## EMC Question of the Week: October 23, 2023

Two high-speed signals are routed parallel to each other on identical microstrip traces. The self-inductance of each trace is 6 nH/cm and the mutual inductance between them is 0.6 nH/cm. If both traces have matched terminations, a reasonable worst-case estimate for the maximum far-end crosstalk between the two signals is

- 0
- 0 dB
- -20 dB
- -20 dB/cm

## Answer

The best answer is “c.” Assuming we have defined crosstalk as the ratio of the coupled voltage in the victim line to the signal voltage in the aggressor line, a good estimate for the maximum crosstalk is L_{12}/L_{22} (or in this case -20 dB).

In two striplines matched at both ends, we would expect the far-end coupling to be close to 0 (-∞ dB), because the electric-field coupling and the magnetic-field coupling would cancel. However, in a microstrip line, the magnetic-field coupling is stronger than the electric-field coupling. And if the source-end of the traces is not matched, cancellation of the two field-coupling mechanisms at the far-end is not achieved.

0 dB of crosstalk means the voltage on the victim trace could equal the voltage in the aggressor trace. This would not occur between any pair of parallel traces that had reasonably matched terminations.

For short traces, the crosstalk is proportional to length. However, as the length approaches a quarter-wavelength at any given frequency, the crosstalk levels out and does not exceed L_{12}/L_{22} at any frequency.

-20 dB/cm? Even at the lower frequencies where crosstalk is proportional to length, it would never be expressible in units of dB/cm. The coupled voltage could be proportional to the length, but not the crosstalk expressed in decibels. Also, the maximum crosstalk would not be expected to decrease with increasing length.

Note that for any given signal expressed in the time domain (or any particular harmonic in the frequency domain), the crosstalk could be much less than L_{12}/L_{22}, but it could not be more.

Have a comment or question regarding this solution? We'd like to hear from you. Email us at