A Practical Comparison of Cabling Effects on Radiated

Emissions Kursat Eroglu

Sr. EMC Engineer Intermec Technologies Corporation

ABSTRACT SOLVING CABLING PROBLEMS

External cables are one of the major sources of radiated emissions. Theoretical and practical effects ‘of cable shielding, use of ferrites have been analyzed in previous studies under test conditions such as noise source, cable, termination setup’s etc. We have looked at cabling effects from a practical point of view. We measured radiated emissions of a typical IT device with Ethernet and RS232 connections under a variety of conditions. We varied cable shielding, shield grounding, used ferrites at different locations and compared the results. The results show that shield type and grounding have a more significant impact on radiated emissions than use of ferrites, with the exception of internal components (filtered connectors).

There are three categories of remedies to unintentional radiated emissions by cables:

1. Reduction of noise at the source This is the ideal remedy since without noise energy, there is no need to find ways of directing the energy to ground or somewhere else. There are countless ways of’ reducing the noise at the source. These range from PCB layering techniques to trace layouts, positioning of critical components, use of dithered clock etc. Most designers implement some of these techniques in their device. However, the world being non-ideal again, the device ends up generating some noise anyway. This brings us to the second type of remedies:

INTRODUCTION 2. Shielding Techniques

From an EMC engineer’s point of view cables are antennas attached to a noise source. Unintentional emissions are generated by internal noise sources such as oscillators and digital circuitry. The noise first radiates directly from the noise source components, through coupling from other components or from traces on the circuit board. This first phase of unintentional emissions can be high enough to fail regulatory requirements in non- conductive enclosures. Many designers opt metal or metallized (conductively coated) enclosures to restrict the board emissions to the device. However, in the non- ideal world of electronics, most devices have to have some form of interconnection with other devices. The most common way of interconnection today is cabling. This brings in the second phase of unintentional emissions, the amplified one. The first phase of the board radiated noise couples through connectors and other components to cables, which in most cases act as amplifying antennas, and radiate the noise out into the world.

An easy way of containing the noise to the signal carrying wires is isolating them in a shield. Ideally a shield around the wires should contain all of the noise and no emissions should be detected. Yet cables are one of the most effective radiators. The explanation is in the non-perfect shielding. Noise carried on the signal wires (or ground wires too) generates a return current on the shield. Discontinuities at the connection points cause disturbance in the flow of the return current, which starts radiating out of the cable. So, for a cable shield to be effective, it must be connected as perfectly as possible to the chassis ground at both ends. In addition good ground connections must be provided at both ends so the common mode currents can find a way to earth ground. Cable layout (flat, twisted pair, coax etc.) is a critical component as well. Each type provides a different degree of flow to noise return current, affecting radiated emissions. Henry Ott and others have shown through experiments that there can be 20 dB or more of radiated noise difference between different cable layout and shielding methods. Shielding material is also important. For cost savings

734

and mechanical reasons a variety of shielding material are used (foil, braided etc.). Each shield material has its own advantages and disadvantages. The emissions characteristic (shield effectivity) of the material should be considered carefully when choosing a cable.

3. Ferrites Shielding techniques work by providing a path to the energy other than radiation, but they do not transform the energy into another form. Ferrites are the solution that transforms the energy. They convert the electrical energy into heat and help reduce the radiation. In theory ferrites are the perfect solution for cable related emissions; however, they have their shortcomings as well. Their effectivity is highly dependent on frequency. Therefore they can only be used for problems in certain frequency bands, not for wide spectrum emissions. Their effectivity also depends on their location and the nature of the emission problem too. In fact our practical experiment showed that use of toroidal ferrites had a smaller effect than different shielding techniques. But, as our results also showed, use of series ferrites on the PCB to filter noise before it has a chance to couple onto the cable had a major impact.

A PRACTICAL EXPERIMENT

Many papers and books have been published demonstrating results of cable emission results. Most of those are under controlled conditions with a known noise source and termination. These are conditions close to theory. We wanted to approach the effectivity comparison from a practical point of view with real life noise sources and terminations.

Test Setup Our test setup consisted of a typical ITE device. The EUT was a network bridge providing connection between Ethernet and wireless networks. It had two Ethernet connections (10 base T and 10 base 2) and an RS232 connection for programming purposes. The 10 base 2 connection used a standard coax cable, the 10 base T a 4 conductor twisted pair cable, and finally the RS 232 connection a 9 conductor shielded cable. The cables were connected to the EUT individually and all together, and terminated with end devices (a hub and a desktop PC). The EUT had a metal (aluminum) enclosure and no openings (vents or other) besides the cable connectors and power inlet.

All cables were at 6 foot standard length. The tests were performed at a registered OATS under ANSI C63.4 compliant conditions in the 30 MHz - 1 GHz range. The EUT had unintentional emissions in more than one region. However two of those regions appeared to be most affected by the changes to the cabling. So, in the comparisons between different cables and modifications the amplitude of those regions was considered.

RESULTS

First plots show the unintentional emissions of the EUT under different conditions. As can be seen in the plots. The noise in the 186 MHz - 232 MHz range and 400 MHz - 480 MHz range was mostly affected by the variations in cabling. Therefore the readings in those regions were taken for comparison of various changes. For simplicity ambient noise was subtracted from the plots. Below is a list of the comparisons made and results:

The first comparison was made on the 10 base 2 Ethernet cable. The connection with standard coax cable was improved by providing better shield to ground connection at the termination end and then at both ends. The first improvement provided - 2 dB improvement in the first region of concern, while improving both ends brought down the emissions by - ‘7 dB. The next set of comparisons were made on the 10 base T cable. It was determined that the ground connection at the termination end was a floating one. This was changed to a chassis ground connection and a 4 dB improvement in both regions was observed. Also the effect of providing earth ground to the PCB of the hub (which was not the case in the standard configuration) was tested for both Ethernet cables. The RS232 cable was subjected to five tests. First an unshielded cable was used. Then a shielded cable with braided shield, and one with foil shield was used. Both types were tested first as is, then with improved chassis ground connections at both ends. The unshielded cable caused approximately 4 dB higher emissions in the first frequency range than the shielded versions and - 6 dB in the second range. A -2-3 dB difference was observed between the shield types with the foil shield showing better characteristics. The emissions

735

. Reduction gained by the grounding improvement at both ends was not significant for either shield type. However, this could be due to the already good quality of the existing ground connection.

. The impact of ferrites on cabling was tested with snap-on type ferrites added to the cables at different locations. The largest effect of this approach was observed with the unshielded RS232 cable at 10 cm distance from the EUT. While 30 cm distance appeared to have less effect than 10 cm, the changes were relatively small (l-3 dB). Ferrite material used had 40 dB loss at 100 MHz.

e ,m _ m a - m v. m K‘ .m I~U~IYW

Figure la Radiated emissions of the EUT in standalone mode

0 c2 m ,a .m SD SK -01 - m I_ PmuYmlYH4

Figure lc Radiated emissions of the EUT in 10 base 2 mode

. While the performance of snap-on ferrites was disappointing, series ferrites on the board right before the connectors had a significant effect. In series, surface mount ferrites provided a 6 dB drop in emissions for the RS232 connection. With the Ethernet cable the drop was less (-4 dB), but the lines were. these ferrites could be applied to was also limited. The biggest drop in the 10 base T connection was observed with the use of a f&ted R.J-45 type connector (- 9 dB).

Below are. the initial emissions plots and comparison graphs demonsaating the differences between different approaches:

r Ioi .m xx +x we c‘.2 1x - - ,w

lm-=vwI(I,

Figure lb Radiated emissions of the EUT in 10 base T mode

0 Irn m w .a Ic‘ - 01 111 m ‘cma ~“v-=Y,““~,

Figure Id Radiated emissions of the EUT in RS232 mode

736

Figure le F-cl ww

Radiated emissions of the EUT with all cables connected

COMPARISONS

Comparison of radiated emissions of Different cabling coniigurations

figure 2b Comparison of the effects of different grounding schemes on 10 base 2 mode.

737

Figure 24 Comparison of the effects of different Grounding schemes on 10 base T mode

Comparison of the effects of different shielding types on RS232 mode.

Figure 2Q Comparison of the effects of Snap-on ftites on the RS232 cable

CONCLUSION

Various cable treatment methods were hied on a typical ITE device. Between different shielding techniques, snap-on ferrites and internal fenites, the internal series ferrites provided best reduction in radiated noise emissions. Snap-on fenited did not have any significant improvement for Ethernet cables, while they provided about the same improvement as the shielding techniques for the RS232 cable. Foil shield

Figure 2d Comparison of the effects of internal series ferrites on different cables (loss)

demonstrated a little better performance than braided shield. Providing earth ground to the termination of the Ethernet connection also proved to be effective. These results contirm mostly prior theoretical or experimental data in a real life system. They indicate for different cabling system a variety of solutions are. available with varying performance. A broad analysis comparing EhfI performance with cost and other factors such as mechanical performance should be done before final selection is made.

738