Crosstalk effects of shielded twisted pairs

This article deals with the modeling and simulation of shielded twisted pairs with CST CABLE STUDIO™. The quality of braided shields is investigated with respect to perfect solid shields. Crosstalk effects are calculated for unshielded twisted pairs, poorly shielded twisted pairs, and twisted pairs with high-quality shields. Explanations are given how to create realistic simulation setups in order to be able to compare them with measurement results.

Cable shielding

Shielded cables are widely used in industrial applications in order to suppress unwanted crosstalk effects between neighbouring wires. A typical shielded cable is the coaxial cable with an inner wire and a concentric outer screen.

Ideal shielding conditions can be achieved by realizing the screen as a solid conductor with a specific thickness. With increasing frequency the penetration depth of the electric field decreases until it is less than one half of the shield’s thickness. At higher frequencies the current tends to flow mainly on the conductor’s surfaces (skin effect), thus completely decoupling the inner part of the screen from the outer part. For a solid screen we can state: the higher the frequency the better the screening.

Due to cost and engineering reasons the use of solid shields is rather limited. More popular are braided shields since they are easier to manufacture, are lighter in weight, and more flexible than solid shields. There are, however, different quality levels. Braided shields have tiny apertures that are permeable to the electric field. As a consequence, the shielding effectiveness decreases with increasing frequency. Hence the goal is to find the optimum between cost saving and shielding effectiveness.

A measure for the shielding effectiveness of shielded cables is the transfer impedance. It describes the frequency dependent transmission of electromagnetic signals from one side of the shield to the other side of the shield. The denser the braid is or the more braid sheets are used, the better the shielding.

 

Figure 1: Transfer impedance of a tubular braid with different number of sheets

Figure 1 illustrates the transfer impedance of various tubular braids - a single braid (top curve), a double braid (middle curve) and a triple braid (bottom curve). As a comparison, the transfer impedance of a solid shield would strongly monotonically decrease as the frequency increases.

The shielding effectiveness of the triple braid shows the best quality since its transfer impedance in the frequency range up to about 50 kHz is quite similar in  behaviour to that of a solid shield. At 50 kHz the transfer impedance reaches a local minimum and rises again with increasing frequency. This means that 50 kHz is the threshold frequency above which the shielding diverges from perfect behavior and falls off in quality at higher frequencies. The double and single tubular braids are even less perfect: they have got higher transfer impedances even at low frequencies, and the local minimum is already reached earlier or does not show up at all.

The question now is: how does this affect the crosstalk behavior? In order to answer this question it is reasonable to perform three simulations of the same cable configuration but different shielding conditions.

 

Cable configuration and simulation tasks

The cable structure of interest can be seen in figure 2. It consists of four shielded twisted pairs in a common outer screen. Twisted pairs are often used since they are easy to manufacture, are inexpensive, and exhibit reasonable crosstalk characteristics. Since twisted pairs are quite susceptible to asymmetric load conditions and might cause certain crosstalk effects they are shielded at times.

In the following four S-parameter simulations are carried out. The first simulation demonstrates the crosstalk effect in case of perfect load conditions. The second simulation deals with the non-shielded twisted pairs and, in the subsequent two simulations, different shields are investigated.

 

Figure 2: Cross-sectional view of shielded twisted pair cable

Perfect load conditions

The aim of this investigation is to demonstrate what happens if one simulates a perfect cable configuration with homogeneously twisted pairs and fully symmetric loads. Figure 3 depicts the simulation configuration.

 

Figure 3: Simulation setup for a perfectly symmetric cable configuration

The screens of the twisted pairs are grounded to zero potential at both sides.  Ports have been defined at the input and output of the four twisted pairs. Figure 4 shows the simulation results.

Figure 4: Simulation result of perfectly symmetric cable configuration. Besides the reflection and transmission to the line’s end there is no crosstalk effect visible. The numeric value for crosstalk is less than -200 dB

As expected, there is a certain reflection at the input and quite good transmission to the end of the line, but there is no perceptible crosstalk into neighboring lines visible. Actually, the numerical value is less than -200 dB and is only limited to this level in order to obtain a reasonable result view.

It is important to note that this result is not caused by an insufficiency of the simulation method but because no crosstalk exists: the twisted pairs are arranged symmetrically, are twisted with identical number of twists per meter, and are loaded in exactly the same way.

Real measurements, on the other hand, always show perturbations and crosstalk effects. When comparing their results with ideal simulation results, there is often an uncertainty because of the discrepancy. It has to be understood, however, that a measurement setup is never as perfect as a simulation setup. It is necessray to reproduce all potential perturbations of the measurement in the simulation order to find the same results. Since such perturbations are often not known this is rather difficult.

In order to overcome this problem tendencies need to be investigated rather than trying to reproduce reality. In the given context it is reasonable to check the crosstalk effect of an unbalanced (asymmetric) termination and to investigate how different shielding conditions contribute to its improvement.

 

Unshielded twisted pairs

It is interesting to understand what might be the crosstalk effect in an unshielded twisted pair cable configuration with asymmetric termination. Figure 5 shows the cable’s cross-section, figure 6 the schematic, and figure 7 the simulation results. The termination is realized by using a 50 Ohm resistor. In order to reproduce production tolerance of about 10% the resistance values have been increased to 55 Ohm and decreased to 45 Ohm, respectively.

 

Figure 5: Twisted pair configuration without shields

Figure 6: Simulation setup with asymmetric load conditions

Figure 7: Crosstalk effect in case of unshielded twisted pairs with asymmetric load conditions. The transmission factor into the neighboring lines lies between -60 dB and about -15 dB in frequency range up to 1 GHz

Shielded twisted pairs – low shielding effectiveness

When returning to the case of shielded twisted pairs the different qualities of shielding can be investigated. A direct measure for the shielding effectiveness and therefore for the crosstalk is the transfer impedance as discussed at the beginning of this article. The higher the transfer impedance the greater the crosstalk effect since surface currents will induce higher voltages inside the screen. As a consequence, the shielding effectiveness worsens.

Figure 8 shows the simulation result of the cable configuration with specific transfer impedance. The chosen values for the simulation are Rt = 0.09 Ohm, Lt = 1e-9 H, and Ct = 1e-14 F.

 

Figure 8: Simulation result in case of low shielding effectiveness (high transfer impedance). The transmission factor into the neighboring lines lies between -200 dB and about -90 dB in frequency range up to 1 GHz

Shielded twisted pairs – high shielding effectiveness

When changing the transfer impedance to a lower value less voltage can be induced, thus yielding higher shielding effectiveness. Figure 9 shows the simulation result of the cable configuration with Rt = 0.001 Ohm, Lt = 1e-10 H, and Ct = 1e-15 F. As expected, the crosstalk effect has been drastically reduced.

Figure 9: Simulation result in case of high shielding effectiveness (low transfer impedance). The transmission factor into the neighboring lines now lies between less than -200 dB and about -120 dB in frequency range up to 1 GHz – a noticeable improvement

Summary

Investigations of (multiple) shielded cables are possible with CST CABLE STUDIO™. A measure of the shielding effectiveness is the transfer impedance value that can be determined by means of the three parameter transfer resistance (Rt), transfer inductance (Lt), and transfer capacitance (Ct). Its should be noted, however, that a direct comparison between simulation and measurement is not possible unless one considers exactly the same conditions in both simulation and measurement setup. Since parasitic effects such as production tolerances are generally not entirely known, it is recommended to investigate tendencies during simulation rather than trying to reproduce measurement results. This article describes how such investigations can be done, and it illustrates how simulation helps to understand basic correlations of various parameters in order to find the optimum solution.