Analysis of Internal RF Interferences
in Mobile Phones
SEVAG BALKORKIAN & HAO ZHANG
Master of Science Thesis in Radio Electronics
Stockholm, Sweden 2005
IMIT/LECS-2005-101
Abstract
Nowadays, mobile phones have greater functionality; a camera, color LCD screen,
wireless LAN, Bluetooth, IrDA and others. In the near future wider variety of new
functionalities will be added, from high quality voice, high definition video to high data
rate wireless channels. As consumer electronics integrate greater functionality and high
operating frequencies, their emissions will exceed the specified limits, most of these
emissions will be a result of the internal interferences in the mobile phone. Moreover
higher operating frequencies will be required to improve the quality of these
functionalities, something that will make it more difficult to control these interferences.
Internal or external sources of electromagnetic interference can degrade the
performance of sensitive analog/digital circuits inside the mobile phone. Moreover the
electronic device must satisfy a host of global regulations that limit it’s susceptibility to
these interferences, as well as the interference emitted by the device itself. Therefore
designing a new electronic device to perform new and exciting functions will not be a
pleasant task if it can not meet certain specifications and function as required to adhere
to certain global regulations.
This thesis project investigates the sources of interference inside a mobile phone;
mainly the electromagnetic interferences and its effect on the radio transceiver focusing
on the GSM receiver sensitivity. This report is a result of intensive research, an
investigation of possible sources of interference, also actual measurements were
performed; RSSI, OTA and sniffing measurements; to identify the physical sources of
interferences, and their effect on the receiver sensitivity. Finally solutions were
recommended and implemented to suppress the interferences due to different sources,
mainly through filtering, shielding or proper grounding of signals and
components/subsystems in the mobile phone.
2
Acknowledgments
We would like to express our gratitude to the following people:
Zareh Mahdessian, our advisor at Sony Ericsson Mobile Communications AB for
proposing the project and for the guidance.
Professor Håkan Olsson, our examiner at KTH, for motivation and advice.
We would like to thank the staff at Sony Ericsson Kista that helped us make this project
successful, thank you all for the support.
Finally we would like to thank Sören Karlsson at Sony Ericsson Mobile Communication
s AB for his continuous support for our project.
3
Abbreviations
3GPP
3rd Generation Partnership Project
ARFCN
Absolute Radio Frequency Channel Number
ASIC
Application Specific Integrated Circuit
BER
Bit Error Rate
CISPR
Comite International Special des Perturbations Radioelectriques
CTIA
Cellular Telecommunications and Internet Association
DUT
Device under Test
EGSM
Extended GSM
EMC
Electromagnetic Compatibility
EMI
Electromagnetic Interference
ESD
Electrostatic Discharge
ETSI
European Telecommunications Standards Institute
FCC
Federal Communications Commission
GSM
Global System for Mobile Communications
IC
Integrated Circuit
IRDA
Infrared Data Association
LAN
Local Area Network
LCD
Liquid Crystal Display
MS
Mobile Station
OTA
Over the air
PCB
Printed Circuit Board
PDA
Personal Digital Assistant
RFI
Radio Frequency Interference
RSSI
Received Signal Strength Indicator
SAM
Standard Anthropomorphic Model
WLAN
Wireless Local Area Network
4
Table of Contents
ABSTRACT............................................................................................................................................................................2
ACKNOWLEDGMENTS .....................................................................................................................................................3
ABBREVIATIONS ................................................................................................................................................................4
TABLE OF CONTENTS.......................................................................................................................................................5
1
INTRODUCTION ........................................................................................................................................................7
2
RELATED WORK.......................................................................................................................................................7
3
TRANSCEIVER STANDARDS..................................................................................................................................8
3.1
GSM STANDARD ...................................................................................................................................................8
3.1.1
GSM 900 / EGSM .............................................................................................................................................8
3.1.2
GSM 1800 / DCS ..............................................................................................................................................8
3.1.3
GSM 1900 / PCS ..............................................................................................................................................8
4
SOURCES OF RF INTERFERENCE........................................................................................................................9
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.7.1
4.7.2
4.8
5
POSSIBLE SOLUTIONS ..........................................................................................................................................15
5.1
5.2
5.3
5.3.1
5.3.2
5.4
6
MEASUREMENT PROCEDURE ...............................................................................................................................25
MEASUREMENT RESULTS AND ANALYSIS ............................................................................................................26
SNIFFING MEASUREMENTS ................................................................................................................................28
8.1
8.2
8.3
8.4
9
MEASUREMENT SCENARIOS .................................................................................................................................21
MEASUREMENTS WHERE CAMERAS ACTIVE/INACTIVE & ANALYSIS ...................................................................22
MEASUREMENTS IN PRESENCE OF WLAN & ANALYSIS.......................................................................................23
OTA SENSITIVITY MEASUREMENTS................................................................................................................24
7.1
7.2
8
LAYOUT & PLACEMENT OF COMPONENTS ...........................................................................................................15
GROUNDING SOLUTIONS ......................................................................................................................................16
EMI SUPPRESSION BY FILTERING ........................................................................................................................17
Noise Suppression using Higher Order Filters ..............................................................................................18
Distortion suppression using filters................................................................................................................19
SHIELDING SOLUTIONS ........................................................................................................................................20
RSSI MEASUREMENTS ..........................................................................................................................................21
6.1
6.2
6.3
7
THERMAL NOISE ....................................................................................................................................................9
ELECTROMAGNETIC INTERFERENCE ....................................................................................................................10
POWER SUPPLY INTERFERENCE ...........................................................................................................................10
IC / PACKAGE DESIGN .........................................................................................................................................11
PCB AND COMPONENT PLACEMENT ....................................................................................................................11
CLOCK AND DATA ...............................................................................................................................................12
GROUNDING DESIGN ............................................................................................................................................12
Single Point Ground System...........................................................................................................................12
Multipoint Ground System..............................................................................................................................13
INTERFERENCE DETECTION PROCEDURES ............................................................................................................14
MEASUREMENT PROCEDURE ...............................................................................................................................29
MEASUREMENT CASES ........................................................................................................................................29
MEASUREMENT RESULTS AND ANALYSIS ............................................................................................................30
SOURCES OF UNCERTAINTY .................................................................................................................................31
RECOMMENDED SOLUTIONS .............................................................................................................................32
9.1
9.2
9.3
9.4
LCD DISPLAY DATA/CONTROL LINES .................................................................................................................32
RADIATED EMISSIONS FROM FLEX CABLES .........................................................................................................32
IMPROPER GROUNDING OF SHIELD BOXES ...........................................................................................................33
2MPX CAM SUBSYSTEM INTERFERENCES ..........................................................................................................34
5
9.4.1
Simulations .....................................................................................................................................................35
9.4.2
Oscilloscope Measurements ...........................................................................................................................36
9.4.3
RSSI Results & Analysis .................................................................................................................................37
9.5
VGA CAM SUBSYSTEM INTERFERENCES ............................................................................................................39
9.5.1
Simulations .....................................................................................................................................................40
9.5.2
Oscilloscope Measurements ...........................................................................................................................41
9.5.3
RSSI Results & Analysis .................................................................................................................................43
9.6
RADIATED EMISSIONS FROM DIODES ...................................................................................................................44
9.6.1
Audio subsystem interferences .......................................................................................................................44
9.6.2
Power subsystem interferences.......................................................................................................................45
CONCLUSION.....................................................................................................................................................................47
FUTURE WORK .................................................................................................................................................................47
APPENDIX A .......................................................................................................................................................................48
APPENDIX B .......................................................................................................................................................................49
REFERENCES.....................................................................................................................................................................59
6
1
Introduction
Nowadays, mobile phones have greater functionality; a camera, color LCD screen,
wireless LAN, Bluetooth, IrDA and others. In the near future wider variety of new
functionalities will be added, from high quality voice, high definition video to high data
rate wireless channels. As consumer electronics integrate greater functionality and high
operating frequencies, their emissions will exceed the specified limits, most of these
emissions will be a result of the internal interferences in the mobile phone. Moreover
higher operating frequencies will be required to improve the quality of these
functionalities which will make it more difficult to control these interferences.
Internal/external sources of electromagnetic interference can degrade the performance
of sensitive analog/digital circuits inside the mobile phone. Moreover the electronic
device must satisfy a host of global regulations that limit its susceptibility to these
interferences, as well as the interference emitted by the device itself. Therefore
designing a new electronic device to perform new and exciting functions will not be a
pleasant task if it can not meet certain specifications and function as required to adhere
to certain global regulations (3GPP, FCC, ETSI, and CISPR).
This thesis report is roughly divided into the following parts: An introduction in chapter 1,
related work or background information about the project described in chapters 2, 3, 4
and 5. Measurement setups and results analysis described in chapters 6, 7 and 8.
Solutions to internal interferences described in chapter 9. Later in the report there is a
conclusion and future work proposed, commonly used terminologies in Appendix A,
higher order filter simulations in Appendix B, and finally the references.
2
Related Work
Throughout the years electronics manufacturers have put serious effort to identify
possible causes of internal interference in mobile phones and other consumer electronic
devices, and ways to solve these issues. The big part of this work was related to
electromagnetic interference caused by the PCB, high rate clocks and other internal
components of devices like oscillators on the analog/digital circuitry.
The authors of [1] discuss radio frequency effects on electronics systems integrated
with high performance chips. Unwanted interferences when superimposed on the
system signals that cause spurious state changes on logic devices and system level
breakdown. The authors of [2] discuss the test requirements that cover OTA
performance of mobile phone antennas; they describe a typical OTA test system and
key parameters extracted from such a test.
7
3
Transceiver Standards
3.1
GSM Standard
The Global System for Mobile communications is a second generation cellular
telecommunication system which was first planned in the early 1980s. Unlike first
generation systems operating at the time, GSM was digital and thus introduced greater
enhancements such as security, capacity, quality and the ability to support integrated
services. [11]
Compared with the existing analog systems, the new system was required to have a
higher capacity, comparable or lower operating costs and comparable or better speech
quality.
Standard
E-GSM 900
DCS 1800
PCS 1900
Lower Band Frequency
RFCN
Upper Band Frequency
Fl(n) = 890 + 0.2*n
Fu(n) = Fl(n) + 45
0 ≤ n ≤ 124
Fl(n) = 890 + 0.2*(n-1024)
Fu(n) = Fl(n) + 45
975 ≤ n ≤ 1023
Fl(n) = 1710.2 + 0.2*(n-512) 512 ≤ n ≤ 885
Fu(n) = Fl(n) + 95
FI(n) = 1850.2 + .2*(n-512)
512 ≤ n ≤ 810
Fu(n) = FI(n) + 80
Table 1: ARFCN and Corresponding Frequency Range of Channels [3]
The carrier spacing is 200 KHz and frequency is designated by the absolute radio
frequency channel number is ARFCN. If we call Fl(n) the frequency value of the carrier
ARFCN n in the lower band, and Fu(n) the corresponding frequency value in the upper
band. Note that all frequencies in the table above are in MHz. [3]
3.1.1
GSM 900 / EGSM
The spectrum range for the GSM 900 operation is between 890 MHz and 915 MHz for
uplink operation and 935 MHz and 960 MHz for downlink operation. EGSM is an
extension to the GSM 900 spectrum, it has additional 10 MHz that provides an
additional 50 channels. Thus the spectrum range for EGSM operation is between 880
MHz and 915 MHz for uplink operation and 925 MHz and 960 MHz for downlink
operation.
3.1.2
GSM 1800 / DCS
GSM 1800 also known as DCS1800 or DCS is a digital network working on a frequency
of 1800 MHz. The spectrum range for the GSM 1800 operation is between 1710 MHz
and 1785 MHz for uplink operation and 1805 MHz and 1880 MHz for downlink
operation.
3.1.3
GSM 1900 / PCS
The spectrum range for the GSM 1900 operation is between 1850 MHz and 1910 MHz
for uplink operation and 1930 MHz and 1990 MHz for downlink operation.
8
4
Sources of RF Interference
In a mobile phone there are many sources of interference which can be in the form of
radiated or conducted electromagnetic interferences and could affect the quality of
signal perceived at the transceiver. These interferences cause signal integrity issues
and improper functionality of sensitive analog/digital circuits inside the mobile phone.
Thermal noise, from power supply or IC packaging, PCB traces, grounding and
crosstalk should be considered as sources of interference inside a mobile phone.
In this section we will discuss the different sources of interference independently; in
later chapters we will propose solutions for these disturbances.
4.1
Thermal Noise
Generally, noise is considered to be an electromagnetic interference. Unwanted
voltages and currents induced by external magnetic fields, electric fields and ground
currents fall into the category of noise and often cause serious errors in low power
circuits.
Usually conducting media generates internal noise without current flow. [4] One
common noise category is resistor thermal noise, which is the noise developed in a
resistor in the absence of current flow, often referred to as Johnson noise. This noise is
generated in a resistor independent of any current flow and has a mean-square voltage
value of 4 KTR (BW ) .
In this expression k is Boltzman’s constant, T is temperature in degrees Kelvin, R is
resistance in Ω, and BW is bandwidth, in Hz.
In practice, there is always some parasitic capacitance across the leads of a resistor
due to the printed circuit board or lead wire connections. For this situation, when the
thermal noise in a resistor is shunted by a non-zero capacitance, the mean-square
voltage value is given by KT .
C
Temperature and resistor values can not always be minimized; however, using signal
conditioning modules with small bandwidth multi-pole low-pass filters will ensure that
external thermal resistor is essentially eliminated. [4]
9
4.2
Electromagnetic Interference
RFI is an electromagnetic radiation or interference which is emitted by electrical circuits
carrying rapidly changing signals as a by-product of their normal operation. It causes
unwanted signals in the form of interference or noise which in turn is to be induced in
other circuits. This interrupts, obstructs degrades or limits the effective performance of
those other circuits inside the system. EMI can be induced intentionally, or
unintentionally, as a result of spurious emissions and responses, inter-modulation
products, and the like.
Figure 1: EMI Propagation Model: (1) Conduction mode, (2) Radiation mode,
(3) Conduction mode – Radiation mode and (4) Radiation mode – Conduction mode
The efficiency of the radiation is dependant on the height above the ground or power
plane; at RF one is as good as the other, and the length of the conductor in relationship
to the wavelength of the signal components either the fundamental, harmonic or
transient (overshoot, undershoot or ringing). At lower frequencies, such as 133 MHz,
radiation is almost exclusively via I/O cables; RF noise gets onto the power planes and
is coupled to the line drivers via the VCC and ground pins. The RF is then coupled to the
cable through the line driver as common mode noise. Since the noise is common mode,
shielding has very little effect, even with differential pairs. The RF energy is capacitively
coupled from the signal pair to the shield and the shield itself does the radiating.
At higher frequencies, usually above 500 MHz, traces get electrically longer and higher
above the plane. Two techniques are used at these frequencies: wave shaping with
series resistors and embedding the traces between the two planes. If all these
measures still leave too much RFI, shielding such as RF gaskets and copper tape can
be used.
4.3
Power Supply Interference
Switching power supplies can be a source of RFI, but have become less of a problem
as design techniques have improved. Noise which is conducted or radiated should be
prevented from returning to the input source, where it can potentially cause havoc on
other devices operating from the same input power. Here EMI filters are utilized to block
this noise and provide a low-impedance path back to the noise source. The larger the
noise interference; the greater is the size, expense, and difficulty of the filter to be
designed. Power supplies that operate at a fixed frequency have their largest EMI
emission at this fundamental, fixed frequency. Emissions also occur at multiples of the
switching frequency but at diminished amplitudes.
10
4.4
IC / Package Design
The IC packaging of a component or an ASIC could cause certain amounts of thermal
noise. The IC package design affects the equivalent thermal resistance generated
noise. Below is a table that shows different packaging designs and a comparison
between different thermal resistances.
Package
DIP16
SOIC24
PLCC44
QFP44
BGA256
RthJC
(K/W)
19
17
10
15
13
RthJC Still Air
(K/W)
48
77
33
48
40
RthJC 0.5 m/s
(K/W)
RthJC 2.0 m/s
(K/W)
31
46
38
27
42
33
Table 2: IC Packaging and Thermal Resistances [5]
4.5
PCB and Component Placement
The PCB connects electronic passive/active components such as transistors, diodes,
capacitors resistors, oscillators and ICs. The routing of the traces on the PCB largely
affects the electromagnetic compatibility performance of the PCB with respect to both
electromagnetic radiations and susceptibility to electromagnetic fields.
In order to get a PCB on which the circuits function properly, the trace routing, the
placement of components/connectors and the decoupling used with certain ICs will
have to be optimized.
Electromagnetic interference can be transferred by electromagnetic waves, conduction,
and inductive/capacitive coupling. This interference must reach the conductors in order
to disturb the components. This means that the loops, long length and large surface of
the conductors are vulnerable to EMI, making the PCB the principle subject of EMC
improvements.
The following equation shows the relationship of current, its loop area, and the
frequency to electric field which is in (V/m):
⎛1⎞
E = kIAf 2 ⎜ ⎟ sin θ
⎝r⎠
Where k is the constant of proportionality, I is the current in amperes, A is  the loop area
in m2, f is frequency in MHz, r is the distance and ֳθ is the angle. And since the distance
to the ground plane is usually fixed due to board stackup requirements, minimizing trace
length on the board layout is crucial to decreasing emissions. [6]
11
4.6
Clock and Data
Controlling clock and data EMI issues has remained a challenge for the electronics
designer, a look at the origin of this noise shows that the digital system clock is the
largest contributor. This is understandable both because the frequency of the system
clock is often the highest of all the signals in the system, and because it is usually a
periodic square wave. The frequency spectrum of such a signal consists of a
fundamental tone of higher amplitude than the harmonic tones, whose amplitudes
diminish as frequency increases. Other signals within the system (e.g., those on the
data and address buses) are updated at the same frequency as the clock, but occur at
irregular intervals and are uncorrelated. This results in a broadband noise spectrum of
much lower amplitude than the clock. Although the total energy in this spectrum is much
larger than the clock energy, it has little effect on the EMI tests. In these tests, the
highest spectral amplitude is what is looked at and not the total radiated energy.
The clock or data signal may cause an electromagnetic interference, transition time of
the pulse from off to on and vise versa is the most important factor in determining the
spectral content of the pulse. Fast transition times generate wider range of frequencies
than do slower transition times. The spectral content of digital devices generally
occupies a wide range of frequencies and can also cause interference in electrical and
electronic devices. [7]
4.7
Grounding Design
It is important to realize that there are several purposes of a ground system. The
concept of a ground as being a zero-potential surface may be appropriate at dc or low
frequencies, but it is never in higher frequencies, since conductors have significant
impedance (inductance) and high frequency currents flow through these impedances,
resulting in points on the ground having different high frequency potentials.
There are basically two major philosophies regarding signal ground schemes: single
point ground systems and multi-point ground systems, there are other types of ground
systems that are used less frequently and in special cases. These are referred to as
hybrid ground systems, and are a combination of the previous two systems over
different frequency ranges.
4.7.1
Single Point Ground System
A single point ground system is one in which subsystem ground returns are tied to a
single point within the subsystem. The intent in using a single point ground system is to
prevent currents of two different subsystems from sharing the same return path and
producing common-impedance coupling. [7]
Typically, single point ground systems are used in analog sub-systems, where low-level
signals are involved. In these cases, milli-volt and microvolt ground drops create
significant common-impedance coupling interference problems for those circuits. Digital
12
sub-systems on the other hand are inherently “immune” to noise from external sources;
however they are susceptible to internal noise. In order to minimize this commonimpedance coupling, the ground system in digital sub-systems tends to be multi-point,
using a large ground plane such as in inner plane such as in inner plane board or
placing numerous alternate ground paths in parallel such as with a ground grid, thus
reducing the impedance of the return path. It is also important to route the signal
conductors in close proximity to the ground returns, since this will also reduce the
impedance of the return.
Figure 2: Single-point ground system:
Common Impedance Coupling in daisy
chain connection
4.7.2
Figure 3: Single-point ground system:
Unintentional coupling between ground
wires
Multipoint Ground System
The other type of ground system philosophy is the multipoint ground system. Typically a
large conductor, often the ground plane, serves as the return in a multipoint ground
system. In such a system the individual grounds of the sub-systems are connected at
different points to the ground conductor. In using a multipoint ground system it is
assumed that the ground return to which the individual grounds are terminated have a
very low impedance between any two points at the frequency of interest. [7]
Figure 4: Multipoint ground system: Ideal case
13
4.8
Interference Detection Procedures
There are several methods used to determine radiated EMI in a system. Some common
methods employed in industry are OTA measurements, EMI screening, probing or
sniffing and actual software tests performed by the mobile phone.
Below is a tabulated summary of the possible measurement and investigation scenarios
proposed for this project. On the right side of each measurement or investigation
scenario is the reason why it should be performed.
Measurements and
Investigations
Reasons
Screening for EMI
To locate possible sources of electromagnetic radiation
and regions of high EMI. A method used to determine
which components contribute to EMI in different modes.
Examining CLK/Data
signals on PCB
To determine possible loops, sources of crosstalk, signal
coupling components and oscillator/clock signal harmonic
interferences.
Spectral analysis of
interferences in
different modes
To find out radiated frequencies and physically locate
components; the sources of interferences. This method is
also called sniffing.
OTA Measurements
(BER RX / Sensitivity)
To find the effect of interference on RX sensitivity / BER in
real life scenarios where the phone is connected to a
network.
Table 3: Different Measurement and Investigation Scenarios
14
5
Possible Solutions
This chapter describes guidelines on how to tackle the problem of EMI internal to the
system. EMC describes the ability of electronic and electrical systems or components to
function correctly when they are close together. In practice this means that the
electromagnetic interference from each device must be limited and also that each
device must have an adequate level of immunity to the interference in its environment
the same holds in micro-level where components inside the device itself interfere with
each other.
There are some emitters whose emissions can serve unintentional radiation; these
emitted signals are regarded as interference signals or EMI. In general there are three
main methods to prevent this interference; other methods are described in the following
sections and in more detail:
•
•
•
Suppress the emission
Make coupling path as efficient as possible.
Make the receptor less susceptible to the emission.
The most important concept is to suppress the emission as much as possible at the
source. For example, we can determine that fast rise/fall times are primary contributors
to the high-frequency spectral content of these signals. In general higher the frequency
of the signal to be passed through the coupling path the more efficient the coupling
path. So we should slow the rise/fall times of the digital signals to improve the efficiency
of the coupling path thus reduce EMI.
One can not slow the rise/fall times without considering the normal operation signaling
conditions required by various devices dependant on those signals. By reducing the
high-frequency spectral content of an emission tends to inherently reduce the efficiency
of the coupling path and hence reduces the signal level at the receptor.
Other brute force methods are viable, but the cost-effectiveness is questionable. For
example, placing the receptor in a metal enclosure or a shield will serve to reduce the
efficiency of the coupling path. But shielded enclosures are more expensive than
reducing the rise/fall time of the emitter, and sometimes the resulting performance is far
less than the ideal. [7]
5.1
Layout & Placement of Components
During the PCB design phase, the layout designer should foresee the need of
suppression for some clock harmonics in the future when the product is eventually
tested for compliance. If the PCB designer places pads at the output of the clock on the
PCB, a capacitor can be easily inserted, across the clock terminals, thus reducing the
emissions of the clock. Also pads may be placed in series with the trace of a clock
output to provide for later insertion of a series resistor to additionally reduce the clock
15
rise/fall times and further reduce the high frequency emissions of the clock signal. In the
initial design the capacitor pads can be left vacant and the series resistor pads can be
wired across with a 0 Ω resistor. If problems with the clock emissions occur during
testing, a capacitor can be inserted and only the PCB artwork and the product parts list
needs to be changed. If this is not done, the entire PCB would need to be re-laid out,
thus by adhering to certain design principles and maintaining the necessary EMC
insight throughout the design will tend to make the necessary suppression easy to apply
and at minimal expense.
For technical reasons, it is best to use a multi-layer printed circuit board with a separate
layer dedicated to the ground and another one to the VDD supply, which results in good
decoupling, as well as good shielding effect. For many applications, economical
requirements prohibit the use of this type of board. In this case, the most important
feature is to ensure good structure for the ground and power supply. [7]
The layout of a digital logic board can have a significant effect on its performance. The
edges of the waveforms being very fast, the frequencies that are contained within
waveforms are particularly high. Accordingly leads should be kept as short as possible if
the circuit is to be able to perform correctly.
A preliminary layout of the PCB must separated the different circuits according to their
EMI contribution in order to reduce cross coupling on the PCB, i.e. noisy, high-current
circuits, low voltage circuits and digital components.
There are also some general guidelines to minimize EMI with a good component
placement, like:
•
Place all components associated with one clock trace closely together to
reduce the trace length and reduce radiation.
•
Keep oscillators, and clock generators away from I/O ports and board edges,
EMI from these devices can be coupled onto the I/O ports.
•
Place high-current devices as closely as possible to the power sources. [6]
5.2
Grounding Solutions
For circuit board grounds, it is almost impossible to get good low-impedance grounds on
two-sided circuit boards, so it is critical to keep ESD currents and high-level radiofrequency interference off such boards. On the other hand, it is easy to achieve low
impedances with the ground plane underneath the traces on multilayer boards. Circuits
built immediately above the ground plane are well protected, regardless of the threat.
EMI problems are frequently the result of high-impedance interconnects. Again,
designers need to keep the ground impedance low; either by connecting the circuit
boards or modules to a common ground plane or by providing a very-low-impedance
ground interconnects, usually by allocating as many connector pins to grounds as
possible. Although the connector space is an important concern, so is functionality. [8]
16
5.3
EMI Suppression by Filtering
The most important means of reducing electromagnetic interference are: the use of
bypass or decoupling capacitors on each active device connected across the power
supply, as close to the device as possible, or control the rise time of high speed signals
using series resistors and VCC filtering. Shielding is usually a last resort after other
techniques have failed because of the added expense of RF gaskets and the like.
EMI suppression filters are used to suppress noise produced by conductors. Noise
radiation can be suppressed, if it is eliminated with a filter in advance. Generally, such
noise suppression is achieved with DC EMI suppression filters, according to the
capacitive and inductive frequency characteristics of the respective conductors in the
circuit.
Filters of this kind can be roughly divided into those; employing a capacitor, an inductor
or a capacitor and inductor combination.
When a decoupling capacitor is connected from a noisy signal line or power line to
ground; the circuit impedance decreases as the frequency increases. Since noise is a
high frequency phenomenon, it flows to ground if a capacitor has been connected to
ground, thereby making it possible to eliminate noise, see figure below. EMI
suppression filters employing a capacitor in this way are used to eliminate this type of
noise (See Figure 6).
Figure 5: Capacitive Noise Suppression
1
where f is the frequency and C is the value of the capacitance.
2πfC
When an inductor is inserted in series in a noise producing circuit, see figure below, its
impedance increases with frequency. In this configuration it is possible to attenuate and
eliminate noise components which are the high frequency components found in the
signal. (See Figure 7)
Z =
Figure 6: Inductive Noise Suppression
Z = 2πfL where f is the frequency and L is the value of the inductance.
If capacitive and inductive suppression characteristics are combined, it is possible to
configure a much higher performance filter. In signal circuit applications where this
17
combination is applied, noise suppression effects which have little influence on the
signal wave form become possible.
This type of filter is also effective in the suppression of high-speed signal circuit noise.
When used in DC power circuits, capacitive-inductive filters prevent resonance from
occurring in peripheral circuits, thus making it possible to achieve significant noise
suppression under normal service conditions.
5.3.1
Noise Suppression using Higher Order Filters
The order of the filter determines the sharpness of the rising edge of the processed
signal, according to the simulations performed it turned out that the higher the filter
order, the slower is the rising edge, thus higher order filters might cause input/ouput
driver not meet specifications and have signal integrity problems. Also in higher order
filter simulations the best rising time in a certain order of filter is different for different
models of the same filter type (e.g. for 3rd order filters: the rising time was the best for
the Butterworth Pi-model filter, while for the T-model filter the Chebyshev I filter was the
best.) To have an idea about different filtering topologies, signal waveforms and EMI
Suppression effect. [See Appendix B]
The table below describes different filtering solutions on a transmission line having a
specific length and driver technology driving the input/output of the transmission line:
Type of Filter
Signal Waveform
1920.0
Signal Rising/Falling
Edge & Noise
Spectrum before/after
filter Mounting
1st Order Chebyshev I
PB Ripple = 1dB
1520.0
1320.0
V
o
l
t
a
g
e
m
V
-
1120.0
100uA
920.0
720.0
520.0
320.0
120.0
10uA
-80.0
0.000
4.000
8.000
12.000
Tim e (ns)
16.000
20.000
Probe
Probe
0
500.000 MHz
1.000 GHz
0
500.000 MHz
1.000 GHz
0
500.000 MHz
1.000 GHz
1mA
1720.0
1520.0
1320.0
V
o
l
t
a
g
e
m
V
-
1120.0
100uA
920.0
720.0
520.0
320.0
120.0
-80.0
0.000
10uA
4.000
8.000
12.000
Time (ns)
16.000
1920.0
Signal Rising/Falling
Edge & Noise
Spectrum before/after
filter Mounting
5th Order Chebyshev I
PB Ripple = 1dB
1mA
1720.0
1920.0
Signal Rising/Falling
Edge & Noise
Spectrum before/after
filter Mounting
3rd Order Chebyshev I
PB Ripple = 1dB
EMI Suppression Effect
Probe
Probe
20.000
Probe
Probe
1mA
1720.0
1520.0
1320.0
V
o
l
t
a
g
e
m
V
-
1120.0
100uA
920.0
720.0
520.0
320.0
120.0
-80.0
0.000
10uA
4.000
8.000
12.000
Time (ns)
16.000
20.000
Table 4: Different Chebyshev Filtering Solutions
18
5.3.2
Distortion suppression using filters
There are many methods that can be used to reduce distortion in a signal. The table
below describes basic methods for distortion suppression: series resistive termination
and RC filtering solutions that reduce the ringing effect and distortion of a clock signal
improving its signal integrity.
Type of Signal
EMI Suppression Effect
7.000
P
P
6.000
5.000
4.000
Initial waveform
V
o
l
t
a
g
e
V
-
3.000
2.000
1.000
0.000
-1.000
-2.000
-3.000
0.000
10.000
20.000
30.000
Time (ns)
40.000
50.00
10.000
20.000
30.000
Time (ns)
40.000
50.00
10.000
20.000
30.000
Time (ns)
40.000
50.00
7.000
6.000
5.000
4.000
Series resistor termination
applied / waveform ringing
effect reduced.
V
o
l
t
a
g
e
V
-
3.000
2.000
1.000
0.000
-1.000
-2.000
-3.000
0.000
7.000
6.000
5.000
4.000
RC filtering solution
applied / waveform
distortion is suppressed
V
o
l
t
a
g
e
V
-
3.000
2.000
1.000
0.000
-1.000
-2.000
-3.000
0.000
Table 5: Different Distortion Suppressing Solutions
19
5.4
Shielding Solutions
It is a form of containment used to preventing RF energy from exiting an enclosure,
generally by shielding a product within a metal enclosure or by using a plastic housing
with RF conductive paint. By reciprocity, we can also speak of containment as
preventing RF energy from entering the enclosure. Therefore a shield is, conceptually, a
barrier to transmission of electromagnetic fields, see figures below.
Figure 7: Shield that Contains
Radiated Emission
Figure 8: Shield that Excludes Radiated
Emission
In order to realize an effective shielding, the shield must enclose the electronics, and
must have no penetrations such as holes, seams or slots. [7]
20
6
RSSI Measurements
In this chapter we examine the critical interference scenarios in different operation
modes; measurement results are presented in the noise spectrum plots. Various
possible scenario combinations were enumerated; a total of 52 cases were investigated;
and only the critical scenarios are presented in noise spectrum plots.
6.1
Measurement Scenarios
Different measurements scenarios were investigated; some cases were discarded since
they do not occur in real life. So after filtering out all the non-occurring scenarios and the
ones that can not be measured OTA, only three cases remained having critical levels of
interference in the GSM band in addition to the reference scenarios.
21
6.2
Measurements where Cameras Active/Inactive & Analysis
EGSM 900 Spectrum
-90
-95
-95
-100
-100
RSSI of channel
RSSI of channel
EGSM 900 Spectrum
-90
-105
X: 936
Y : -110
-105
-110
X: 936
Y : -111
-110
-112.7874
-114.1149
-115
-115
-120
925
930
935
940
945
950
Channel Frequency in MHz
955
-120
960
925
Figure 9: Noise Spectrum
(Cameras Inactive – Backlight ON – Flash OFF)
930
955
960
Figure 10: Noise Spectrum
(Cameras Inactive – Backlight OFF – Flash OFF)
EGSM 900 Spectrum
EGSM 900 Spectrum
-90
-90
-95
-95
X: 936
Y : -100
X: 936
Y: -101
-100
RSSI of channel
-100
RSSI of channel
935
940
945
950
Channel Frequency in MHz
X: 949
Y: -107
-105
-110
X: 949
Y : -105
X: 955.8
Y : -106
-105
-110
-112.0575
-112.954
-115
-115
-120
925
930
935
940
945
950
Channel Frequency in MHz
955
-120
960
925
930
935
940
945
950
Channel Frequency in MHz
955
960
Figure 12: Noise Spectrum
(2MPx CAM Active – Backlight Off – Flash OFF)
Figure 11: Noise Spectrum
(VGA CAM Active - Backlight OFF - Flash OFF)
EGSM 900 Spectrum
-90
-95
X: 936
Y : -101
RSSI of channel
-100
X: 949
Y : -103
X: 955.6
Y : -107
-105
-110
-112.7759
-115
-120
925
930
935
940
945
950
Channel Frequency in MHz
955
960
Figure 13: Noise Spectrum (2MPx CAM Active – Backlight On – Flash Off)
In the first two cases there were no high levels of interference, only a single 13MHz
harmonic appeared at 936MHz but it had negligible interference strength. From several
scenarios only these three cases were chosen to perform further investigations on. (See
figures 12, 13 and 14)
22
6.3
Measurements in presence of WLAN & Analysis
In presence of WLAN transmissions an interference was noticed in the DCS band at
lower WLAN transmission channels (channels 1 to 7), this interference level diminished
in strength as transmission was at higher channels. The levels of interference are much
below the threshold level.
DCS 1800 Spectrum
EGSM 900 Spectrum
-80
-95
-90
RSSI of channel in dBm
-90
RSSI of channel
-100
-105
-110
-113.4598
-115
-120
X: 1872
Y: -104
-100
-110
-112.7914
-120
-130
925
930
935
940
945
950
Channel Frequency in MHz
955
960
1810
Figure 14: Noise Spectrum (WLAN TX1 Backlight OFF - Flash OFF in EGSM900)
1820
1830
1840
1850
1860
Channel Frequency in MHz
1870
1880
Figure 15: Noise Spectrum (WLAN TX1 Backlight OFF - Flash OFF in DCS1800)
EGSM 900 Spectrum
DCS 1800 Spectrum
-90
-85
-90
-95
RSSI of channel in dBm
-95
RSSI of channel
-100
-105
X: 936
Y : -111
-110
X: 1872
Y : -105
-100
-105
-110
-112.8048
-115
-120
-125
-114.0805
-115
-130
-135
-120
925
930
935
940
945
950
Channel Frequency in MHz
955
960
1810
Figure 16: Noise Spectrum (WLAN TX7 Backlight OFF- Flash OFF in EGSM900)
1820
1830
1840
1850
1860
Channel Frequency in MHz
1870
1880
Figure 17: Noise Spectrum (WLAN TX7 Backlight OFF - Flash OFF in DCS1800)
EGSM 900 Spectrum
DCS 1800 Spectrum
-90
-90
RSSI of channel in dBm
-95
RSSI of channel
-100
-105
X: 936
Y : -109
-110
-113.954
-115
-120
-100
X: 1872
Y : -107
-110
-112.615
-120
-130
-140
925
930
935
940
945
950
Channel Frequency in MHz
955
960
1810
Figure 18: Noise Spectrum (WLAN TX13 Backlight OFF - Flash OFF in EGSM900)
23
1820
1830
1840
1850
1860
Channel Frequency in MHz
1870
1880
Figure 19: Noise Spectrum (WLAN TX13
- Backlight OFF - Flash OFF in
DCS1800)
7
OTA Sensitivity Measurements
In this chapter we investigate the sensitivity of the receiver in the three different scenarios as
selected in the previous chapter where high levels of interference exist. Over-the-air tests
determine how a specific network will influence the connectivity performance of a mobile
handset. It can yield data that can be used to demonstrate that a product meets
performance criteria.
Over-the-air performance tests measure the magnitude and direction of
transmitted/received energy, sensitivity and BER to determine the performance of the
wireless device. A typical polar pattern configuration is obtained with the measurement
antenna(s) fixed the antenna under test is rotated through 360°. [9]
In OTA tests it is important to determine the influence of the user’s body on the
transceiver properties and performance and the differences between free-space and the
standard anthropomorphic model. SAM phantom tests determine the blocking effect of
the human head on the antenna pattern, but do not address radiation absorption,
hazard or health and safety issues [9]. Below is typical OTA test system.
Network
Analyzer
Device Under
Test (DUT)
Spectrum
Analyzer
Diagonal dual
polarized horn
Universal Radio
Communication
Tester
Relay Switch
Unit
walkway
`
MAPS
Controller
Fibre Optics MAPS system
Figure 20: DUT in a Typical OTA Test System
To have a reliable channel sensitivity scan, we acquired the average of 5000 samples of
data; a threshold BER of 2.439 percent in order for the call not to be disconnected.
Transmission from the MS was performed at minimal power to reduce battery
consumption for longer hours of measurement. The start level of scanning was -100
dBm for the software to be able to locate interference peaks below this level in a faster
seek time.
24
The DUT was positioned in “talk position” so as to emulate actual situation where there
are reflections from the human head, although we could not perform measurements
considering the human “hand effect”. We can assume that the obtained results in “talk
position” will be the worst case, since reflections will be directly heading towards the
antenna, so no need to consider the other case.
Figure 21: DUT in Talk Position
7.1
Measurement Procedure
Initially we started with a basic scan of every tenth channel of the EGSM band, after
obtaining the results of the first pass we made a second pass taking more samples
adjacent to the interference channels. Finally we scanned extra channels as a result of
the sniffing results discussed in the following chapter.
We obtained about 42 channels out of 174 channels which constitutes 24 percent of the
total channels for the first case. We had 27 channels which constitute 16 percent of the
channels in the band and for the second case, but for the last case only 17 channels
which constitute about 10 percent of the channels were required since this case is a
reference. These percentages were fair enough to cover all the possible interference
frequencies. The first and second cases had the 2MPx camera active, and the backlight
on or off respectively. The third was a reference case used to determine the average
sensitivity of the band when the cameras and backlight are off.
25
7.2
Measurement Results and Analysis
For different DUTs we obtained different interference levels at 955.4 MHz frequency on
the EGSM900 band (Figures 24 & 25). This was due to the loose shield of the 2MPx
camera, which was not containing radiated emissions of the camera MCLK_2M and its
internal oscillator. Resistance measurements were performed between the shielded
boxes and the interference at 955.6 MHz disappeared when more pressure was applied
or 2MPx shield box changed. This interference and the other 13MHz harmonic
frequency interferences were blocked and suppressed in the solutions chapter.
EGSM 900 Spectrum
EGSM 900 Spectrum
-90
-90
-95
-95
X: 955.4
Y : -97.94
-100
-100
X: 936
Y : -102.5
Sensitivity of channel
Sensitivity of channel
X: 936
Y: -102.5
X: 949
Y: -105.8
-105
X: 936.6
Y: -107.1
X: 943.4
Y : -109
X: 949
Y : -105.8
-105
X: 936.6
Y : -107.1
-108.3219
-109.2088
-110
-110
-115
-115
-120
925
930
935
940
Channel Frequency in MHz
945
950
-120
955
Figure 22: Noise Spectrum with Good Shield
(2MPx CAM Active – Backlight ON – Flash OFF)
925
930
935
940
Channel Frequency in MHz
945
950
955
Figure 23: Noise Spectrum with Loose Shield
(2MPx CAM Active – Backlight ON – Flash OFF)
EGSM 900 Spectrum
-90
-95
Sensitivity of channel
-100
X: 936
Y : -104.5
-105
X: 948
Y : -108.5
X: 943.4
Y : -110.3
-110.0085
-110
-115
-120
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 24: Noise Spectrum
(Cameras Inactive – Backlight ON – Flash OFF)
In the measurement results above high levels of interference were noticed at 936 MHz
and 949 MHz; the backlight had some contribution in the interference at these
frequencies.
26
EGSM 900 Spectrum
-90
-95
RSSI of channel
-100
-105
X: 936
Y: -106.7
-109.4071
-110
-115
-120
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 25: Noise Spectrum
(Cameras Inactive – Backlight OFF – Flash OFF)
This case is a reference used to determine the average sensitivity levels at the receiver
when the cameras, backlight and flash are off.
27
8
Sniffing Measurements
In an electronic system, the primary emission sources are currents flowing in circuits
like clocks, data drivers, oscillators and other components that are mounted on printed
circuit boards.
Although the radiated measurements in the laboratory environment can not be accurate
enough, it is possible to establish a minimum set-up in the lab at which one can perform
emissions diagnosis and carry out comparative tests.
Analyzing the data obtained from the sniffing measurements, we noticed that several
components emitting interference signals above a certain threshold level. In this chapter
we try to highlight these components and describe the key interferences.
Probing or “sniffing” is a measurement methodology used to physically locate the
source of emissions from a product. After detaching external mechanics of the DUT,
and scanning its internal components using a probe or sniffer loop antenna.
The emission spectrum can be observed through the spectrum analyzer, and relative
emission strength can be concluded. The advantage of sniffing is that the source of
interference can be directly located on the PCB, the disadvantage is that the
interference amplitude is dependant on the distance between the probe and the
component, so any variation in the distance causes an uncertainty in the amplitude of
the actual interference, causing the peak value of the interfering signal vary.
Figure 26: Spectrum Analyzer
A near-field or “sniffer” probe was used to physically locate the source of emissions
from the product, since we only want to detect the field strengths in the near field. The
probe is connected to the spectrum analyzer for frequency domain display.
Probe design is a trade-off between sensitivity and special accuracy. The smaller the
probe, the more accurately it can locate signals but the less sensitive it will be. The
sensitivity can be increased with a preamplifier if working with low power circuits.
The sum of radiating sources will differ between near and far fields and the probe will
itself distort the field it is measuring. Perhaps one might mistake a particular hot spot
28
found on the circuit board for the actual radiating point, whereas the radiation is in fact
coming from cables or other structures that are coupled to this point via an often
complex path. Thus probes are best used for tracing and for comparative rather than
absolute measurements [10].
8.1
Measurement Procedure
Sniffing measurements are performed in a shielded chamber to avoid external
interferences. The device has its external mechanics detached to expose the primary
and secondary layers. Before the detailed sniffing measurements, a fast pre-scan
through is performed to locate major peaks or interference frequencies. This
enormously speeds up the sweep rate for a whole frequency scan.
8.2
Measurement Cases
Based on the RSSI measurement results of the previous phases of the project, and
after dropping the unnecessary cases, we were left with three cases for the secondary
layer and two cases for the primary layer as shown in the table below.
Device sides
Secondary
Primary
2MPx
Camera
Inactive
Inactive
Active
Inactive
Active
Backlight
Flash
ON
OFF
ON
ON
ON
OFF
OFF
OFF
OFF
OFF
Table 6: Sniffing Measurements on Primary and Secondary Sides
29
8.3
Measurement Results and Analysis
On the primary side of the PCB a new frequency of interference at 943.6MHz was found
that did not appear in the previous measurement results, this frequency appeared after
the pre-scan performed on different components of this side of the PCB. Thus the new
sniffing spectral scans were based on four frequencies, 936MHz, 943.6 MHz, 949MHz
and 955.6MHz.
On this side most of components were covered with shielding boxes which were quite
effective to block emissions of internal components to the exterior and vice versa. So
the measurements were mainly done on the uncovered components located on the
PCB and the flex cables.
Many adjacent peaks appear around the center frequency, especially 936MHz and
955.4MHz in (2MPx Cam active – Backlight On – Flash OFF) case. But those adjacent
peaks disappeared when cameras and backlight were off.
Also on this layer, a new peak frequency was also found at 943.6MHz. This interference
was mainly contributed by a capacitor, an inductor and the flex cable between PCB
/LCD and the capacitors that are placed on the flex cable. The two mega pixel camera’s
shielding box was loose, so while sniffing high levels of emission were observed.
While on the secondary side and after examining the obtained results we noticed that
the 2 mega-pixel camera when active will cause more interference; not only on the key
interference channels, but also its adjacent channels. Similarly the backlight will
contribute in raising the interference to higher levels, but will not introduce other
interference frequencies in the spectrum.
On the secondary side there are two key components that contribute to high levels of
interference, these components are diodes. The first one causes very high peak at 936
MHz (-88.56dBm ~ -95.03dBm) for all three cases, a high peak (-93.29dBm ~ 98.3dBm) at 949 MHz and also high peak (-95.73dBm) at 955.4 MHz. The second one
causes very high peak and noise floor in certain cases. Only when camera, backlight
and flash are all off, the noise floor falls back to normal value (around -110dBm). Note
that the noise floor can not be measured from the spectrum analyzer directly, so this
average values are only estimates.
30
8.4
Sources of Uncertainty
Radiated interference, whether intentional or not, decreases in strength with distance
from the source. For radiated fields in free space, the decrease is inversely proportional
dis tan ce < λ
2π .
to the distance provided that the measurement is made in the near field
EMC measurements are inherently less accurate than most other types of
measurements. It is always wise to allow a margin of about 10dB between
measurements and the specification limits, not only to cover measurement uncertainty
but also tolerances arising in production. These uncertainties could arise from different
factors such as: instrument and cable errors, mismatch errors in cable impedance,
antenna calibration issues, reflections not only from DUT but also from the ground plane
and antenna cable. Thus we can say that the radiated measurements in the laboratory
environment can not be accurate enough, but it is possible to perform emissions
diagnosis and carry out comparative tests based on the results obtained through these
measurements.
Performing sniffing measurements on the product helps to physically locate the sources
of emission. Although the radiated measurements can not be accurate enough, one can
carry out comparative tests and observe the key interference components. [10]
31
9
Recommended Solutions
9.1
LCD Display Data/Control lines
A cause of EMI could be the CLK signal going in/out of connector from PCB to flex
cable. For this case we tried to put decoupling capacitors on the data lines, but the EMI
contribution of this section seemed to be quite negligible on the GSM receiver due to
the distance from the receiver antenna. Thus this solution did not solve the interference
problem. Also we tried to use a copper tape to shield the flex cable, but no improvement
in the sensitivity of the receiver was noticed.
Figure 27: A solution for Data/Control lines between PCB and LCD display
9.2
Radiated Emissions from Flex Cables
While sniffing the LCD, VGA CAM and 2MPx CAM flex cable, high levels of interference
was found on 936 MHz frequency. The interference contribution of the flex cables was
noticeable in OTA sensitivity measurements also. The solutions provided could not be
implemented on the flex cables, only solutions were done on the PCB.
32
9.3
Improper Grounding of Shield Boxes
For the measurements performed in the table below we used a digital multi-meter, two
probes and the DUT. We had the external mechanical cover of the device removed in
order to access the internal shield boxes and perform detailed measurements.
Compared to the older DUTs the new shields required more pressure to be applied in
order to get better contact with ground, thus for any reason lesser pressure resulted in
higher interference levels at certain frequencies.
DUT 1
High resistance between the
two points: 22.5 Ω (2 MPx
Camera Shield box – RF
Shield box)
DUT 2
High resistance between the
two points: 46 ~ 68 Ω (2 MPx
Camera Shield box – RF
Shield box)
DUT 3
Lower resistance between
the two points:2 ~ 3 Ω
(2 MPx Camera Shield box –
RF Shield box)
Table 7: Improper Grounding of 2MPx CAM Shield Box
The problem of interference at 955.6 MHz was solved by changing the shield box in
later stages.
33
9.4
2MPx CAM Subsystem Interferences
This section includes simulation results using Mentor Graphics Hyperlynx software,
possible solutions to the MCLK_2M trace line were proposed and these solutions were
implemented and verified to be consistent with the simulation results.
Meeting the specifications of the two mega pixel camera (rising time/slew rate
limitations) was the challenge. In this case the specifications were not met after applying
solutions, but the camera was functional.
CAM_DAT[i]
CAM_DAT[i]
CAM_DAT[i]
CAM_DAT[i]
CAM_DAT[i]
CAM_DAT[i]
CAM_DAT[i]
CAM_DAT[i]
n
MCLK_2M
PIX_CLK
Other_CLK
Figure 28: Data/CLK lines between PCB and 2MPx CAM
We tried to apply different filtering solutions to the MCLK_2M in order to reduce the EMI
resulting from the clock line and its harmonics. The appendix includes different filtering
types (Butterworth, Chebyshev, Bessel and Legendre filters), topologies (t-model and
pi-model) and orders (1st, 3rd and 5th). These results were compared and a conclusion
was made in section 5.3.1.
34
9.4.1
Simulations
Type of Filter
Signal Waveform
EMI Suppression Effect
1mA
1960.0
1760.0
1560.0
MCLK_2M with
decoupling
capacitor of
33pF
1360.0
V
o
l
t
a
g
e
m
V
-
100uA
1160.0
960.0
760.0
560.0
360.0
10uA
0
160.0
-40.0
0.000
4.000
8.000
12.000
Time (ns)
16.000
500.000MHz
1.000GHz
20.0
1mA
1960.0
MCLK_2M with
a series
termination of
22 Ω and
decoupling
capacitor of
22pF.
1760.0
1560.0
1360.0
V
o
l
t
a
g
e
m
V
-
100uA
1160.0
960.0
760.0
560.0
360.0
10uA
0
160.0
-40.0
0.000
4.000
8.000
12.000
Time (ns)
16.000
20.00
500.000 MHz
1.000 GHz
500.000MHz
1.000GHz
1mA
1960.0
Pr
Pr
1760.0
1560.0
MCLK_2M with
a resistive
termination of
220 Ω
1360.0
V
o
l
t
a
g
e
m
V
-
100uA
1160.0
960.0
760.0
560.0
360.0
160.0
-40.0
0.000
4.000
8.000
12.000
Time (ns)
16.000
20.000
Table 8: Solutions Verification for MCLK_2M
35
10uA
0
9.4.2
Oscilloscope Measurements
The oscilloscope measurements results were similar to the simulation results, only the
falling edge had a 1ns delay. In the figure below the fastest rising edge resembles the
original MCLK_2M signal; the other two are the rising edges after applying the 33 pF
capacitor and the one with the 22 Ω and 22 pF RC filtering solutions.
Figure 29: Oscilloscope Measurements of MCLK_2M Rise Times
36
9.4.3
RSSI Results & Analysis
Original MCLK_2M Signal
EGSM 900 Spectrum
EGSM 900 Spectrum
-90
-90
-95
-95
X: 936
Y: -98
X: 936
Y : -99
X: 955.6
Y: -101
-100
X: 948.8
Y: -104
RSSI of channel
RSSI of channel
-100
-105
X: 948.8
Y : -106
-105
X: 955.6
Y : -106
-110
-110
-111.6724
-112.5977
-115
-115
-120
-120
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 30: Noise Spectrum (2MPx CAM
Active – Backlight Off – Flash OFF)
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 31: Noise Spectrum (2MPx CAM
Active – Backlight ON – Flash OFF)
For both cases the interference level at 936MHz was -98 ~ -99 dBm and at 949 MHz
was -104 ~ -106 dBm, the interference at 955.6 Mhz was excluded since the shield box
covering the 2MPx camera totally discards it when properly grounded.
MCLK_2M with Decoupling Capacitor
EGSM 900 Spectrum
-90
-95
-95
-100
-100
RSSI of channel
RSSI of channel
EGSM 900 Spectrum
-90
X: 936
Y : -106
-105
-105
X: 936
Y: -107
X: 949
Y : -110
X: 949
Y: -110
X: 957
Y : -111
-110
X: 951.4
Y: -111
-110
-112.3391
-113.1724
-115
-120
-115
925
930
935
940
945
Channel Frequency in MHz
950
955
-120
960
Figure 32: Noise Spectrum (2MPx CAM
Active – Backlight Off – Flash OFF)
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 33: Noise Spectrum (2MPx CAM
Active – Backlight On – Flash OFF)
For both cases the interference level at 936 MHz was -106 ~ -107 dBm and at 949 MHz
was -110 dBm, thus resulting in interference reduction of 8dB at 936 MHz and 4 ~ 6 dB
at 949 MHz.
37
MCLK_2M with series Resistive Termination and Decoupling Capacitor
EGSM 900 Spectrum
-90
-95
-95
-100
-100
RSSI of channel
RSSI of channel
EGSM 900 Spectrum
-90
X: 936
Y: -106
-105
X: 949
Y : -108
-105
X: 936
Y: -107
X: 949
Y: -109
-110
-110
-112.3966
-112.9655
-115
-120
-115
925
930
935
940
945
Channel Frequency in MHz
950
955
-120
960
Figure 34: Noise Spectrum (2MPx CAM
Active – Backlight Off – Flash OFF)
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 35: Noise Spectrum (2MPx CAM
Active – Backlight On – Flash OFF)
For both cases the interference level at 936 MHz was -106 ~ -107 dBm and at 949 MHz
was -108 ~ -109 dBm, thus resulting in interference reduction of 6dB at 936 MHz and 3
~ 6 dB at 949 MHz.
MCLK_2M with series termination
EGSM 900 Spectrum
EGSM 900 Spectrum
-90
-90
-95
-95
-100
-100
RSSI of channel
RSSI of channel
X: 936
Y: -103
X: 936
Y: -104
-105
X: 949
Y : -108
-105
X: 949
Y: -108
-110
-110
-112.4253
-112.4713
-115
-115
-120
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 36: Noise Spectrum (2MPx CAM
Active – Backlight Off – Flash OFF)
-120
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 37: Noise Spectrum (2MPx CAM
Active – Backlight On – Flash OFF)
For both cases the interference level at 936 MHz was -103 ~ -104 dBm and at 949 MHz
was -108 dBm, thus resulting in interference reduction of 3 dB at 936 MHz and 3 ~ 5 dB
at 949 MHz.
According to the experimental results obtained, the solution of having a decoupling
capacitor of 33pF was found to be the best. Note that the relative measurements were
performed on two different devices.
38
9.5
VGA CAM Subsystem Interferences
A cause of EMI could be the CLK signals going in/out of connector from PCB to flex
cable.
Figure 38: Data/CLK lines between PCB and VGA CAM
Filtering is an efficient way to decrease clock rising edge, since longer the rising time,
lesser the interference at higher frequencies above 500 MHz. It is important to choose
suitable values for the resistor and capacitor, since there is a trade-off between rising
time and noise spectrum of the signal. Below is the schematic of the VGA CAM
connector, and the proposed solution to suppress the EMI.
Figure 39: A Solution for Data/CLK lines between PCB and VGA CAM
39
9.5.1
Simulations
Figures 41, 42, 43 and 44 present a comparison of the noise spectrum before and after
adding RC filtering solutions to the trace involved.
1mA
1mA
100uA
100uA
10uA
10uA
0
500.000 MHz
1.000 GHz
Figure 40: Noise Spectrum before adding
RC filter
0
500.000 MHz
1.000 GHz
Figure 41: Noise Spectrum after adding
RC filter (R=33Ω, C=22pf)
1mA
1mA
100uA
100uA
10uA
10uA
0
500.000 MHz
1.000 GHz
Figure 42: Noise Spectrum after adding
RC filter (R=220Ω, C=22pf)
40
0
500.000 MHz
1.000 GHz
Figure 43: Noise Spectrum after adding
RC filter (R=0Ω, C=33pf)
9.5.2
Oscilloscope Measurements
Figures 45, 46, 47 and 48 present the results of the oscilloscope measurement of
MCLK / PCLK, it is evident how the resistor and capacitor values effect the rising time of
both clocks. Note that the rising time is read from 10% ~ 90% rising time of the signal.
Figure 44: VGA_MCLK on test point ¾
(R = 0Ω, C = 33pf)
Figure 45: VGA_MCLK on test point 4
(R = 33Ω, C = 22pf)
Figure 46: VGA_MCLK on test point 4
(R = 220Ω, C = 22pf)
Figure 47: VGA_MCLK on test point ¾
(R = 0Ω, C = 22pf)
Test
point
Rising time(ns)
R = 0Ω, C = 33pf
Rising time(ns)
R = 33Ω, C = 22pf
Rising time(ns)
R = 220Ω, C = 22pf
Rising time(ns)
R = 0Ω, C = 22pf
3
4
5.347
------9.739
------11.13
8.464
Table 9: Rising Time Summary of MCLK_VGA
41
Figure 48: VGA_PCLK on test point ½
(Initially R = 0Ω, no capacitor)
Figure 49: VGA_PCLK on test point 2
(R = 33 Ω, no capacitor)
Figure 50: VGA_PCLK on test point 2
(R = 220 Ω, no capacitor)
Figure 51: VGA_PCLK on test point ½
(R = 0Ω, C = 22 pf)
Figure 52: VGA_PCLK on test point 2
(R = 220Ω, C = 22pf)
Test
point
Rising Time(ns)
R = 0Ω, C = 0pf
1
2
6.291
Rising time(ns)
R = 33Ω, C =
0pf
Rising time(ns)
R = 220Ω, C =
0pf
------6.599
Rising time(ns)
R = 0Ω, C =
22pf
------9.621
Table 10: Rising Time Summary of PCLK_VGA
42
8.709
Rising time(ns)
R = 220Ω, C =
22pf
------11.4
9.5.3
RSSI Results & Analysis
EGSM 900 Spectrum
EGSM 900 Spectrum
-90
-90
-95
-95
X: 936
Y : -99
-100
RSSI of channel
RSSI of channel
-100
X: 949
Y : -107
-105
-110
X: 936
Y : -104
-105
X: 949
Y : -108
-110
-112.2989
-113.6149
-115
-120
-115
925
930
935
940
945
950
Channel Frequency in MHz
955
-120
960
Figure 53: Noise Spectrum of Reference
925
930
EGSM 900 Spectrum
-95
-95
-100
RSSI of channel
RSSI of channel
-100
X: 949
Y : -105
-105
-110
-110
-113.1379
-115
-115
930
935
940
945
950
Channel Frequency in MHz
955
-120
960
X: 949
Y : -105
925
930
935
940
945
950
Channel Frequency in MHz
955
960
Figure 56: Noise Spectrum (2M pixel camera inactive
– VGA camera ON – Backlight OFF)
Figure 55: Noise Spectrum after adding RC
filter on PCLK (MCLK-R 0Ω&C 33pf, PCLK-R
220Ω&C 22pf)
Reference
After decoupling
R = 33Ω&C = 22pf
After decoupling
R = 220Ω&C = 22pf
After decoupling on PCLK
Rpclk = 220Ω&Cpclk = 22pf
X: 936
Y : -103
-105
-112.8851
925
960
EGSM 900 Spectrum
-90
-120
955
Figure 54: Noise Spectrum after adding RC filter_1
(MCLK-R 33Ω & C 22pf, PCLK-R 33Ω & C 22pf)
-90
X: 936
Y : -105
935
940
945
950
Channel Frequency in MHz
Peak/channel
-99dBm / 5
Peak(dBm)/channel
-107dBm / 70
-104dBm / 5
-108dBm / 70
-105dBm / 5
-105dBm / 70
-103dBm / 5
-105dBm / 70
Table 11: Summary of MCLK_VGA & PCLK_VGA Solutions
Larger resistor values can make more effective EMI suppression, by further damping
the rising edge of the signal to reduce higher frequency interference. Though, the timing
requirements for the clock rising edge should be acceptable. According to the
oscilloscope plots, almost no timing margin was left for both clocks with 220 Ω resistor
that will lead to the difficulty to meet hold / setup time requirements for the sensor.
43
From the noise spectrum measurement plots we can deduce that, lower resistor values
will be a better choice, the 33Ω for resistor and 22pf for capacitor is a feasible choice.
This RC filter can cause reduction of 5dB at 936 MHz.
9.6
Radiated Emissions from Diodes
While sniffing the PCB for radiated emissions, two diodes near the RF antenna were
suspected to be possible sources of interference.
By performing over the air measurements of the receiver sensitivity, the radiation
frequencies from the device were detected, and by sniffing the approximate location of
the emitting components were verified.
9.6.1
Audio subsystem interferences
As a result of the sniffing measurements, two other sources of interference were located
which indicated the presence of a RF radiation problem on the audio circuitry which
connects the back speakers to the audio power amplifier. (Figure 59)
Figure 57: A Solution for the Audio Subsystem Interference
These two diodes are used for avoiding ESD. A proposed solution was to add four
capacitors C1, C2, C3 and C4 of 33 pf on the diodes in parallel. These four capacitors
are supposed to work to keep the frequency as original because high frequency signal
will go through capacitor instead of diode. That will be helpful to decrease interference
on high frequency.
44
EGSM 900 Spectrum
EGSM 900 Spectrum
-90
-90
-95
-95
-100
-100
X: 936
Y: -104
RSSI of channel
RSSI of channel
X: 949
Y: -102
-105
X: 949
Y: -105
X: 936
Y: -106
-105
-110
-110
-113.0287
-114.6552
-115
-115
-120
-120
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 58: Noise Spectrum of Reference
Reference
After decoupling
925
930
935
940
945
Channel Frequency in MHz
950
955
960
Figure 59: Noise Spectrum after decoupling
Peak/channel
-104dBm / 5
-106dBm / 5
Peak(dBm)/channel
-102dBm / 70
-105dBm / 70
Table 12: Summary of Audio Diode Solution
Before adding these capacitors, most of the peak levels of interference were below -102
dBm. After adding capacitors, both the noise floor and peak value decreased by 2~3 dB.
9.6.2
Power subsystem interferences
A proposed solution was to add one decoupling capacitor C5 of 33 pF and later to put a
series resistor 33 Ω.
Figure 60: A Solution for Backlight DC-DC Converter Interference
45
EGSM 900 Spectrum
EGSM 900 Spectrum
-90
-90
-95
-95
X: 936
Y : -102
X: 949
Y : -105
X: 955.6
Y : -107
-105
X: 949
Y: -103
-105
-110
-111.7011
-110
-113.2931
-115
-120
X: 936
Y: -102
-100
RSSI of channel
RSSI of channel
-100
-115
-120
925
930
935
940
945
950
Channel Frequency in MHz
955
960
925
930
935
940
945
950
Channel Frequency in MHz
955
960
Figure 62: Noise Spectrum after Solution
Figure 61: Noise Spectrum of Reference
Peak/channel
Peak(dBm)/channel
Reference
-102dBm / 5
-105dBm / 70
After decoupling
-102dBm / 5
-103dBm / 70
Table 13: Summary of Power Diode Solution
(2MPx CAM Active – VGA CAM OFF – Backlight OFF)
EGSM 900 Spectrum
EGSM 900 Spectrum
-90
-90
-95
-95
-100
RSSI of channel
RSSI of channel
-100
X: 936
Y : -104
-105
X: 949
Y : -108
-110
-110
-113.5057
-115
-113.1034
-120
X: 936
Y : -106
-105
X: 949
Y : -108
-115
925
930
935
940
945
950
Channel Frequency in MHz
955
-120
960
Figure 63: Noise Spectrum of Reference
Reference
After decoupling
925
930
935
940
945
950
Channel Frequency in MHz
955
960
Figure 64: Noise Spectrum after Decoupling
Peak/channel
-104dBm / 5
-106dBm / 5
Peak(dBm)/channel
-108dBm / 70
-108dBm / 70
Table 14: Summary of Power Diode Solution
(2MPx CAM Inactive – VGA CAM OFF – Backlight ON)
For this particular solution, no obvious improvement was noticed by adding a
decoupling capacitor.
46
Conclusion
Detecting and solving internal interference issues in a mobile phone is a formidable and
interesting task. It is formidable since it involves an overhead of measurements and
physical implementation of solutions. It is also interesting since the approach of
detecting and suppressing these interferences requires understanding of the overall
receiver functionality and requirements.
This report is a result of intensive research, theoretical background of possible sources
of interference. Actual measurements were preformed; RSSI measurements were used
to quickly determine the key interference frequencies, OTA measurements uses a more
complex setup to determine the receiver sensitivity through BER evaluation of the
received signal and sniffing measurements to physically locate the sources of
interference; to identify the physical sources of interferences, their effect on the receiver
sensitivity and overall receiver performance. Finally solutions were recommended and
implemented to suppress these interferences.
Some of the solutions resulted in good interference suppression, while others had no
obvious improvement in the sensitivity of the receiver. Lower order filters were preferred
since they had better EMI suppression results, with less EMI at the resonance
frequencies. There is always a tradeoff between the signal integrity and EMI
suppression, one should keep in mind to meet the overall system requirements and
provide solutions accordingly.
Special attention should be given to layout, the layout engineer has to adhere to certain
design guidelines, and be able to foresee futuristic EMI suppression needs. Shield
boxes should be verified to be well grounded so as not to act as antennas.
Future Work
In the future investigations, simulations and experimentation on newer technologies with
faster and more aggressive driver inputs/outputs should be performed to determine
signal integrity, coupling and EMI issues arising due to advances in technology.
Also other functionalities of the mobile phone should be investigated, like the IrDA,
Bluetooth, USB interface and others.
47
Appendix A
Conducted susceptibility – The relative inability of a product to withstand
electromagnetic energy that reaches it through external cables, power cords, and other
I/O interconnects.
Containment – Preventing RF energy from exiting an enclosure, generally by shielding a
product within a metal enclosure (faraday cage) or by using a plastic housing with RF
conductive paint. By reciprocity, we can also speak of containment as preventing RF
energy from entering the enclosure.
Electromagnetic compatibility (EMC) – The ability of a product to coexist in its intended
electromagnetic environment without causing or suffering functional degradation or
damage.
Electromagnetic interference (EMI) – A process by which disruptive electromagnetic
energy is transmitted from one electronic device to another via radiated or conducted
paths or both. In common usage, the term refers particularly to RF signals, but EMI can
occur in the frequency range from “DC to daylight.”
Immunity – A relative measure of device or system’s ability to withstand EMI exposure.
Radiated susceptibility – The relative inability of a product to withstand EMI that arrives
via free space propagation.
Radio frequency (RF) – The frequency range within which coherent electromagnetic
radiation is useful for communication purposes: roughly from 10 KHz to 100GHz. This
energy may be generated intentionally, as by a radio transmitter, or unintentionally as a
byproduct of an electronic device’s operation.
RSSI – Received Signal Strength Indication is a measurement of the strength;
not necessarily the quality; of the received signal strength in a wireless environment; it
is represented in arbitrary units.
Suppression – Designing a product to reduce or eliminate RF energy at the source
without relying on a secondary method such as a metal housing or chassis.
Susceptibility – A relative measure of device or system’s propensity to be disrupted or
damaged by EMI exposure.
48
Appendix B
PI - Model Filter
T - Model Filter
Butterworth Filter Parameters in Hyperlynx
Butterworth Filter Parameters
st
1 Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
3rd Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
5th Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
1st Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
3rd Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
5th Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
49
PI - Model Filter
T - Model Filter
Chebyshev I Filter Parameters in Hyperlynx
Chebyshev I Filter Parameters
st
1 Order
Chebyshev I
Cutoff Frequency
= 245 MHz
Pass band Ripple
= 1 dB
3rd Order
Chebyshev I
Cutoff Frequency
= 245 MHz
Pass band Ripple
= 1 dB
5th Order
Chebyshev I
Cutoff Frequency
= 245 MHz
Pass band Ripple
= 1 dB
1st Order
Chebyshev I
Cutoff Frequency
= 245 MHz
Pass band Ripple
= 1 dB
3rd Order
Chebyshev I
Cutoff Frequency
= 245 MHz
Pass band Ripple
= 1 dB
5th Order
Chebyshev I
Cutoff Frequency
= 245 MHz
Pass band Ripple
= 1 dB
Rg
11.2 pH
RL
50
PI - Model Filter
T - Model Filter
Bessel Filter Parameters in Hyperlynx
Bessel Filter Parameters
st
1 Order
Bessel
Cutoff
Frequenc
y = 245
MHz
3rd Order
Bessel
Cutoff
Frequenc
y = 245
MHz
5th Order
Bessel
Cutoff
Frequenc
y = 245
MHz
1st Order
Bessel
Cutoff
Frequenc
y = 245
MHz
3rd Order
Bessel
Cutoff
Frequenc
y = 245
MHz
5th Order
Bessel
Cutoff
Frequenc
y = 245
MHz
51
PI - Model Filter
T - Model Filter
Legendre Filter Parameters in Hyperlynx
Legendre Filter Parameters
st
1 Order
Bessel
Legendre
Frequenc
y = 245
MHz
3rd Order
Legendre
Cutoff
Frequenc
y = 245
MHz
5th Order
Legendre
Cutoff
Frequenc
y = 245
MHz
1st Order
Legendre
Cutoff
Frequenc
y = 245
MHz
3rd Order
Bessel
Cutoff
Frequenc
y = 245
MHz
5th Order
Legendre
Cutoff
Frequenc
y = 245
MHz
52
PI - Model Filter
T - Model Filter
Butterworth Filter Signal (Rising Time of 40MHz CLK)
1920.0
1st Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
Probe
Probe
1720.0
1720.0
1520.0
1520.0
1320.0
V
o
l
t
a
g
e
m
V
-
920.0
e
m
V
-
720.0
320.0
120.0
120.0
4.000
8.000
12.000
Time (ns)
16.000
-80.0
0.000
20.000
1720.0
1720.0
1520.0
1520.0
920.0
720.0
16.000
20.000
Probe
Probe
1120.0
920.0
720.0
520.0
520.0
320.0
320.0
120.0
4.000
8.000
12.000
Time (ns)
16.000
-80.0
0.000
20.000
4.000
8.000
12.000
Time (ns)
16.000
1920.0
Probe
Probe
1720.0
1720.0
1520.0
1520.0
20.000
Probe
Probe
1320.0
1320.0
V
o
l
t
a
g
e
m
V
-
1120.0
920.0
720.0
1120.0
920.0
720.0
520.0
520.0
320.0
320.0
120.0
120.0
-80.0
0.000
8.000
12.000
Time (ns)
1320.0
V
o
l
t
a
g
e
m
V
-
1120.0
-80.0
0.000
4.000
1920.0
Probe
Probe
120.0
V
o
l
t
a
g
e
m
V
-
720.0
520.0
1920.0
5th Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
920.0
320.0
1320.0
V
o
l
t
a
g
e
m
V
-
1120.0
520.0
1920.0
rd
Probe
Probe
1320.0
V
o
l
t
a
g
1120.0
-80.0
0.000
3
Order
Butterwo
rth
Cutoff
Frequen
cy = 245
MHz
1920.0
4.000
8.000
12.000
Time (ns)
16.000
20.000
53
-80.0
0.000
4.000
8.000
12.000
Time (ns)
16.000
20.000
PI - Model Filter
T - Model Filter
Chebyshev I Filter Signal (Rising Time of 40MHz CLK)
1920.0
1st Order
Chebysh
ev I
Cutoff
Frequen
cy = 245
MHz
1720.0
1720.0
1520.0
1520.0
V
o
l
t
a
g
e
m
V
-
V
o
l
t
a
g
e
m
V
-
1120.0
920.0
720.0
e
m
V
-
320.0
120.0
120.0
4.000
8.000
12.000
Time (ns)
16.000
-80.0
0.000
20.000
1720.0
1520.0
1520.0
920.0
e
m
V
-
720.0
16.000
20.000
Probe
Probe
1120.0
920.0
720.0
520.0
520.0
320.0
320.0
120.0
120.0
4.000
8.000
12.000
Time (ns)
16.000
-80.0
0.000
20.000
4.000
8.000
12.000
Time (ns)
16.000
1920.0
Probe
Probe
1720.0
1720.0
1520.0
1520.0
20.000
Probe
Probe
1320.0
V
o
l
t
a
g
1120.0
920.0
e
m
V
-
720.0
1120.0
920.0
720.0
520.0
520.0
320.0
320.0
120.0
120.0
-80.0
0.000
8.000
12.000
Time (ns)
1320.0
V
o
l
t
a
g
1120.0
1320.0
e
m
V
-
4.000
1920.0
Probe
Probe
1720.0
-80.0
0.000
V
o
l
t
a
g
720.0
520.0
1920.0
5th Order
Chebysh
ev I
Cutoff
Frequen
cy = 245
MHz
920.0
320.0
1320.0
V
o
l
t
a
g
1120.0
520.0
1920.0
rd
Probe
Probe
1320.0
1320.0
-80.0
0.000
3
Order
Chebysh
ev I
Cutoff
Frequen
cy = 245
MHz
1920.0
Probe
Probe
4.000
8.000
12.000
Time (ns)
16.000
20.000
54
-80.0
0.000
4.000
8.000
12.000
Time (ns)
16.000
20.000
PI - Model Filter
T - Model Filter
Bessel Filter Signal (Rising Time of 40MHz CLK)
1920.0
1st Order
Bessel
Cutoff
Frequen
cy = 245
MHz
1720.0
1720.0
1520.0
1520.0
1320.0
V
o
l
t
a
g
e
m
V
-
920.0
e
m
V
-
720.0
320.0
120.0
120.0
4.000
8.000
12.000
Time (ns)
16.000
-80.0
0.000
20.000
1720.0
1720.0
1520.0
1520.0
8.000
12.000
Time (ns)
16.000
20.000
Probe
Probe
1320.0
V
o
l
t
a
g
1120.0
920.0
e
m
V
-
720.0
1120.0
920.0
720.0
520.0
520.0
320.0
320.0
120.0
120.0
4.000
8.000
12.000
Time (ns)
16.000
-80.0
0.000
20.000
4.000
8.000
12.000
Time (ns)
16.000
1920.0
Probe
Probe
1720.0
1720.0
1520.0
1520.0
1320.0
20.000
Probe
Probe
1320.0
V
o
l
t
a
g
e
m
V
-
1120.0
920.0
720.0
1120.0
920.0
720.0
520.0
520.0
320.0
320.0
120.0
120.0
-80.0
0.000
4.000
1920.0
Probe
Probe
1920.0
V
o
l
t
a
g
e
m
V
-
720.0
520.0
-80.0
0.000
5th Order
Bessel
Cutoff
Frequen
cy = 245
MHz
920.0
320.0
1320.0
e
m
V
-
1120.0
520.0
1920.0
V
o
l
t
a
g
Probe
Probe
1320.0
V
o
l
t
a
g
1120.0
-80.0
0.000
3rd
Order
Bessel
Cutoff
Frequen
cy = 245
MHz
1920.0
Probe
Probe
4.000
8.000
12.000
Time (ns)
16.000
20.000
55
-80.0
0.000
4.000
8.000
12.000
Time (ns)
16.000
20.000
PI - Model Filter
T - Model Filter
Legendre Filter Signal (Rising Time of 40MHz CLK)
1920.0
1st Order
Legendr
e
Cutoff
Frequen
cy = 245
MHz
1720.0
1720.0
1520.0
1520.0
1320.0
V
o
l
t
a
g
e
m
V
-
920.0
e
m
V
-
720.0
e
m
V
-
320.0
120.0
120.0
4.000
8.000
12.000
Time (ns)
16.000
-80.0
0.000
20.000
1720.0
1720.0
1520.0
1520.0
920.0
720.0
20.000
Probe
Probe
920.0
720.0
520.0
320.0
320.0
120.0
4.000
8.000
12.000
Time (ns)
16.000
-80.0
0.000
20.000
4.000
8.000
12.000
Time (ns)
16.000
1920.0
Probe
Probe
1720.0
1720.0
1520.0
1520.0
20.000
Probe
Probe
1320.0
V
o
l
t
a
g
1120.0
920.0
e
m
V
-
720.0
1120.0
920.0
720.0
520.0
520.0
320.0
320.0
120.0
120.0
-80.0
0.000
16.000
1120.0
520.0
1320.0
e
m
V
-
8.000
12.000
Time (ns)
1320.0
V
o
l
t
a
g
e
m
V
-
1120.0
-80.0
0.000
4.000
1920.0
Probe
Probe
120.0
V
o
l
t
a
g
720.0
520.0
1920.0
5th Order
Legendr
e
Cutoff
Frequen
cy = 245
MHz
920.0
320.0
1320.0
V
o
l
t
a
g
1120.0
520.0
1920.0
rd
Probe
Probe
1320.0
V
o
l
t
a
g
1120.0
-80.0
0.000
3
Order
Legendr
e
Cutoff
Frequen
cy = 245
MHz
1920.0
Probe
Probe
4.000
8.000
12.000
Time (ns)
16.000
20.000
56
-80.0
0.000
4.000
8.000
12.000
Time (ns)
16.000
20.000
Butterworth Filter (EMI Levels)
PI - Model Filter
1st Order
Butterworth
Cutoff
Frequency
= 245 MHz
T - Model Filter
1mA
1mA
100uA
100uA
10uA
10uA
0
3rd Order
Butterworth
Cutoff
Frequency
= 245 MHz
500.000 MHz
1.000 G Hz
1mA
1mA
100uA
100uA
500.000 MHz
1.000 G Hz
0
500.000 MHz
1.000 G Hz
500.000 MHz
1.000 G Hz
10uA
10uA
0
5th Order
Butterworth
Cutoff
Frequency
= 245 MHz
0
500.000 MHz
1.000 G Hz
1mA
1mA
100uA
100uA
10uA
10uA
0
500.000 MHz
0
1.000 G Hz
PI - Model Filter
T - Model Filter
1mA
1mA
100uA
100uA
Chebyshev I Filter (EMI Levels)
st
1 Order
Chebyshev
I
Cutoff
Frequency
= 245 MHz
10uA
10uA
0
500.000 MHz
1.000 G Hz
1mA
3rd Order
Chebyshev
I
Cutoff
Frequency
= 245 MHz
100uA
500.000 MHz
1.000
75 nsG Hz
0
500.000 MHz
1.000 G Hz
0
500.000 MHz
1.000 G Hz
100uA
10uA
10uA
0
5th Order
Chebyshev
I
Cutoff
Frequency
= 245 MHz
00ns
1mA
500.000 MHz
1.000 G Hz
1mA
1mA
100uA
100uA
10uA
10uA
0
500.000 MHz
1.000 G Hz
57
Bessel Filter (EMI Levels)
PI - Model Filter
1st Order
Bessel
Cutoff
Frequency
= 245
MHz
T - Model Filter
1mA
1mA
100uA
100uA
10uA
10uA
0
500.000 MHz
1.000 G Hz
1mA
3rd Order
Bessel
Cutoff
Frequency
= 245
MHz
100uA
500.000 MHz
1.000 G Hz
75 ns
500.000 MHz
1.000
G Hz
75 ns
500.000 MHz
1.000 G Hz
100uA
10uA
10uA
0
500.000 MHz
1.000 G Hz
1mA
5th Order
Bessel
Cutoff
Frequency
= 245
MHz
0
0 ns
1mA
100uA
0 0ns
1mA
100uA
10uA
10uA
0
500.000 MHz
1.000 G Hz
0
PI - Model Filter
T - Model Filter
1mA
1mA
100uA
100uA
Legendre Filter (EMI Levels)
st
1 Order
Legendre
Cutoff
Frequency
= 245
MHz
10uA
10uA
0
500.000 MHz
1.000 G Hz
1mA
3rd Order
Legendre
Cutoff
Frequency
= 245
MHz
100uA
500.000 MHz
1.000 G Hz
75 ns
0
500.000 MHz
1.000 G Hz
0
500.000 MHz
1.000 G Hz
100uA
10uA
10uA
0
5th Order
Legendre
Cutoff
Frequency
= 245
MHz
0
0 ns
1mA
500.000 MHz
1.000 G Hz
1mA
1mA
100uA
100uA
10uA
10uA
0
500.000 MHz
1.000 G Hz
58
References
[1] Hongxia Wang et Al., Radio Frequency Effects on the Clock Networks of Digital
Circuits, IEEE 2004
[2] Matin Wiles, CTIA test requirements cover over the air performance, Wireless
Europe, October – November 2004
[3] 3rd Generation Partnership Project; Technical Specification Group GSM/EDGE
V8.17.0, 2004-11
[4] William J. Dally & John W. Poulton, Digital System Engineering, Cambridge
University Press 1998
[5] Karl Rinne, Dept. Electronics and Computer Engineering, University of Limerick,
http://www.ul.ie, August 2005
[6] Intel, Design for EMI Application Note AP-589, February, 1999.
[7] Clayton R. Paul, Introduction to Electromagnetic Compatibility, Wiley-Interscience
Publication, 1992
[8] William D. Kimmel and Daryl D. Gerke are principals in the EMI consulting firm
Kimmel Gerke Associates, Ltd., based in St. Paul, MN.
[9] Martin Wiles, CTIA test requirements cover over-the-air performance, Institute of
Physics and Institute of Physics Publishing Ltd. 2005
[10] Tim Williams, EMC for Product Designers, Third Edition, Newnes, 2001
[11] Siegmund M. Redl et Al, GSM and Personal Communications Handbook, Artech
House, 1998
59