Shielded Cables Tame Servo System EMI
Electromagnetic interference (EMI) in motion control system cabling can be a major source of frustration for servo motor system designers and users. Left unsuppressed, it interferes with system control, creates data errors, and can even turn devices on and off. EMI frequencies can range from dc and the lowest audio frequencies through the entire radio-frequency band.
Electromagnetic interference (EMI) in motion control system cabling can be a major source of frustration for servo motor system designers and users. Left unsuppressed, it interferes with system control, creates data errors, and can even turn devices on and off. EMI frequencies can range from dc and the lowest audio frequencies through the entire radio-frequency band. They can be either conducted through the cable or radiated from the cable. Conducted emissions normally occupy the frequency band below 30 MHz, while radiated emissions are above 30 MHz.
Electronic cable shields protect signals in the cable conductors from external sources of EMI and also reduce the level of interference radiated by the cable, which could affect other conductors, equipment, and wiring surrounding the cable.
Types of EMI
To understand proper shielding, one must first understand the coupling mechanisms of EMI, since a shield for one type of coupling may be totally ineffective for another. Moreover, improperly referencing (grounding) the shield can be worse than no shielding at all. Four conductive mechanisms for noise can degrade a signal:
Inductive (magnetic) coupling;
Direct coupling; and
Radiated (RF) coupling.
Capacitive-coupled noise tends to be the least problematic and is the easiest to suppress; however it can cause data errors with high frequency, high impedance transmissions.
Characteristics: This high-frequency voltage interference is not current related. It can be spotted on an oscilloscope with a characteristic zero-voltage reference. Mathematically the formula can differentiate it from the radiated coupling mechanism.
Solutions: Use a foil outer conductor surrounding the cable referenced to ground. Grounding is particularly important as the capacitance of the shield to the signal wires provides a coupling mechanism into the system.
Inductive (magnetic) coupled noise is the result of a strong magnetic field acting as a power mechanism. This can induce a signal across relatively low impedance and disrupt transmissions. This form of EMI and the disruptions implied on the system can have sufficient power to turn devices on or off.
Characteristics: Inductive (magnetic) coupled noise consists of a zero voltage average (will not have a dc offset) and can range from lowest frequencies to the upper range of measurement (>500 MHz). The same formula applies for inductive coupled noise as for capacitive coupled noise.
Solutions: Twisted pair conductors surrounded by a braided shield referenced to ground used with differential driver/receiver combinations are generally effective. This type of shielding reduces noise from sources and receivers by suppressing both emission and reception of inductive EMI due to its close proximity to the conductors and electrical reference to ground. Inducted noise will follow the path of least inductance, so the shielding will capture it before it can affect other cables. Foil shielding is not nearly as effective due to magnetic eddy currents.
Direct-coupled noise is where noise currents have a direct connection to the system, such as where a power supply is creating noise spikes on an ac line.
Characteristics: Direct-coupled noise can have a non-zero voltage average. A dc offset is a certain indicator of a direct coupling mechanism. It can be extremely low frequency (e.g., 60 Hz noise) and will not be dictated by any impedance formula other than the transfer of power limitations.
Solutions: Direct-coupled noise must be eliminated with isolation, filtering, or other impedance matching. Shielding is ineffective for mitigation, but can at least help to contain the transmission within the system. Large direct-coupled noise spikes can be magnetically coupled in an unshielded system.
Radiated coupling is the most complex mechanism and has some peculiar limitations related to the frequencies involved that must be understood.
Characteristics: Typically, the device in question must be 1/2ë (wavelength) from the transmission source and have at least a ë/20 antenna length. As a result, RF interference has a source outside a specific device. When hunting for the EMI source, one will probably need to look farther away than might be initially expected. Even at 5 GHz clock frequency, adevice would have to be 8 m from the interfering signal source for the RF to be received and the antenna approximately 400 mm (15.75 in.) long.
Solutions: Foil shielding is inadequate; however, braided shielding can be effective, although its application is the most critical when dealing with radiated EMI. First, an RF shield must never be terminated inside the circuitry being protected. Complete 360° shield coverage is a must. Given the extremely high frequencies that can be involved, small holes or paths that might normally be overlooked can create significant impedance issues. Something as innocuous as a slot in a case that allows a cable to enter and exit may be a source of RF input. When dealing with the transmission effects, the efficiency of transmission is guaranteed if Anyone familiar with stereo tuners knows the proper shielding for RF and can point out the waveguides and feedthrough capacitors that are foreign to others.
Selecting cable type
The right type of shielded wires for a specific application and frequency range is a major step to protect low-level signal integrity in control circuits and ensure accurate and reliable servo system positioning. Additionally, shielded cables used for high-power drive circuits ensure that a motion system does not interfere with surrounding equipment. For example, a properly terminated shield can prevent ground noise current (sometimes called common mode current) produced by the servo drive from reaching the ground grid. Braided, spiral, and foil shields provide a low-impedance path for conducted low- and high-frequency noise current to return back to the drive.
The primary purpose of shielding is to attenuate radiated emissions. Part of the noise energy reaching a cable shield is reflected and part is redirected through a low-impedance shield, but some residual noise energy penetrates the shield and corrupts lower-level signals in adjacent cable conductors. The goal is to select the most effective shield to minimize noise penetration.
Servo motor cables (feedback and power) are affected by emission as well as susceptibility to ambient EMI. In addition to interference generated by equipment and wiring in the vicinity of a servo motion system, motor drives are the main sources of electrical noise. To minimize this, servo cables connecting feedback, analog, digital, and other low-voltage circuits benefit from a shield's ability to reduce incoming as well as outgoing EMI. Often, twisted pairs are placed inside a shield to reduce crosstalk between the pairs. Twisting the signal and return wires reduces the radiation. Adding a shield over the pair adds another layer of protection and cuts down crosstalk between conductors and wire pairs within the cable. The outer shield protects all conductors from external EMI and reduces interference radiated by the cable.
Servo power cables produce high levels of EMI radiation because of the system's ability to switch the drive current on and off at very fast rates. However, these rapid cur-rent changes create significant high-frequency interference, which is radiated by the power cable both capacitively and inductively. Shielding the power cables reduces the level of radiation and protects feedback signals and equipment in the system. A good practice guideline for servo connections is available at the Danaher Motion website. ( ftp://dagobah.kollmorgen.com/dist/TechInfo/AppNotes/noisechecklist.pdf )
Many factors enter into the selection process for the proper shield construction, including shield materials. Among them are:
Some cables have an overall shield that encloses all conductors.
Other cables have individual conductors or individual pairs covered with shielding.
Cables intended for harsh environments have individual shielding and overall shielding combined.
Double shields separated by a layer of insulation increase noise protection, but reduce cable flexibility. For example, the first shield layer made of aluminum foil provides 100% coverage and protection from high-frequency interference. The second, braided copper shield (over an insulation layer between the shields), improves low-frequency protection and adds considerable strength and flexing life.
Three types of shields
Cables using braid, spiral, and foil shields are sufficient for the full range of voltages common to servo systems.
A braid shield is normally made of copper strands braided into a mesh covering individual conductors, twisted pairs, or all conductors in a cable. Coverage is determined by the tightness of strands in the braid and is typically 60-95% for braided shields. Higher percentage of coverage means better EMI protection and lower radiation.
Size of braid strands, typically between 32 and 40 AWG, directly affects the shield's flexibility, flexing life, and coverage. Tinned braid strands have greater resistance to corrosion, better electrical contact, and better solderability, but are less flexible than bare strands of equal size. Shields of continuous-flex servo motor cables normally have extremely fine bare copper strands. Signal cables should include twisted pairs enclosed in a foil shield with an overall braid shield.
Spiral shields offer even higher flexibility and flexing life than braided shields. They are made of bare or tinned copper wire wound in a helical pattern around the conductors. Spiral shields perform best at lower frequencies and often have coverage over 95%. They are used for most high-flex or continuous-flex applications, such as torsional flex where braided, or especially foil shields, can be damaged by the twisting motion of the cable. Grounding a spiral shield is more difficult when the 360° contact must be maintained.
Foil shields are typically made of aluminum foil bonded to a polyester backing. The backing is needed to support the foil mechanically. Aluminum offers effective high-frequency EMI protection of capacitively coupled noise. Copper foil is used less frequently and covers a lower frequency range. Foil shield can be applied around conductors in one of three ways: foil facing the conductors, foil facing out, or shield edges folded in a shortening, Z-shaped fold.
The first two arrangements allow some noise leakage because no metal-to-metal contact exists where the foil/backing edges overlap. The Z-shaped fold, however, creates 360° (100%) coverage of the conductor since the foil sides face each other without interruption. A drain wire alongside the foil provides a means to ground the foil shield reliably.
UL, CSA, CE requirements
In addition to reducing or eliminating interference with the operation of the servo system and the surrounding equipment, shielding cables may be required to comply with certain regulatory standards such as European Conformity (CE).
Size of conductor, type of insulation, and listing mark (if any) is normally printed on the cable's jacket, along with voltage and temperature ratings. Underwriters Laboratories (UL) and Canadian Standard Association (CSA) marks indicate that the cable was evaluated by one or both organizations and found to be safe when used according to the manufacturer's specifications. However, presence of such listing does not indicate a cable with superior shielding properties. CE imposes limits on the amount of noise conducted to the line, but just using a CE-marked cable is no guarantee that the whole system is CE compliant. Compliance also depends on how the cable is used, so there is no substitute for carefully studying cable specifications and testing in a real-life installation.
Understanding flexing requirements
Cables designed for flexible applications typically have:
Fine copper-strand conductors;
Non-slip compound over each layer of conductors;
Non-wicking wrapping tape;
An inner jacket between the conductor bunch and shield;
Extremely fine copper strand shields;
Foil with braid over feedback conductor shields; and
A flexible outer jacket.
Shields of flex-rated cables have extremely fine bare copper strands that easily conform to the shape of the conductor bunch. When a cable is flexed, the shield should be able to slide back and forth along the conductor bunch with low friction and without catching on peaks and valleys formed by individual conductors and spaces between them. Lack of smooth, round surfaces underneath the shield can permanently deform the cable in the shape of a corkscrew. A thin inner jacket is placed between the shield and conductor bunch during fabrication to fill in gaps between the conductors and to create a smooth, round shape for the shield to slide over.
One common practice is to add fillers and fleece tape to form a round, smooth surface beneath the shield. Another cable fabrication process forms an extruded inner jacket that, due to its construction, forms an almost perfect round shape, which cannot unwind during cable flexing. This construction is more costly than the filler and taping design but offers greater reliability.
The shield is braided or wound over the inner jacket and then the shield is covered with the outer jacket.
Movement of cable components relative to each other during flexing produces triboelectric noise resulting in static and piezoelectric interference. Carefully designed flex cables minimize this noise effect, but the phenomenon still deserves careful consideration.
Different kinds of bending and flexing affect the choice of cable components, including shields. Continuous flex, bending flex, and torsional flex represent typical cable motions. Cable specifications indicate the type of motion the cable was designed to withstand without damage.
Continuous-flex refers to a linear (one plane) back and forth flexing or rolling of a cable placed in a C track. Bending flex involves one end of the cable fixed in place and the other end being bent back and forth. Cables rated for one of these linear-flex applications should not be twisted.
For example, torsional flex or twisting is present in robotic equipment where the robot arm containing cables twists clockwise and counter clockwise. Spiral shields are best suited for such applications.
Shield property comparison
Up to 100%
Protection frequency range
Low to high frequencies
Ease of termination
Lee Stephens is a systems engineer at Danaher Motion;
National Electrical Code (NEC)
The National Fire Protection Association, as stated in its National Electrical Code (NEC) publication, separates all cables into two groups. The Building-Wiring Group comprises cables used in fixed installations where cables are not exposed to any movement and often are routed in conduits, behind walls, or in other areas where visual inspection is not possible.
The second group is called Flexible Cords and Cables. These cables connect electrical components that can move relative to one another. Applications include industrial machinery, material handling equipment, tools, and other equipment having moving parts and containing electrical cables.
All integral cable components, including conductors, insulation, shields, and jacket, must be carefully designed to ensure that they meet stated service life under the worst flexing conditions. However, flex-rated cables are usually not recommended for fixed installations since they are not designed and tested for conduit and other contained areas. The cables are normally located in plain view where a damaged cable can be easily seen and replaced.
Guidelines for cabling methods to reduce EMI
Here are several common sense techniques that will improve the servo system's EMI noise immunity.
Length of cables, especially feedback cables, should be selected (or fabricated) to allow for some slack to avoid small bending radii, but not so excessive as to increase EMI noise levels, which corrupts the signals. Motor/drive cables radiate more interference when their length is greater than required; keep cables as short as possible.
Separate power and feedback cables reduce cross talk between high-current conductors connecting the motor to the drive, and conductors carrying low-voltage feedback signals and other analog and digital signals. Power and signal cables should be run in separate conduits where possible, or kept at least 4 in. apart for drive currents up to 20 A; 6 in. apart for currents up to 40 A; and 8 in. apart for drive currents up to 80 A. Power and signal cables should be dressed perpendicular to each other if they need to cross.
Composite (power/feedback) cables offer space savings and easier routing, but greatly increase the possibility of feedback conductors subjected to EMI noise produced by the power conductors. Quality composite cables contain signal wires grouped in twisted pairs surrounded with double shields.
Many servo motor manufacturers offer prefabricated motor cables for use with their systems. These cables save time and usually yield much better results than cables fabricated in house.
Common-mode chokes for motors are usually recommended for cables longer than 80 ft.
When using ac power-line filters, input and output leads must be kept separated.
Differential inputs offer much greater noise immunity compared to single-ended inputs in analog circuits. Signal lines should be made using shielded cables with shields connected to ground on both sides. Power cable shields should also be grounded on the drive and the motor sides to prevent noise produced by the motor windings from entering the circuits.
Shields on spliced cables should also maintain 360° electrical contact. Making shield 'pigtails' for termination exposes part of the conductor to the EMI noise and should be avoided. Cables should not be divided across a terminal strip. Conductive braids should be used to connect all metal parts of cabinets. To provide good electrical contact, paint should be removed from the area on the electrical panel where the drive will be mounted.