| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
I realize there's
a fundamental lack of DC-to-CD duality, and things get complex quickly. The core idea is
to quantify or constrain what’s missing in COM that leads to interoperability failures. COM evaluates differential-mode (DM) performance but can overlook the impact of common-mode (CM) impairments, especially those introduced at the system periphery. Take skew as a simple
concrete example. If the periphery introduces no skew, then COM typically evaluates the channel correctly. But if skew is present — say, from a transmitter — it produces common-mode voltage. That CM voltage, affected by CM return loss (CM RL), can convert
into differential-mode energy and reach the receiver. This effect is not captured in COM, which assumes ideal DM signaling. In this way, CM-to-DM
conversion is a key contributor to channel interoperability. The reverse — DM to CM — is primarily a concern for emissions or receiver sensitivity. Returning to the
skew example: I considered incorporating
equalization into the analysis. However, that would require rethinking the CM RL spec. Specifically, we'd need to shift from a static CM RL mask to a CM RL power-based SNR metric, analogous to how SNR_MDFEXT (proposed) models crosstalk noise. This would allow
defining an SNR_CMRL. SNR_CMRL is easily converted into modal gammas which are used to create the channel's CM transfer function, yielding a modal pulse response for which equalization could be applied meaningfully. But — one step at
a time. … Rich
From: Hansel DSilva <00004708d0869f3e-dmarc-request@xxxxxxxxxxxxxxxxx>
Hi Richard, In the attached proposal regarding using SCMR_CH (for the channel) to control skew and other imbalances, I noticed that only SCD21 is mentioned. Would the proposal include computing
SCMR_CH using SDC21 as well? Below is my reasoning behind this question.
- Equation 178-1 in D2P0.
- D2p0 mentions VCM_FB as the full-band peak-to-peak AC common-mode voltage defined by the method specified in 179.9.4.1 and measured with the transmitter equalization set to “no Equalization”.
- I interpret this as calling out the differential-mode to common-mode insertion loss (SCD21).
- Equation 178-6 and 178-7 in D2p0.
- It calls out differential-mode to common insertion loss (SCD21) and common-mode to differential-mode insertion loss (SDC21).
- Assuming the throughs are (1)---(2) and (3)---(4).
-Differential insertion loss is reciprocal, meaning SDD21 ≈ SDD12.
a.
SDD21= 0.5*(S21 - S23+ S43- S41).
SDD12= 0.5*(S12 - S32+ S34- S14).
4.
- Mode conversion terms are not strictly reciprocal:
a.
SCD21 ≈ SDC12 and SCD12 ≈ SDC21.
5.
- However, SCD21 ≠ SCD12, as shown in the following equations:
a.
SCD21= 0.5*(S21 - S23 + S41 - S43).
a.
SCD12 = 0.5*(S21 + S32 - S14 - S34).
a.
SDC21 = 0.5*(S12 + S23 -S41 -S43).
a.
SDC12 = 0.5*(S12 - S32 + S14 - S34). Regards, Hansel D'Silva From: Richard Mellitz <richard.mellitz@xxxxxxxxx> CAUTION: EXTERNAL EMAIL
Hi All, The has been presented some this before. This this attached is a proposal for using SCMR_CH (for the channel) to control skew and other imbalances. The first half, SCMR and V_CM was controlled
in the transmitters. This proposal it to complete the process for the channels. 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 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 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 |