1) The document discusses power loss in 4G handsets due to increased complexity of RF front-end architectures supporting multiple frequency bands for 4G LTE. 2) It provides background on the evolution of wireless technologies from 1G to 4G, noting increasing data speeds and frequency band allocations have driven complexity. 3) A typical 4G handset RF front-end is described as consisting of many components including antennas, switches, filters, power amplifiers and more to support 16+ frequency bands, introducing challenges like higher power loss.

1. Power Loss in 4G Handsets
By: Igor Lalicevic – Director RF Systems and Application DelfMEMS
Tel: +33 607 50 38 21 igor.lalicevic@delfmems.com
Abstract
The focus of this paper is on the handset power loss that is experienced with LTE wireless technology. Power loss is
observed through different aspects of mobile phone performance degradation and from the reduction of mobile battery
life point of view.
The history of wireless technology development will be discussed from the data speed improvement aspect and its
effect on complexity of today’s RF front-end architectures. All major components in a typical mobile phone RF FE will be
considered and we will focus on their impact on performance with respect to the power loss.
Special attention will be placed on the next generation of RF switching requirements for the higher end of the LTA
performance scale and a particular emphasis on the need for high performance, high quality and highly reliable RF
MEMS switches. The paper will include a description of DelfMEMS RF MEMS switch technology and how it addresses
the need for high performance RF switching, while minimizing power loss.
Paper
The headline findings of Cisco’s latest Visual Networking Index (VNI) states that by the year 2018, for the first time in the
history of the internet, mobile and portable devices will generate more than half of the global IP traffic. This is a good
way to describe history of wireless technology development.
We have witnessed tremendous evolution and change in cellular standards over the past 35 years. First Generation of
wireless telephone technology, so called 1G, was launched in Japan by NTT (Nippon Telegraph and Telephone) in 1979.
1G technology used analog transmission techniques for transmitting voice signals. This voice only standard, used a
frequency modulation technique (FDMA: Frequency-Division Multiple Access) where voice call are modulated to a
higher frequency of about 150MHz and up as it is transmitted between radio towers.
Low capacity, unreliable handoff, poor voice links and no security at all, resulted in development of early digital systems
which became known as 2G: Second Generation of wireless telephone technology. These became available in the 1990s
and used a digital multiple access modulation method, such as TDMA (time division multiple access) and CDMA (code
division multiple access) which enabled the first low bit rate data services.
2G systems offered higher spectrum efficiency, first data services, and more advanced roaming and for the first time a
single unified standard (GSM: Global System for Mobile Communications) was provided. First employed in Europe in
1991, today GSM is utilized across the whole world.
As the requirements for sending data over the air-interface increased, GPRS (General Packet Radio Service) and WAP
(Wireless Application Protocol) technologies were added to the existing GSM systems. This is considered an
advancement from 2G to a 2.5G technologies, where packet-switching wireless application protocols enabled wireless
access to the internet.
Even though 2G supports data over the voice paths, 2G data speeds were typically 9.6 Kb/s or 14.4 Kb/s which
effectively made 2G voice-centric system. Data speed improvement that 2.5G technology introduced with more
advanced coding methods provided data rates of up to 384 kbps. However, limitations of packet transfers technology
which behave in a similar way to a circuit switch call over the air combined with low system efficiency and non-
standardized networks across the world led to a birth of a the 3G standard.

2. The initial planning for Third Generation 3G standards started in the 1980s and was focused on multimedia applications
such as video-conferencing for mobile phones. Over the years 3G’s focus has moved to personal wireless internet
access, the need to roam worldwide staying connected and to achieving increased system and network capacity.
In year 2001 the first commercial 3G network based on W-CDMA technology was launched in Japan by NTT DoCoMo.
The need to create a globally applicable mobile phone system specification resulted in 3rd Generation Partnership
Project (3GPP) which unites telecommunications standard development organizations known as “Organizational
Partners”. 3GPP has become the focal point for mobile systems beyond 3G and provides requirements for 3G data speed
specifications of up to 2 Mb/s for stationary users.
To additionally enhance UMTS (Universal Mobile Telecommunications System) based 3G networks to enable higher data
speeds, features like HSPA (High Speed Packet Access) have been implemented to provide data transmission capabilities
able to deliver speeds up to 14.4 Mbps on the downlink and 5.8 Mbps on the uplink.
HSPA+ (High Speed Packet Access) or Evolved High-Speed Packet Access was a further evolution of HSPA that is capable
of delivering theoretical peak data speeds of 168 Mbit/s (downlink) and 22 Mbit/s (uplink). HSPA+ provided a migration
method towards the next wireless standard, 4G, data speeds without actually deploying a new 4G radio interface. The
latest Fourth Generation LTE (Long-Term Evolution) standard uses a new air interface based on OFDMA (Orthogonal
Frequency-Division Multiple Access) digital modulation scheme. The LTE cellular standard is divided into six categories,
each of which caters for different requirements (Table1).
Table 1. (LTE Cat5 was not implemented due to the 4x2 MIMO requirement)
The new LTE Cat6 category is capable of providing up to 300 Mbps downlink data rate and 50Mbps uplink data rate.
Even so, the current 300Mbps/50Mbps is not close to customer needs and market requirements. The peak data rate
target for the most recent LTE-Advanced technology is up to approximately 1 Gbps for Up Link connections. Even this is
just a first step and target that will have to be improved rapidly and constantly. The high data rates required for 4G LTE
Advanced are achieved by a bandwidth increase methodology that is called Carrier Aggregation (CA). Carrier Aggregation
or channel aggregation enables multiple LTE carriers to be used together to provide bandwidth increase.
In just the last few years, we have witnessed an explosion of growth in the number of allocated frequency bands.
Coming from the 2G first digital cellular standard that introduced two mode quad bands system solutions, we have seen
the trend of increasing number of frequency bands allocations has continued further, with 3G typically supporting up to
8 frequency bands in order to support global roaming and higher data sped needs. LTE development dominated by the
need for global roaming and wider frequency bandwidth brought the latest explosion of allocated frequency bands.
Currently we have over 40 bands allocated to LTE FDD and LTE TDD applications.
Increased complexity and mobile handset market price erosion pressure made RF Front End (RF FE) architecture designs
for high end smartphones today exceptionally challenging. RF FE’s that have to support over 16 frequency bands are not
uncommon any more. Original Equipment Manufacturers (OEM’s) are looking to minimize their model variants and even
more, for some of them, the trend is to build one world global phone model, which would by default mean 20 or more
bands that will be have to be supported in RF FE.

3. A typical state of the art RF FE consist of; two antennas, (main & diversity needed to achieve 2x1 RX MIMO), antenna
tuner and couplers, main front end module (FEMid) which includes antenna switch and filters, diversity front end
module (divFEM) which includes antenna switch and Rx filters to accomplish 2x1 MIMO downlink configuration. It
further includes a Multimode Multiband PA module and Power Management Unit with Envelope Tracking Modulator,
RFIC transceiver and multiple additional add-on components such as standalone filters, power amplifiers, switches etc.
Example of LTE RF Front End architecture
The list of needed RF FE components has become increasingly long and they all have to meet required specifications for
this “high number of bands” environment. This complex RF environment introduces many challenges for components,
particularly power loss (or insertion loss), isolation and linearity which are key for high performance. Additional
complications are introduced with Inter-band CA requirement, which requires the use of multiple active Tx/Rx paths
within the single RF FE, with the usual impact on cost, performance and power.
These supplementary complexities result from the requirements to reduce Intermodulation and cross modulation from
the two or more receiver and transmitter paths. In this environment, for all RF components and particularly for the, RF
Antenna Switch linearity performance is becoming a crucial specification. Basic RF performances such as insertion loss,
isolation, linearity and power efficiency in LTE-A RF Front End have again become the main driver in component
selection.
An excellent example of component performance impact on the customer is on battery life. For the next generation
smart phones, battery life will be the most important customer consideration for over 70% of users according to the
International Wireless Industry Consortium. High RF FE insertion loss, which is creating power loss, has a direct negative
impact on smartphone battery life. Previously customers were used to the 2G phone experience and assumed that
MIPI RFFE
Serial Control
Master
Diversity
Antenna Port
Diversity
RX LNAs
RFIC
Main
RX LNAs
RFIC
TX FE
RFFE
Control
Interface
Diversity
FEM
SPnT
SPnT
mipi RFFE
mipi RFFE
High
Bands
FEMid
mipi RFFE
mipi RFFE
Envelope Tracking
Modulator
VPA
...mipi RFFE
mipi RFFE
PA
mipi RFFE
VHF 4G PAM
SPnT
mipi RFFE
mipi RFFE
SPnT
mipi RFFE
Main
Antenna Port
PA
3G/4G
HB
mipi RFFE
SPnT
PA
3G/4G
LB
GSM
PA
PA
MMMB
PAM
SPnT
mipi RFFE
mipi RFFE
Antenna
Tuner
MIPI RFFE
SPnT
SPnT

4. mobile handsets are typically used without re-charging the battery for days. The explanation is fairly simple; the 2G
modulation scheme allows usage of very efficient saturated PA’s (Power Amplifiers) since amplitude modulation is not
necessary. In this low data rate speed environment the vast majority of RF current is spend on PAs.
The situation becomes progressively worse with 3G standards. Phones can still be used without re-charging the battery
for days, but battery life became shorter than with a 2G phone. PAs are not working anymore in saturated mode,
instead 3G requires PAs working in linear mode operating mode. The requirement is coming from increased data speed
requests, this assumes introduction of amplitude modulation methods, which will require linear mode for PAs which are
by default inefficient from a current drain point of view. Battery life for 3G phones is explained on next two graphs.
Graph 1 Graph 2
Graph1 is an example of the current drain (battery life with Linear PA) vs. mobile phone transmitted power where we
see that current drain is increasing exponentially vs. output power. Henceforward it is obvious that at very high or
maximum output power, our battery will not last for a long time.
Graph 2 is an example of user’s power distribution vs. handset output power. We can see that 3G phones are used very
rarely at a high power levels, where the current consumption is high. Voice average output power is around -10dBm
(blue curve = low data rate speed), and data usage average output power is around +10dBm (Pink curve = high data rate
speed).
In conclusion, we can state that 3G phones have limited time use at high power level where current drain is high, and
the normal operating point for 3G phones is at lower power levels which allows longer battery life. At the same time this
explains why our battery will last shorter when we use our 3G phone for data applications, compared just to an simple
voice communication.
From Graph1 we can see that at output power of +10dBm, the current drain is only 1/3 of the current drain level at
maximum transmitted output power for 3G (which is +23dBm). This will provide battery life around 60% longer at
+10dBm compared to battery life at +23dBm transmitted power.
All the positive experiences with long battery life for 2G and 3G standards disappeared with fourth generation phones.
4G (LTE and LTE Advanced or LTA CA) phones have to be charged almost every day, and in a typical user case, a couple of
times every day. 4G PAs use exactly the same or very similar designs as 3G Pas, with the same linear PAs today being
used for both 3G and 4G applications. The crucial difference is that LTE networks typically require short bursts of high RF
output power levels for data transmission. Going back to Graph1 this means 4G phones are almost always used at
maximum (+23dBm) output power level which implies maximum current drain for 4G applications and short battery life.
LTE-A with carrier aggregation and global roaming requirements are making things even worse. The necessity of LTE-A to
introduce additional LTE bands resulted in tremendous RF FE power loss for 4G handsets. Typical RF FE power loss for 2G
was around 0.5dB, and with 3G standard this loss was increased to around 1.5dB. The power loss for 4G RF FE due to

5. additional filtering needs and complex switching requirement can easily reach over 4dB. To offset and cover increased
power losses and to meet maximum antenna radiated power requirements, 4G PA’s have to be capable of providing
additional power which will make 4G PAs even less efficient.
All this emphasizes and makes the impact of power loss on battery life painfully obvious.
The most effort recently, however with only limited success, to improve PA current drain was spent on PMU (Power
Management Unit) development with Envelope Tracking being the latest and greatest that the industry can offer today.
Efficiency improvements of around 10% have so far been achieved, of course any improvement is welcomed and more
will have to come.
It has become increasingly clear that high performing RF FE component will bring crucial benefits to existing RF FE
architecture and will help to simplify the same. Our company, DelfMEMS, focuses is on RF MEMS switches with inherent
high linearity, high operating frequency, ultra-low insertion loss and very high port-to-port isolations, making it a perfect
candidate for high performance LTE-A applications. Even though the inherent high linearity of a MEMS switch that can
enable uplink CA is probably the most obvious benefit of RF MEMS switching, the focus in this paper is placed on the
switch ultra-low insertion loss that will reduce greatly 4G RF FE power loss. To compare MEMS switches vs existing
switch solid-state technology, the Figure of Merit (FoM) graph is shown below.
Switch Figure of Merit is expressed in femto seconds and is defined as a Ron*Coff product, where Ron is switch on
resistance affecting insertion losses, and Coff is off switch Capacitance affecting isolation. The MEMS FoM of less than 10
clearly shows the switch benefit in the frequency domain for power loss and isolation over existing solid-state switch
technologies.
The DelfMEMS RF MEMS switches further retain ultra-low insertion loss even with an increasing number of switch
throws for high multi-throw switch configurations and at higher frequency operation. This makes the RF MEMS switch a
perfect candidate for the 4G multi band environment where RF components have to demonstrate minimum power loss
degradation at frequencies of up to 3.5GHz. The graph below compares DelfMEMS and solid state switch technologies
for a SP12T RF switch insertion loss vs. frequency. Benefits of the RF MEMS switch in high throw switch configuration at
higher frequencies are impressive.

6. LTE-A RF Front End
The power loss reduction which can be realised with the DelfMEMS switch in a 4G, high frequency environment by
replacing existing solid state switches, leads directly to a significant battery power saving and increased reception
sensitivity. These exceptional improvements become even greater at 3.5GHz (LTE TDD bands 42 & 43) for the bands that
will be introduced in Japan in 2015, with others to follow.
The block diagram presented below is assessing power savings when comparing the DelfMEMS RF MEMS switch to
existing solid-state antenna switches.
Transmission current drain as the dominant contributor to smartphone battery life and current savings percentages
across bands has been considered as equivalent to percentages of longer talk time.

7. A simple calculation where Insertion Loss (IL) is max power loss for the switch expressed in dB, and Current Drain Loss
(CDL) is calculated current drain loss expressed in percentages demonstrates up to 17% longer talk time when solid state
switches are replaced by ultra-low loss RF MEMS switch.
The same logic applies on LTE-A Call Quality and Data Reception Quality.
As stated earlier, Global Roaming and single World phone requirements for smartphones is creating high power loss RF
FE environment for both diversity and main receiver RF paths.
Receiver sensitivity or Rx sensitivity is one of the key specifications for any radio receiver. Considering that call quality is
equal to Rx sensitivity, increased RF FE Insertion Loss degrades Rx Sensitivity by the same amount and Rx Sensitivity
degradation directly impacts phone call quality and data reception quality.
The additional block diagram presented below assesses Rx sensitivity improvements comparing DelfMEMS RF MEMS
switches to existing solid-state switches for both main and diversity antenna switch paths. Decreased RF FE Insertion
Loss will improve Rx Sensitivity by the exactly same amount.
MIPI RFFE
Serial Control
Master
Diversity
Antenna Port
Diversity
RX LNAs
RFIC
Main
RX LNAs
RFIC
TX FE
RFFE
Control
Interface
Diversity
FEM
SPnT
SPnT
mipi RFFE
mipi RFFE
High
Bands
FEMid
mipi RFFE
mipi RFFE
Envelope Tracking
Modulator
VPA
...mipi RFFE
mipi RFFE
PA
mipi RFFE
VHF 4G PAM
SPnT
mipi RFFE
mipi RFFE
SPnT
mipi RFFE
Main
Antenna Port
PA
3G/4G
HB
mipi RFFE
SPnT
PA
3G/4G
LB
GSM
PA
PA
MMMB
PAM
SPnT
mipi RFFE
mipi RFFE
Antenna
Tuner
MIPI RFFE
SPnT
SPnT
Battery Power Savings
using MEMS Antenna Switch
B 13,17… SOI RF MEMS Current
IL [dB] -0.85 -0.25 Savings
CDL 18% 6% 12%
B 5,8 … SOI RF MEMS Current
IL [dB] -0.9 -0.25 Savings
CDL 19% 6% 13%
B 1,2,3,4 SOI RF MEMS Current
IL [dB] -1.2 -0.3 Savings
CDL 24% 7% 17%
B 7,40, 41… SOI RF MEMS Current
IL [dB] -1 -0.35 Savings
CDL 21% 8% 13%

8. The simple calculation where IL is max power loss for the switch expressed in dB, and RxSI is calculated receiver
sensitivity improvement expressed in dB, demonstrates up to 1.1dB RX sensitivity improvement which is equivalent to
an impressive 29% Rx Sensitivity improvement.
The typical RF MEMS switch structure uses either a cantilever beam or a bridge and features highly conductive
electrodes, which are electrostatically actuated in order to create an ohmic contact on a conducting line. The result is
mechanical switching. These typical basic structures carry several serious issues such as stress on the anchors, a
tendency for stiction, low switching speed and metallic creep in the beam.
DelfMEMS’s RF MEMS switch design offers an innovative approach to get around these problems, instead of trying to
reduce their effect. As a result, the switch simultaneously offers increased performance in terms of isolation and
insertion losses. This has been achieved through the development of a unique anchorless structure for mechanical RF
switching.
DelfMEMS’s switch solution features a free moving flexible membrane, known as Free-Flex™, which carries the contact
area and is held and positioned by two sets of pillars and stoppers. The membrane is electrostatically actuated by 2 sets
of electrodes enabling two controlled states of the switch. ON state is achieved by making physical contact between the
membrane contact area and the transmission line and electrostatically controlled OFF state is achieved by keeping a
physical distance between membrane contact area and the transmission line. This means that the switch contact area
will be either attracted to the conductive line or repelled from it.
Complete control of MEMS membrane allows for an increased gap between contact area and transmission line in the
OFF-state, which is directly linked to the switch isolation and allows for switch resetting in the unlikely event of stiction.
MIPI RFFE
Serial Control
Master
Diversity
Antenna Port
Diversity
RX LNAs
RFIC
Main
RX LNAs
RFIC
TX FE
RFFE
Control
Interface
Diversity
FEM
SPnT
SPnT
mipi RFFE
mipi RFFE
High
Bands
FEMid
mipi RFFE
mipi RFFE
Envelope Tracking
Modulator
VPA
...mipi RFFE
mipi RFFE
PA
mipi RFFE
VHF 4G PAM
SPnT
mipi RFFE
mipi RFFE
SPnT
mipi RFFE
Main
Antenna Port
PA
3G/4G
HB
mipi RFFE
SPnT
PA
3G/4G
LB
GSM
PA
PA
MMMB
PAM
SPnT
mipi RFFE
mipi RFFE
Antenna
Tuner
MIPI RFFE
SPnT
SPnT
B 13,17… SOI RF MEMS RxSI
IL [dB] -0.85 -0.25 0.6
B 5,8 … SOI RF MEMS RxSI
IL [dB] -0.9 -0.25 0.65
B 2,3 Rx SOI RF MEMS RxSI
IL [dB] -1.2 -0.3 0.9
B 1,4 Rx SOI RF MEMS RxSI
IL [dB] -1.4 -0.3 1.1
B 7,40, 41… SOI RF MEMS RxSI
IL [dB] -1 -0.35 0.65
B 13,17… SOI RF MEMS RxSI
IL [dB] -0.85 -0.25 0.6
B 5,8 … SOI RF MEMS RxSI
IL [dB] -0.9 -0.25 0.65
B 2,3 Rx SOI RF MEMS RxSI
IL [dB] -1.2 -0.3 0.9
B 1,4 Rx SOI RF MEMS RxSI
IL [dB] -1.4 -0.3 1.1
B 7,40, 41… SOI RF MEMS RxSI
IL [dB] -1 -0.35 0.65

9. Moving from ON-state to OFF-state is made through an electrostatic active actuation, which de-correlates between
restoring forces, contact forces and the membrane mechanical properties. Importantly this doesn’t depend only on the
elastic restoration forces. This advanced electrostatic actuation results in a very short switching time, less than 3µs.
The ability to have a reduced gap between the membrane and the transmission line is a major advantage of DelfMEMS
switch structures. It ensures increased ON-state contact force is achieved with reduced actuation voltages and
consequently delivers ultra-low insertion losses. Due to the reduced gap, maximum deflection of the membrane will be
reduced as well, which as a result decrease membrane mechanical stress and the creep effect.
Thanks to the DelfMEMS’s original design approach, RF MEMS switches can be used effectively for the first time as an RF
FE switching solution.
Igor Lalicevic