Guard Trace Impact on Crosstalk Between PCB Traces

This article discusses the crosstalk reduction between PCB traces by utilizing a guard trace between the traces and investigating the effect of the guard trace grounding. This discussion is an extension of the investigation presented in the March 2017 issue of the In Compliance magazine [1] where the impact of varying the PCB board geometry (distance between the traces and distance from the traces to the ground plane) was described.

Figure 1: Microstrip PCB and its circuit model

Generator circuit is driven by a time-varying voltage source, VS with the impedance RS, and terminated by a load resistor, RL. The receptor circuit is terminated by the load resistors, RNE and RFE, on the near- and the far-end, respectively. Guard trace separates the two microstrip traces and can be connected to ground on either or both ends, or left floating.

The near- and far-end crosstalk voltages induced in the receptor circuit are due to the superposition of the capacitive and inductive coupling between the circuits. Let’s discuss each coupling separately.

Capacitive Coupling and Shielding

Figure 2 shows a field model of the capacitively-coupled circuits [2], [3], driven by a sinusoidal source.

Figure 2: Capacitively-coupled circuits

In this model, is the capacitively-induced noise voltage, CGR – is the mutual capacitance between the generator and receptor circuits. It is assumed that both the generator and the receptor circuit are electrically short, and thus the mutual capacitance can be represented as a lumped parameter. In this field model of the capacitive coupling, the electric field lines, , (created by the voltage in the generator circuit, ) originate on the generator circuit and terminate on the receptor circuit.

The circuit model of the capacitively-coupled circuits is shown in Figure 3.

Figure 3: Circuit model of the capacitively-coupled circuits

The noise voltage induced in the receptor circuit due to the capacitive coupling is:

(1)

If the generator voltage and frequency cannot be changed, and the termination resistances are fixed, then the noise voltage can be lowered by reducing the mutual capacitance CGR. This can be accomplished by: (a) moving the conductors further apart (see [1] for details), (b) changing their orientation, and (c) shielding, discussed next.

Effect of the Shield on Capacitive Coupling

Figure 4 shows a field model of the capacitively coupled circuits with a guard trace (acting as a shield) between the generator and the receptor circuits. In order for the guard trace to reduce capacitive coupling, it must be grounded at least at one end. We will verify this in the measurement section.

Figure 4: Capacitively-coupled circuits separated by a guard trace

Note that some of the electric field lines terminate on the guard trace giving rise to the mutual capacitance, CGS, between the generator circuit and the guard trace. Since the guard trace is grounded, however, this capacitance does not impact the induced noise voltage in the receptor circuit. Thus the equivalent circuit model is the same as shown in Figure 3, and the noise voltage induced in the receptor circuit is given by Eq. (1).

Even though the expressions for the noise voltage with and without the guard trace are identical there is one major difference. The difference is that CGR with a guard trace is much smaller than CGR without a guard trace. We will show this effect in the measurement section.

Inductive Coupling and Shielding

Figure 5 shows a field model of the inductively-coupled circuits [2], [3].

Figure 5: Inductively-coupled circuits

In this model, and are the inductively-induced noise voltages across the near- and far-end resistors, respectively. LGR – is the mutual inductance
between the generator and receptor circuits. For electrically short circuits, this mutual inductance can be represented by a lumped parameter. Field model of the inductive coupling consists of the magnetic field lines, , (created by the current in the generator circuit,) crossing the loop area of the receptor circuit.

The circuit model of the inductively-coupled circuits is shown in Figure 6.

Figure 6: Circuit model of the inductively-coupled circuits

The noise voltages induced in the receptor circuit due to the inductive coupling are:

(2a)

(2b)

If the generator current and frequency cannot be changed, and the termination resistances are fixed, then the noise voltage can be lowered by reducing the mutual inductance LGR. This can be accomplished by; (a) bringing the ground plane closer to the signal plane (see [1] for details), (b) changing the trace orientation, and (c) shielding, discussed next.

Effect of the Shield on Inductive Coupling

Figure 7 shows a field model of the inductively coupled circuits with a guard trace (acting as a shield) between the generator and the receptor circuits. In order for the guard trace to reduce inductive coupling, it must be grounded at both ends. We will verify this in the measurement section.

Figure 7: Inductively-coupled circuits separated by a guard trace

Field model of the inductive coupling consists of the magnetic field lines, , (created by the current in the generator circuit, ) crossing the loop area of two circuits: the receptor circuit (modeled by the mutual inductance LGR) and the guard trace circuit (modeled by the mutual inductance LGS). Additionally, the shield current, , creates the magnetic field lines crossing the loop area of the receptor circuit. This is modeled by the mutual inductance LRS.

The corresponding circuit model is shown in Figure 8.

Figure 8: Circuit model of the inductively-coupled circuits separated by a guard trace

In this model: LSH is the self-inductance of the shield, RSH is the shield resistance, and is the shield current. The shield current induces an additional voltage in the receptor circuit which opposes the generator-current induced voltage. This is the essence of the shield effectiveness. If the shield is not grounded at both ends then it has no effect on the inductive coupling. We will verify this in the measurement section.

Total Coupling

The total coupling in the circuit is obtained by the superposition of the capacitive and inductive couplings.

Without the guard trace the total coupling is:

(3a)

(3b)

When the guard trace is grounded at one end only (NE or FE), it reduces the capacitive coupling only; it does not reduce the inductive coupling.

Since the expression for the near-end voltage inside the brackets of Eq. (3a) is always positive, (regardless of which coupling type dominates), it follows that reducing the capacitive coupling also reduces the total coupling.

The situation is different for the far-end voltage. When the inductive coupling dominates, the expression inside the brackets of Eq. (3b) is negative. Therefore, reducing the capacitive coupling makes the total coupling even larger (in the negative sense). This only happens when the inductive coupling dominates the capacitive one and we only reduce the capacitive coupling.

This somewhat counterintuitive result will be verified in the measurement section (compare Case 1 vs. Cases 2 and 3). Guard trace grounded at both ends, reduces both the capacitive and the inductive coupling.

Verification

The experimental setup for crosstalk measurements is shown in Figure 9.

Figure 9: Experimental setup

The generator circuit is driven by a 1 V – amplitude, 1 MHz trapezoidal signal with a rise and fall times of 100 ns. All resistive terminations are 50 Ω; generator and receptor traces are separated by 30 mils, and are 54.8 mils from the common ground plane.

Four different cases were investigated, and are described in Table 1, and shown in Figures 10-13.

 Case Guard Trace Termination VNE VFE 1 Floating 1.28 mV 580 µV 2 Grounded at NE 1.04 mV 760 µV 3 Grounded at FE 1.10 mV 740 µV 4 Grounded at Both Ends 720 µV 400 µV

Table 1: Guard trace terminations and induced voltages

Figure 10: Case 1 – Floating guard trace

Figure 11: Case 2 – NE grounded, FE open

Figure 12: Case 3 – NE open, FE grounded

Figure 13: Case 4 – NE and FE grounded

Figure 10 (Case 1) shows the induced voltages when the guard trace is absent or floating. Figures 11 and 12 show the voltages when the guard trace is grounded at one end only. Note that the near-end total voltage always decreases. The far-end total voltage, however, increases (in the negative sense) since the capacitive coupling was lowered. These results are consistent with the discussion of the previous section.

Grounding the guard trace at both end reduces both the capacitive and inductive couplings, as shown in Figure 13, and therefore reduces the total coupling the most.

References

1. Adamczyk, B., Teune, J., Crosstalk Reduction between PCB Traces, In Compliance Magazine, March 2017.
2. Clayton R. Paul, Introduction to Electromagnetic Compatibility, Wiley, 2006.
3. Henry W. Ott, Electromagnetic Compatibility Engineering, Wiley, 2009.

Dr. Bogdan Adamczyk is a professor and the director of the EMC Center at Grand Valley State University (http://www.gvsu.edu/emccenter) where he performs EMC precompliance testing for industry and develops EMC educational material. He is an iNARTE certified EMC Master Design Engineer, a founding member and the chair of the IEEE EMC West Michigan Chapter. Prof. Adamczyk is the author of the textbook “Foundations of Electromagnetic Compatibility with Practical Applications” (Wiley, 2017). He can be reached at adamczyb@gvsu.edu.

Ryan Aldridge is the Engineering Laboratories Supervisor at Grand Valley State University, and the vice chair of the IEEE EMC Chapter of West Michigan.  He works closely with Prof. Adamczyk developing EMC educational material and assists him with GVSU’s EMC Center.  He is currently working on his Master’s Degree in Electrical and Computer Engineering at Grand Valley State University.