The victim's near end noise (NEXT) is dominated by the even and odd impedance difference, and the victim's far end noise (FEXT) is directly proportional to the odd and even time delay difference. Therefore, when the crosstalk model is considered, it is important to obtain a good correlation between even and odd time delays and impedances. The best suitable models for this condition are distributed four-line coupled and symmetric coupled lossy line models.
To model the crosstalk, the differential fully coupled model, similar to the one shown in Figure 1, is built first, and then we change the probing setup to measure the crosstalk values (Figure 2). The IConnect partitioning in the distributed four-line coupled model should take into account the time delay difference; otherwise the FEXT model will underestimate the induced voltage. When the desired correlation between the even and odd impedances and time delays is obtained, the differential model could be could easily be converted to a single-ended one by replacing the PWL source on the “victim” with the source probe set to zero volts. IConnect simulation of this setup will produce four waveforms where the reflected waveform of the probe port will give NEXT value, and the waveform transmitted from the same port will give FEXT.
Figure 1. Fully coupled differential model topology.
Figure 2. Crosstalk model. An ideal probe to measure the crosstalk replaced one of the PWL sources.
This FAQ Applies to:
Product: CSA8000B, CSA8000, TDS8000, TDS8000B, CSA8200, TDS8200, 80SICON, 80SSPAR, 80SICMX
FAQ ID : 56526View all FAQs »