Chapter 6 Lab Exercise: Measuring Frequency-Domain Crosstalk Between Microstrip Traces

Crosstalk between printed circuit board traces is matter of great importance to both EMC and signal integrity engineers. Many circuit board layouts require high-speed traces to be routed parallel to each other. If the traces are too close, the crosstalk might be unacceptable. But how close is too close?

Preparation:  This exercise requires a special test board with SMA connectors and parallel traces. These boards are relatively easy to design and layout. PCB fabricators can build them in small quantities are a cost of a few hundred dollars per board. They can also be purchased from various companies that specialize in EMC or SI education. 

Equipment Required:

• a test board with parallel microstrip traces and low-inductance connectors
• vector network analyzer
• 2 50-Ω terminators

Procedure:

Step 1: Students should measure the board and trace geometry including the length and width of the traces, distance between traces, and trace height.

Step 2: Students should calculate the expected crosstalk due to common-impedance coupling at DC and 100 MHz. Both should be extremely small and won't affect the measurements in this exercise.

Step 3: Using the VNA, students can drive one trace through Port 1 and measure the voltage coupled to the second trace through Port 2. 

Step 4: With an open termination on the far-end of the source trace and a matched termination on the near end of the victim trace, plot S₂₁. Choose any frequency where S₂₁ increases linearly with frequency to determine the amount of capacitive crosstalk. Calculate the value of C₁₂.

Step 5: The crosstalk should level off at a value equal to C₁₂/C₂₂. Use this to estimate the value of C₂₂.

Step 6: Short the far end of the source trace and the near end of the victim trace. Use the measured plot of S₂₁ to calculate L₁₂ and L₂₂. 

Step 7: Note that the traces have the same geometry, so L₁₁=L₂₂ and C₁₁=C₂₂. Using the measured values of L₂₂, L₁₂, C₂₂ and C₁₂, calculate the expected near-end and far-end crosstalk when both traces have matched 50-Ω terminations. 

Step 8: Terminating the unconnected ends of the traces with 50-Ω terminators, measure the near-end and far-end crosstalk. Compare your results to the expected values. Why isn't the far-end crosstalk zero?

Notes: 

Once the L and C parameters are known, the time-domain coupling can be determined. A pulse generator and a 2-channel oscilloscope can be used to measure the near-end and far-end coupling in the time domain.