 9:59 AM
To: STDS-802-3-B400G@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_B400G] Proposal for Impedance change for IEEE P802.3dj d2.0
Rich may I suggest a small change to the order of operations in COM for s-parameters.
I would suggest changing the order of s-parameter manipulations (if they are still the same as a couple of years ago).
I believe you currently read in the channel s4p file and immediately convert to Sdd and then add in package and die s-parameters.
This works if both Rdies are 50 ohms. However if they are not, then to get COM to agree with ADS (which is really realy nice to have)
then you should:
Read in S4p’s, Convert all addition S-parameters to S4p’s, Cascade S-parameters, then re-normalize the s-parameters from 50 ohms to Rdie (both ends have to be the same), then finally
convert to Sdd before computing the frequency domain transfer function.
The error in COM is not particularly large but is noticeable under certain conditions when trying to compare COM SBR with those from ADS when not using Rd = 50.
Below slide is taken from a presentation I was working on in 2020.
From: Richard Mellitz <000014533bad0b9c-dmarc-request@xxxxxxxxxxxxxxxxx>
Sent: Thursday, June 12, 2025 4:49 PM
To: STDS-802-3-B400G@xxxxxxxxxxxxxxxxx
Subject: [802.3_B400G] Proposal for Impedance change for IEEE P802.3dj d2.0
|
Caution: This email originated outside of Semtech.
|
Hi All,
Here is a proposal for changing the reference impedance for d2.0 to 92.5 ohms differential and 23.125 ohms common mode since the target impedance for test fixture was change.
The is no change to the COM computation algorithm required. There are several sections that are touched by this.
Comments? Thoughts?
Regards,
Richard Mellitz
Samtec
|
To unsubscribe from the STDS-802-3-B400G list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1
To view our privacy policy, including the types of personal information we collect, process and share, and the rights and options you have in this respect, see
www.semtech.com/legal.
To unsubscribe from the STDS-802-3-B400G list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1
To view our privacy policy, including the types of personal information we collect, process and share, and the rights and options you have in this respect, see
www.semtech.com/legal.
To unsubscribe from the STDS-802-3-B400G list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1
To view our privacy policy, including the types of personal information we collect, process and share, and the rights and options you have in this respect, see www.semtech.com/legal.
To unsubscribe from the STDS-802-3-B400G list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-B400G&A=1