Exploring the Characteristics of Field Coupled Crosstalk

Demonstrating the Differences in Closely Coupled Cabling

Interference between circuits is a common occurrence, which is increasingly problematic with our ever condensing technology operating in a wide range of operating frequencies. This interference, known as crosstalk, can be conducted or radiated, and we refer to the circuits generating and receiving this interference as either the source or receptor. When examining the many types of crosstalk, the focus is generally on the source and path part of the source-path-receiver generalized EMC model. Therefore, a variety of mechanisms are responsible for allowing this to occur. In this article, we’ll break down some of the important characteristics of crosstalk and discuss some common ways of eliminating it.

We first start by separating crosstalk into the following categories, starting with conducted and radiated coupling between two circuits, as shown in Figure 1.

In all levels of system design, it is imperative to be able to identify the differences between each type of coupling mechanism, and then understand how to positively impact the performance of the circuit or system. Next, we’ll cover the background of each type of coupling mechanism, giving some common names an engineer might hear when discussing it and how to properly identify it.

Radiated Coupling

Radiated or contactless coupling is a key issue in circuit design. Since wires, cable looms and PCB lands are more than just a conductive path, they’re essentially small radiating components of the system in which they reside. When engineers mention this type of coupling mechanism, they’re referring to the path which describes how the electromagnetic radiation travels between the source and receptor circuits. We further break this down to the following types:

• The far field normally referred to as electromagnetic field coupling;
• The near field electric field dominant coupling normally referred to as capacitive coupling;
• The near field magnetic field dominant coupling normally referred to as inductive coupling.

To determine the type of radiated crosstalk present in the system, you must first determine how far away the source of the noise is. Thus, the first thing to understanding crosstalk is to determine how close you are to the source by knowing the frequency of the emission, and how far the device being negatively affected is away from the source. For the purposes of this article, this distance is defined as:

If you’re close enough to the source of the noise to be in the near field, the electromagnetic wave has yet to become a plane wave, loosely defined where the electric and magnetic field parts form an electromagnetic wave. Here the wave takes on unique characteristics in that the impedance of the wave is not a constant value, and the source can be considered either a high voltage source or high in current.

In higher voltage applications, which usually arise with high impedances and lower currents, the type of noise is referred to as capacitive coupling. Capacitive coupling can occur anytime two conductors with a voltage differential exist with a dielectric between them. The electric fields can negatively interact with each other and exist as parasitic capacitance which will vary with frequency. Capacitance of two parallel wires, assuming distance between them is at least three times greater than the wire diameter, can be reduced to the following equation, which takes into account the dielectrics of the two mediums:

The important parameters to take note of is that the parasitic capacitance is drastically reduced with just a little separation between the two wires. To model this type of unwanted parasitic capacitance the victim circuit can be modeled as having a frequency dependent current source injected into the victim circuit.

Conversely, when dealing with lower resistance or high current applications in the near field, the main concern instead falls to magnetic field or inductive coupling, where the mutual inductance characterized by the magnetic flux generated by the source allows the source and victim to become linked together. While the complete physics behind the variety of inductances is involved, we can simplify the mechanism allowing crosstalk to a form of inductance referred to as mutual inductance, L_m, described by the magnetic flux linking two circuits together. This type of coupling can occur between two neighboring circuits, and results in a frequency dependent voltage source being induced onto the receptor circuit from the source circuit, much like a transformer couples two windings close together. The voltage source created in the neighboring wire is created based on Faraday’s law of induction, as follows:

Using this relationship, the generalized derivation linking together the mutual inductance, flux, magnetic field and noise voltage together can be found below. The key factor here is the mutual inductance the two wires share, much like the mutual capacitance the two wires share in the case of capacitive coupling. The magnitude of L_m is dependent on the magnetic flux (ϕ), between the source and receptor, ϕ₁₂, the source current and the geometric angle between the receiving circuit and how the magnetic field pierces the neighboring loop.

Understanding these equations allows us to understand that the magnitude of noise voltage, and ultimately the mutual inductance shared between two circuits, is based on how fast the current switches, as well as the area and magnetic field generated by the source.

Transforming the time domain equation into the frequency domain further understates that frequency of the noise source, just as capacitive coupling plays a large role in stray fields.

To understand the mechanics behind mutual inductance and the magnetic flux linking two circuits together, see Figure 3, which depicts the magnetic field lines curling around the current in the source circuit (loop) passing through the receptor circuit (loop).

Since those magnetic field lines are encased by the neighboring circuit, we can now refer to them with the term flux density, ϕ, which is simply the result of the integration of the magnetic field through a loop surface area. Using that relationship, we can then apply it to a simple twin wire setup and see that the current in the source pair of wires creates a magnetic field linking the source wire to the neighboring circuit. An electromotive force (voltage) is then produced based on the frequency and loop area (or how tightly coupled the source and victim circuit is).

This frequency dependent voltage source is often referred to as noise voltage being induced in the receptor circuit. In practice, we say that the two wires are inductively coupled together via their shared mutual inductance.

Based on the preceding information, we can summarize key elements that affect our radiated emissions regarding both capacitive and inductive coupling:

• Frequency
• Wire distance, or separation
• Voltage
• Impedance

Conducted Coupling

Conversely, there are often times where a design engineer is forced to use a common return path for their design either to save space, weight or money. This paves the way for conducted coupling, which occurs when the source and victim circuits share common impedance in either the return or supply path. This type of coupling showcases a common misconception, that common paths have zero impedance. In general, people are most familiar with the role that increasing frequency plays in field coupled cross talk. However it’s often a surprise to the designer that at any frequency, including DC, the disadvantage of common impedance coupling results in a voltage drop across the common power or return path. This is due to the impedance that the common coupled paths share, which at lower frequencies is predominantly the DC resistance.

The unexpected impedance will cause a voltage drop, relative to the impedance of the path, to develop across the impedance, resulting in a noise voltage in the circuit. This means that you could be supplying a reference or a supply voltage but, due to the common voltage drop across the supply, the actual voltage may be less than originally thought. Situations where this is especially problematic include the following:

• Analog to digital converters are especially susceptible to this as they rely on a common return (or ground) to reference their reference voltage to;
• Power electronics or any circuitry drawing large amounts of current;
• Automotive/aerospace applications that use large wire bundles which utilize a common return for signals;
• The return path is improperly sized (incorrect wire gauge) for the current/application in which it is being used.

It will be shown during experimentation exactly how big of a problem this is across a wide range of frequencies, when compared to the radiated type of crosstalk.

Designing the Test Setups

To be able to show the different characteristics that affect the radiated and conducted crosstalk of your product or design, test fixtures using common twin lead appliance cord are used to create multiple simple test setups and collect results. The goal of these test fixtures is to separate and keep constant all the noise factors between each test setup except the specific characteristic being tested. For this reason, the test fixtures are broken apart into five fixtures described below, the first four keeping the far end load constant and varying the test setup and frequency, and the last setup by varying the far end loads and keeping the frequency constant:

• Varying wire spacing of unshielded wires, where the wire spacing between source and victim are varied by a small amount;
• Common impedance return path, where the return path is joined between the source and victim;
• Employing coaxial cable to reduce the radiated emissions from the source;
• Employing twisted wire pair to reduce the radiated emissions from the source;
• Changing the far end load and its effect on field coupled crosstalk.

After analyzing the results of each of these five test fixtures, the basics of how to not only identify the wide variety of crosstalk but how to control and test for them can be explored and used. The same basic physical setup is used for each of the five experiments, which consist of about six feet of terminated twin lead wire and a far end load (except for the last fixture).

Wire Spacing of Unshielded Wires

In order to introduce the concept of unshielded wires and how distance can reduce the noise voltage coupled from source to receptor conductor, the wire spacing in the first setup is varied. This experiment had two sets of wires, the first with the wires’ insulation touching, leaving no space between source and victim, and the second leaving a ½” gap between the two. A sinewave is place onto the source wire with a 10Ω load, and the coupled voltage created across the receptor wire is measured. The experiment demonstrates that, in electric or magnetic field coupling, distance plays a role in reducing the coupled energy. In inductive coupling, the mutual inductance coupling factor is being reduced, while in capacitive coupling it’s the mutual capacitance the two circuits share.

The physical setup is demonstrated in Figure 7.

Common Impedance Coupling

As previously stated, a common technique employed by many designers is to share a common return or source for their supplies. The circuit diagram used to demonstrate this situation is shown in Figure 8, by using unshielded wires and a common return path.

To demonstrate the voltage drop across the common return path, a fixed amplitude sine wave is applied onto the source wire, and the output is recorded. The frequencies chosen are similar to the other variable frequency experiments, but also continue lower below the 60Hz range to demonstrate the voltage drop across what’s normally considered a 0 Ω return path.

In reality, the return path does create a finite voltage drop which is based on frequency through that conductor, the wire radius of the conductor and the physical shape of the wire (a cylinder in this case). To make the determination if the skin effect (which explains where the majority of the current is in the wire) has any effect on the resistance of the wire, we first need to determine the skin depth. The governing equation for a cylinder is:

The important parameter in determining the skin depth is the frequency of the current in the conductor. As the skin depth of the current approaches the radius of the wire (which occurs as the frequency approaches DC), then the nominal voltage drop of the wire is characterized by the following relationship:

In the above setup, the wire gauge is designed so that the skin depth below 1KH allows the current to be evenly distributed throughout the wire, allowing it not to affect the return path resistance.

Coaxial Cable Setup

One of two very common but often expensive techniques to reduce crosstalk between wires is employing the use coaxial cabling. Coaxial cable reduces unwanted noise in a neighboring conductors or system by reducing capacitive coupling by having an outer shield/return path that confines the electric field to its interior. When properly terminated with a ground at one end, any noise current coupled to the outside of the conductor is fed to ground, and not allowed to capacitive couple to the neighboring conductor.

The three pictures shown in Figure 9 describe the electric and magnetic fields before shielding, after shielding with a terminated shield, and then with the proper return current on the shield.

Additionally, coaxial cabling reduces amount of magnetic flux being allowed to link the source and receptor circuits through via an inductive coupling path. This is due to the very small loop area between the center conductor and shield, much of the differential mode operating flux linkage will cancel.

As in the previous experiments, measurements are taken at a range of frequencies in order to determine the effect that twisted pairs has on the generated noise voltage. In the results section of this article, we’ll compare the two different types of self-shielding and comment on its effectiveness over a range of frequencies. Figure 10 describes the test setup, which differs from the previous ones in that it will be tested with higher frequencies to demonstrate the effect of the terminated shield with a differential mode current.

Twisted Wire Pair Setup

The other popular way of reducing radiated emissions from a source circuit is employing the use of twisted wire pair. This shielding effect is increased with the number of twists per length, transforming the circuit into what can be considered a series of smaller loops, allowing for the mutual inductance coupling factor between the two circuits to be maximized. This can be thought of as small voltage sources being created, each equal in magnitude but opposite in polarity. The real physical effect being observed here is that while the magnetic field vector is penetrating each loop in the same direction, the normal vector changes with the loop, thus making the induced voltage equal but opposite in polarity.

Later we’ll compare the coupled voltage from previous experiments with coaxial cabling, and characterize the effect that this setup has on the coupled noise voltage. Figure 11 describes the test fixture as designed.

Varying the Far End Load

As previously stated, different types of coupling mechanisms between two twin wire systems is caused by either large amounts of voltage or large amounts of current in a system. Previously, the coupling mechanism has been strictly a function of frequency, and the amount of current or voltage in the source wire was limited by the output of the function generator.

In this test setup, we’ve designed a system where the load is variable which allows us to characterize the coupling mechanism as a function of load instead of frequency. The source circuit, described in Figure 12 allows for shorted end load (through a resistor) or an open circuit. This allows the source circuit to be completely magnetic field or electric field dominant.

The resistance in the victim loop is controlled via a far end potentiometer that allows for a variable load. In an electric field dominant system, the coupled noise voltage will increase with the impedance. The equivalent circuit diagram is shown in Figure 13 for an electric field dominant circuit.

The result of the circuit analysis is easily obtainable, and with two resistors in parallel with the current source, will equate to a noise voltage at the near end that increases with resistance. In a magnetic field dominant system, the opposite is true. As the impedance decreases, the amount of current allowed to flow in the system will increase. The equivalent circuit diagram for the mode of inductive coupling is shown in Figure 14.

Analyzing the Results

The results of this experiment cover the various factors that affect near field crosstalk between wires. To ensure that the results are well understood, they’re separated into three separate sections, with control on specific factors, as follows:

• Comparing the geometric spacing and configuration of two wires two individual wires with separation with a fixed load and
  varying frequency;
• Keeping the far end load impedance, geometric spacing, constant but instead joining wires together to share a return path and varies the frequency;
• Constant geometric spacing, geometry and frequency, while varying the far end load impedance.

This allows the results to show what type of modifications affect the amount of coupled voltage from receptor to receiver.

In the first situation, where the geometry of the wires is examined, Table 1 presents the results. The expected results correlate well, and demonstrate the reduction of the amount of coupled voltage based first on wire separation, then twisted wire pair, and then coaxial cable.

Table 1: Chart tabulating the normalized crosstalk based on frequency and wire type

The chart, shows graphically:

• The scale of the coupled voltage as frequency increases;
• The dramatic reduction in coupled voltage by simply separating two unshielded wires;
• How well coaxial cable is in shielding at higher frequencies when compared to twisted wire pair (TWP).

Of special note is how TWP compares to coaxial cable, as both are very common in practice. In general, the twisted wire pair is unable to suppress coupled noise as effectively as coaxial cable, especially in the MHz and above frequency range. This is demonstrated when comparing the coupled noise at .5 MHz and then again at 2MHz.

Next, instead of examining crosstalk between two wireless coupled wires, the return paths are made common between the two wires. The results contained in Table 2 demonstrate that a common return path is worse than individual returns paths, as the two charts clearly show.

Table 2: Chart tabulating the normalized crosstalk based on frequency and wire type for common impedance coupling

It also demonstrates that, even though we’ve discussed the angular frequency being a strong contributor to crosstalk, there is a finite resistance in the wire that’s based on the physical parameters of the wire itself, and that finite resistance causes a coupled voltage at and below 10kHz. This is an order of about one magnitude greater than when the wires are separated.

Lastly, after demonstrating some common wire geometries and return path configuration, the far end load impedance’s effect on field coupled crosstalk is then examined. The results of this section succeeded in showing two main results:

1. The far end impedance when matched to the victim circuit denotes the largest amount of coupled voltage;
2. Near field magnetic fields couple best when the impedance in the system is low, and near field electric fields couple best when the impedance in the system is high.

These results correlate with the expected results described in the experiment section. The results are obtained by closing a switch as load on the source wire to create the shorted load and opening it up to create the open ended load. Then a voltmeter is used to measure the far end loop resistance, then removed as to not add a DC bias to the results. The frequency and amplitude of excitation signal is kept constant for this, and just the load impedance is changed to examine that particular effect on near field coupled crosstalk.

Table 3: Chart tabulating the normalized crosstalk based on far end impedance

These results clearly show that:

• Wire spacing and various other methods of self-shielding material affect the amount of coupled crosstalk differently;
• The type of field coupled crosstalk scales as frequency increases, but in the case of a shared return there will still be an amount of coupled crosstalk because of the finite resistance in the shared return wire.

The amount of coupled crosstalk correlates well with the type of source and the amount of impedance in the victim circuit.

Together these results demonstrate the basic characteristics of near field coupled crosstalk in a closely coupled system, and the relationship they have with frequency, wire spacing and type, and impedance.

Conclusion

In circuit design, wires and PCB traces are a hidden component which can create unwanted crosstalk between circuits. Crosstalk can be conducted through shared supply or return paths, or radiated via capacitive or inductive coupling. While the effect of crosstalk is increased with higher frequencies, conducted crosstalk can occur with DC voltage since the wire resistance of the shared path, along with the current in one circuit, can create a voltage offset in the other circuit. Radiated crosstalk, on the other hand, can be greatly reduced when lowering frequency, but that’s typically not an option. Instead, when laying out circuit paths, some key methods for mitigating inductive and capacitive coupling are:

• For each circuit, keep supply and return path close in order to maximize mutual inductance, to reduce the chances of inductive coupling with another circuit.
• Choose the proper method of shielding. Twisted pair can reduce magnetic coupling, whereas coaxial cable (if terminated properly) is an effective means of reducing capacitive coupling.

And lastly, it’s important to remember that, because of the reciprocity antennas share between sending and receiving, any fixes or characterizations that work at reducing radiated emissions from the source work to shield the receiver as well.

Lisa Linna works at Bosch. She has her Bachelors in Electrical and Computer Engineering and her Masters in Electrical Engineering from the University of Michigan Dearborn. She can be reached at lisa.linna@gmail.com.

Christopher Semanson works at Ford Motor Company in Dearborn, MI working on driver assistance software and applications as well as powertrain controls. He also assists Mark Steffka in teaching at the University Of Michigan Dearborn and often University of Detroit Mercy on topics regarding all facets of EMC. He has his Bachelors in Electrical and Computer Engineering and his Masters in Electrical Engineering from the University of Michigan Dearborn. Chris can be reached at csemanso@gmail.com.