SCREENING OF CABLES IN THE MHZ TO GHZ FREQUENCY RANGE

EXTENDED APPLICATION OF A SIMPLE MEASURING METHOD

Otto Breitenbach Thomas Hähner Bernhard Mund

Alcatel Kabel Alcatel Kabel Bedea

Summary:
Although cables are passive components, which can’t disturb the circumference by its own, their electro-
magnetic compatibility (EMC) is as important as the EMC of active devices, which are connected with those
cables.
The increasing use of cables, especially in the field of information technologies and cable distributed
television systems up to very high frequencies requires new measuring methods for determination of the
screening effectiveness of those cables. This methods shall be cheap, easy to use, applicable up to the GHz
range and with good reproducibility.
An investigation shows, that the well known triaxial measuring method with a one-sided short circuit
commonly used for lower frequencies retains its merits also in the GHz range. The extension of this triaxial
method up to higher frequencies was first investigated in 1993 in Germany [1]. After two years of discussion
and improvement in the international working group IEC TC46/WG5 it became a Committee Draft, IEC
46A/320/CDV, „shielded screening attenuation test method“ [2] [3].
1. Introduction
In many cases, above all in the lower frequency range, the screening effectiveness of cables is described by
the transfer impedance ZT. It is, for an electrically short peace of cable, defined as the quotient of the
longitudinal voltage measured on the secondary side of the screen to the current in the screen, caused by a
primary inducing circuit, related to unit length [4]. Although the transfer impedance ZT covers only the
galvanic and magnetic couplings it is common practice to use it also as a quantity which includes the effect of
the coupling capacitance CT through the cable screen [5]. In this case it is named equivalent transfer
impedance ZTE which includes the effects of galvanic, magnetic and capacitive coupling.
For the determination of the proper coupling capacitance there is, as standardised quantity, the capacitance
coupling admittance YT. The coupling admittance, for an electrically short peace of cable, is defined as the
quotient of the current in the screen caused by the capacitive coupling in the secondary circuit to the voltage
in the primary circuit related to unit length [4].

I1 I1 l << λ/4
l << λ/4

U1 CT
I2
U2

Fig. 1: Definition of transfer impedance Fig. 2: Definition of coupling admittance

U2 I2
ZT = (1-1) YT = = jω ⋅CT (1-2)
I 1 ⋅l U 1 ⋅l
With electrically short cables, where wave propagation can be neglected, the screening quantities related to
unit length can directly be used to calculate an induced disturbing voltage. In the higher frequency range the
implications get similar complicated as the transmission characteristics of a simple line, dependent on the
impedance and admittance per unit length as well as on the terminating resistors.
For the measurement of the transfer impedance two methods are standardised, the well known triaxial
measuring method according to IEC 61196-1, clause 12.2 and the line injection method according to IEC
61196-1 clause 12.1. The triaxial method is limited to a frequency range in which the coupling length is short
(less than approximately one sixth of the wave length). For example the upper frequency limit is appr. 30
MHz for a coupling length of 1m. The line injection method is applicable in a frequency range up to 3 GHz
[6]. This procedure requires that the characteristic impedance and propagation velocity of the induced loop
are matched to the cable under test. This complicates its use up to the GHz range considerably.
The measuring methods for the capacitance coupling admittance according to IEC 61196-1, clause 12.3 are
limited to frequencies below 1 MHz. According to experience the coupling capacitance CT is independent of
frequency, at least up to 1000 MHz [5]. Therefore it is common practice to convert the measured results of YT
proportionally to high frequencies.
2. Objectives
It is desirable to measure and evaluate the screening efficiency of cable screens also in the wave propagation
frequency range such that its characteristics can be directly applied. This requires to look closer at the condi-
tions of such applications.
In general, a system of electromagnetic induction consists of a transmission circuit in the cable, which is
assumed to be fully defined, and of a surrounding transmission system, which is assumed to be universal with
respect to the definition of cable screening. The screening effectiveness may be universally described by the
maximum power output into the surroundings of the cable related to the power propagating in the cable. The
power ratio is best expressed logarithmically as screening attenuation.
An often used procedure to determine the screening attenuation is the well known „absorbing clamp method“
according IEC 61196-1 clause 12.4. The drawback of this method is, that its set-up requires relatively much
space, does not exclude environmental effects - unless the measuring area is enclosed in a shielded cabin -,
and that the available absorbing clamp transformers considerably limit the measurement sensitivity.
It suggests itself to limit the free space such that the said problems don’t occur but wave propagation near the
cable surface is not significantly changed. A triaxial measuring set-up is the solution. It has a one-sided short
circuit between the metal tube and the cable screen. Power is fed into the terminated inner circuit of the cable
and the disturbing power is measured at the opposite end of the outer circuit.
3. Theory of the triaxial measuring method
On the basis of the known reversibility of primary and secondary measuring circuits, the proposed measuring
set-up, presented in Fig. 3, actually meets the IEC standard despite of the interchange of generator and
receiver. The benefits of feeding the inner system, which is terminated by its characteristic impedance, are the
matching of the generator and reflection free wave propagation over the cable length.
The characteristic impedance of the outer circuit depends on the diameter of the measuring tube and the cable
design. The effect of the mismatch in the outer circuit is discussed later on.
The equivalent circuit using concentrated elements (shown in Fig. 4) facilitates the understanding of the theo-
retical relationships.

cable sheath cable screen inner conductor

cable under test coupling length terminating
input voltage U app. 2m resistor R1= Z 1
1
tube
U1 Z1
calibrated receiver
or network analyzer ZT⋅l
U1 screen
signal
generator
U2
YT⋅l R U2

tube

Fig. 3: Triaxial measuring set-up for screening

attenuation Fig. 4: Equivalent circuit of the
triaxial measuring set-up
7/2
The following symbols are used:
• as screening attenuation
• asn normalised screening attenuation
( Zs=150Ω and |∆v/v1 |=10% or εr1/εr2n=1,21)
• α1, α2 attenuation constants of primary and secondary circuit
• β1, β2 phase constants of primary and secondary circuit
• c = 3 ⋅108 m propagation velocity in free space
0
s
• CT through capacitance per unit length
• Da outer diameter of cable screen
• Dm inner diameter of measuring tube
• εr1, εr2 rel. dielectric permittivities of primary and secondary circuit
• εr2,n normalised value of the rel. dielectric permittivity of the environment of the cable
• f measuring frequency
• l effective coupling length
• λ0, λ1, λ2 wave length in free space, primary and secondary circuit
• MT effective mutual inductance per unit length
for braided screens M T = M 12' − M 12' '
where M’12 relates to the direct leakage of the magnetic flux and M’’12 relates to the
magnetic flux in the braid [5]
• P1 feeding power of the primary circuit (cable under test)
• P2 measured power received on the input resistance R of the receiver in the secondary
circuit
• Pr radiated power in the environment of the cable, which is comparable to P2,n+P2,f of
the absorbing clamp method of IEC 61196-1, 12.4
• Ps radiated power in the normalised environment of the cable under test
( Zs=150Ω and |∆v/v1 |=10% or εr1/εr2n=1,21)
•R load resistance of secondary circuit (input resistance of receiver)
• RT screen resistance per unit length
• U1, U2 input and output voltage of the primary and secondary circuit
• 1 2
v , v propagation velocity in primary and secondary circuit
•ω radian frequency
• YT = j 2π f CT capacitance coupling admittance per unit length
• FZ = Z Z Y
1 2 C capacitive coupling impedance per unit length
• Z T = RT + j 2π f M T transfer impedance per unit length
• Z1, Z2 characteristic impedance of primary and secondary circuit
• Zs =150 Ω normalised value of the characteristic impedance of the environment of the cable
under test
Based on the conditions of the objects to be measured it is assumed that the transfer impedance ZT is low and
the reciprocal quantity of the coupling admittance YT is high in comparison with the characteristic
impedances Z1 and Z2 and the load resistance R. Therefore the feedback of the secondary circuit on primary
circuit can be neglected.
When the frequency is low one may consider the primary circuit shown in Fig. 4 as voltage divider and read
the disturbing voltage ratio directly. The one-sided short circuit in the measuring circuit prevents the
efficiency of the capacitance coupling admittance YT.
U 2 Z T ⋅l
≈ (3-1)
U1 Z1
In the high frequency range, where wave propagation has to be considered, one may expect the transfer
impedance to be proportional to the frequency in most cases. Therefore it is expedient to use the following
equation:
Z T = R T + j ω M T ≈j ω M T (3-2)
and consider the effective mutual inductance per unit length MT at high frequencies as an approximated
constant quantity as it is usually done with the through capacitance CT.
7/3
It is common practice to describe the capacitive coupling in the form of the capacitive coupling impedance
ZF., which is nearly invariant with respect to the geometry of the outer circuit (tube). [5, 9]
Z F = Z1 Z 2YT = Z 1 Z 2 jω CT (3-3)
Furthermore, the attenuation constants α1 and α2 of the circuits may generally be neglected as, for example,
the value of nearly 1 dB/m of the common cable type RG 58 at 3 GHz is relatively small compared to the
usual measuring uncertainty.
In the relevant literature it is common practice to describe wave propagation in the form of phase constant [5,
7]. If the ratio between effective length and wave length is used instead of the phase constant, the periodic
phenomena become clearer. With wave length λ0 in free space or λ1, λ2 in the circuits 1 and 2, the following
relation exists:
l l
β1,2 ⋅l = 2π ⋅ εr 1,2 ⋅ = 2π (3-4)
λ0 λ1,2
According to the theory of wave propagation [7] and line crosstalk [8], a wave propagates in the matched
inner circuit towards the matched end. In the outer circuit a part of the induced wave propagates forwards to
the measuring receiver and the other part is moving backwards to the short circuit. The total reflection at the
short circuit reverses this backward wave and superposes it to the original forward wave, i.e. the sum can be
obtained as measured value.
If the second circuit is matched at both ends the backward wave would be measured at the generator end (near
end) and the forward wave at the opposite end (far end) separately.
Hence for the near end derives from [5]
U 2 n ZT + Z F c0 
 − j 2π ( εr 1 + εr 2 )λl0 

= 1 − e 
( )
(3-5)
U1 2Z1 jω εr1 + εr 2 
 

and for the far end
U 2 f Z F − ZT c0 
 − j 2π ( εr 1 − εr 2 )λl  − j 2π
l
λ2
= ⋅1 − e  ⋅e
( ) (3-6)
0

U1 2 Z1 jω εr1 − εr 2 
 


0
Fig 5.: calculated voltage ratio for a typical
U 2n, f
20 ⋅log in dB braided cable screen, quantities used:
20 U1
CT = 0,02 pF/m
40
MT = 0,4 nH/m
far end
R = 50 Ω l = 2m
60 Z1 = 50 Ω εr1 = 2,3
Z2 = 120 Ω εr2 = 1,1
80
near end

100
10 KHz 100 KHz 1 MHz 10 MHz 100 MHz 1 GHz

With a short circuit and an unmatched measuring receiver these original voltage waves cause additional
voltage portions. The sum of all voltage portions is zero at the shorted end (near end) and U2 at the receiver
end (far end). By use of the wave parameter and reflection factors or terminating resistors it is possible to
calculate all voltage portions and the voltage U2 from the primary induced voltage waves (3-5) and (3-6) as
follows:

U2
U1
Z − ZF
≈ T
εr1 − εr 2
⋅1 − e − [ jϕ1
]+ ZT + Z F
εr1 + εr 2
⋅1 − e − [ jϕ2
]⋅ ω ⋅1Z ⋅
c0
(
2 + (Z 2 / R − 1)⋅1 − e − jϕ3
) (3-7)
1

or in consideration of equation (3-2), (3-3)

7/4
 MT MT
− CT Z 2 + CT Z 2
U2
≈
Z1
[1 − e ]+
− j ϕ1 Z1
[1 − e ϕ ]⋅ 2 + (Z
− j 2
c0
( ) (3-8)
2 / R − 1)⋅ 1 − e
− jϕ3
U1 εr1 − εr 2 εr1 + εr 2

where
ϕ1 = 2π ( εr 1 − εr 2 )λl 0
ϕ2 = 2π ( εr 1 + εr 2 )λl 0
ϕ3 = ϕ2 − ϕ1 = 4π εr 2
l
λ0
Another way to obtain the related induced voltage shows [10].
The functional equation (Fig. 6)
1 − e− jϕ
= 2 sin ϕ 2 with ϕ = ϕ1, ϕ2, ϕ3 (3-9)

shows, that the equation of the voltage ratio contains three periodic partial functions of the ratio effective
length l to wave length λ0:
2

1 − e− jϕ
ϕ3
1

ϕ2 ϕ1
2
0 1 2 3 4 5

l/λ0

Fig. 6.: calculated periodic functions for εr1 = 2,3 and εr2 = 1,1

For low frequencies, when l<<λ0 and, consequently, sinϕ ≈ϕ, equation (3-7) changes into (3-1), the result of
the common measuring method for the transfer impedance.
An example of the theoretical curve of the voltage ratio is shown in fig. 7 in two diagrams. The left one with
a logarithmic scale to extend the lower frequency range and the right one with a linear scale up to very high
frequencies.

U2
40
40 20 ⋅log in dB
U1
U2
20 ⋅log in dB
50
U1 50

60
60

70
70

80
80 0 500 1000 1500 2000 2500 3000
0.01 0.1 1 10 100 1000

f/MHz f/MHz

Fig 7.: calculated voltage ratio, quantities used:

CT = 0,02 pF/m MT = 0,4 nH/m
R = 50 Ω l = 2 m
Z1 = 50 Ω εr1 = 2,3
Z2 = 120 Ω εr2 = 1,1

7/5
It is not expedient to value the induced power for an exact length of cable at a single frequency, anywhere
between a minimum and maximum of the function. Only the periodic maximum voltage is important for the
evaluation of the screening effectiveness. In the outer circuit the wave propagation shall be nearly the same as
in free space. Therefore the characteristic impedance Z2 is higher than the common input resistance R of the
measuring receiver, i.e. 50 Ω or sometimes 75 Ω .
Consequently periodic maximum values of the voltage ratio are obtained from (3-7), (3-8) which are inde-
pendent of the input resistance of the receiver R and of effective cable length l:

U2 c Z −Z ZT + ZF
≈ 0 ⋅ T F + (3-10)
U1 max
ω Z1 εr1 − εr2 εr1 + εr2

or in consideration of equation (3-2), (3-3)

MT MT
U2 Z1 − CT Z 2 Z1 + CT Z 2
≈ + ⋅c0 (3-11)
U1 max
εr 1 − εr 2 εr 1 + εr 2

At first sight CT, Z2, εr2 and ZF appear as random quantities which depend on freely chosen dimensions of the
measuring tube. In reality, however, the voltage ratio is independent of the characteristic impedance of the
outer circuit since CT⋅Z2 and ZF are practically invariant with respect to the dimensions of the measuring tube
[5,9]. Furthermore, the influence of the cable sheath on the resulting relative permittivity εr2 is negligible if
the design of the measuring tube takes into account the requirement for a wave propagation which is
approximately the same as in the free space, in consequence εr2 ≈1,0.
The periodic maximum value is independent of the effective length l and frequency f or wave length λ. A
measured frequency response would hint at a frequency related quantity rather than the pure mutual
inductance MT.
As it is seen from fig. 6 and 7 the envelope rise is reached with the first maximum of the wide period at:
λo c0
≤ 2 ⋅ εr1 − εr2 or f> (3-12)
l 2 ⋅l ⋅ εr1 − εr2
In this frequency range ZT can be calculated if ZF is negligible:

ω ⋅Z1 ⋅εr1 − εr 2 U 2
ZT ≈ ⋅ (3-13)
2 ⋅c0 ⋅ εr1 U1 max

4. Screening Attenuation
The screening attenuation is defined as the logarithmical ratio of the maximum power in the secondary
(outer) circuit to the power propagating in the primary (inner) circuit.
 Pr ,max 
a s = − 10 ⋅log 10 Env  (4-1)
 P1 
The power coupled into the outer circuit depends on Z2 although the peak voltage is independent of it. Thus a
normalised value of the characteristic impedance of the outer circuit Zs has to be defined. It is common
practice to define Zs = 150 Ω [5].
In the standardised „absorbing clamp method“ (IEC 61196-1 clause 12.4) the outer circuit is matched with Z2,
and the radiated power is the sum of the near end and far end crosstalk. From the comparison of that
measuring circuit with the measuring circuit of the triaxial method result the relation of the measured power
to the radiated power.
The equivalent circuit for an electrical short part of the length ∆l and for a negligible capacitive coupling
illustrates the circumstances.

7/6
 I1

Fig. 8 Equivalent circuit for an

electrical short part of the length ∆l and
U1 Z1
ZT⋅∆l
negligible capacitive coupling

∆U2*
Z2 Z2

∆I2*

The power in the primary circuit is:

2
U1
P1 = U 1 ⋅I1 = = I1 ⋅Z1
2
(4-2)
Z1
The power in the secondary circuit, which is coupled by the transfer impedance ZT is
∆U 2*
P2* = ∆U2* ⋅∆I2* ∆U2* = I1 ⋅ZT ⋅∆l ∆I 2* = (4-3)
2 ⋅Z 2
thus
( ) ( Z T ⋅∆l )
2
P2* ∆U 2* 1
2

= ⋅ = (4-4)
P1 2 ⋅Z 2 I 12 ⋅Z1 2 ⋅Z1 ⋅Z 2

If the secondary circuit is short circuited at one end and terminated by R at the other end the power measured
at R is

P2 ( Z T ⋅∆l ) 2
= (4-5)
P1 Z1 ⋅R
thus
*
P2 R
= (4-6)
P2 2Z2
or in the case of radiation due to the normalised characteristic impedance of the environment

Pr Pr ,max R
= = (4-7)
P2 P2 ,max 2 Z s

Thus the screening attenuation has to be calculated by:

P1 P1 2 ⋅Z s
a s = 10 ⋅log 10 = 10 ⋅log 10 ⋅
Pr ,max P2,max R
2
 U1  2 ⋅Z s
= 10 ⋅log 10 
U
 ⋅

(4-8)
 2 ,max  Z1

U1 300Ω
= 20 ⋅log 10 + 10 ⋅log 10
U 2 ,max Z1

7/7
5. Normalised screening attenuation
From eq. (3-10) one may recognise, that the maximum voltage ratio and therefore the screening attenuation is
depending on the velocity difference between the primary and secondary circuit. Therefore the test results
may also be presented in normalised conditions where Zs = 150 Ω and the velocity difference |∆v/v1| = 10 %
or εr1/εr2,n = 1,21.
The normalised screening attenuation is calculated by:

ω ⋅ Z1 ⋅Z s ⋅ εr 1 − εr 2 ,n
a s ,n = 20 ⋅log 10 (5-1)
Z T ⋅c0

With respect to eq. (3-10), (4-8) and (5-1) and with neglect of ZF, the difference ∆a of the normalised and the
measured screening attenuation is given by:

 εr 2, n 
 1− 
 εr1 
∆a = asn − as = 20 ⋅log10  2 ⋅  (5-2)
 ε 
 1 − r 2,t 
 εr1 
 
where εr2,t ≈1,1 is the relative dielectric permittivity of the outer circuit (tube) during measurement.
The following table shows the difference ∆a for typical cable dielectric:

εr1 2,3 2,1 1,6 1,3

εr2,n 1,9 1,7 1,3 1,1
∆a in dB -12 -11 -8 -2

6. Measuring results
The measuring curve of common type of cables prove the validity of the theoretical basis. The voltage ratio
U2 / U1 is measured by means of a network analyser having an internal resistance of 50 Ω . The screening
attenuation as is presented in Figures 9 to 13 for three types of cables as a function of frequency.

• RG 58 according to MIL-C-17 with single copper braid

• HF 75 0,7/4,8 2YCY with a dielectric of solid PE and a single copper braid
• HF 75 1,0/4,8 02YCY with a dielectric of foamed PE and a single copper braid
• RG 223 according to MIL-C-17 with double copper braid

The theoretical relations of the transitions from low to medium and high frequencies - appearing in the cal-
culated curve in Fig. 5 - become most evident with the single copper braid (see Fig. 9). Here the voltage ratio
is independent of the frequency up to approx. 0,4 MHz but proportional to the effective length of the
measuring tube like the transfer impedance. At high frequencies, higher than approx. 100 MHz,
superpositioned periodic functions occur showing maximum values of approximately equal magnitude
independent of frequency and effective length. The frequency, at which the superposition appears is
reciprocal to the effective length just as the frequency spacing of the peak values (see Fig. 9,10). In contrast
to the effective length of 2 m, the effective length of 0,5 m does not allow to plot the screening envelope
curve with sufficient accuracy any more, due to the wide spacing of the long period maximum values.

7/8
 U2
U2 U1 as
U1 -32 -32 40

-52 -52 60

-72 -72 80

-92 -92 100

-112 -112 120

0.1 1 10 100 1000 0 500 1000 1500 2000 2500 3000
f/MHz f/MHz

Fig. 9: Frequency response of logarithmic voltage ratio |U2/U1| in dB (left hand scale) and screening
attenuation as (right hand scale) of single braid screen, Cable Type RG 58, coupling length l = 2 m

U2
U2
U1
as
U1 -32 -32 40

-52 -52 60

-72 -72 80

-92 -92 100

-112 -112 120

0.1 1 10 100 1000 0 500 1000 1500 2000 2500 3000

f/MHz f/MHz

Fig. 10: Frequency response of logarithmic voltage ratio |U2/U1| in dB (left hand scale) and screening
attenuation as (right hand scale) of single braid screen, Cable Type RG 58, coupling length l = 0,5 m

The periodic frequency spacing in the measured curve and the screening attenuation are dependent on the ve-
locity difference between primary and secondary circuit (eq. 3-7, 3-10). This theoretical relation become most
evident in Fig 11 and 12. Where the cable screen of both cables are equal, but the relative permittivities of the
cable dielectric εr1 and thus the velocity difference in the test set-up differ. In Fig 11 we have εr1=2,3 and a ve-
locity difference |∆v/v1| ≈45% whereas in Fig 12 εr2=1,7 and |∆v/v1| ≈24%. Thus in Fig 12 we have a larger
frequency spacing of the wide period and also a lower screening attenuation. But the normalised screening
attenuation of both cable screens are equal, as≈43 dB.

For the cable with double copper braid (Fig. 13) the theoretical relations become apparent only if the
measurement is very accurate and the receiver is sensitive enough for low induced voltage. Apart from its
level and distinct function of frequency, the screening attenuation of the double copper braid is obviously
similar to that of the single copper braid.

7/9
U2 U2
U1 U1
-34
as
-34 40

-54 -54 60

-74 -74 80

-94 -94 100

0,01 0,1 1 10 100 1000 0 800 1600

f/MHz f/MHz

Fig. 11: Frequency response of logarithmic voltage ratio |U2/U1| in dB (left hand scale) and screening
attenuation as (right hand scale) of Cable Type HF 75 0,7/4,8 2YCY, er1=2,3 , |∆v/v1|=45% ,
coupling length l = 2 m

U2 U2
U1 U1 as
-34 -34 -40

-54 -54 -60

-74 -74 -80

-94 -94 -100

0,01 0,1 1 10 100 1000 0 800 1600

f/MHz f/MHz

Fig. 12: Frequency response of logarithmic voltage ratio |U2/U1| in dB (left hand scale) and screening
attenuation as (right hand scale) of Cable Type HF 75 1,0/4,8 02YCY, er1=1,7 , |∆v/v1|=24% , coupling
length l = 2 m

U2
U1
U2
U1
as
-32 -32 40

-52 -52 60

-72 -72 80

-92 -92 100

-112 -112 120

0.1 1 10 100 1000 0 500 1000 1500 2000 2500 3000

f/MHz f/MHz

Fig. 13: Frequency response of logarithmic voltage ratio |U2/U1| in dB (left hand scale) and screening
attenuation as (right hand scale) of double braid screen, Cable RG 223, coupling length l = 2 m

7/10
7. Comparison with absorbing clamp method
In the absorbing clamp method according to IEC 61196-1 clause 12.4 , in principle, the current on the outside
of the cable under test is measured. The matched outer circuit is directly induced by the inner circuit. The
power in the outer circuit is related to the current by calibration.
Comparison results are available of some coaxial cables of different screen designs. They show a difference
of max. 3dB

screening attenuation as in dB
cable type, frequency absorbing triaxial
screen GHz clamp method
method
RG 58, 0,2 51 48
single 0,8 52 50
braid 3,0 - 50
RG 214, 0,2 51 50
single 0,8 54 51
braided 3,0 - 53
RG 214, 0,2 79 79
double 0,8 82 81
braid 3,0 - 83
RG 223, 0,2 86 88
double 0,8 90 90
braid 3,0 - 83

8. Practical design of the test set-up

The set-up to measure the screening attenuation aS is in principle the set-up to measure the attenuation of RF
devices, where the voltage ratio U2/U1 is measured. The cable under test is connected to the output of a
RF-generator, the output of the coupling tube is connected to the measuring input of a RF-receiver. Generator
and receiver may be included in a sensitive network analyser (see Fig. 3 and 14).
The measuring tube shall be of a material, which is not ferromagnetic and good conductive (for example
brass), with an inner diameter of about 40 mm to 50 mm and a length of 2 m to 4 m or more, where the total
length of 2 m or more may be achieved by screwing together single parts of tubes (RF-tight).
One way to realise the short circuit at the near end of the CUT is to solder a braid of silvered copper wires to
a punched disk of copper. This "contacting braid" is fixed on the outer conductor of the cable sample where
the sheath is removed, e.g. with cable clamps. The electrical contact between this contacting braid and the
measuring tube may then be achieved by a jam-disk, which is fixed by the clasp cap, which is screwed to the
tube (see Fig. 15).
The contacting braid, which is prepared once may be used several times. Soldering of the screen of the cable
sample to the tube - as usual at the classic triaxial - set-up is no longer required and the time to prepare the
CUT is minimised.
The termination at the far end of the CUT is achieved by a resistor of the same value as the characteristic
impedance of the CUT. Experience has shown, that best results are obtained with SMD resistors respectively
so called "Mini-Melf-Resistors" with low mechanical dimensions and good RF-characteristics, which are
soldered directly between the inner and the outer conductor of the CUT. To avoid radiation and to contact the
outer conductor of the CUT, this termination is shielded by a case, which is well conductive (see Fig. 3).

7/11
 1 2
1

OUT R A B
3
4
4 5

6
7
3

1 tube of brass 5 sheath of cable sample

1 Network Analyser 3 Power Divider 2 clasp cap 6 cable under test (CUT)
2 Attenuator 20 dB 4 Measuring tube 3 jam disk 7 contacting braid
4 contact to the cable screen of CUT
Fig. 14: schematically arrangement for the Fig. 15: Short circuit between tube and cable
measurement of the screening attenuation aS screen of the CUT

Important for clear and reproducible results is the centring of the sample in the measuring tube. A slacked
hanging cable under test in the measuring tube will led to deviations of the characteristic impedance Z2 of the
outer system over the coupling length and thus to additional reflections. Centring may be achieved by
punched polyethylene disks which are placed in the measuring tube, or better by stretching the sample under
test, e.g. with a desk vice. Also a vertical mounting of the measuring tube is useful.
9. Influence of mismatches
There may be mismatches in the inner or outer circuit of the test set-up which influence the results
significant. Theoretical and practical investigations [11] show that a mismatch of the terminating resistor in
the inner circuit is of low influence as long as:
Rter min ation − Z1
⋅100% ≤ 10% .
Z1
Additional mismatches in the outer circuit however result in significant errors. With the screening case of the
terminating resistor a mismatch is inserted into the outer circuit, which affect the results significantly
depending on the mechanical dimensions [11]. The mean characteristic impedance of the outer circuit,
formed by the cable screen and the measuring tube, respectively in the outer circuit at the screening case is
given by:
60Ω D  60Ω D 
Z2 ≈ ⋅ln m  (9-1) Z3 ≈ ⋅ln m  (9-2)
εr 2  Da  εr 2  Dcase 

where
Da outer diameter of cable screen
Dcase outer diameter of screening case
Dm inner diameter of measuring tube

A deviation between Dcase and Da thus results in different impedance’s and therefore in additional reflections
in the outer circuit. For example a screening case with a outer diameter of Dcase = 1,2 ⋅ Da results in a
impedance Z3 which is 11 Ω less than Z2 (εr2=1,0).

7/12
Fig. 16 facilitates the understanding of the theoretical relationships.

r2n=-1 r23 r32 = -r23 r3f

r2n = -1 (due to the short circuit)
Z3 ; γ3=γ2 Z − Z2
Z2 ; γ2 r23 = 3
from
Z1 ; γ1
to Z3 + Z2
generator receiver
r32 = − r23
Rtermination=Z1

R − Z2
r3 f =
l2 l3
R + Z2

Fig. 16: real triaxial set-up

The outer circuit thus consist of two lines with different characteristic impedance’s. To calculate the voltage
at the receiver some additional variables have to be defined.
Uh is the voltage, which is coupled from the cable under test into the outer circuit (Z2, γ2, l2), propagation to
the far end, including the total reflection at the near end.

U h U 2 f U 2n
= + ⋅r2 n ⋅e − γ 2 ⋅l2 (9-3)
U1 U1 U1

Where U2f, U2n are the voltages in a matched outer circuit according eq. (3-5) and (3-7).
Multiple reflections of this wave between the short circuit at the near end of the outer circuit and the
transition from Z2 to Z3 are described by T2f.

1 + r23
T2 f = (9-4)
1 − r2 n ⋅r23 ⋅e − 2⋅γ 2 ⋅l2

The superposition of the wave which is propagating from the line Z2, γ2, l2 to the far end (receiver) of the line
Z3, γ3, l3 - including the multiple reflections between the transitions from Z3 to Z2 and Z3 to R (receiver input)
- is described by T3f.

1 + r3 f
T3 f = ⋅e − γ 3 ⋅l3 (9-5)
1 − r32 ⋅r3 f ⋅e − 2⋅γ 3 ⋅l3

The superposition of the wave which is propagating from line Z3, γ3, l3 to line Z2, γ2, l2 is described by T32.

1 + r32
T32 = − 2⋅γ 3 ⋅l3
⋅r3 f ⋅e − 2⋅γ 3 ⋅l3 (9-6)
1 − r32 ⋅r3 f ⋅e

The superposition of the wave which is propagating from line Z2, γ2, l2 to line Z3, γ3, l3 is described by T23.

1 + r23
T23 = ⋅r2 n ⋅e − 2⋅γ 2 ⋅l2 (9-7)
1 − r2 n ⋅r23 ⋅e − 2⋅γ 2 ⋅l2

In consideration of all these reflections the voltage at the receiver is calculated by:

U receiver U h T2 f ⋅T3 f
= ⋅ (9-8)
U1 U 1 1 − T32 ⋅T23
Fig. 17 and 18 show the calculated voltage ratio for a cable screen with the same characteristics as in Fig. 7

7/13
with different dimensions of the screening case.

U receiver U receiver
40 20 ⋅log 10 in dB 40 20 ⋅log in dB
U1 U1

50 50

60 60

70 70

80 80
0 500 1000 1500 2000 2500 3000
0 500 1000 1500 2000 2500 3000

f/MHz f/MHz

Fig. 17: calculated voltage ratio including Fig. 18: calculated voltage ratio including
multiple reflections caused by the screening case, multiple reflections caused by the screening case,
quantities used: quantities used:
CT = 0,02 pF/m CT = 0,02 pF/m
MT= 0,4 nH/m MT= 0,4 nH/m
R = 50 Ω Z1 = 50 Ω εr1 = 2,3 R = 50 Ω Z1 = 50 Ω εr1 = 2,3
Z2 = 120 Ω εr2 = 1,1 l2 = 2 m Z2 = 120 Ω εr2 = 1,1 l2 = 2 m
Z3 = 90 Ω εr2 = 1,1 l3 = 0,03 m Z3 = 90 Ω εr2 = 1,1 l3 = 0,1 m
To avoid the disturbing reflections at the screening case one have to minimise the reflection factor r23 or (and)
r3f. A worthwhile solution in praxis is to design the screening case in a way that the characteristic impedance
Z3 is approximately of the same value as the input resistance of the receiver. In this case the reflection factor
r3f ≈0 and thus T3f=1, T32=0. That results in a voltage ratio which is equal to the ideal frequency response of
eq. (3-9).

10. Preview
In the field of digital data transmission and telecommunication the use of symmetrical multi pair/quad cables
increases. Therefore it is also necessary to determine their EMC behaviour.

The EMC behaviour of symmetrical cables depends on both, the unbalance attenuation aU of the pairs and the
screening attenuation aS of their screen. The sum of the unbalance attenuation aU of the pairs and the
screening attenuation aS of their screen is named coupling attenuation aC.
With the new triaxial measuring procedure explained above, also the coupling attenuation aC may be
measured, when the cable is driven in the differential mode, see figure 19. This procedure is still under
consideration [12].

cable under test symmetrical/

measuring tube asymmetrical
termination
generator

balun transformer receiver

Fig. 19: test set-up for the determination of the coupled power attenuation of a shielded twisted pair.

7/14
[1] Breitenbach O., Hähner T.
„Kabelschirmung im Übergang von MHz- zu GHz-Frequenzen“
NTZ Bd. 46 Heft 8 1993
[2] IEC 46A(Ger)62
[3] IEC 46A/320/CDV
[4] IEC-Standard 61196-1 (1995-05), 12. Screening effectiveness
[5] Halme, L., Szentkuti, B.
„The background for electromagnetic screening measurements of cylindrical screens“
Techn. Mitteilung PTT Nr. 3, 1988
[6] Eicher, B., Staeger, C., Szentkuti, B., Fahrni, H.
„Simple and accurate screening measurements on RF-cables up to 3 GHz“
Techn. Mitteilung PTT Nr. 4, 1988
[7] Meinke, H.
„Einführung in die Elektrotechnik höherer Frequenzen“
Springer Verlag 1961
[8] Klein, W.
„Die Theorie des Nebensprechens auf Leitungen“
Springer Verlag 1965
[9] Kaden, H.
„Die elektromagnetische Schirmung in der Fernmelde- und HF-Technik.“
Springer Verlag 1950
[10] Jungfer, H.
„Die Messung des Kopplungswiderstandes von Kabelabschirmungen bei hohen Frequenzen“
NTZ 1956 Heft 12
[11] Merz, C.
„Untersuchung des Einflusses verschiedener Störfaktoren bei der Messung der Schirmdämpfung mit
Hilfe des Triaxialen Meßverfahren“
Praktikumsbericht WS 95/96
FH Gießen-Friedberg
[12] Hähner T., Mund B.
EMV-Verhalten symmetrischer Kabel
emc journal 4/97

 1998 The Institution of Electrical Engineers

Printed and published by the IEE, Savoy Place, London WC2R 0BL, UK.

7/15