In PCBs, ICs, and cable assemblies, the most commonly referenced crosstalk is far-end crosstalk observed at a receiver component.
Band-reject filters are the inverse of a bandpass filter: they filter out unwanted power within a specific frequency range.
The transfer function of a band reject filter can be calculated by hand or by using a SPICE simulator.
Old analog telephones were highly susceptible to different types of crosstalk
Remember those old phones that plugged into the wall? Before the phone system went digital, it was possible to hear faint echoes of someone else’s conversation during a phone call. This noise, where signals bleed between different wires in a cable, is better known as crosstalk, and the experienced electronics designer should be familiar with this phenomenon.
When layout designers and engineers refer to crosstalk, we don’t typically differentiate between different types of crosstalk. In fact, there are two types of crosstalk and there are two possible causes of crosstalk in any system. The two types of crosstalk are near-end and far-end crosstalk, both of which create unwanted interference between signals on different interconnects. Let’s look closer at the differences between near-end vs. far-end crosstalk to see how they arise and how they interact with components on an interconnect.
Comparing Different Types of Crosstalk
Near-end and far-end crosstalk both arise when a driven signal couples some of its power into a nearby interconnect on a PCB, in a cable, or in an IC. Crosstalk is undesired, as we want to ensure interconnects only carry power launched from the driver component. In general, you can never totally eliminate crosstalk; you can only reduce its strength to the point where it is un-measurable.
In understanding these types of crosstalk, we need to consider where the crosstalk is measured and what direction it travels once induced in an interconnect.
Near-End vs. Far-End Crosstalk
Near-end and far-end crosstalk define the location (or polarity) where a crosstalk signal is measured in an interconnect: near-end refers to the driver side of the victim interconnect, while far-end refers to the receiver side. The crosstalk signal can be induced from the aggressor interconnect into the victim interconnect as the signal travels, and where the signal is detected depends on the coupling mechanism and signal swing direction (rising vs. falling edge).
Forward vs. Backward Crosstalk
Because near-end crosstalk is measured at the driver end of the victim interconnect, it is often called backward crosstalk. This is because the crosstalk signal had to travel “backwards” (back towards the driver, or in the reverse direction of the aggressor signal) along the victim interconnect. The same idea applies to forward crosstalk (far-end crosstalk)—the signal needs to travel forwards to the receiver where it is detected.
Example showing crosstalk induced on a victim interconnect, leading to backward and forward crosstalk on the rising and falling edges, respectively
Forward and backward crosstalk are not specific types of crosstalk; these terms only refer to the direction that some signal travels on the interconnect. Backward crosstalk could also be far-end crosstalk, depending on the switching direction and the relative orientation of the aggressor/victim interconnects.
Mutual Inductance and Mutual Capacitance
All crosstalk is created inductively (through parasitic mutual inductance) or capacitively (through parasitic mutual capacitance). Inductive crosstalk is only induced when the aggressor signal changes levels, thus higher speed signals create stronger crosstalk. Capacitive crosstalk is created due to a changing potential difference between the two interconnects. In circuit models describing interconnects, the mutual inductance and capacitance are used to describe coupling between the aggressor and victim interconnects.
In this cross-section, we can see how parasitics between two traces produce mutual capacitance and mutual inductance, which are responsible for coupling noise between traces as crosstalk
The image above should show how crosstalk can be suppressed or eliminated in a physical layout—we need to address parasitic coupling between the victim and aggressor interconnects. Some methods for reducing crosstalk in a physical layout include:
Applying shielding structures between victim and aggressor interconnects, such as additional ground pour or via fences.
Increasing the spacing between victim and aggressor interconnects.
Increasing the width or decreasing the distance to the reference plane to lower loop inductance.
Other Types of Crosstalk
The telecommunications world has taken to naming multiple types of crosstalk beyond near-end and far-end crosstalk. Note that these different types of crosstalk do not have special coupling mechanisms, coupling of noise between aggressor and victim interconnects always occurs inductively or capacitively. Just like in a PCB, IC, or cable assembly, crosstalk can appear as forward or backward crosstalk. Here are other terms often encountered in the telecommunications industry that are used to describe crosstalk:
Alien crosstalk (AXT): This term refers specifically to crosstalk between wire strands in unshielded twisted pair (UTP) cabling, but it can also be used to describe crosstalk in PCBs. Alien crosstalk refers to crosstalk between cables that are not part of the same assembly.
Power-sum NEXT and FEXT: These terms refer to the power carried by an induced crosstalk signal.
Power-sum equal-level crosstalk (PS-ELFEXT). This is equal to PS-FEXT + PS-NEXT.
Of these terms, only one describes a particular situation involving crosstalk: alien crosstalk. On a PCB, this is less common than crosstalk between traces and may not be noticeable for several reasons. First, cables may include some ground wiring, which creates some isolation between traces and other cables. Second, cables in PCBs tend to be located far from a trace that can induce crosstalk, so any crosstalk would be quite weak. Finally, the signal level in digital PCBs is often too low to produce a strong crosstalk signal, regardless of the position of the cable.
One possible exception can occur at very high frequencies, such as in wireless/RF systems that carry high power. In these systems, the frequency is high enough and the power in a cable assembly might be strong enough to produce strong crosstalk to a nearby trace. RF systems with unique cable or connector assemblies often need to be simulated with a 3D field solver utility to properly quantify crosstalk and identify when it will exceed allowed limits.
When you need to evaluate near-end vs. far-end crosstalk in your PCB and implement solutions to reduce crosstalk, use PCB design and analysis software with an integrated 3D EM field solver. Cadence provides a powerful set of software tools that help automate many important tasks in systems analysis, including a suite of pre-layout and post-layout simulation features to evaluate your system.