Estimated reading time: 8 minutes

Crosstalk, Part 2: What It Looks Like

In Part 1 of this series (Note 1), I pointed out that people seemed to find the concept of crosstalk difficult to wrap their arms around. I pointed out that this might be because of the following reasons:

1:  It has two different fundamental causes

2:  These causes generate two different signals

3: These two signals flow in opposite directions

4:  These signals can interact with each other

5:  These two signals have significantly different shapes

6:  These shapes behave differently as a function of coupled length, and

7:  Neither shape resembles the “aggressor” signal that caused the crosstalk in the first place!

Part 1 of the series dealt with the first four items above. Crosstalk is caused by electromagnetic coupling, which has two components, an electric (capacitive) coupling and magnetic (inductive) coupling. These components result in two independent crosstalk components, forward and backward crosstalk. The two coupling components tend to cancel in the forward direction and to reinforce each other in the backward direction. And the two components, forward and backward crosstalk, have decidedly different characteristics.

This column, Part 2, of the series, we will look more closely at the shapes of the two crosstalk components. In Part 3 of the series we will look even more closely at the shape and the magnitude of the backward crosstalk component.

Forward Crosstalk

The fundamental forward crosstalk components (the capacitively and inductively coupled components) flow in opposite directions (again, see Part 1). As such, they tend to cancel. They exactly cancel in a purely homogeneous environment, i.e., an environment that is uniform everywhere around the coupled traces – a stripline environment where the relative dielectric coefficient of the materials is constant and uniform everywhere between the two planes is such an environment. In this special case, there is no forward crosstalk to worry about, at least in most practical cases.

If the environment is not homogeneous – as in a microstrip environment or a stripline environment where there are different material properties present between the two planes – forward crosstalk signal can be generated. The magnitude of this signal is almost always small unless there are special conditions or unless the coupled region is relatively long. Nevertheless, it is instructive to look at what a forward crosstalk signal looks like if one is generated.

Consider Figure 1. In the bottom part of the figure we have an aggressor driver driving a step-function signal toward a receiver. The victim trace is shown above the aggressor trace (Note 2).

Figure 1. Forward crosstalk animation.

Recall that crosstalk is a point concept. That is, crosstalk coupling occurs at a point, and as the aggressor signal propagates along the trace, the point of coupling propagates along the victim trace in sync with the aggressor signal. The forward crosstalk-coupled signal on the victim trace also propagates along the victim trace in the same (forward) direction and at the same propagation speed. So at every coupled point along the trace we have the effects of the coupling at that point PLUS the effects of the coupling at all the prior points along the victim trace that are propagating along at the same speed.

The coupling at any point might be small. But the accumulated signal becomes larger and larger as the coupled region increases. Therefore, in cases where there is a relatively long coupled region (i.e., long, parallel traces), the signal might actually grow to a significant size.

We can make these two generalities about forward crosstalk:

1. The magnitude of the forward crosstalk signal increases with increasing coupled length.
2. The width of the forward crosstalk pulse is approximately equal to the rise time of the aggressor signal.

There will be a sharp contrast between these generalities and those that we will draw for backward crosstalk later on!

Backward Crosstalk

Consider now the backward crosstalk shown in Figure 2. It shows the same configuration as Figure 1 but this time we will focus on the backward crosstalk signal.

Figure 2. Backward crosstalk animation.

At the beginning of the coupling region, the aggressor signal couples into the victim trace at that point. At the next point in time, two things have happened:

1. The coupled signal in the victim trace propagates backwards.
2. The coupling point between the two traces propagates forward.

So as the aggressor signal propagates along its trace, a coupled signal couples into the victim trace at the same time and in sync with it. But at the same time, the signals that have already been coupled into the victim trace continue to propagate in the backward direction. This continues until the aggressor signal reaches the end of the coupled region.

Note in particular that there is no buildup of the victim signal (as there is in the forward crosstalk case). The backward crosstalk signal does not grow (increase) as the coupled region get longer (see Note 3.) It simply gets wider.

So now we can make these two generalizations regarding backward crosstalk:

1. The magnitude of the backward crosstalk signal remains constant even with increases in the coupled length.
2. The width of the backward crosstalk signal continues to increase as the coupled length increases.

Note how these two generalizations are exactly opposite those for forward crosstalk!

Backward Crosstalk Width

Figure 3 shows the situation when the aggressor signal just reaches the end of the coupled region. Notice that a backward crosstalk signal is being coupled into the victim trace at that point. But the portion of the coupled signal that was created when the aggressor signal was just starting down the coupled region has propagated backward along the victim trace from that coupling point. This portion of the coupled signal has been propagating backward at the same speed as the aggressor signal has been propagating forward. The distance that that component of the backward crosstalk signal has been traveling, therefore, is the same distance as the distance along the coupled region.

Figure 3. Condition when the aggressor signal is at the end of the coupled region.

Therefore, the second generalization above can be refined and expanded as follows:

2. The width of the backward crosstalk signal continues to increase as the coupled length increases and is equal to twice the length of the coupled region (see Note 4.)

Backward Crosstalk Reflections

The story does not end with Figure 2, however. The backward crosstalk signal shown propagating toward the near end (see Note 5) may reflect off whatever is there and then propagate toward the far end. The reflected signal at the far end is often what is of concern in our circuits.

Figure 4 is what happens next. This animation follows immediately after the animation in Figure 2. It shows two different conditions. The top victim trace has no termination at the near end, so there is a 100% positive reflection of the backward crosstalk signal. This signal component propagates back up the victim trace toward the far end. The bottom victim trace has a terminating resistor at the near end. This absorbs the backward crosstalk signal and there is no reflection of the signal back toward the far end. Importantly, this second case produces NO (backward) crosstalk signal at the far end!

Figure 4. How the backward crosstalk component might be reflected.

Consider the implications of this. If the crosstalk signal at the near end is important to us, then any backward crosstalk signal that may be generated is of some importance. But if it is only the far-end crosstalk signal that important to us, then we can completely eliminate it by properly terminating the victim trace at the near end!

Not only that, but we have already discussed how the forward crosstalk component can be minimized by putting the traces in a homogeneous environment (i.e., in a stripline environment.) Therefore, these two steps (putting the traces in a stripline environment and properly terminating the near end of the victim trace) can completely eliminate a crosstalk problem at the far end of the victim trace, regardless of the geometry! That is a powerful implication.

Now, it may not be possible to terminate the near end of the victim trace without causing other undesirable circuit performance effects. If so, then terminating the near end may not be an option in some cases. In those cases, the primary way to minimize crosstalk is to minimize the coupling between the two traces. That is done using the two design strategies:

1. Route the traces as close as practical to the underlying plane, and

2. Separate the traces as far apart as practical.

Why Troubleshooting Can Be So Difficult

By now, you may have some idea as to why troubleshooting crosstalk issues can be so difficult. Here are some of the reasons:

1. The backward and forward crosstalk signals look entirely different from each other and neither one resembles the aggressor signal that created it (more on this in Part 3 of this series.)

2. The backward and forward signals may occur (at the far end) at different points in time, depending on the relative lengths of the aggressor and victim traces and whether the coupling region occurs over the entire length of the traces or just a portion of their length(s).

3. The forward component may or may not exist depending on the degree of homogeneity of the environment.

4. The backward component may or may not exist (at the far end), and its magnitude may be unpredictable, depending on the reflection coefficient at the near end of the victim trace.

In Part 3 of this series we will look more closely at the shape of the backward crosstalk signal component and try to estimate its magnitude.

Notes

1. See "Crosstalk: Why It's Difficult to Understand, Part 1"
2. I have taken some artistic license here. In most practical situations the forward crosstalk signal is opposite in sign to the aggressor signal. So the forward crosstalk signal in this case might actually be a negative (in sign) signal.
3. We will expand on this point in Part 3 of this series.
4. We will refine this statement in Part 3 to add “plus one rise time.”
5. The near-end and far-end crosstalk signals are often referred to as NEXT and FEXT, respectively.

Douglas Brooks has an MS/EE from Stanford University and a Ph.D. from the University of Washington. He has spent most of his career in the electronics industry in positions of engineering, marketing, general management, and as CEO of several companies. He has owned UltraCAD Design Inc. since 1992. He is the author of numerous articles in several disciplines, and has written articles and given seminars all over the world on Signal Integrity issues since founding UltraCAD. His book, Printed Circuit Board Design and Signal Integrity Issues was published by Prentice Hall in 2003. Visit his Web site at www.ultracad.com.