TECHNICAL

CISPR
REPORT
16-3
2003

AMENDMENT 2
2006-11

INTERNATIONAL SPECIAL COMMITTEE ON RADIO INTERFERENCE

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Amendment 2
Specification for radio disturbance and immunity
measuring apparatus and methods –
Part 3:
CISPR technical reports

© IEC 2006 Droits de reproduction réservés ⎯ Copyright - all rights reserved

International Electrotechnical Commission, 3, rue de Varembé, PO Box 131, CH-1211 Geneva 20, Switzerland
Telephone: +41 22 919 02 11 Telefax: +41 22 919 03 00 E-mail: [email protected] Web: www.iec.ch

Com mission Electrotechnique Internationale

International Electrotechnical Com m ission
PRICE CODE
X
Международная Электротехническая Комиссия
For price, see current catalogue
 –2– TR CISPR 16-3 Amend. 2 © IEC:2006(E)

FOREWORD

This amendment has been prepared by CISPR subcommittee A: Radio interference

measurements and statistical methods.

The text of this amendment is based on the following documents:

DTR Report on voting

CISPR/A/659/DTR CISPR/A/681/RVC
CISPR/A/662/DTR CISPR/A/678/RVC

Full information on the voting for the approval of this amendment can be found in the report on
voting indicated in the above table.

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
The committee has decided that the contents of this amendment and the base publication will
remain unchanged until the maintenance result date indicated on the IEC web site under
"http://webstore.iec.ch" in the data related to the specific publication. At this date, the publication
will be

• reconfirmed,
• withdrawn,
• replaced by a revised edition, or
• amended.

_____________

Page 7

3 Definitions

Add, on page 9, after 3.11, the following new definitions:

3.12
weighting (e.g. of impulsive disturbance)
the pulse-repetition-frequency (PRF) dependent conversion (mostly reduction) of a peak-
detected impulse voltage level to an indication which corresponds to the interference effect on
radio reception
NOTE 1 For the analog receiver, the interference effect is the psychophysical annoyance, i.e. a subjective quantity
(audible or visual, usually not a certain number of misunderstandings of a spoken text).
For the digital receiver, the interference effect may be defined by the critical Bit Error Ratio (BER) (or Bit Error
Probability (BEP)), for which perfect error correction can still occur, or by another objective and reproducible
parameter.

3.13
weighting characteristic
the peak voltage level as a function of PRF for a constant effect on a specific radio-
communication system, i.e., the disturbance is weighted by the radio communication system
itself
TR CISPR 16-3 Amend. 2 © IEC:2006(E) –3–

3.14
weighting function
weighting curve
the relationship between input peak voltage level and PRF for constant level indication of a
measuring receiver with a weighting detector, i.e. the curve of response of a measuring receiver
to repeated pulses

3.15
weighting factor
the value in dB of the weighting function relative to a reference PRF or relative to the peak value

3.16
weighting detector
detector which provides an agreed weighting function

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
3.17
weighted disturbance measurement
measurement of disturbance using a weighting detector

Page 10

4 Technical Reports

Add, after the existing subclause 4.7 published in Amendment 1, the following new subclauses
4.8 and 4.9:

4.8 Background material on the definition of the r.m.s.-average weighting detector

for measuring receivers

4.8.1 Introduction – purpose of weighted measurement of disturbance

Generally, a weighted measurement of impulsive disturbance serves the purpose of minimizing

the cost of disturbance suppression, while keeping an agreed level of radio protection. The
weighting of a disturbance for its effect on modern digital radiocommunication services is
important for the definition of emission limits that will protect these services. Amendment 1 of
CISPR 16-1-1 defines a detector that is a combination of an r.m.s. and an average detector. The
selection of the type of detector and of the transition between these detector functions is based
on measurements and theoretical investigations.

4.8.2 General principle of weighting – the CISPR quasi-peak detector

The effect on radiocommunication services depends on the type of interference (e.g. broadband
or narrowband, pulse rate etc.) and on the type of service itself. The effect of the pulse rate was
recognized a short time after the CISPR was founded in 1933. As a result, the quasi-peak
weighting receiver for the frequency range of 150 kHz to 1 605 kHz was defined as shown for
band B in Figure 4.8.1. However in CISPR 1 [1] it was already accepted that “Subsequent
experience has shown that the r.m.s. voltmeter might give a more accurate assessment” but the
quasi-peak type of voltmeter has been retained for certain reasons – mainly for continuity.
 –4– TR CISPR 16-3 Amend. 2 © IEC:2006(E)

34

32 30 MHz-1000 MHz (band C and D)

28
dB

24 0,15 MHz-30 MHz (band B)

Rel. input level for constant indication

20

16 9 kHz-150 kHz (band A)

12 43,5 dB
8

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
0

–4

–8

–12
Single pulse 1 10 100 Pulse rate 1 kHz
IEC 2010/06

Figure 4.8.1 – Weighting curves of quasi-peak measuring receivers

for the different frequency ranges as defined in CISPR 16-1-1.
The weighting factor is shown relative to a reference pulse rate (25 Hz or 100 Hz)

4.8.3 Other detectors defined in CISPR 16-1-1

• Peak detector

The peak detector follows the signal at the output of the IF envelope detector and holds the
maximum value during the measurement time (also called dwell time) until its discharge is
forced. This indication is independent of the pulse repetition frequency (PRF).

• Average detector

The average detector determines the linear average of the signal at the output of the IF envelope
detector. It should be kept in mind that for low PRFs, CISPR 16-1-1 specifies the average
detector measurement result as the maximum scale deflection of a meter with a time constant
specified for the quasi-peak detector. This is necessary to avoid reduced level indication for a
pulse modulated disturbance by using long measurement times. The weighting function varies
with 20 dB per decade of the PRF (see Figure 4.8.2).

• RMS detector

The r.m.s. detector determines the r.m.s. value of the signal at the output of the IF envelope
detector. Despite being mentioned in [1] and being described in CISPR 16-1-1, at the time of
writing of this report it has not been put to practical use in CISPR product standards. The
weighting function varies with 10 dB per decade of the PRF (see Figure 4.8.2). Up to now, no
meter time constant applies for the r.m.s. detector for intermittent, unsteady and drifting
narrowband disturbances.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) –5–

Comparison of detector weighting functions

(example for bands C and D with 120 kHz bandwidth)
70

60 Average
RMS
Quasi-Peak
50 Peak
Weighting factor/dB

40

30

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20

10

0
1 10 100 1 000 10 000 100 000 1 000 000

fp/Hz
IEC 2011/06

Figure 4.8.2 – Weighting curves for peak, quasi-peak, r.m.s. and linear average detectors
for CISPR bands C and D

4.8.4 Procedures for measuring pulse weighting characteristics of digital

radiocommunications services

All modern radio services use digital modulation schemes. This is not only true for mobile radio
but also for audio and TV. Procedures for data compression and processing of analog signals
(voice and picture) are used together with data redundancy for error correction. Usually, up to a
certain critical bit-error ratio (BER) the system can correct errors so that perfect reception
occurs.

Whereas analog radio systems require signal-to-noise ratios of as much as 50 dB for satisfactory
operation, in general, digital radio communication systems allow error-free operation down to
signal-to-noise ratios of approximately 10 dB. However the transition region from error-free
operation to malfunction is small. Therefore planning guidelines for digital radio are based on
almost 100 % coverage. When a digital radio receiver operates at low input levels, the
susceptibility to radio disturbance is important. In mobile radio reception, the susceptibility to
radio disturbance is combined with the problem of multi-path propagation.

4.8.4.1 Principles of measurement

The significance of the weighting curve for band B in Figure 4.8.1 is as follows: to a listener the
degradation of reception quality, caused by a 100-Hz pulse, is equivalent to the degradation from
a 10-Hz pulse, if the pulse level is increased by an amount of 10 dB. In analogy to the above, an
interference source with certain characteristics will produce a certain BER, e.g. 10 –3 in a digital
radiocommunication system, when the interfering signal is received in addition to the radio
signal. The BER will depend e.g. on the pulse repetition frequency (PRF) and the level of the
interfering signal. In order to keep the BER constant, the level of the interfering signal will have
to be readjusted while the PRF is varied. This level variation vs. PRF determines the weighting
characteristics. Measurement systems with BER indication are needed to determine the required
level of the interfering signal for a constant BER as e.g. shown in Figure 4.8.3.
 –6– TR CISPR 16-3 Amend. 2 © IEC:2006(E)

BER

Radio signal
generator

Radio
receiver

Interference
source
IEC 2012/06

Figure 4.8.3 – Test setup for the measurement of the pulse weighting characteristics
of a digital radiocommunication system

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
The test setup shown in Figure 4.8.3 consists of a radio signal generator that transmits the
wanted radio signal to the receiver. For the determination of the BER, the radio receiver either
has to know the original bit sequence for comparison with the detected bit sequence or the latter
must be looped back to the radio signal generator for comparison with the original. Both systems
are available and have been used for tests. Mobile radio testers, e.g., apply the loop-back
principle.

4.8.4.2 Generation of the interference signal

A signal generator with pulse-modulation capability can be used to generate the interference
signal. For correct measurements, the pulse modulator requires a high ON/OFF ratio of more
than 60 dB. Using the appropriate pulse width, the interference spectrum can be broadband or
narrowband, where the definition of broadband and narrowband is relative to the communication
channel bandwidth. Figure 4.8.4 gives an example of an interference spectrum used for the
determination of weighting characteristics.

* RBW 9 kHz Marker 1 [T1]

VBW 30 kHz 61,89 dBμV
Ref. 90 dBμV * Att. 0 dB SWT 3,1 s 128 000 000 000 MHz
90

80 B
1 PK*
CLRWR 70
1
60

50
PRN
40

30

20

10

–10
Center 128 MHz 5 MHz Span 50 MHz
IEC 2013/06

Figure 4.8.4 – Example of an interference spectrum: pulse modulated carrier

with a pulse duration of 0,2 μs and a PRF < 10 kHz
TR CISPR 16-3 Amend. 2 © IEC:2006(E) –7–

With increasing pulse duration, the main lobe of the spectrum becomes narrower. This is also
used to study the effect of narrowband pulses on radiocommunication systems. The advantage
of using a band-limited pulse spectrum instead of a broadband pulse generator is to avoid
overloading the receiver under test. Otherwise non-linearity effects could cause deterioration of
the weighting characteristics. In addition to pulse-modulated carriers, unmodulated carriers can
be used to determine the sensitivity of different systems to narrowband (CW signal) EMI.

Extensive measurements have also been presented in [2] with on/off-keying of a QPSK-
modulated signal, thus keeping the spectrum width wider than the system bandwidth even with
longer pulse durations. Since actual receivers do not provide BER indication, the method
described in the ITU Recommendation 1368 was used as the failure criteria: DVB-T reception
was regarded as distorted when more than one visible erroneous block was shown on the screen
within an observation period of 20 s. Alternatively, any picture-freeze, also for short periods, was
regarded as a failure. For DRM, the reception was considered as distorted when the system
showed more than one dropout in a 20 s observation time.

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Further measurements have been made with spread-spectrum modulated carriers in order to
study the effect of spread-spectrum clock interference on wideband radiocommunication services
(see [3] and [4]).

Table 4.8.1 – Overview of types of interference used in the experimental study

of weighting characteristics

Interference signals Pulse-modulated On/Off-keyed QPSK- Spread-spectrum

modulated modulated

Pulse width in relation T < 1/B to 100/B T < 1/B to 100/B Continuous
to signal bandwidth

T = pulse width, B = radio signal bandwidth

4.8.4.3 Other principles of measurement

The receiver under test should receive a signal that is just sufficient to give quasi error-free
reception (e.g. a BER = 10 –7 or a factor of 10 –3 lower than the critical BER). Thus the receiver
operates like a receiver at the rim of a coverage area, where a disturbance above the emission
limit can easily cause interference.

For radio telephone systems, where the downlink (to the mobile) and uplink (to the base station)
frequencies are in different bands, the use of a pulse modulated carrier helps to concentrate the
interference on the mobile receiver and thus avoids interference with the loop-back connection.

4.8.5 Theoretical studies

The work of developing measurement procedures considering a digital radio receiver as a
disturbance victim, is a very complex problem since there are many different modulation and
coding schemes to consider as digital communication services are undergoing rapid
development. The results of theoretical studies for radio systems using error correction have
been presented in [5] and [6]. These studies are based on the same fundamental assumptions
that are explained above:

• the BER is the performance parameter of interest for the digital communication system;
• the repetitive pulsed disturbance is the waveform of particular interest;
• the disturbance pulses have a pulse duration that is short compared to the digital symbols
transmitted.

Results for some selected convolutional codes (for more details, see [5]):
 –8– TR CISPR 16-3 Amend. 2 © IEC:2006(E)

A convolutional code is generated by passing the information sequence through a linear finite-
state shift register. In general, the shift register consists of K stages and n algebraic function
generators. The input data to the channel encoder is shifted into and along the shift register k
bits at a time. The number of output bits for each k-input sequence is n bits. The rate R of the
code is defined as n/k. The parameter K is called the constraint length of the convolutional code.
In Figures 4.8.5 a) and b) as well as 4.8.6 a) and b) the r.m.s. and peak values corresponding to
a constant BER of 10 –3 are shown for different convolutional codes and binary phase shift keying
(BPSK) modulation. These results have been simulated with ACOLADE© 1) (Advanced
Communication Link Analysis and Design Environment). In the graphs, the pulse repetition
frequency of the disturbance is presented as related (normalized) to the gross-bit rate (or symbol
rate) R s of the communication system. The simulation is done in the band-pass domain. This
means that the results can be transformed to an arbitrary carrier frequency. The disturbance
pulse width is 10 % of the bit duration time. For the lowest rate R = ¼, the r.m.s. value is
approximately constant down to the critical point where it increases rapidly. Thus, for a well-
protected system, the r.m.s. value corresponding to a constant BER is constant with respect to

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
the pulse repetition frequency of the repetitive pulsed disturbance.

50 50

40 40
dBV

dBV
BEP[dBV]

BEP [dBV]
constant BEP

constant BEP

30 30
for constant

for constant

20 20
value for

value for
RMSvalue

RMSvalue

10 10
Uncoded
RMS

RMS

Conv. R=2/3, K=3

Conv. R=1/2, K=3 Uncoded
0 Conv. R=1/2, K=3
Conv. R=1/4, K=3 0
Conv. R=1/2, K=5

−10 −3
10 10
−2
10
−1
10
0 1
10 −10 −3 −2 −1 0 1
Normalized Pulse Repetition Frequency 10 10 10 10 10
Normalized pulse repetition frequency IEC 2014/06 Normalized Pulse Repetition Frequency
Normalized pulse repetition frequency IEC 2015/06

Figure 4.8.5 a) – The r.m.s. level for constant

BEP for three K=3, convolutional codes of Figure 4.8.6 a) –The r.m.s. level for constant
different rate BEP for two rate ½, convolutional codes

120
120
Uncoded Uncoded
Conv. R=2/3, K=3 Conv. R=1/2, K=3
Conv. R=1/2, K=3 Conv. R=1/2, K=5
100
Peak value for constant BEP dBV

100 Conv. R=1/4, K=3

dBV

Peak value for constant BEP [dBV]

BEP[dBV]
constantBEP

80 80
for constant

60 60
value for
Peakvalue

40 40
Peak

20 20

0 −3 −2 −1 0 1
0 −3 −2 −1 0 1
10 10 10 10 10 10 10 10 10 10
Normalized Pulse Repetition Frequency Normalized Pulse Repetition Frequency
Normalized pulse repetition frequency Normalized pulse repetition frequency IEC 2017/06
IEC 2016/06

Figure 4.8.5 b) – The peak level for constant Figure 4.8.6 b) –The peak level for constant BEP
BEP for three K=3, convolutional codes of for two rate ½, convolutional code
different rate
—————————
1) ACOLADE© is an example of a suitable product available commercially. This information is given for the
convenience of users of this Technical Report and does not constitute an endorsement by IEC of this product.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) –9–

The results in Figure 4.8.5 show the following: above the symbol rate R s , the weighting
characteristic follows the r.m.s. value of the impulsive signal that causes the interference.
Below R s , the weighting characteristic depends on the amount of coding: for the uncoded
signal, the peak value increases with less than 10 dB per decade as the PRF decreases. With
better coding, the part of the weighting characteristic with flat response becomes shorter.
Therefore, it is important to characterize real radiocommunication systems in order to obtain
meaningful results.

4.8.6 Experimental results

The methods described in 4.8.4 have been used for the measurement results in this part. The
test signals are described where necessary.

4.8.6.1 Weighting in band A

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
For band A, i.e. below 150 kHz, no measurement results of digital radiocommunication
systems are available.
NOTE Weighting of radio disturbance generally requires a consideration of intermittent, unsteady and drifting
narrowband disturbances. Therefore the concept of defining a corner frequency, below which the average detector
becomes effective has been applied to band A as well, using the corner frequency proposed for band B, since the
original CISPR specification of the r.m.s. detector does not apply a meter time constant.

4.8.6.2 Weighting in band B

Weighting of interference to the Digital Radio Mondial (DRM) Broadcast System

At the World Radio Conference (WRC) in June 2003, the new Digital Radio Mondial was
officially started. During the four week duration of the conference, a great number of special
DRM transmissions became available from many radio stations. The measurements reported
below, were taken on 8 July, 2003, when a great number of transmissions were still available.

DRM uses OFDM (Orthogonal Frequency Division Multiplex) with 200 carriers. The occupied
bandwidth of each transmission is 10 kHz. In addition to the digitized audio signal, a certain
amount of data (radio station information etc.) is transmitted. A conventional AM receiver can
be used to downconvert the signal to an IF of 12 kHz, which is then decoded using a digital
signal processor and a special DRM software radio.

During the time of measurement, the radio stations in table 4.8.2 were received at the station
near Munich, with amateur dipole antennas mounted on the roof with a higher receive input
voltage (50 to 60 dBµV) than required for the experiment.
 – 10 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

Table 4.8.2 – DRM radio stations received for the measurement

of the weighting characteristics

Frequency Beam Target Av. DRM power Program Transmit site

kHz kW
5975 060 W Europe 40 T-Systems Jülich
Media Broadcast
6095 ND Europe 35 RTL/music and Junglinster,
short Luxembourg
announcements
6140 ND W & C Europe 40 DW English Jülich
7320 105 W & C Europe 33 BBCWS Rampisham
13605 037 C Europe 6 IBB/R. Sawa Morocco
15440 040 W & C Europe 80 DW English Sines

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
“W & C” means West and Central (Europe)

The various transmissions were available for 1h or 2 h . The measurement results (weighting
characteristics) were essentially the same for all frequencies, even if the amount of data
transmitted in addition to the audio signal was not the same. Time dependent fading of the
input signal had to be compensated for manually using a step attenuator that was inserted in
the antenna connection, see Figure 4.8.7.

Principally the same type of interference signal was generated as in Figure 4.8.4. However,
for a signal with an occupied bandwidth of 10 kHz, it is possible to use a longer pulse duration
(10 μs o r more) .

Step Attenuator
Antenna for manual
fading compensation

IF Out
AM Rec. PC

Sig. Gen.

With pulse
modulation IEC 2018/06

Figure 4.8.7 – Test setup for the measurement of weighting curves for Digital Radio
Mondial (DRM). The received signal was downconverted to an IF of 12 kHz for decoding
by special hard and software in a personal computer (PC)

Since no indication of BER was available, the “Audio” status indication on the PC (DRM
software radio display) was used as a criterion. As soon as the interference becomes too
high, the “Audio” status indication will turn from green to red.

As explained earlier, the signal level is attenuated so that the reception quality is just enough.
The weighting characteristic (see Figure 4.8.8) shows a 10 dB/decade increase of the
interference signal for PRFs between 1 kHz and 5 Hz. The nonlinearities are mainly due to
uncompensated fading of the input signal. A detailed weighting curve is shown for a pulse
width of 10 μs. For higher pulse widths, the weighting curve was measured only at three
(resp. four) points to verify the 10 dB/decade behaviour. Below a PRF of 5 Hz, the weighting
curve rises suddenly. And below about 2 Hz, the signal cannot be disturbed by the pulse
width of 500 μs. However lightning strokes are reported to generate longer dropouts, which
indicates that longer clicks might cause such dropouts as well.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 11 –

DRM at 5,975 MHz; 6,095 MHz; 6,140 MHz; 7,320 MHz; 13,605 MHz;
data rate 20,9 kBit/s; signal level kept at constant SNR

70
width 1E-05s
60 width 5E-05s
width 1E-04s
50 width 5E-04s

40
dBμV

30

20

10

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
0
1 10 100 1 000 10 000
fp/Hz IEC 2019/06

Figure 4.8.8 – Weighting characteristics for DRM signals for various pulse widths
of the pulse-modulated carrier. Since the DRM signals are actual radio signals,
the exact modulation scheme is not known

The report [2] describes the following DRM signals and two receiver types for the
measurements:

• Mode B, Modulation 16/64QAM, Interleave 2 sec, protection level 1 / 0

• Mode B, Modulation 16/64QAM, Interleave 2 sec, protection level 0 / 0

The interference signal for Figures 4.8.9 and 4.9.10 is a pulse-modulated carrier with
additional QPSK modulation in order to generate a wide bandwidth of the interference
spectrum as explained in 4.8.4.2.

DRM Mode B 16/64QAM prot. level 0: Trend

140 dBμV

200 ns
1 μs
120 dBμV
10 μs
100 μs
1 ms
100 dBμV
10 ms
Pulse level (Pk)

80 dBμV

60 dBμV

40 dBμV

20 dBμV

0 dBμV
1 Hz 10 Hz 100 Hz 1 000 Hz 10 000 Hz
Pulse repetition frequency IEC 2020/06

Figure 4.8.9 – Weighting characteristics for DRM protection level 0:

average of results for two receivers
 – 12 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

DRM Mode B 16/64QAM prot. level 1: Trend

140 dBμV

200 ns
120 dBμV 1 μs
10 μs
100 μs
100 dBμV 1 ms
10 ms
Pulse level (Pk)

80 dBμV

60 dBμV

40 dBμV

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20 dBμV

0 dBμV
1 Hz 10 Hz 100 Hz 1 000 Hz 10 000 Hz
Pulse repetition frequency
IEC 2021/06

Figure 4.8.10 – Weighting characteristics for DRM protection level 1:

average of results for two receivers

The weighting characteristics in Figures 4.8.9 and 10 show a 10 dB/decade slope down to
approx. 100 Hz. Since there is no other digital radio system in band B, the corner frequency of
the proposed RMS/AV detector between r.m.s. and linear average detection for this frequency
band can only be based on the results of DRM (see 4.8.7). A corner frequency of 10 Hz is
therefore proposed for band B as a compromise between the two results.

4.8.6.3 Weighting in bands C/D

4.8.6.3.1 Weighting of impulsive interference to Digital Video Broadcast Terrestrial

(DVB-T)

• Test setup

One test setup for DVB-T consists of a DVB-T signal generator and a DVB-T measuring
receiver. The components are connected via coaxial cables. The interference signal (a pulse-
modulated carrier, see Figure 4.8.4 for an example of the spectrum) is fed into the signalling
connection via a combiner.

The parameters used are the following.

DVB-T uses COFDM (Coded Orthogonal Frequency Division Multiplex) with 6817 (8k) or 1705
(2k) carriers. The OFDM carriers may be modulated either with QPSK (Quadrature Phase
Shift Keying) or with 64 QAM (Quadrature Amplitude Modulation), resp. 16 QAM. QAM is
preferred to QPSK as QAM allows higher data transfer rates. The transmission code rate CR
is defined by CR = number of information bits/(number of information bits + error protection
bits). Values of CR = 2/3 and 3/4 are used in actual systems. Each COFDM symbol is
followed by a guard interval GI which is GI = 1/8 in actual systems. The DVB-T modulation
and coding system allows many combinations, of which only a few are relevant. Therefore the
parameters used in systems operating in some European countries have been selected.
These allow transmission rates between 14,745 Mbit/s and 24,88 Mbit/s (see Table 4.8.4)
depending on modulation and code rate. Different coders and decoders are used in the
system. The bit-error ratio (BER) reading can be taken before the Viterbi decoder as well as
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 13 –

before and after the Reed Solomon decoder of the measuring receiver. A comparison is given
in Table 4.8.3. The transmission level is set so that the BER after the Reed Solomon decoder
without interference is just below 10 –8 . This results in different signal levels depending on the
system parameters. The interference levels have then been adjusted to a critical value of BER
= 2,0*10 –4 before the Reed Solomon decoder.

For the BER measurement, the modulator generates a Pseudo Random Binary Sequence
(PRBS) as data stream. The evaluation of the data stream is done in the receiver in two
different procedures. The BER before Viterbi and before Reed-Solomon is evaluated by
correlation. Flags in the bit stream are used to determine the BER after Reed-Solomon. If the
decoder does not recognize a flag as correct, the following bit combination is determined to
be false.

The relationship in Table 4.8.3 was found experimentally between the bit error ratios before
and after the Viterbi and Reed Solomon decoders for two pulse rates.

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Table 4.8.3 – Comparison of BER values for the same interference level

Pulse rate 10 k 500 k

Hz
BER before Viterbi decoder 1,5*10 –2 4,4*10 –3
BER before Reed Solomon 2,0*10 –4 2,0*10 –4
BER after Reed Solomon 1,0*10 –6 1,0*10 –8

So, the results with BER measured before Reed Solomon (with 2,0*10 –4 ) and after Reed
Solomon (with 1,0*10 –6 ) are roughly comparable.

Table 4.8.4 – Transmission parameters of DVB-T systems used in various countries

Country Modulation Code rate Guard interval Transfer rate

France/UK 64QAM 2k 3/4 1/8 24,88 Mbit/s
Spain 64QAM 8k 3/4 1/8 24,88 Mbit/s
Germany 16QAM 8k 2/3 1/8 14,745 Mbit/s

The measurement results are presented in Figures 4.8.11, 4.8.12 and 4.8.13. In all tests, the
interference signal leading to these results are pulse-modulated carriers.
 – 14 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

–4
DVB-T f = 500 MHz, 64 QAM 2k, CR 3/4, GI 1/8, BER before RS = 2 × 10 ,
-61,5 dBm, 24,88 Mbit/s (FR, UK)
140

width 0,1E-06s
120 width 0,2E-06s
width 0,5E-06s
width 1,0E-06
100 width 2,0E-06
width 5,0E-06s
width 10E-06
80
dBμV

60

40

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20

0
1 10 100 1 000 10 000 100 000 1 000 000 10 000 000
fp/Hz IEC 2022/06

Figure 4.8.11 – Weighting characteristics for DVB-T with 64QAM 2k, CR 3/4
(as used in France and United Kingdom)
–4
DVB-T f = 500 MHz, 64 QAM 8k, CR 3/4, GI 1/8, BER before RS = 2 × 10 ,
-61,7 dBm, 24,88 Mbit/s (ES)
140

120
width 0,1E-06s
width 0,2E-06s
100 width 0,5E-06s
width 1,0E-06
width 2,0E-06
width 5,0E-06s
80 width 10E-06
dBμV

60

40

20

0
1 10 100 1 000 10 000 100 000 1 000 000 10 000 000
fp/Hz IEC 2023/06

Figure 4.8.12 – Weighting characteristics for DVB-T with 64QAM 8k, CR 3/4
(as used in Spain)
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 15 –

–4
DVB-T f = 500 MHz, 16 QAM 8 k, CR 2/3, GI 1/8, BER before RS = 2 × 10 ,
-61,8 dBm, 14,745 Mbit/s (DE)
140

width 0,1E-06s
120 width 0,2E-06s
width 0,5E-06s
width 1,0E-06
100 width 2,0E-06
width 5,0E-06s
width 10E-06
80
dBμV

60

40

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20

0
1 10 100 1 000 10 000 100 000 1 000 000 10 000 000
fp/Hz IEC 2024/06

Figure 4.8.13 – Weighting characteristics for DVB-T with 16QAM 8k, CR 2/3
(as used in Germany)

A number of 6 different receiver types were tested in report [2] for DVB-T with 16QAM 8k, CR
2/3 and for DVB-T with 64QAM 8k, CR 2/3. To get receiver independent results, the individual
characteristics were combined using average values inside the range where all receivers
offered a result. Excluded were two receivers in PRF ranges, where they showed a non-
typical behavior. These combined results are shown in the “trend” characteristics in
Figures 4.8.14 and 4.8.15. The interference signal for both figures is a pulse-modulated
carrier with additional QPSK modulation in order to generate bandwidth of the interference
spectrum at least as wide as the DVB-T signal spectrum as explained in 4.8.4.2.
 – 16 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

DVB-T 8k 16QAM 2/3: Trend

140 dBμV
50 ns
200 ns

120 dBμV 0,5 μs

1 μs
5 μs
100 dBμV 10 μs
Pulse level (Pk)

80 dBμV

60 dBμV

40 dBμV

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20 dBμV

0 dBμV

10 Hz 100 Hz 1 000 Hz 10 000 Hz 100 000 Hz

Pulse repetition frequency IEC 2025/06

Figure 4.8.14 – Average weighting characteristics of 6 receiver types

for DVB-T with 16QAM

DVB-T 8k 64QAM 2/3: Trend

120 dBμV
50 ns
200 ns
0,5 μs
100 dBμV 1 μs
5 μs
10 μs
Pulse level (Pk)

80 dBμV

60 dBμV

40 dBμV

20 dBμV

0 dBμV
10 Hz 100 Hz 1 000 Hz 10 000 Hz 100 000 Hz
Pulse repetition frequency IEC 2026/06

Figure 4.8.15 – Average weighting characteristics of 6 receiver types

for DVB-T with 64QAM
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 17 –

• Interpretation of the results

In Figure 4.8.11, the corner frequency can only be assumed to be approx. 100 Hz, whereas in
Figures 4.8.12 and 4.8.13, the corner frequencies can clearly be seen. They however depend
on the interference pulse width as in Figures 4.8.11 and 4.8.13. Since all weighting curves are
given for the shortest pulse (see Figure 4.8.1), also for the corner frequency, the shortest
pulse is always relevant. The system used in Germany shows the most robust performance
against impulsive interference due to its lower code rate and 16QAM 8k modulation.

4.8.6.3.2 Weighting of impulsive interference to other digital radiocommunication

systems operating in CISPR bands C and D

• Digital Audio Broadcasting (DAB)

DAB operates in the VHF (174 MHz to 230 MHz) and the L (1 452 MHz to 1 492 MHz) bands

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
with a bandwidth of 1,5 MHz per channel using Coded Orthogonal Frequency Division
Multiplex (COFDM) to minimise multipath fading. The audio signal data rate is reduced by
MUSICAM (a masking pattern adapted for Universal Coding and Multiplexing), which is a part
of the MPEG-2 (Moving Picture Expert Group) standard. The total transmitted bit rate is 2,4
Mbit/s. The 1500 subcarriers are modulated using Differential QPSK (DQPSK). The weighting
characteristics in Figure 4.8.16 were measured using a test version of a DAB receiver.
Weighting characteristics of commercial DAB receivers have been presented in report [2].

–4
DAB DQPSK BER = 1,0 × 10
120

width 0,5E-06s
100
width 1,0E-06s
width 2,0E-06s
width 5,0E-06s
80 width 10,0E-06s
dBμV

60

40

20

100 1 000 10 000 100 000 1 000 000

f/Hz IEC 2027/06

Figure 4.8.16 – Weighting characteristics for DAB (signal level -71 dBm)
with a flat response down to approximately 1 kHz
 – 18 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

DAB (QPSK 2/3): Trend

120 dBμV
50 ns
200 ns
0,5 μs
100 dBμV 1 μs
5 μs
10 μs

80 dBμV
Pulse level (Pk)

60 dBμV

40 dBμV

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20 dBμV

0 dBμV
10 Hz 100 Hz 1 000 Hz 10 000 Hz 100 000 Hz

Pulse repetition frequency IEC 2028/06

Figure 4.8.17 – Weighting characteristics for DAB:

average of two different commercial receiver types

The differences between the results in Figures 4.8.16 and 4.8.17 are possibly due to the
different types of the impulsive signal: for Figure 4.8.16 a simple pulse-modulated carrier was
used, whereas for Figure 4.8.17 an on/off-keyed QPSK-modulated signal was used.

• Terrestrial Trunked Radio (TETRA) system

TETRA is used in workshops, the building and construction industries, airports,

transportation/trucking and safety services. It operates in the frequency range 380 MHz to
520 MHz (in some areas also in 870 MHz to 990 MHz) with a data rate of 36 kbit/s per carrier,
an occupied bandwidth of ≈ 25 kHz and channel separations of 12,5, 20 or 25 kHz. Speech
data reduction is done using Algebraic Code Excited Linear Prediction (ACELP) to 4,8 kbit/s
per traffic channel. Up to four traffic channels are normally transmitted on one carrier. The
error protection may be high or low, depending on the code rate. The modulation procedure is
π/4-DQPSK. Figure 4.8.18 shows the measured weighting characteristics for a high code rate
= 1 (low error protection).
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 19 –

TETRA downlink f = 394,0 MHz, BER = 2 %

140

width 0,1E-06s
120 width 0,5E-06s
width 1,0E-06s
width 5,0E-06s
100 width 10,0E-06s

80
dBμV

60

40

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20

0
10 100 1 000 10 000 100 000 1 000 000 10 000 000
f/Hz IEC 2029/06

Figure 4.8.18 – Weighting characteristics for TETRA (signal level – 80 dBm)

for a code rate of 1

Since the pulse spectrum is much wider than the channel bandwidth, all weighting
characteristics are separated by the PRF ratio in dB. Above a PRF of 10 kHz, the slope of
curves is 20 dB/decade, corresponding to the increase of the voltage of the center line of the
interference spectrum. Therefore the weighting characteristics below 10 kHz PRF should be
regarded as relevant.

• Global System for Mobile Communication (GSM)

This digital cellular telecommunication system operates in the 900 MHz (GSM 900) and
1 800 MHz (GSM 1800) frequency bands. The offset between uplink (mobile to base station)
and downlink is 45 MHz (GSM 900) and 95 MHz (GSM 1800) respectively. The occupied
bandwidth is 300 kHz and channel spacing is 200 kHz. Modulation for constant spectrum
envelope is achieved with Gaussian Minimum Shift Keying (GMSK). The error correction
mechanisms applied are different for traffic channels (1b bits) and other bits (Class 2 bits).
Therefore different bit error rates apply: BER, RBER 1b and 2 (residual BER) and FER
(Frame error rates). The test setup and signals of Figures 4.8.3 and 4.8.4 have been used,
with a mobile communication tester as a signal source.
 – 20 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

GSM 900 Type 1 downlink f = 947,4 MHz RBER 1b = 0,4 % 400 frames -90 dBm
120

width 0,1E-06s
100 width 0,5E-06s
width 1,0E-06s
width 2,0E-06s
width 5,0E-06s
80 width 10,0E-06s

60
dBμV

40

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20

0
100 1 000 10 000 100 000 1 000 000 10 000 000
f/Hz
IEC 2030/06

Figure 4.8.19 – Weighting characteristics for RBER 1b of GSM (signal level –90 dBm)

GSM 900 Type 1 downlink f = 947,4 MHz RBER 2 = 2 % 400 frames -90 dBm
120

width 0,1E-06s
width 0,5E-06s
100 width 1,0E-06s
width 2,0E-06s
width 5,0E-06s
80 width 10,0E-06s
dBμV

60

40

20

0
100 1 000 10 000 100 000 1 000 000 10 000 000
f/Hz IEC 2031/06

Figure 4.8.20 – Weighting characteristics for RBER 2 of GSM

The characteristics typically rise at 10 dB/decade between 100 kHz and 2 kHz with a steeper
slope below about 2 kHz PRF. Unfortunately measurements below a PRF of 1 kHz were not
possible due to instability of the test system. The results shown in Figures 4.8.19 and 4.8.20
are very similar to the BER and RBER 1b curves of Figure 4.8.21 similar to those published in
[7] and [8] using the simulation software COSSAP. The values obtained in Figure 4.8.21 have
been calculated assuming a pulse-modulated carrier with a pulse duration of 2 μs as the
interference signal.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 21 –

C/I Improvement for τ = 10 μs

70
BER
60 FER
RBER 1b
RBER 2
50

40
dB

30

20

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
10

0
100 1 000 10 000 100 000
f/Hz IEC 2032/06

Figure 4.8.21 – Carrier-to-interference improvements with decreasing PRF in dB

computed for GSM using COSSAP

• Frequency Modulation (FM) Broadcast System

Based on the assumption that FM broadcast will survive past the transition from analog to
digital radio systems for some time, measurements have been made based on the methods of
report [2] resulting in Figure 4.8.22. The FM signal contained a pilot carrier only; the increase
of noise due to the interference was measured in the demodulated signal. The interference
signal is a pulse-modulated carrier with additional QPSK modulation in order to generate
bandwidth of the interference spectrum at least as wide as the FM signal spectrum as
explained in 4.8.4.2.

2 µs pulses

60

50
QP
Pulse level dBμV

40

RMS
30

20

10

0
100 1 000 10 000 100 000
Pulse repetition frequency Hz IEC 2033/06

Figure 4.8.22 – RMS and quasi-peak values of pulse level

for constant effect on FM radio reception
 – 22 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

Figure 4.8.22 is not a weighting characteristic! It shows that the r.m.s. value of the pulse level
with 2 μs width is closer to being constant than the quasi-peak value. This has been shown
for other pulse widths as well but is not presented here for reasons of space.

4.8.6.4 Weighting for Band E (1 through 18 GHz)

• GSM system

The weighting characteristics found for a mobile operating in the 1 800 MHz (GSM 1800)
frequency band is very similar to the system operating in the 900 MHz (GSM 900) frequency
band (compare Figure 4.8.23 with Figures 4.8.19 through 4.8.21). The offset between uplink
(mobile to base station) and downlink is 95 MHz for GSM 1800. As in Figures 4.8.19 through
4.8.21, the curves are rising below 2 kHz PRF with a slope of more than 20 dB/decade.

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
GSM 1800 Type 2 downlink f = 1850,8 MHz RBER 1b = 0,4 % 400 frames -90 dBm

120

width 0,1E-06s
100 width 0,5E-06s
width 1,0E-06s
width 2,0E-06s
width 5,0E-06s
80 width 10,0E-06s

60
dBμV

40

20

0
100 1 000 10 000 100 000 1 000 000 10 000 000
f/Hz IEC 2034/06

Figure 4.8.23 – Weighting characteristics for RBER 1b of GSM

(signal level –90 dBm)

• Digitally Enhanced Cordless Telephone (DECT) system

DECT is used in homes and offices for distances up to 300 m (in picocells). It provides 10
channels spaced 1,728 MHz apart in the frequency range 1,88 to 1,90 GHz. The occupied
bandwidth is ≈ 1,5 MHz. For speech data reduction Adaptive Differential Pulse Code
Modulation (ADPCM) is used. Modulation is done with Gaussian Mean Shift Keying (GMSK).
The data stream for testing is Pseudo Random Binary Sequence (PRBS).
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 23 –

DECT FP f = 1897,344 MHz, BER = 2 %, Evaluation time = 5,0 s

120

width 0,1E-06s
100 width 0,5E-06s
width 1,0E-06s
width 2,0E-06s
width 5,0E-06s
80 width 10,0E-06s
dBμV

60

40

20

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
0
100 1 000 10 000 100 000 1 000 000 10 000 000
f/Hz IEC 2035/06

Figure 4.8.24 – Weighting characteristics for DECT (signal level –83 dBm)

The weighting characteristics for DECT show a response near 10 dB/decade in the range
between 50 kHz and 500 kHz PRF in the upper PRF areas for narrow pulses and a steep
slope below about 10 kHz PRF. Only for longer pulse widths, the weighting characteristic is
flat.

• Code Division Multiple Access (CDMA) systems IS-95 and J-STD 008

IS-95/J-STD 008 have been specified by TIA (US Telecommunications Industry Association)
and are used in the frequency ranges 825 MHz to 900 MHz (IS-95) and 1,8 GHz to 2,0 GHz.
The occupied bandwidth is ≈ 1,4 MHz (3 dB: 1,23 MHz). The modulation is done with
Quadrature Phase Shift Keying (QPSK). For the uplink (mobile to base station) the optimum
setting of the receive power at the base station is controlled via power control bits.

IS-95 forward f = 878,49 MHz, FER = 2 %, -83 dBm, full rate

140

width 0,1E-06s
120 width 0,5E-06s
width 1,0E-06s
width 2,0E-06s
100 width 5,0E-06s
width 10,0E-06s

80
dBμV

60

40

20

0
100 1 000 10 000 100 000 1 000 000 10 000 000
f/Hz IEC 2036/06

Figure 4.8.25 – Weighting characteristics for IS-95 (signal level -97 dBm)
with comparatively high immunity to interference
 – 24 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

J-STD 008 forward f = 1955,0 MHz FER = 2 % full rate -97 dBm
140

width 0,1E-06s
120 width 0,5E-06s
width 1,0E-06s
width 2,0E-06s
100 width 5,0E-06s
width 10,0E-06s

80
dBμV

60

40

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
20

0
100 1 000 10 000 100 000 1 000 000 10 000 000
f/Hz IEC 2037/06

Figure 4.8.26 – Weighting characteristics for J-STD 008 (signal level –97 dBm)

• 3 rd Generation Digital Radiocommunication Systems

Two different systems have been investigated:

– Wideband CDMA (W-CDMA), which is going to be deployed in Europe, and

– CDMA2000, which is mainly going to be applied in North America and some other areas.

Tests have been made on both systems. However at the time of testing, available mobile
phones for W-CDMA did not give stable BER results in the test setup (loop back) with the
mobile testers. So, only results for cdma2000 are available now. Results for W-CDMA will
certainly become available at a later date. They have been used later with success for
evaluating the interference effect to spread-spectrum clock signals (see [3] and [4]).

CDMA2000 as described by Third Generation Partnership 2 (3GPP2) is an access method

intended for use in the IMT-2000 proposal for Third Generation (3G) cellular telephone
systems. The system is based on spread-spectrum codes and provides high and variable data
rates. It is an evolutionary development of IS-95 (cdmaOne) which is also based on Code
Domain Multiple Access (CDMA) to the air interface. This means that the individual channels
are separated from each other by individual codes. The basic chip rate is 1,2288 MHz. All
IMT-2000 compatible systems feature transmitted bit rates of up to 384 kb/s up to a cruising
speed of 500 km/h, in urban areas up to 120 km/h.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 25 –

CDMA2000 forward f = 1955 MHz; FER = 0,5 %; data rate 9,6 kBit/s; signal level -112 dBm

100
90 width 0,1E-06s
width 0,2E-06s
80 width 0,5E-06s
70 width 1,0E-06s
width 2,0E-06s
60
dBμV

50
40
30
20
10
0

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
100 1 000 10 000 100 000 1 000 000
f/Hz IEC 2038/06

Figure 4.8.27 – Weighting characteristics for the Frame Error Ratio (FER) of cdma2000
(measured at a receive signal level of –112 dBm) for a low data rate of 9,6 kb/s.
The curves are rising increasingly fast at lower PRFs

CDMA2000 forward f = 1955,0 MHz; FER = 0,5 %; different data rates;

signal level -106 dBm; pulse width 0,1 μs

100
90 data rate 9,6 kB/s
data rate 76,8 kB/s
80
70
dBμV

60
50
40
30
20
10
0
100 1 000 10 000 100 000 1 000 000
f/Hz IEC 2039/06

Figure 4.8.28 – Weighting characteristics for the Frame Error Ratio (FER) of cdma2000
(measured at a receive signal level of – 106 dBm) for two different data rates
(9,6 and 76,8 kb/s)

For higher data rates (e.g. 384 kb/s, which was not available for the test), the system would
still be more susceptible to impulsive disturbance. For the higher data rates, the faster rising
knee (corner frequency) is shifted to lower PRFs but remains high compared to 1 kHz.

4.8.7 Effects of spread-spectrum clock interference on wideband radiocommunication

signal reception

It was argued that the classical impulsive (i.e. broadband) and unmodulated or pulse-
modulated narrowband interference may not completely reflect today’s sources of
interference. In this context, the question on the effect of spread-spectrum-clock signals came
up. This type of emission has in the recent past raised the concern of spectrum regulators and
some studies have been published. Spread-spectrum clocking reduces the measured
emission level of computer clocks, but what is the effect of spread-spectrum clocking on
victim radiocommunication systems? Spread-spectrum clock interference was expected to
 – 26 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

have especially a severe effect on wideband radiocommunication services. Therefore DVB-T

and W-CDMA as well as CDMA2000 were selected for tests and presented in [3] and [4].

The application of frequency modulation or other spread spectrum modulation to the clock
signal distributes the clock power over a frequency band wider than the EMI measurement
bandwidth and thus reduces the emission level, when measured with a bandwidth as narrow
as e.g. 120 kHz. In table 4.8.5 measurements are shown for a frequency-modulated clock
signal spectrum and of the corresponding unmodulated clock signal with f centre = 500 MHz,
f mod = 30 kHz (sinewave), spread amount δ = 3,5 MHz (i.e. the spectrum width due to
modulation) and a peak level reduction Δ = 5,0 dB.

Table 4.8.5 – Example of measurement results in dB(μV) of unmodulated

and FM modulated carriers for various detectors (bandwidth 120 kHz)

FM modulated carrier FM modulated carrier

Detector Unmodulated carrier

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
for highest peak for centre frequency
PK in dB(µV) 55,6 50,39 44,3
QPK in dB(µV) 55,4 49,30 43,16
AV in dB(µV) 55,38 38,38 37,12
RMS in dB(µV) 55,38 42,50 38,87

Using the measurement bandwidth of 1 MHz at 2 GHz (with a proportionally higher spread
amount) reduces the differences between unmodulated and FM modulated carriers to 1,2 dB
for the peak detector, 17 dB for the average detector and to 11 dB for the r.m.s. detector.

The measurement results in [3] and [4] show that the high immunity of the DVB-T and W-
CDMA systems to unmodulated carriers (i.e. clock signals) is lost to frequency modulation.
Considering

– that due to a lower measured emission level, the interference level may be increased by
the amount of Δ (e.g. 6 dB) to reach the same emission limit and
– that frequency modulation causes a reduction of up to 25 dB of the original immunity of
the system to clock signals,
frequency modulation causes a total increase of possible interference effect of up to 25 +
6 dB = 31 dB.

It is agreed that some digital modulation systems have been especially designed to suppress
unmodulated interference. In general, however it must be pointed out that the EMI
measurement bandwidth should match the bandwidth of possible victim radiocommunication
receivers. It might therefore be an advantage to describe the detector function for various
measurement bandwidths. If an r.m.s. detector is used for the measurement of spread-
spectrum modulated emissions, the measured value will be proportional to the square-root of
the measurement bandwidth. For the frequency range of CISPR bands C and D, the
radiocommunication signal bandwidths have always had a wide range of values. Narrowband
FM with as few as 7,5 kHz on the one hand and the amplitude-modulated TV signal spectrum
including the residual sideband with as much as 6 MHz on the other were in use until recently
and the 120 kHz was used as the measurement bandwidth. This situation has not changed
very much with the introduction of TETRA (bandwidth approx. 25 kHz) and DVB-T (bandwidth
6,6 MHz (VHF) and 7,6 MHz (UHF)).

4.8.8 Analysis of the various weighting characteristics and proposal of a weighting

detector

Looking at various results of weighting characteristics in the sections above, we can see that
above a certain corner frequency, the weighting function decreases with approximately 10 dB
per decade of pulse repetition frequency. A decrease of 10 dB per decade corresponds to the
weighting function of an r.m.s. detector (see Figure 4.8.2). Below this corner frequency, the
weighting function decreases with a higher rate. A higher rate of decrease (20 dB/decade)
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 27 –

can be achieved using the linear average detector function. This behaviour can be
approximated by a combination of two detectors, the r.m.s. and the linear average detector.
The average detector applies the meter time constant as described in CISPR 16-1-1 for
intermittent, unsteady and drifting narrowband disturbances. Figure 4.8.29 serves to
understand the meaning of the corner frequency. It is not possible to satisfy the protection
requirements of all services with the same perfection, therefore the selection of the various
corner frequencies between the proposed average and r.m.s. weighting functions in each
band can be regarded as a compromise. Where corner frequencies for different pulse widths
are different, the corner frequency for the shorter pulse widths apply, as the detector
weighting always applies to the shortest possible pulse width, which is determined by the
measurement bandwidth. It is proposed to keep the measurement bandwidths specified in
CISPR 16-1-1 for the CISPR bands A through E .

Table 4.8.6 – Survey of the corner frequencies found

in the various measurement results

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
System Reference fc Comment
kHz

DRM 2, 10 0,1/0,005 10 Hz chosen for feasibility

DVB-T 2, 11 0,1 - 10 f c depending on w p , modulation and coding

DAB 2, 9 5 f c partially depending on w p

TETRA 9 0,5 narrowband system, mainly used below 1 GHz

GSM 900 7, 8, 9 1,5 above f c :very close to r.m.s.

FM (2) unpubl. < 0,5 weighting characteristics follows r.m.s. down to 0,5 kHz

GSM 1800 7, 8, 9 1,5 above f c :very close to r.m.s.

DECT 9 50 above f c : flatter than r.m.s.

IS-95 9 2 very similar to J-STD 008; above f c close to r.m.s.

J-STD 008 9 5 very similar to IS-95; above f c close to r.m.s.

CDMA2000 9 50 data rate 9,6 kb/s; above f c , curves are very close to r.m.s.

CDMA2000 9 10 data rate 76,8 kb/s; above f c , curves are very close to r.m.s.

Where f c is the corner frequency and w p is pulse width

As a result of the values found in Table 4.8.6, the following corner frequencies were selected:

Band A: 10 Hz (same as band B)

Band B: 10 Hz
Bands C/D: 100 Hz
Band E: 1000 Hz
NOTE The corner frequency of 10 Hz was selected for band A, in order to give the r.m.s. detector a function
similar to the one of band B and in addition to allow the use of the meter time constant in order to provide an
asymptote for the purpose of measuring intermittent, unsteady and drifting narrowband disturbances with the r.m.s.
detector.
 – 28 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

RMS+Average weighting detector compared to existing detectors

(example as proposed for bands C and D)

70

60
RMS-AV Average
RMS-AV
50
Weighting factor/dB

20 dB/decade Quasi-Peak
Peak

40

30 Corner
10 dB/decade
frequency

20

10

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
0
1 10 100 1 000 10 000 100 000 1 000 000
fp/Hz IEC 2040/06

The asymptote of 58,7 dB near 1 Hz is due to the average detector meter time constant.

Figure 4.8.29 – The proposed r.m.s.-average detector for CISPR bands C and D
with a corner frequency of 100 Hz

Linear
average Peak
RMS detector reading
detector

IEC 2041/06

The linear average detector has an inherent meter time constant. The maximum output of the average
detector is taken using a peak reading function.

Figure 4.8.30 – RMS-average detector function by using an r.m.s. detector

followed by a linear average detector and peak reading

A digital r.m.s. detector with r.m.s. computing times of 10 ms, followed by a digital linear
average detector, results in the r.m.s.-average weighting curve of Figure 4.8.29 for the
shortest pulse width allowed by the measurement bandwidth of 120 kHz. RMS computing
times of 10 ms will give r.m.s. values of the disturbance signal within 10 ms. The 10-ms
packets are then weighted using a linear average function. The peak reading function after a
meter time constant of 100 ms is effective then for low repetition pulses (f p below 10 Hz)
which causes the weighting curve to approximate the asymptote of 58,7 dB.

Conclusion: it has been shown experimentally and partly numerically that some detector
functions that are currently in use in CISPR product standards

– either indicate a higher interference potential of impulsive disturbance than the interferer
actually represents (i.e. they overweigh the disturbance) if “peak” and “quasi-peak”
detectors are used, or
– indicate a lower interference potential of impulsive disturbance than the interferer really
represents (i.e. underweigh or de-emphasize the disturbance) for the “average” detector
with respect to the possible interference effect on digital radiocommunication systems,
whereas using the r.m.s.-average detector represents the interference effect rather well.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 29 –

4.8.9 Properties of the r.m.s.-average weighting detector

For CISPR weighting functions, the pulse width is always assumed to be defined as the
inverse of the measurement bandwidth, corresponding to the response to the Dirac pulse.
Therefore the weighting functions of the r.m.s.-average detector are shown in Figure 4.8.31
for the shortest pulse widths allowed by the bandwidths specified.

RMS+ average weighting functions for bands A, B, C/D and E

80

RMS-AV Bands C/D

70 RMS-AV Band E
RMS-AV Band A
RMS-AV Band B
60
Weighting factor/dB

50

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
40

30

20

10

0
1 10 100 1 000 10 000 100 000 1 000 000
fp/Hz IEC 2042/06

Figure 4.8.31 – RMS-average weighting functions for CISPR bands A, B, C/D and E
for the shortest pulse widths allowed by the measurement bandwidths

Figure 4.8.31 shows theoretical weighting curves. In practice, the weighting factors apply up
to values of approximately 40 dB for broadband emissions. If wider pulse widths, e.g. pulse-
modulated carriers are measured, then the weighting function will change depending on the
pulse width. If e.g. the pulse width is 10 times the shortest pulse width, this will shift the
weighting curve by an amount of 10 dB, if the PRF is above the corner frequency, and by
20 dB, if PRF and reciprocal of pulse width are below the corner frequency.

If the r.m.s.-average weighting detector is used with a wider measurement bandwidth than the
one specified, then the weighting curve will be shifted due to the shorter pulse width as shown
in Figure 4.8.32.
 – 30 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

RMS+average weighting detector for bands C/D used

with 120 kHz and 1 MHz bandwidths
70

RMS-AV 120 k
60 RMS-AV 1 MHz
Peak

50
Weighting factor/dB

40

30

20

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
10

0
1 10 100 1 000 10 000 100 000 1 000 000

fp/Hz IEC 2043/06

Figure 4.8.32 – Shift of the r.m.s.-average weighting function

for CISPR band C/D by using a bandwidth of 1 MHz instead of 120 kHz,
if the shortest possible pulse widths are applied

Measurement speeds: Measurement times and scan rates for the r.m.s.-average measuring
receiver can be made similar to those of measuring receivers using the average detector (see
CISPR 16-2-1, 16-2-2 and 16-2-3), i.e. measurements can be made substantially faster than
with the quasi-peak measuring receiver. The definition of measurement times will have to take
the characteristics of the disturbance into account.

The process of defining limits that are based on the r.m.s.-average detector has to take into
account the effects of different types of disturbances on digital communication systems. For
unmodulated sinewave signals, all detectors will yield the same result. For Gaussian noise,
the r.m.s.-average measuring receiver will indicate a level approximately 1 dB higher than the
average detector level, 6 dB lower than the quasi-peak detector (for bands C and D) level and
10 dB lower than the peak detector indication. Measurements of impulsive noise will result in
levels between the average detector level and the quasi-peak detector or peak detector
indications. In general, the r.m.s.-average detector is applicable for all types of continuous
disturbance.

Table 4.8.7 shows examples of measurement results for some broadband disturbance
sources, measured with the average, r.m.s.-average and quasi-peak detectors at frequencies
in bands B and C. The measurements were taken in a small round-robin test, conducted in
Germany in 2004.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 31 –

Table 4.8.7: Measurement results for broadband disturbance sources (measurements

with r.m.s.-average and quasi-peak detectors are normalized to average detector
values)

EUT f Average value RMS-average Quasi-peak

minus average minus average
MHz dB(μV) dB dB
Hairdryer 1,0 32,75 +3,39 +11,81
Hairdryer 35 33,80 +8,49 +26,84
Washing machine 1 0,768 20,67 +4,74 +21,79
Washing machine 1 124 13,68 +3,80 +19,91
Washing machine 2 0,71 26,98 +1,71 +9,22
Washing machine 2 116 18,90 +3,92 +22,04

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Taking into account that the r.m.s.-average measuring receiver addresses disturbance effects
of all types of continuous emissions, it is possible to define one limit only, i.e., a single limit
could be used and the limits for average and quasi-peak (or peak) detectors could be merged
into one single limit, except for cases of discontinuous disturbances and disturbances like
microwave oven emissions. However, this decision is ultimately to be made by product
committees i.e., the committees responsible for the definition of emission limits. The
application of the r.m.s.-average detector for measurements of discontinuous disturbances is
to be discussed.

4.8.10 References
[1] CISPR publication 1, 2 nd edition 1972, Specification for C.I.S.P.R. radio interference
measuring apparatus for the frequency range 0,15 MHz to 30 MHz.
[2] CISPR/A/WG1(Hasenpusch/Dunker)04-01 Investigations of broadcast radio services in
Germany
[3] CISPR/A/WG12(Ad hoc Weight Convenor)05-01: Possible Effects of Spread-Spectrum-
Clock Interference on Wideband Radiocommunication Services
[4] STECHER, M. Possible Effects of Spread-Spectrum-Clock Interference on Wideband
Radiocommunication Services. In Proc. IEEE EMC Symposium Chicago, USA, 2005.
[5] CISPR/A/WG1(Stecher-Stenumgaard)03-01 Weighting of interference according to its
effect on digital communication services – results for convolutional codes.
[6] STENUMGAARD, P. A Promising Solution of How to Evolve Present Radiated Emission
Standards in Order to Protect Digital Communication Services. Proc. IEEE EMC
International Symposium on EMC, Istanbul, Turkey May 2003.
[7] CISPR/A/WG1(Stecher/Neibig)01-01: Weighting of Disturbance taking the Effect on
Digital Radiocommunication Systems into account. Example: The Effect on GSM-900.
[8] Neibig, U. Störempfindlichkeit digitaler Mobilfunkempfänger am Beispiel von GSM-900”
(Susceptibility of digital mobile receivers for the example of GSM 900). In Frequenz, 54
(2000), pp. 257 – 262.
[9] Measurement results published in CISPR/A/WG1(Stecher)00-1
[10] Measurement results published in CISPR/A/WG1(Stecher)03-1
[11] Measurement results published in CISPR/A/WG1(Stecher)04-1
[12] CISPR 16-1-1, Specification for radio disturbance and immunity measuring apparatus
and methods – Part 1-1: Radio disturbance and immunity measuring apparatus –
Measuring apparatus
Amendment 1 2)

—————————
2) To be published.
 – 32 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

[13] CISPR 16-2-1, Specification for radio disturbance and immunity measuring apparatus
and methods – Part 2-1: Methods of measurement of disturbances and immunity –
Conducted disturbance measurements
[14] CISPR 16-2-2, Specification for radio disturbance and immunity measuring apparatus
and methods – Part 2-2: Methods of measurement of disturbances and immunity –
Measurement of disturbance power
[15] CISPR 16-2-3, Specification for radio disturbance and immunity measuring apparatus
and methods – Part 2-3: Methods of measurement of disturbances and immunity –
Radiated disturbance measurements
[16] ITU Recommendation BT.1368, Planning criteria for digital terrestrial television services
in the VHF/UHF bands

4.9 Common mode absorption devices (CMAD)

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
4.9.1 Introduction

4.9.1.1 Purpose and application of CMAD

Common mode absorption devices (CMAD) are applied on cables leaving the test volume
during radiated disturbance tests at a test site. The purpose of this part of CISPR 16 is to
define the common mode (CM) impedance at the point where the cable leaves the test
volume.

Figure 4.9.1 shows an example of an EUT in a radiated emission measurement test set-up for
table-top equipment. The cables (e.g. the power supply cable, telecom cables, or other
external connections used to exercise the EUT during the test) leave the test volume at the
centre of the turntable. Radiation at frequencies between 30 MHz and about 200 MHz is from
the cable acting as an antenna. The EUT together with the cable and the ground plane of the
test site behave like a “top loaded monopole antenna.” The radiation of such an antenna
depends on the HF source (at the EUT connection of the cable), on the current distribution
(CM current on the cable), and length of this unintended antenna. The current distribution
depends upon the CM impedance at both ends of the antenna.

EUT

External Point A Top loaded

cable monopole
antenna

Point B

Ground plane Ground plane

IEC 2044/06

Figure 4.9.1 – Example of a simple EUT model

The CM impedance is unknown at both ends of the cable (points A and B in Figure 4.9.1). The
CM impedance at point A is determined by the EUT and can have any value. For a given EUT,
the value is fixed whenever the test is performed with an identical EUT. However, the CM
impedance at point B (Z apparent ) may vary at each test laboratory, and can have any value
because actual test procedures give no requirements for the CM impedance at this point. The
actual value depends on the construction and layout of the test laboratory outside the test
volume. Examples are given in [1].
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 33 –

It has been shown [1, 2] that the variation of the undefined CM impedance at point B can lead
to variations as shown in Table 1 for radiated emissions measured from small EUTs. The
sizes of these small EUTs were between 20 cm × 10 cm × 10 cm and 50 cm × 30 cm × 30 cm.

Table 1 – Expected deviations between different laboratories for small EUTs

due to variations of the impedance Z apparent at point B

Frequency range Possible maximum deviations of the

radiated emission results between
different laboratories
30 MHz – 50 MHz 10 dB – 25 dB
50 MHz – 120 MHz 5 dB – 15 dB
120 MHz – 200 MHz 2 dB – 7 dB

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
NOTE The variations of the cable layout in the test volume are not considered in this context. Table 1 does not
include the variations of the radiation emission results due to variations in the cable layout.

The purpose of a CMAD is to reduce the influence of the CM impedance at point B upon the
compliance uncertainty to a negligible amount.

4.9.1.2 Important properties of CMADs

The main purpose of a CMAD is to ensure that the CM impedance Z apparent at the point B of
Figure 4.9.1 is always the same, independent of the undefined impedance at the cable
entrance to the test volume in the different laboratories. Therefore the following two properties
are important:

• The cable including the CMAD should have a CM impedance Z apparent (or S 11apparent)
within a specified tolerance.
• The CMAD impedance Z apparent (or S 11apparent ) should be independent of the CM
impedance at the other end of the CMAD.

An additional purpose of a CMAD can be to attenuate disturbance signals not produced by the
EUT, in order to distinguish between the EUT as a disturbance source and other disturbance
sources. For this purpose, the insertion loss A IL of the CMAD can be used as a figure-of-
merit.
NOTE 1 The insertion loss A IL is comprised of two components:
a) loss due to dissipation inside the device, and
b) loss due to mismatch between CMAD and line.

If two CMADs are used in cascade, the resulting insertion loss in general is not the sum of the
individual insertion losses.
NOTE 2 The primary function of the absorbing clamp described in Clause 4 and Annex B of CISPR 16-1-3, is for
the measurement of interference power. Depending on the ferrite material used, some types of absorbing clamps
are suitable as CMAD.
NOTE 3 The EM clamp defined in IEC 61000-4-6 for RF-injection immunity testing is not suitable for use as a
CMAD as described herein.

4.9.2 CMAD as a two-port device

4.9.2.1 Simple model of a CMAD

Usually CMADs are constructed using multiple ferrite clamps. Ferrite clamps have the
advantage of being applicable to any types of cables, within a range of diameters. For
measurements of the CMAD characteristics, the test cable is replaced by a well-defined test
conductor. In CISPR 16-1-4, Clause 9, a test conductor of 4 mm diameter is defined, located
above a ground plane at the height defined by the dimensions of the CMAD (typically 30 mm).
The CMAD (ferrite clamp) along with the test conductor above the ground plane is regarded
as a two-port device – see Figure 4.9.2.
 – 34 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

Reference planes
close to the mechanical end of
the CMAD under test

CMAD under test

Ground plane

Metal rod of 4 mm in diameter

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
as test conductor

Reference Reference
plane plane

2-port device

Port 1 Port 2
IEC 2045/06

Figure 4.9.2 – Representation of a CMAD as a two-port device

A two-port device is completely described by the S-parameters measured at ports 1 and 2.

The S-parameters are referenced to the characteristic impedance, Z ref , of the test conductor
above the ground plane without the CMAD in place, given by

Z0 ⎛ 2h ⎞
Z ref = cosh −1 ⎜ ⎟ in Ω
2π ⎝ d ⎠

where

Z 0 is the free-space impedance (120 π) in Ω;

d is the test conductor diameter (defined to be 4 mm);
h is the height of the centre of the test conductor above the ground plane.

EXAMPLE Typical values of Z ref for various heights h are:

h = 30 mm >> Zref = 204 Ω

h = 65 mm >> Zref = 248 Ω
h = 90 mm >> Zref = 270 Ω

Any two-port network may be represented using various sets of parameters; each of these
gives a complete characterisation of the two-port device. Examples of two-port parameter
sets are:

• S11 , S21 , S 12 and S 22 – S-parameters: four complex numbers, related to a reference

impedance Z ref ;
• A, B, C, D (ABCD matrix: 4 complex numbers);
• other types of two-port parameter representations are described in the literature, but do
not offer any advantages in the present context.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 35 –

4.9.2.2 Parameters of a CMAD represented as a two-port device

The performance of a CMAD can basically be defined by the four complex S-parameters when
measured as a two-port device in a test jig. The test conductor in the test jig has a diameter of
4 mm. The height above the ground plane, h, is defined by the dimensions of the CMAD.
These two parameters define the reference impedance, Z ref , for the S-parameter
measurements. If the CMAD is symmetrical, S 11 and S 22 have the same value. If the device is
not symmetrical, the test report must describe which port was used for the S 11 test (the end
closed to the EUT to be used for radiated emissions measurements), or the results must be
reported for both ports of the CMAD.

4.9.2.3 Conversion between S-parameters and ABCD-parameters for a two-port

network element

The conversion from S-parameters to ABCD-matrix representation is given by the following

equations (Z ref is the reference impedance to which the S-parameters are referred):

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
A=
(1 + S11 )(1 − S22 ) + S12S21 (4.9.1)
2S21

B=
(1 + S11 )(1 + S22 ) − S12S21 Z (4.9.2)
ref
2S21

C=
(1 − S11 )(1 − S22 ) − S12S21 Zref (4.9.3)
2S 21

D=
(1 − S11 )(1 + S22 ) + S12S21 (4.9.4)
2S21

The inverse equations are:

2
S21 =
A + B′ + C ′ + D (4.9.5)

A + B′ − C ′ − D
S11 =
A + B′ + C ′ + D (4.9.6)

2(A ⋅ D − B ⋅ C )
S12 =
A + B′ + C ′ + D (4.9.7)

− A + B′ − C ′ + D
S22 =
A + B′ + C ′ + D (4.9.8)

where

B ′ = B Z ref
(4.9.9)

C ′ = C ⋅ Z ref
(4.9.10)
NOTE All operations in preceding equations are for complex numbers. All parameters are functions of frequency.
The equations are valid at each frequency point.
 – 36 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

4.9.2.4 Range of variations for S11 due to undefined impedance at the far end of a
CMAD

The apparent impedance of a two-port network element characterized by its ABCD-

parameters is given by:

A ⋅ Zend + B
Zapparent =
C ⋅ Zend + D

From this equation the S 11 parameter can be calculated using:

Zapparent − Z0 (A − C ⋅ Z0 )Zend + (B − D ⋅ Z0 )
S11apparent = =
Zapparent + Z0 (A + C ⋅ Z0 )Zend + (B + D ⋅ Z0 )

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Z apparent and S 11apparent are the values seen at port 1 if port 2 is connected to an impedance
of Z end .

Both quantities Z apparent and S 11apparent are a conformal mapping of Z end , expressed as:

a ⋅ Zend + b
f (Zend ) =
c ⋅ Zend + d

The general form of the equation for this type of conformal mapping is:

a⋅z+b
f (z ) =
c⋅z+d

This type of function has the property that it transforms straight lines and circles in the z-
plane into either straight lines or circles in the f -plane. In particular, if the values of z are
restricted to positive real values, the transformation of this half plane results in a circle in the
f -plane, as shown in Figure 4.9.3.

Δf
f0

z-plane f-plane
IEC 2046/06

Figure 4.9.3 – Conformal mapping between z-plane and f-plane

The centre of this circle is at:

b⋅c −a⋅d a
f0 = + (complex value)
2c Re(d / c )
2 c

The radius of this circle is:

TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 37 –

b⋅c − a⋅d
Δf = (scalar value)
2c 2 Re(d / c )

The maximum value of f is then:

f max
= f 0 + Δf (scalar value)

The minimum value of f is then:

f min
= f 0 − Δf if f0 > Δf else f min
=0

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Using these relations for Z apparent gives the following parameters:

Position of the centre of the circle:

B ⋅C − A ⋅ D A
Z apparent/center = + (complex value)
2C Re(D / C )
2 C

Radius of the circle:

B ⋅C − A ⋅ D
ΔZ apparent = (scalar value)
2C 2 Re(D / C )

Maximum value of Z apparent :

Z apparent = Z apparent/c enter + ΔZ apparent

max

Minimum value of Z apparent :

Z apparent = Z apparent/c enter − ΔZ apparent if Zapparent/center > ΔZapparent else

min
Z apparent =0
min

For S 11 the relevant parameters are given by:

Position of the centre of the circle:

S11/ center =
(B ⋅ C − A ⋅ D ) ⋅ Z 0 +
A − C ⋅ Z0
⎛ ⎞ A + C ⋅ Z0
(A + C ⋅ Z 0 )2 Re⎜⎜ B + D ⋅ Z0 ⎟⎟
⎝ A + C ⋅ Z0 ⎠
 – 38 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

Radius of the circle:

ΔS11 =
(B ⋅ C − A ⋅ D ) ⋅ Z 0
⎛ ⎞
(A + C ⋅ Z 0 )2 Re⎜⎜ B + D ⋅ Z 0 ⎟⎟
⎝ A + C ⋅ Z0 ⎠

Maximum value of S11apparent :

S11apparent = S11 / center + ΔS11

max

Minimum value of S11apparent :

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
S11apparent = S11 / center − ΔS11 if S11 / center > ΔS11 else S11apparent =0
min min

4.9.3 Measurement of CMAD

4.9.3.1 Introduction

The CMAD parameters are defined as parameters of the two-port network measured at the
reference planes with the reference impedance Z ref given by the dimensions of the test jig
cross section at the reference plane. Vector network analysers (VNA) used to measure the S -
parameters operate with coaxial connectors having 50 Ω characteristic impedance. Between
this 50 Ω coaxial connection and the non-coaxial configuration of the two-port device to be
measured, an adaptor is needed to convert the 50 Ω coaxial connector to the geometry of the
two-port device to be measured. Figure 4.9.4 illustrates the relevant set-up.

Reference Reference
plane plane

2-port device

Reference Reference
Reference plane 1 plane 1 Reference
plane 2 plane 2

50 Ω 50 Ω
2-port device

Adapter Adapter

IEC 2047/06

Figure 4.9.4 – Conversion from 50 Ω coaxial system to the geometry

of the two-port device-under-test
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 39 –

If the TRL (thru-reflect-line) calibration method is used, it is possible to define calibration at

reference plane 1 of Figure 4.9.4. The measurement result is directly referred to the
connections of the two-port device at reference plane 1, and the measurement result does not
include the adaptors. Measurements based on TRL calibrations are therefore recommended
for accurate measurements of CMAD characteristics. The details of the TRL calibration
method are described in 3.2.

The classical SOLT (short-open-load-thru) calibration method for a VNA is made at the
reference plane 2 of Figure 4.9.4, for which the necessary calibration kits are commercially
available. If this calibration at reference plane 2 is used, the properties of the adaptors are
included in the measurement result.

The effect of the adaptors can be compensated partially using other “simplified” methods –
two alternative methods are described in 3.3 and 3.4:

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
a) Measurement with SOLT calibration and position shifting (matching adaptors)
b) Measurement with SOLT calibration and transformation to Z ref (lossless 50 Ω adaptors)

4.9.3.2 Measurement with TRL calibration method

4.9.3.2.1 Introduction

The TRL calibration method is based on the model shown in Figure 4.9.5 [3, 4].

Reference Reference
plane plane

50 Ω 50 Ω
2-port device

e10 a1 S21 b2 e32

a0 b3

e00 e11 S11 S22 e22 e33

e01 S12 e23

b0 a3
b1 a2
Error due to Error due to
adapter A including adapter B including
cables cables
(Port 1) (Port 2) IEC 2048/06

Figure 4.9.5 – Basic model for the TRL calibration

The four S -parameters are the true values of the measured two-port device. The parameters
of the two adapters A and B are unknown and need to be derived from the calibration
measurements with different external connections.

Four calibration configurations are necessary for the TRL calibration:

a) “Reflect” (Port 1): Measuring the complex value S 11 of the adaptor section and adapter at
port 1 without any other connection (simulating an open) – see Figure 4.9.6 a)
b) “Reflect” (Port 2): Measuring the complex value S 22 of the adapter section and adapter at
port 2 without any other connection (simulating an open) – see Figure 4.9.6 b)
 – 40 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

c) “Through”: Measuring the complex values S 11 , S 12 , S 21 , S 22 with the two adapter sections
directly connected together (without the transmission line section in between) – see Figure
4.9.6 c)
d) “Line”: Measuring the complex values S 11 , S 12 , S 21 , S 22 with the transmission line section
introduced – see Figure 4.9.6 d).

The results of these calibration measurements are 10 complex numbers for each frequency.
Many VNAs have firmware for the TRL calibration included. If the VNA includes firmware for
TRL calibration, it will use these reference measurements to calculate the proper corrections
for the TRL measurement. If the VNA does not support the TRL calibration, the necessary
corrections can be made externally according to the procedure described below (4.9.3.2.2 to
4.9.3.2.5).

The characteristic impedance of the “line” section has to be known exactly and is introduced
into the calibration data used by the firmware of the VNA. Some firmware also asks for the

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
electrical length of the “line” section, but theoretically only the impedance is needed. The
properties of the adaptor section and adaptors outside the calibration plane do not need to be
known for the TRL calibration. These properties are measured in the calibration procedure
and are compensated directly by the TRL calibration. Any type of adaptors may be used.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 41 –

Reference
plane

50 Ω

a) Configuration for the calibration measurement “reflect port 1” IEC 2049/06

Reference
plane

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
50 Ω

IEC 2050/06
b) Configuration for the calibration measurement “reflect port 2”

Reference
plane

50 Ω 50 Ω

IEC 2051/06
c) Configuration for the calibration measurement “through”

Reference Reference
plane plane

Line
50 Ω 50 Ω

IEC 2052/06
d) Configuration for the calibration measurement “line”
NOTE The length L of the reference line for the calibration needs not to be the same as the length
used for the measurement of the CMAD. The length of the reference line for the calibration
procedure has to be selected according to the frequency range needed.

Figure 4.9.6 – The four calibration configurations necessary for the TRL calibration

The length of the “line” section during the TRL calibration establishes the frequency range for
which the TRL calibration may be used. This limitation results from the mathematical
procedure used in the TRL calibration, where for some frequencies a divide-by-zero (or very
small values) is possible and must be avoided.

If the length of the “line” reference is L , the frequency range shall be limited to between low
and high frequencies f L and f H as follows:

c
fL = 0 ,05
L
 – 42 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

c
fH = 0 ,45
L

where c is 3 × 10 8 m/s. A “line” length of 0,6 m is appropriate for calibration in the frequency
range 30 MHz to 200 MHz. If the measurement has to be extended to higher frequencies, a
second “Line” calibration is necessary. A second calibration with a “Line” length of 0,12 m
would be appropriate for the frequency range 150 MHz to 1 000 MHz.

4.9.3.2.2 Measurements

The results of the four calibration measurements shall be recorded in the following format
shown in Table 2, where the subscript M denotes measured value.

Table 2 – Calibration measurement results format

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Reflect Port 1 S 11,M
Reflect Port 2 S 22,M

⎡T 11,Mt T 12,Mt ⎤ 1 ⎡S 12,Mt S 21,Mt −S 11,Mt S 22,Mt S 11,Mt ⎤

Thru TMt = ⎢ ⎥= ⎢ ⎥
T
⎣ 21,Mt T 22,Mt ⎦ S 21,Mt ⎣ −S 22,Mt 1 ⎦

⎡T 11,Md T 12,Md ⎤ 1 ⎡S 12,Md S 21,Md −S 11,Md S 22,Md S 11,Md ⎤

Delay TMd = ⎢ ⎥= ⎢ ⎥
⎣T 21,Md T 22,Md ⎦ S 21Md ⎣ −S 22,Md 1 ⎦

The conversion between T -parameters and S -parameters is given by:

⎡T 11 T 12 ⎤ 1 ⎡S 12 S 21−S 11S 22 S 11 ⎤
⎢ ⎥= ⎢ ⎥
⎣T 21 T 22 ⎦ S 21 ⎣ −S 22 1 ⎦

⎡S 11 S 12 ⎤ 1 ⎡T 12 T 11T 22−T 12T 21 ⎤

⎢ ⎥= ⎢ ⎥
⎣S 21 S 22 ⎦ T 22 ⎣ 1 −T 21 ⎦

4.9.3.2.3 Calculation of the conversion parameters

From the calibration measurements, the conversion parameters are derived according to the
following procedure:

a) Define the Matrix M as:

⎛m m12 ⎞
M = ⎜⎜ 11 ⎟⎟ ≡ TMd ⋅ TMt −1
m
⎝ 21 m 22 ⎠

b) Define the Matrix N as:

⎛n n12 ⎞
N = ⎜⎜ 11 ⎟⎟ ≡ TMt −1 ⋅ TMd
n
⎝ 21 n 22 ⎠

c) Solve the preceding equations with respect to x :

m 21x 2 + (m 22 − m11 )x − m12 = 0

This gives two solutions. Give the two solutions the names x 1 and x 2 such that x1 > x 2 .

d) Solve the equations with respect to y :

TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 43 –

n12 y 2 + (n 22 − n11 )y − n 21 = 0

NOTE Take note of the indices – not the same as previous equation

This gives two solutions. Give the two solutions the names y 1 and y 2 such that y 1 > y 2 .

e) Define a , b , c , and d as follows:

a = x1
b = x2
c = y1
d = y2

f) Define also the following variables:

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
e00 = b
e33 = −d
S11,M − b
f1 =
S11,M − a
S22,M + c
f2 =
S22,M + d
S11,Mt − b
f3 =
S11,Mt − a

g) Now compute the following variables:

e11 = f1f 2 f3
e10 e 01 = e11 (b − a )

We now have to distinguish which root to take and therefore we compute the real part of
the reflection coefficient:

1 S11,M − b
Γa =
e11 S11,M − a

It should be near a value of 1 for OPEN and near a value of -1 for SHORT. If not, take the
other root for e 11 , and redefine e 10 e 01 .
h) Perform the further computations:

e 22 = f 3 e11
e 23 e 32 = e 22 (c − d )
e10 e32 = S 21,Mt (1 − e11e 22 )

i) Determine the forward conversion parameters e 00 , e 11 , e 22 , e 33 , as well as:

Δ x = e00 e11 − e10 e01

Δ y = e22e33 − e32 e23
e10 e32
k = e10 e23 =
e23 e32

j) Determine the inverse conversion parameters:

 – 44 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

inv
e00 = e00 Δ x
inv
e11 = e11 Δ x
inv
e22 = e22 Δ y
inv
e33 = e33 Δ y
Δinv
x = 1 Δx

Δinv
y = 1 Δy

e01Δ y 1 e00 e11 − Δ x Δy 1 e00 e11 Δ x − 1

k inv = = =
e32 Δ x k Δx e22 e33 − Δ x k e22 e33 Δ y − 1

4.9.3.2.4 Forward conversion (from S effective to S measured )

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
If Sij are the effective S-parameters of a network, then the measured S-parameters, Mij , are
computed using:

N = (1 − S11e11 )(1 − S22 e22 ) − S21S12e11e22

M11 = [(e00 − S11Δ x )(1 − S22 e22 ) − S21S12e22 Δ x ] N

M12 = (S12 k )(e00 e11 − Δ x ) N
M 21 = (S21k )(e22 e33 − Δ y ) N
[
M 22 = (e33 − S22 Δ y )(1 − S11e11 ) − S12S21e11Δ y N ]
4.9.3.2.5 Inverse conversion (from S measured to S effective )

The effective S-parameters S ij are determined from measured S-parameters Mij using:

(
N inv = 1 − M11e11
inv inv
1 − M 22e22 )(inv inv
− M 21M12e11 e22 )
inv
S11 = e00 [(
− M11Δinv inv
)(
inv inv
x 1 − M 22 e22 − M 21M12 e 22 Δ x N inv ) ]
S12 = (M 12 k inv
)(e inv inv
00 11 e −Δ inv
x )N inv

S21 = (M 21k inv

)(e inv
22 e inv
33 −Δ inv
y )N inv

S22 = [(e inv

33 − M 22 Δinv inv
)( inv inv
y 1 − M 11e11 − M 12 M 21e11 Δ y )
N inv ]
These four S-parameters are the result of measurements using the TRL calibration method.

4.9.3.3 Measurement with SOLT calibration and position shifting - simplified method

4.9.3.3.1 General

The SOLT (Short-Open-Load-Through) calibration method is only possible at the reference

planes 1 and 6 shown in Figure 4.9.7. The influence of the adaptor section (between
reference planes 1 and 2, and between reference planes 5 and 6, of Figure 7) and the open
length of the test conductor (between reference planes 2 and 3, and between reference
planes 4 and 5, of Figure 4.9.7) upon the final result may be partially compensated by making
two measurements with the CMAD at two different positions inside the jig, then combining
results from these two measurements. This compensation is optimal if matching adaptors are
used (to match the impedance of the 50 Ω coaxial connection to the reference impedance Z ref
of the test jig). The following simplified method may be applied in the frequency range from
30 MHz to 1 000 MHz.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 45 –

4.9.3.3.2 Scattering parameter representation

Figure 4.9.7 shows a scattering parameter representation of the CMAD measurement. A

network analyzer is connected to the reference planes 1 and 6. Hence the measurement can
yield only the S-parameters, Sij_T , of the total system, while the S-parameters of a CMAD,
Sij_c , are to be determined. The total S-parameters are express in terms of the S-parameters
of the various sections as

⋅ S11_c exp( − j 2 β ⋅ LA )(1 − Y ) + S 21_c ⋅ S12 _c ⋅ S11_b exp[− j 2 β (L A + LB )]

(S )
11 _ T CMAD = S11_a + S 21_a ⋅ S12 _a
(4.9.11)
(1 − X )(1 − Y ) − Z
S 21_a ⋅ S12 _a ⋅ S11_c exp( − j 2 β ⋅ LA )
≈ S11_a +
1− X

S 21_a ⋅ S 21_c ⋅ S 21_b exp[− jβ (L A + LB )]

(S )
21 _ T CMAD =
(1 − X )(1 − Y ) − Z (4.9.12)

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
S 21_a ⋅ S 21_c ⋅ S 21_b exp[− jβ (L A + LB )]
≈
(1 − X )(1 − Y )

with

X ≡ S 22 _a ⋅ S11_c exp( − j 2 β ⋅ L A )
Y ≡ S 22 _c ⋅ S11_b exp( − j 2 β ⋅ LB )
Z ≡ S 22 _a ⋅ S 21_c ⋅ S12 _ c ⋅ S11_b exp[− j 2 β (L A + LB )]

The approximate expressions in the above equations are derived assuming that the insertion
loss of a CMAD is relatively large, i.e. |S 21_c |, |S 12_c | << 1, because usually the insertion loss
is more than 20 dB. Due to the symmetrical structure of a CMAD, the unknown parameters to
be determined are S 11_c (=S22_c ) and S21_c (=S 12_c ) only.

L1 LCMAD L1

IA LA LB IB

50 Ω Z0 50 Ω

ST

S21_a exp(-jβLA) S21_c exp(-jβLB) S21_b

S11_a S22_a S11_c S22_c S11_b S22_b

S12_a exp(-jβLA) S12_c exp(-jβLB) S12_b

Reference
plane 1 2 3 4 5 6

Adapter A CMAD Adapter B

IEC 2053/06

Figure 4.9.7 – Measurement of CMAD characteristics

4.9.3.3.3 Estimation of |S 21_c |

The |S 21_c | of a CMAD under test is approximately given by

 – 46 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

S21 _ T 1 − (S 22 _a ) 2 ⋅ exp[− j 2 β ⋅ (LA + LB + LCMAD )]

S21_c = CMAD
≈ S 21 _ c (4.9.13)
simple
S 21_ T
empty
[1 − S 22 _a ][
⋅ S11_c ⋅ exp(− j 2 β ⋅ LA ) 1 − S 22 _a ⋅ S11_c ⋅ exp(− j 2 β ⋅ LB ) ]
where

S21_a ⋅ S21_b ⋅ exp(− jβ ⋅ L) S21_a ⋅ S21_b

S21_T = = (4.9.14)
empty 1− S22_a ⋅ S11_b ⋅ exp(− j 2β ⋅ L) 1− S22_a ⋅ S11_b ⋅ exp(− j 2β ⋅ L)

This equation apparently demonstrates that if the adapters have a good match with the line
section (i.e. S22_a =0), the simplified method may yield an accurate value of |S21_c |.

4.9.3.3.4 Estimation of Γ (= |S 11_c |)

The |S 11_c | of a CMAD under test is approximately given by

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Γsimple =
[(S ) − (S
11 _ T 1 ) ]exp(− jβ ⋅ L)
11 _ T 2 (4.9.15)
(S 21 _ T )
empty
[exp(− j 2 β ⋅ L11 ) − exp(− j 2β ⋅ L12 )]

where (S 11_T ) 1 and (S 11_T ) 2 denote S 11_T parameters measured with L A = L A1 and L A2 ,
respectively. Both adapters are assumed to be identical, i.e. |S 21_b |=|S 12_a |.

4.9.3.3.5 Uncertainty of |S 21_c |

Since the adapters may cause mismatch to some extent, uncertainty in the estimated
insertion loss AIL_simple (=1/|S 21_c |simple ) should be evaluated using the following equation
derived from equation (4.9.13) with L A , L B << λ assumed:

1 − 2 S 22 _a ⋅ S11_c 1 + 2 S 22 _a ⋅ S11_c
< error in AIL_simple < (4.9.16)
2 2
1 + S 22 _a 1 − S 22 _a

Equation (4.9.16) can provide a good estimate for a test set-up having matching adapters.
However, it overestimates the uncertainty in the case of simple adapters without matching
circuit. Hence, the following expression should apply to a set-up that does not have matching
adapters:

(1 − S 22 _a ⋅ S11_c )2

< error in AIL_simple <

(1 + S 22 _a ⋅ S11_c )
2

(4.9.17)
2 2
1 + S 22 _a 1 − S 22 _a

The following preliminary check of the test set-up shall be made with the SOLT calibration
before the CMAD measurement:

a) |S 21_T | empty is the magnitude of S21_T measured in the test set-up without a CMAD
introduced, as shown in Figure 8 a).

b) |S 21_T | open is the magnitude of S 21_T measured in the test set-up with truncated metal
rods as shown in Figure 8 b). The lengths of the rods, L1 and L2, are determined
according to Figure 7. The |S 21_T |open data provides information about possible adverse
effects of direct coupling between the adapters.

During the above measurements, the distance between the flanges, L, shall be the same as in
the CMAD measurement.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 47 –

Ω
50 Ω
50 Z
Z00 50Ω
50 Ω

IEC 2054/06
a) Empty jig

L
L1
L1 L2
L2

Ω
50 Ω
50 Z00
Z 50 Ω
50 Ω

b) Open jig IEC 2055/06

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Figure 4.9.8 – Preliminary measurements of the test set-up

4.9.3.3.6 Uncertainty of Γ (= |S 11_c |)

Equation (4.9.16) provides approximations for the uncertainty in the estimated |Γ| caused by
mismatch. Theoretical and experimental analysis yields the same expressions for the
uncertainty as shown in Equation (4.9.16) or Equation (4.9.17). Estimation of the uncertainty
in |Γ simple | caused by mismatch shall be done using Equation (4.9.16) or Equation (4.9.17),
assuming a U-shaped probability distribution.

Another error is introduced by the position of a CMAD during actual measurements. Errors in
|Γ simple | can be estimated using the equation

sin(β ( ΔL + δ ))
error in Γ = (4.9.18)
sin( β ( ΔL )

for the case of an error δ in ΔL=|L A1 -L A2 |. Uncertainty shall be estimated using Equation
(4.9.18), assuming a uniform distribution.

4.9.3.3.7 Mismatch of the adapters

The ratio of the maximum and minimum values of |(S21_T ) empty | may be expressed as

2
(S 21_ T ) empty 1 + S 22 _a
max
≅ 2
(4.9.19)
(S 21 _ T ) empty 1 − S 22 _a
min

in which the adapters A and B are assumed to be identical. Thus, variations in |S 21_T |empty
versus frequency gives information about the reflection coefficients of the adapters.

4.9.3.4 Measurement with SOLT calibration and ABCD transformation to Z ref level

The SOLT calibration is only possible at the reference planes 1 and 6 shown in Figure 4.9.7
and in Figure 4.9.9. If lossless 50 Ω adaptors are used, it is possible to measure the two-port
parameters at the 50 Ω level then convert the results to the Z ref level by applying the ABCD
matrix method described in 2.3. The reference planes in Figure 4.9.9 have slightly different
positions to those shown in Figure 4.9.7.
 – 48 – TR CISPR 16-3 Amend. 2 © IEC:2006(E)

LCMAD

LA LB

50 Ω Zref 50 Ω

Reference
plane 1 2 3 4 5 6
Adapter A CMAD Adapter B
IEC 2056/06

Figure 4.9.9 – Position of the reference planes for the measurement

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
with SOLT calibration and ABCD transformation to Z ref level

The VNA is calibrated at reference planes 1 and 6. The S-parameters of interest are the S-
parameters at reference planes 3 and 4. The error due to this difference of reference shall be
kept as small as possible, considering the following. If a lossless 50 Ω adapter is used, and if
the construction of the adapter is such that the 50 Ω impedance is maintained in the section
between reference planes 1 and 2, it is possible to compensate the electrical delay between
the two reference planes. Most VNAs allow an internal compensation for an electrical delay.
The length may be determined by calculation, or by measurement of the S 11 with the open
adaptor and adjusting the delay by observing the Smith chart until the result approaches the
result of an open circuit. The S-parameters measured by the VNA with electrical delay
adjusted and activated are then the S-parameters at reference planes 2 and 5.

Using the Equations (4.9.1) to (4.9.4) with Z ref = 50 Ω, the ABCD parameters of the two-port
network between reference planes 2 and 5 are obtained. Equations (4.9.5) to (4.9.10) are
used to transform the ABCD parameters back to S-parameters, as referred to a new Z ref (for
example 204 Ω for typical CMAD dimensions). These S-parameters are the values referred to
the reference planes 2 and 5. If the lengths L A and L B , between the reference planes 2 and 3
and the reference planes 4 and 5, are small (for example less than 30 mm for typical CMAD
dimensions), the influence of the line elements L A and L B may be neglected (at least for
frequencies up to about 300 MHz).

The most significant contributions to the uncertainty in this measurement are:

• tolerance of the calibration kit used for the calibration at the reference plane 1 and 6;
• tolerance of the mechanical dimensions of the test jig;
• impedance value of the adapter section between reference plane 1 and 2 or 5 and 6;
• electrical length of the adapter section between reference plane 1 and 2 or 5 and 6;
• tolerance of the delay compensation for the adapter section between reference
plane 1 and 2 or 5 and 6;
• distance between adaptor and CMAD (LA and L B ).

4.9.4 References

[1] RYSER, H. Using ferrite clamps on the cables leaving the turntable,
CISPR/G/WG1(Ryser)96-1, September 1996.

[2] OSABE, K. et al. A correlation test among measurement sites for radiated EMI using an
actual machine and a stabilized power line impedance. Zurich EMC Symposium, 2001.

[3] GRONAU, G. Höchstfrequenztechnik, Springer Verlag 2001, ISBN 3-540-41790-7,

chapter 8.1.4.3 "TRL Kalibrierung,“ pp. 389 – 408.
TR CISPR 16-3 Amend. 2 © IEC:2006(E) – 49 –

[4] ENGEN GF. and HOER, CA. Thru-Reflect-Line: an improved technique for calibrating
the dual six-port automatic network analyzer. IEEE Trans. Microwave Theory and
Techniques, 1979, vol. 27, pp. 987-998.

[5] POZAR, DM. Microwave Engineering, 2nd ed., John Wiley & Sons, Inc., New York,
1998, pp. 217-221.

[6] YEO SP. and CHENG, M. Network analyzer calibration for measurements near
waveguide cutoff frequency. Microwave Journal, March 1999, pp. 134-142.

[7] IEC 61000-4-6, Electromagnetic compatibility (EMC) – Part 4-6: Testing and
measurement techniques – Immunity to conducted disturbances, induced by radio-
frequency fields

[8] CISPR 16-1-3, Specification for radio disturbance and immunity measuring apparatus
and methods – Part 1-3: Radio disturbance and immunity measuring apparatus –

FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE
Ancillary equipment – Disturbance power

[9] CISPR 16-1-4, Specification for radio disturbance and immunity measuring apparatus
and methods – Part 1-4: Radio disturbance and immunity measuring apparatus –
Ancillary equipment – Radiated disturbances

___________
 FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.
LICENSED TO MECON Limited. - RANCHI/BANGALORE

ISBN 2-8318-8885-9

-:HSMINB=] ] Z^:
ICS 33.100.10; 33.100.20

Typeset and printed by the IEC Central Office

GENEVA, SWITZERLAND