Analog Channel Loss in SerDes System
Limiting factors in high-speed data transmission includes cross talk, attenuation, and
reflection noise. The Analog Channel block
System object™ parameterize a channel model that represents a lossy transmission line typical
in high-speed SerDes application. The loss model is constructed either from a parameterized
channel loss model or from an impulse response from another source.
Loss Model from Channel Loss Metric
A discrete time, band-limited analog impulse response characterizes the serdes.ChannelLoss
System object. It represents the response of a system to an impulse response vector with an
impulse magnitude of , where dt is the sample interval.
To calculate the impulse response, serdes.ChannelLoss
first calculates the S-parameter component S21 according to
channel loss at frequencies ranging from 0 to
fmax, maximum frequency of interest, where . This is done by determining the loss at the target frequency, and then
linearly extrapolating required channel length to achieve target channel loss. Then
transmitter and receiver termination S-parameter are then calculated according to the
equations 93A-17 and 93A-18 from the IEEE 802.3bj-2014 specifications .
After calculating S21, the System object adds the negative frequency data points based on the expected even symmetry of the real components of S21 and the odd symmetry of the imaginary components of S21 of the frequency response. The impulse response is calculated from the inverse Fourier transform of S21. Finally, the impulse response is resampled so that the sample interval is dt.
Loss Model from Impulse Response
To construct a loss model from an impulse response vector, input the impulse response vector from another source. You can also define the impulse sample interval. Changing the symbol time and number of samples per symbol changes the data rate of the SerDes system.
Introducing Cross Talk
You can include crosstalk in your simulation from the SerDes Designer app,
or using the Analog Channel block in Simulink®. If the parameterized channel loss model is used, you can specify the strength
of the near and far end crosstalk aggressors according to specification standards or you can
specify your own custom integrated crosstalk noise (ICN) levels. If a custom impulse
response is used, then up to 6 additional columns can be used to represent the crosstalk
impulse response. For more information, see Analog Channel and
 IEEE 802.3bj-2014. "IEEE Standard for Ethernet Amendment 2: Physical Layer Specifications and Management Parameters for 100 Gb/s Operation Over Backplanes and Copper Cables." https://standards.ieee.org/standard/802_3bj-2014.html.