Addressing Forward and Backward Crosstalk in PCB Design
If you’ve ever been at a raucous party in your college days, then you know how tough it is to talk over everyone else in the room. Signals in your electronics can propagate into neighboring traces, making it difficult for messages to reach their destinations.
With faster signal speeds comes more crosstalk in your PCB. But crosstalk actually comes in two varieties. Each can have a serious effect on signal integrity at each end of a trace, but you can suppress crosstalk with some simple design strategies.
Capacitive and Inductive Coupling
While few PCB designers state this explicitly, there are actually two types of crosstalk, and each type manifests itself differently. These two types of crosstalk arise due to the way in which neighboring traces couple to each other. As much as we might like to imagine, circuit traces in a PCB are not perfectly isolated from each other.
In thinking about the way an aggressor and a victim trace interact on a PCB, the two traces act as a pair of inductors, where a signal in the aggressor trace induces a signal in the victim trace according to Faraday’s law. The effects of forward and backward crosstalk in PCBs are particularly prominent in high speed digital signals, where the fast rise time induces a momentary burst of current.
Neighboring traces are really parallel conductors separated by the PCB dielectric, so they also form a capacitor. The impedance between the traces depends on the length and depth of the traces, the separation between traces, and the frequency spectrum in the propagating digital signals. Each type of coupling has particular effects on crosstalk, and designers must devise better solutions for crosstalk suppression, especially if you are working with a system where differential signaling is not viable.
In high speed signaling, traces are normally terminated as they can easily behave as transmission lines. Although trace termination eliminates impedance discontinuity, thus preventing signal integrity problems due to reflection, termination has no effect on forward or backward crosstalk. Other design strategies will need to be used to suppress crosstalk.
Traces on a green PCB=
Two Types of Crosstalk
Given the inherent coupling between two traces, how do these two types of crosstalk arise and how do they affect signals in a PCB?
First, consider a case where a digital signal in the aggressor trace switches from OFF to ON. Looking at Faraday’s law, one finds that the current induced by an increasing magnetic field always propagates in the opposite direction as the drive current. This induced current is called inductively coupled crosstalk. This is one component in backward crosstalk.
The other component in backward crosstalk is due to capacitive coupling. When a signal propagates through the aggressor trace, the pulse occupies a small region within the trace that is equal to the speed of light in the trace multiplied by the signal rise/fall time. This region can be called a coupling region. Once the signal switches off, the victim trace discharges, causing current to flow in the victim trace. The current flows in both the forward and backward directions.
The backward crosstalk pulse has a nearly constant amplitude in magnitude but is twice as wide as the propagation delay of the aggressor pulse as it travels through the coupling region. What is so interesting about this pulse is that the width of the pulse has nothing to do with coupling strength; in other words, it is only a function of the size of the coupling region. In contrast, the amplitude of a forward crosstalk signal increases as the length of the coupled region increases. Its pulse width is equal to the rise/fall time of the aggressor signal.
If you examine the output from a network , forward and backward crosstalk during digital switching will decrease the opening in an eye diagram, indicating a higher bit error rate. Current spikes and drops prevent the reference signal from rising to the correct level on these traces and contribute to the overall bit error rate. If you perform some signal simulations in your PCB, or you connect your device to an oscilloscope, you can spot the effects of each type of crosstalk if you know what to look for.
Backward crosstalk manifests itself as a burst of current in the negative direction; this looks like a digital pulse that has double the width of the aggressor pulse. Forward crosstalk appears to be a Gaussian-like pulse that appears during the rise/fall time of the aggressor signal. The two components will point in opposite directions, thus they will appear as voltage pulses with opposite polarity.
Signal measurement with an oscilloscope
The only time traces do not couple is when there is no current present in the traces, or when DC current flows in the traces. In the case of switching off a DC current, there is still a momentary voltage/current transient that can outlast the switching time if the RC time constant is long. This issue with transients only becomes really apparent in traces connected to loads with high resistance or that are shunted to ground with a resistor. Traces also have stronger coupling if they are closer together as this decreases capacitance.
This immediately provides one solution to reducing coupling between traces: do not space them too close together. Using logic ICs with slightly slower rise/fall times also helps, as you can still maintain the same data transfer rate (i.e., switching speed). The downside to slowing down the rise time is that it reduces the tolerance range on trace lengths in high speed signaling.
The great thing about the two types of crosstalk is that they almost completely cancel each other if the aggressor and victim traces are laid out in stripline configuration with controlled impedance. Routing multiple signals as differential striplines with controlled impedance helps reduce crosstalk between an aggressor pair and victim pair even further.
Crosstalk suppression in your circuits means paying attention to a number of design rules and best practices. Fortunately, a PCB design package like Altium Designer® uses a heavily rules-driven design engine that ensures your routing strategy can keep crosstalk within acceptable limits. You’ll have access to simulation tools that allow you to diagnose a range of signal integrity issues and verify that your device functions as designed.