This is the 5th article . of our ”EMC awareness” series. At this point, before addressing the coupling path, occuring from (or to) the power mains, it was in order to review a solution that is widely involved in controlling Conducted, Radiated and Crosstalk EMI situations: the use of shielded conductors. The subject is not that simple and requires some insight. This article will explain as clearly as possible for the non-specialist how a cable shield works, how much EMI reduction can be expected, and why the choice of certain cables or installation practices will result in mediocre results.
The former articles, after a broad overview of the EMC subject, covered the principal Military & Civilian Norms and Test methods, insisting on the legal inforcement of these verifications in European countries, who turned them into mandatory laws.
We also introduced the source/coupling-path/ victim concept, essential to a comprehensive approach of EMC. Most of the time, it is the coupling path between the culprit source and the victim equipment that is the crux of the problem, hence of its solutions, so the 5 essential coupling mechanisms wer listed, by which EM Interference take place. Although any equipment can be alternately the victim, or the source, of an EMI problem, we focused on EM susceptibility as being the manifestation that appears first in the designer’s or field engineer’s worries.
Nevertheless, emissions problems sooner or later may show-up, but since coupling mechanisms are reciprocal, the author has taken the choice of always reviewing susceptibility situations first, because once understood, the comprehension of emission mechanisms would follow easily. The 3rd article treated the frequent mechanism of Common Impedance Coupling, and the 4th article described two coupling paths where the interference occurs through Radio propagation or near-induction coupling (Crosstalk).
At this point, before addressing the coupling path, occuring from (or to) the power mains, it was in order to review a solution that is widely involved in controlling Conducted, Radiated and Crosstalk EMI situations: the use of shielded conductors. The subject is not that simple and requires some insight. This article will explain as clearly as possible for the non-specialist how a cable shield works, how much EMI reduction can be expected, and why the choice of certain cables or installation practices will result in mediocre results.
1. Basic Role of a Shield over a Cable Link
As soon as an equipment is fitted with external cables whose length exceeds the largest box dimension, it is highly probable that these cables will be the largest contributors to radiated susceptibility and emissions, at least up to several hundred MHz. To some extent, cables are also involved in the Common Impedance Coupling, conducted path. Although shielding a cable may appear as the obvious, solid barrier to radiated coupling to or from the wiring inside, application may not be so easy. Throwing-in shielded cables at the last minute may give disappointing or even disastrous results.
The author has even seen odd cases where shielded cables increased the radiated EMI levels at some frequencies. There are explanations to this, of course, as will be seen. The basic principle for a shield to work against all types of EMI, with the widest coverage of situations (E field and H field, Low and High Frequency, Diff. Mode and Com.Mode, etc.), is to create a continuous barrier that encloses the conductors and is 360° bonded to the conductive boxes at both ends. No matter which theory is applied to model this shield (reflection loss, absorption loss, Faraday cage effect, mutual inductance etc. ad infinitum), calculations and experiments show that when an entire system is enclosed in a continuous barrier, its sensitivity to EMI is reduced.
This is true, regardless this barrier is earthed or not (Fig.1). If the boxes are not full metallic envelopes, the principle still can work, provided there is at least one large metal face or ground plane to connect the shield on both ends, closing the cable-to-shield return path for CM currents. Otherwise, as in the case of solid plastic boxes, a cable shield without a reference plate for grounding its ends will not be efficient against radiated susceptibility or emissions. For such case, I/O port decoupling and ferrite loading would be more appropriate if no more than 20 to 30dB reduction is needed.
2. Principal types of cable shields
Cables shield are seldom solid tubes, welded at each end to the system boxes or sub assemblies (exception accounted for semi-rigid coax and some very specific military systems). Yet, the concept of a conductive sheath surrounding the wires can still be achieved by other constructions. Many technologies are available for cable shields:
- Tinned copper braid (single or double layer)
- Aluminium foil or aluminized mylar folded over the cable like cigarette paper
- Aluminium foil + copper braid
- Thin metallic tissue (silver, stainless steel, copper)
- Stretched metal foil
- Tinned copper or tinned steel spiral wrap
- High permeability wrap, associated with one or several layers of copper braid
- Corrugated, bellow-style cable shield
- Semi-rigid copper shield (essentially used for some RF coaxial links)
What are their respective merits? How much attenuation can we expect, and in what applications? How much is enough? What is the impact of the shield termination hardware at the equipment barrier, and can it be predicted? These will now be explained.
2.1 The Two basic Families of shielded cables
Although any conductor(s) slipped in a metallic sheat can be labelled as ”shielded”, there are two basic types of shielded cables: coaxial cables and shielded pairs or multipairs. Both types reduce the interference received, or generated by the active conductors (HF ground loop coupling, Xtalk, field induction), but they present a fundamental difference.
2.2 Coaxial Cables
In a coaxial cable, the shield is altogether:
- the return path for the intended signal
- an alternate, preferred path for the undesired noise current, whatever it is received (Susceptibility) or generated (Emission) by the system.
This brings a specific constraint for a coaxial link, that is expressed in the following rule:
RS-1: With a coaxial cable, the shield must be connected to the signal reference at both ends, for functional reasons, and to the equipment chassis, for EMC reasons.
Thus, the normal termination of a coaxial cable is forcing the 0v to be grounded to the equipment chassis through the I/O port. Although this is generally the recommended configuration against high frequency EMI (Multi-Point Ground, see our Article #3), there are cases where the designer has opted for a Single Point (or Star) Ground arrangement. In such cases, an additional rule has to be followed.
RS-2: If the design of a signal interface requires an isolated 0v, a coaxial link was not the best choice, unless a galvanic isolation device (signal transformer or opto-isolator) is used.
Some equipment designers, trying to stick to the Single Point Gnd rule (see Article.#3, SGP) keep the coaxial shield grounded to the 0v, but floated from the chassis in order to prevent a ground loop. Although this opens the loop at low frequency, it turns out as a disaster in case of high frequency EMI, since the coaxial shield will collect the EMI currents and dump them onto the signal reference, that is generally a critical conductor. Fig. 4 shows a trade-off solution to this dilemma. Closest to the I/O port, the coaxial shield is connected to the chassis via a ”zero-inductance”, leadless ceramic capacitor of low value, generally a few nanoFarads.
A few specific advantages beneficial to EMC performance can be mentioned for the coaxial cable.
Thanks to tight manufacturing tolerances, parameters like low HF line losses and characteristic impedance are specified with a good accuracy. Since the coaxial cable (invented around the 1920’s) has a long history of intensive use in RF engineering and instrumentation, a large inventory of good quality connectors (BNC, N, SMA etc.) is available with shielding performances ranging from good to excellent.
2.3 Shielded Twisted Pairs and Multi conductors shielded cables
With a Shielded Pair, there is a noticeable difference: the shield is no longer an active return conductor. Against susceptibility, the shield is there to neutralize the EMI ambient currents instead of letting them flow in the protected wire pair. Reciprocally, for preventing the signals carried by the pair from circulating in the external cable-to-ground loop, and eventually radiate unacceptable EMI, the shield will collect these Common Mode currents escaping from the pair and give them a low impedance return path back to their source. Since practically all discrete signal pairs are twisted – a simple, efficient way for preventing field-to-cable Diff. Mode pick-up and Crosstalk, shielded twisted pairs are usually designated as STP, by contrast to unshielded ones (UTP).
Advantage of the STP:
Given that the signal current is flowing back and forth in the two wires of the pair, the shield plays no role in the return of intended signal. The designer has all freedom to ground the shield of the STP to the equipment frame, and still keep his 0V reference isolated, if he so wishes.
A few specific STP disadvantages:
Due to the twisting, the accuracy of the cores-to-shield distance is not as perfect as with a coaxial cable, causing more line losses and impairments due to the fluctuations of the characteristic impedance. For reasons related with the above, the symmetry of the two wires of the pair vs ground (that is the shield) is not perfect, causing some % of the signal current to flow in the shield, hence in the system ground, generatingm EMI emissions. The same is true for susceptibility.
3. Evaluating the Merits of a Shielded Cable
For long, people were using shielded cables more or less casually, assuming that if an interconnection was shielded, it would no longer be a cause of EMI concern. In fact, like any element of a system, the global quality of a shielded link must be quantified. This includes not only the shield intrinseque performance, but also its terminations, i.e. the connector/receptacle assemby. Measurement techniques exist that allow to evaluate the effectiveness of a cable shield, along with calculation models for predicting the performance of a given cable, once installed.
The simplest, intuitive way would be to illuminate the shielded cable with a given electromagnetic field at several frequencies, and record the induced current or voltage on the inner conductor. Then to repeat the test with an unshielded version of the same cable. The comparison of the induced currents (or voltages) with and without the shield would give a figure of the shield performance. Unfortunately, this is an expensive test, requiring sets of antennas and an anechoic room, bearing the uncertainty inherent to any radiated measurement.
Furthermore, the results for a same cable sample would vary depending on the type of radiating antenna used in the test (H-field loop, Dipole etc.), the near field or far-field condtions of the set-up, and the height above ground for the tested item. A better method consist in measuring the shield transfer impedance, Zt, as explained next.
3.1 Shield Transfer Impedan ce, Zt
A convenient way of characterizing the merit of a cable shield (Ref. 2, 3, 5, 6) is its Transfer Impedance, Zt. It relates the current flowing on a shield surface to the voltage it develops on the other side of this surface (Fig.5). This voltage is due to a diffusion current through the shield thickness (if the shield is a solid tube, this diffusion rapidly becomes unmeasurable as frequency increases, due to skin effect) and to the leakage inductance through the braid’s holes. The better the quality of the braid, the less the longitudinal shield’s voltage.
Let us start with the simple configuration of a coaxial cable exposed to an EMI threat. As a result, an undesirable current is flowing along the shield. Since the shield wall is not a perfect tubular conductor, the flow of current is encountering two mechanisms:
a) the braid dc resistance (typ. 5 to 20mΩ per meter for a single braid). Fig. 5, shows the current density decreasing progressively in the thickness of the shield, such as when frequency increases, the skin effect tends to concentrate more current on the shield surface that is looking toward the source, leaving less and less current on the opposite surface.
b) the maze of small leakages caused by the holes in the braid weaving. This effect is defined as a leakage inductance in nanoHenry/meter. As a result, a small voltage is appearing along the internal core-toshield space.
If used for susceptibility calculations and normalized to a 1 meter sample of our coaxial cable, Zt is defined as:
- Vi = longitudinal voltage induced inside the shield, causing a noise current to circulate in the center conductor.
- Ish = current injected on the shield by the external EMI source
The term Zt itself contains the shield resistance Rsh and the shield transfer inductance Lt, regarded as the leakage inductance from the insideout (or the reverse), such as:
Typical values of Rsh, Lt for a decent quality, single braid shield are 10- 15mΩ/m and 1-2 2nH/m respectively.
The above description is a gross approximation. The actual mechanism is more complex: in fact it is the mutual inductance between the shield and the inner conductor that plays a major role. If we call L1 the loop formed by the center conductor alone and the ground, and L2 the shield-to- ground loop, there is a strong mutual inductance M1-2 between these two loops.
Due to the tight coupling between the inner conductor and its surrounding shield, this mutual inductance is almost equal to the self inductance of the center conductor vs ground. The result is that the current in the shield will induce in the inner conductor an opposite current that tends to cancel the initial EMI current. The cancellation is never 100% but can come pretty close. With a good quality single braid, it reaches 99.7 %, that is only 0.3% of the initial noise current remains in the inner wire.
This cancelling effect leads to a third rule, essential to the functioning of a shielded cable:
RS-3: For its good operation, a cable shield must carry a current equal and opposite to the total, net, current carried by the inner conductor. Therefore it must be grounded (not necessarily earthed) to the equipment boxes at both ends.
This is essential for the shield to work. A good test is: with a perfect shield, a current probe clamped around the whole cable should read NO current, meaning that the net current flowing in the inner conductor is perfectly balanced by an equal and opposite current in the shield.
Typical values of Zt for various cables are shown in Fig.6 . If the shield is grounded by pigtails (a poor practice), the pigtails’ and other impedances must be added to the Zt and loop impedances calculations. Once the external EMI current in the loop (Ish) is known (measured or calculated), the noise voltage induced internally can be derived for any length of this cable by:
(*) Caution: One must check that the physical length of the exposed cable does not exceed λ/2. If ℓ ≥ λ /2, the voltage calculation must be limited to an half-wave lenght, that is ℓ ≈150/F(MHz).
The principle is perfectly reciprocal and applied to emissions as well (Fig.5). For the 1m sample, the internal signal current Io returning by the shield’s inner surface causes an EMI voltage to appear along the outer side of the shield, that can be expressed as:
- Vext = external voltage appearing along the shield-to-ground loop
- V0, I0 = signal voltage and current on load side (ZL) of the coaxial cable
This voltage, in turn, excites the antenna formed by the external cableto-ground loop.
The key advantages of the Zt concept are:
- It is perfectly reciprocal (susceptibility <–> emission)
- It does not matter if the shield current is due to a field illuminating the cable ( Radiated Susceptibility scenario) or to a conducted interference (Ground shift between the two equipments).
- Zt is an intrinseque parameter to the shielded cable, independent of the radiated or conducted nature of an actual EMI threat
- Being a conducted measurement (current injection over the shield), it does not suffer the uncertainties of a radiated measurement.
3.2 An alternate way for characterizing a shielded cable: Shield Reduction factor Kr
Although Transfer Impedance Zt is a widely used and dependable parameter, Shielding Effectiveness (SE) or Reduction Factor (Kr) as figures of merit are often preferred by designers, because they can relate it directly to the whole shielding performance required for the system. It would be a nonsense to require 60dB of shielding for a system boxes if the associated cables and connecting hardware provide only 20dB, and vice-versa.
Practical formulas, directly expressing the shielding factor Kr of a cable, given its Zt (Ω/m) have been devised (Ref.2). This shielding factor Kr becomes a dimensionless number in dB that incorporates Zt, but allows for a direct prediction for an installed shielded cable.
Regarding susceptibility: Shield Reduction factor (Kr) is the ratio of the Differential Mode Voltage (Vd) appearing, core-to shield at the receiving end of the cable, to the external Common Mode Voltage (Vcm) applied in series into the loop. It can be expressed by:
Regarding emission: a reciprocal definition, similar to the basis of Equ. (4) can be used for characterizing a shielded cable with respect to emission, simply by the ratio of the Common Mode Voltage (Vcm) appearing in series into the external loop, to the Differential Mode Signal Voltage (Vd) applied, core-to shield at one end of the cable.
Calculations and experiments have shown that, except for the sign, the Kr factor is the same in the two above cases. Kr could also be regarded as the Mode Conversion Ratio between the internal circuit (center conductor and shield) and the external one (the shield-to-ground line). One could also compare the current in the loop if the shield was not there, to the remaining inner circuit current when the shield is in place, grounded both ends.
A complete demonstration of the rationale leading to the expression of Kr can be found in Ref (2). We will just give the end results, expressing Kr:
- Rsh = shield resistance in Ω/m
- Lext = self-inductance of the external shield-to-ground loop
- Lt = Transfer inductance of shield
This expression unveils three frequency domains;
a) Very Low Frequencies: the term ωLt is negligible, Zt is dominated by resistance Rsh:
Kr =Rsh / (Rsh + jω Lext ) ≈ 1 (0dB) below few kHz, since the lower term, loop impedance reduces to Rssh
b) Medium Frequencies (typically above 5-10 kHz for ordinary braided shield):
Kr = (Rsh + jω Lt ) / ( jω Lext )
Here the Reduction Factor increases linearly with frequency.
c) Higher frequencies (typically above 1 MHz), up to first λ/2 resonance), the reduction factor stays constant, independent of length and frequency:
Kr = Lt / Lext
A handy formula is derived, valid for any frequency from 10kHz up to first < λ/2 resonance:
The value for Zt must be the one taken at the frequency of concern.
Kr values when cable length is approaching or exceeding λ/2
As already mentioned for Zt, when the dimension of the cable reaches a half-wave length, the shield is no longer carrying a uniform current. The “electrically short line” assumption becomes progressively less and less acceptable when cable length ℓ exceeds λ/10. The shield grounded both ends behaves as a dipole exhibiting self-resonance and anti-resonance for every odd and even multiple of λ/2, respectively.
Knowing that actual wave propagation in the loop is slightly slower than in free space, the effective wavelength is recalculated to find the actual resonances. At these frequencies, the shield current will exhibit peaks, resulting in approximately 10dB periodic degradations of factor Kr.
This reflects the actual situation where, for a uniform EMI stimulus, the resulting interference will show periodic ”humps” beyond the first resonance point. Taking typical values for the shield-to-ground characteristic impedance with a conservative approach aligned on the asymptote of the humps, we reach a simple expression for worst case Kr beyond the first resonance:
For both Susceptibility and Emission cases, above λ/2 resonance:
As a recap of Kr for below and above resonance conditions:
A Few Practical Results for Shield factor Kr, below and above first λ/2 Resonance.
Fig. 8 shows calculated results for 3 coaxial cables, 1 meter above ground, with 360° contact at connector backshell. Curves are valid for any length, provided that the resonance region is adjusted if length is different from 1m. Fig. 9 shows test results for a 5m coaxial cable whose shield is intentionnaly spoiled by a 10cm pigtail.
Deterioration of Kr above 8 MHz is spectacular. Notice that below the MHz region, Kr degrades progressively down to almost 0dB around a few kHz: a shielded cable has no effect against low-frequency Comm. Mode coupling.
3.3 Field Radiated by a Coaxial Cable
Most RF signals, baseband video, some LAN links and other high-frequency signals are carried on coaxial cables. Provided that the shield is correctly tied to the signal ground reference at both ends, and preferably also to the chassis by the coaxial connectors, only a very little current (typically 0.3 to 0.1%, above a few MHz) returns by paths other than the shield itself (Fig.10). This external current radiates a small electromagnetic field, associated with the quality of the shield and its installation. It is related to the external voltage appearing along the shield due its Zt.
An RG-174 coax is connecting two racks. Useful signal: 15 MHz analog video, with 3mV detection sensitivity.
The cable parameters are:
Length: 2 m, Average height above the metallic frame = 10 cm
Good quality coaxial connectors are used at both ends (2.5mΩ/connector).
When exposed to a 50V/m field at 30MHz, a Comm. Mode voltage of 6 volts is induced in the shield-to-frame loop (see Article #4 on radiated coupling). What is the voltage induced internally?
We use the equation (6) for Kr. Curve Fig. 6 indicates 0.3Ω/m for the RG174 @ 30MHz.
Kr (dB) = – 20 Log [1 + (6 F(MHz) / Zt (Ω/m) ]
= -20log (1 + 6 x 30 /0.3) = -55dB corresponding to ≈ 1/600 reduction factor.
The voltage on the center conductor will be 6V/600 = 10mV. This is 3.3 times above the threshold of video sensitivity.
Several solutions can reduce the coupled interference:
- Select a coaxial cable with a lower Zt, i.e., Zt < 0.1W/m at 30 MHz. Such performance is achievable with optimized braided shield (thicker, denser braid) or more easily, with more costly double-braid shield.
- Slip a large ferrite bead over the cable shield. It will take an added series impedance of about 1,000Ω to achieve the required attenuation. Passing the cable three times into a large bead will provide such impedance.
- Decrease cable height above chassis ground by at least 3 times.
4. EMI Reduction by Shielded Pairs or Multiconductor Shielded Cables
The concept of transfer impedance, shown for a coaxial cable, is transposable to shielded twisted pairs (STP), keeping in mind that the shield is no longer an active return conductor (Fig. 3). With balanced interfaces and wire pairs, the current returning by the shield is only prorated to the percentage of asymmetry in the link (Ref.3). If the transmission line is balanced to X% symmetry, the current returning by the shield is, for the worst possible combination of tolerances, only X% of the total current in the loop impedance ZEXT. In this case, Eq. 1 applied to radiated suceptibility becomes:
As a result, the radiated field is reduced by a factor equal to X%, compared to an ordinary coaxial cable situation. Depending on the quality of the balanced link, X may range from 1 to 10%, a typical (default) value being 5%. Recent progress have been made with high quality (Class #5 or #6 STP), but their best balance is generally in the 2- 3% range. If the wire pairs are interfacing circuits that are not balanced (e.g., the signal references grounded at both ends), a larger portion of the signal current will use the shield as an alternate return path. This portion is difficult to predict: at worst, this unbalanced scheme cannot cause more interference than the coaxial case.
Shields grounded One End only
If, for legitimate reasons (e.g., low frequency ground loops between distant boxes, upsetting a sensitive analog input), a shield has to be grounded at one end only, it will be only effective against LF electric fields and capacitive crosstalk. It has virtually no effect on CM immunity or radiation, as the CM loop current does not return by the shield but, rather by the chassis and ground plane, as if there were no shield.
If the frequency of the electric field (or capacitive) threat increases, the capacitive current captured by the STP will increase, and will be flowing on a shield whose impedance also increases (Fig.12). Thus the shield voltage versus ground increases with (F)2), becoming a significant fraction of VCM. The floated end of the shield becomes the “hot” tip of a receiving monopole, and we have just replaced a noisy pair by a noisy shield. So, exceptions acknowledged, a cable shield must be connected at both ends to the boxes, whether these are grounded or not.
Exceptions are low-level analog instrumentation (strain gages, thermocouples, etc.) and audio interface cables, where only an electrostatic shield is needed. Ground loops are supressed by galvanic isolation amplifiers, differential amplifiers, and so forth, and grounding a shield at both ends could inject LF (few kHz) noise into the cable. A few millivolts injected this way are harmless for digital interfaces but can constitute strong interference for low-level analog signals.
Shielded Flat Cables
One specific case of shielded multiconductor cable is the shielded flat cable. A few typical versions are shown in Fig 13. Version (a), sometimes advertised as “shielded”, is merely a flat cable with an embedded ground plane. Although offering some advantages, its reduction is often insufficient because CM current can still flow on the single-side foil edges and radiate ( Ref.4). The (b) version, also marketed as “shielded” is leaky at HF due to the long, unclosed seam that runs over the entire length.
The drain wire is acceptable as a low frequency shield connection, but absolutely inadequate at HF. Versions (c) and (d) deserve to be called “shielded” as the shield totally encircles the wires. Yet, with (c), there is no access to outer metal surface, 360° bonding is not easily made, and the drain wire is still there.
Importance of the Shield Connections
As important as a good shield with low Zt is its low-impedance termination (Ref.1) to the equipment metal boxes. The connection impedance Zct is directly in the signal current return path, in series with Zt (Fig.14). Therefore, Zct can increase the coupled voltage for a given shield current. The following values are typical transfer impedances of one shield end connection:
We saw that grounding a shield by pigtails will ruin cable shield inefficiency. The same is true for internal pigtails gathering the individual shields of STPs. They can picks up PCB radiation and drives the resulting current over the pairs shields, causing them to radiate. Fixing this simple detail can reduce by 20dB the emission level. Obviously, it is vital that a cable shield be terminated by a low-impedance connection (lower than Zt of the cable itself). Most connector styles are available in shielded versions allowing a 360° contact on the braid (Fig.15).
Discussion regarding STP vs. UTP
There is frequent controversy regarding the possibility of using unshielded twisted pairs (UTP) for high speed data links inside buildings, instead of more expensive STP. Measured data, and radiation models, have shown that in the 30-200 MHz range, a 20-25 dB reduction factor was needed for an ordinary wire pair to satisfy FCC class B. If, instead of the typical 10-30% unbalance of an ordinary singleended link, a differential link with a high grade UTP is used, a total 2% unbalance can be achieved, meeting the 20 dB reduction goal. This goal could also be achieved with the help of balancing transformers.
Furthermore, the symmetry with a STP cable is slightly inferior to that of the same cable in UTP version; this is due to the fact that a perfect geometry of two wires twisted inside a shield is more difficult to control: a 0.05mm un-evenness in a 0.5mm dielectric thickness results in 10% assymetry of each wire-to-shield (impedances Z1/Z2 and Z3/Z4 in Fig.11). High grade, UTP with 2% unbalance are available, while the corresponding STP exhibits > 3%. In addition, symmetry impairment is often aggravated by mediocre practices for shield continuity in building wiring.
Therefore, many articles exist, suggesting substantial savings by not using STP. But radiated emission is not the only EMC constraint. In industrial sites, hospitals or high-rise commercial buildings, immunity to RF fields of 10V/m and to Electrical Fast Transients require added CM protection, that even high grade UTP cannot provide. Thus, on the basis of immunity, STP is often mandatory in harsh environments.
Michel Mardiguian EMC Consultant, France
- Hoeft, L. Transfer Impedance of backshells, 1982 IEEE/EMC Symposium
- Mardiguian,M. Simple prediction method for cable shielding factor, based on Zt, ITEM, 2012 Design guide
- Mardiguian, M. Differential Transfer Impedance of shielded twisted pairs -ITEM, 2010 Design guide
- Palmgreen, C. Shielded Flat Cables for EMI/ESD reduction, 1981 IEEE/EMC Symposium
- Schelkunoff, S. Electromagn. Theory of coaxial lines and cylindrical shells, Bell Technical Journal, 1934
- Vance, E. Coupling to Shielded cables, Wiley, 1978