DESIGN CONSIDERATIONS FOR
MULTI-CARRIER CDMA BTS POWER AMPLIFIERS
Osama W. Ata
Core RF Engineering, Sprint, Overland Park, KS, USA. Currently, CTO at BC, P.O.Box 2385, Ramallah, Palestine
Keywords: Power amplifiers, amplifier linearization, multi-carrier CDMA, multi-tone testing, feed forward, pre
distortion, cellular design, 1 dB compression point, PAPR, BTS, emission mask, Gaussian distribution.
Abstract: System manufacturers that provide wireless telecommunications equipment to cellular operators, often do
not manufacture their own transceivers. They get their own supplies from various base station power
amplifier vendors. Unfortunately cellular operators and system providers take design considerations of those
primary power amplifiers for granted. The consequences are: Operators are overpaying for base station
amplifiers to meet power and capacity demands. Base stations are complicated systems and the optimum
technical design considerations of the power amplifier sub-systems have a major impact on multi-carrier
Code Division Multiple Access (CDMA) Base Transceiver Station (BTS) performance. This paper
discusses those power amplifier vendor claims, demonstrates design considerations and economic design
solutions of consequent savings for the operator, based on a realistic market case. We realize that upfront
savings, in power amplifier units, in a US market of approximately 550 multi carrier basestation sectors
would exceed $ 10 M.
1 INTRODUCTION
Capacity enhancement of the 3G CDMA 2000
network rests essentially upon the optimized
performance of the multi-carrier amplifier in the
BTS. This is because the CDMA channel capacity,
is primarily power dependent. The input 1dB
compression point is a quantity measure of the
amplifier linearization. It is defined as the input
power value where the corresponding output power
falls by 1 dB from the linear slope of the amplifier
output characteristic curve. An underestimated
power backoff from the 1 dB compression point,
necessary to restrain occurring peaks from entering
the saturation region in the BTS amplifier would
result in a non-linear output performance, an
increase in the BER and consequently an increase in
the dropped call rate. An overestimated power
backoff on the other hand, has two major drawbacks:
• Inefficient power utilization of the BTS
amplifier.
• Possible increase in amplifier size and
uneconomic cost impact (Price roughly
doubles for every extra 3 dB requirement in
the amplifier’s max. power output)
The problem here is that some vendors are
reluctant to provide essential information on BTS
amplifier specifications pertaining to backoff power
requirement.
An assumption of 12 dB backoff power
requirement is taken for granted by many vendors.
Indeed, a large 12 dB backoff is a major drawback in
OFDM (Orthogonal Frequency Division
Multiplexing) technology which is definitely not the
case in multi-carrier CDMA 2000. Unlike
multicarrier CDMA amplifiers, the required power
backoff in OFDM amplifiers is research and
experimentation proven. Unfortunately, CDMA
amplifier vendors, take the OFDM backoff power of
12 dB for granted and apply it on CDMA amplifier
design. On a different front, CDMA operators aspire
for having up to 6 carriers per BTS amplifier,
despite the CDMA vendors accepted restriction on
power backoff in CDMA amplifier design.
Hence the objectives of this paper are three
folds:
• To get a better understanding of the backoff
power BTS amplifier design requirements and
promote consequent awareness amongst major
cellular operators and vendors.
• To shed some light on current multi-carrier
BTS amplifier considerations of a CDMA
134
W. Ata O. (2007).
DESIGN CONSIDERATIONS FOR MULTI-CARRIER CDMA BTS POWER AMPLIFIERS.
In Proceedings of the Second International Conference on Wireless Information Networks and Systems, pages 134-139
DOI: 10.5220/0002145301340139
Copyright
c
 SciTePress

2000 cellular operator in an interesting US
market.
• To simulate a multi-carrier example and
compare it to a single carrier case.
An example will be shown to demonstrate an actual
case in a real multi-carrier CDMA 2000 scenario.
2 MODELING
2.1 Assumption
Two cases representing a two-tone signal and a nine-
tone signal were considered. The two-tone signal is a
representation of a real-time sinusoidal signal in the
frequency spectrum domain. The nine-tone signal,
on the other hand, may be considered as a real-time
sum of four sinusoidal signals in the frequency
domain with a DC component in the middle of the
spectral symmetry.
The frequency deviation between two successive
tones was considered to be 1.25 MHz. The effect of
amplitude variation of the individual tones was,
however, not considered since the application of a
suitable modulation scheme would take care of that.
Consequently, the isolation of the amplitude change
of the individual tones would only allow for the
effect of the phase variation together with the effect
of the peak to average power ratio of the input signal
to be studied.
2.2 Simulation
The baseband nine-tone signal is simulated by
considering an array of size (257x1), with zero
values except for the middle part representing the
nine tones i.e. nine spectral lines of equal amplitude
and uniformly random phase, in the frequency
domain. A random generator in the MATLAB
software is used to generate nine uniformly
distributed random numbers between zero and one.
These numbers are manipulated to convert the
original range to a modified one between -1 and +1.
The numbers are then multiplied by 180 to simulate
the random phases of the nine tones (i.e. four
carriers) between -180 and +180 degrees. The
Fourier transform of the array is computed and the
absolute values of the resulting array are plotted,
after normalization, to represent a clip of the nine-
tone signal in the time domain.
Figure 1a shows the normalized nine-tone signal
in the time domain while the uniformly random
phase values generated in the composition of the
resulting time varying signal are shown in Figure 1b.
A histogram of a close fit to a Rayleigh
distribution of the multi-carrier signal resulted, as in
Figure 2, after repeated generation of the uniform
random phases. The Signal in Figure 1a was hence
determined as relatively the nearest representative
signal for multi-carrier BTS applications. The
variation of input average power for two-tone and
nine-tone signals versus same scale of input peak
voltage, could then be determined, as shown in
Figure 3. A constant power drop of 5.35 dB resulted,
with the nine-tone variation of input peak voltage,
indicating the relative back-off of input average
power away from the 1dB compression point
towards the amplifier linear region.
Figure 1: (a) Normalized nine-tone signal in time domain.
(b) Uniformly random phase values.
3 TWO-TONE AND MULTI-TONE
TESTING
3.1 Two-Tone Testing
The two-tone test has been the most common one for
measuring the inter modulation distortion (IMD) of a
power amplifier for many years. The test consists of
two unmodulated carriers at the input port of the
non-linerar amplifier (NLA) and the resulting Peak
/Average Power Ratio (PAPR) is only 3dB.
In a multicarrier composite signal, the
modulation of each tone determines the distribution
of power within the peak/average range. The
absolute PAPR of “n” modulated signals may be
obtained using
P
n
= 10log(n) + P (1)
Where, P= PAPR of an individual signal in dBs
P
n
= PAPR of the composite signal.
DESIGN CONSIDERATIONS FOR MULTI-CARRIER CDMA BTS POWER AMPLIFIERS
135

Figure 2: Histogram of multi-tone signal in Figure 1a.
Figure 3: Comparative average signal input power versus
signal peak voltage.
Figure 4 shows the 1 dB compression point, defined
as the input power value; 6 dBm, at which the
corresponding output power value drops by 1 dB
from the extended slope solid slope line.
Figure 4: A modeled power amplifier power curve.
3.2 Multi-Tone Testing
It is important to note that PAPR of a CW multitone
is a probability distribution, not a unique number.
This is demonstrated in Figure 5 which shows a 16-
tone CW signal distribution in which all carriers are
locked to a common reference
7
(i.e. phase aligned).
The same Figure shows a distribution of 16 tones but
randomly phase modulated (much like a set of
AMPS carriers). While the relative PAPR in both
cases is 12 dB (i.e.10 log
10
(16)), the random phase
case shows a much lower normalized power value at
any shown probability of peak occurrence.
Furthermore, because the carriers are independent
random variables, the central limit theorem implies a
Gaussian distribution for a large number of carriers.
Therefore the addition of more carriers would not
increase the PAPR at a particular probability.
Figure 5: PAPR probability of 16-tone continuous signal.
The PAPR of a single-carrier CDMA signal may
be rigorously calculated. The calculation becomes
exceedingly complicated for multi carrier CDMA
signals. For the purpose of this article, Figure 6
shows three different curves for CDMA PAPR
distribution , comparing 3 different scenarios .
Figure 6: PAPR probability of a multicarrier CDMA
signal.
It is interesting to observe that the three
scenarios of a single CDMA carrier, 5 CDMA
carriers and a dual mode live traffic of a CDMA
carrier and 15 AMPS signals have close
distributions. Below 1E-4 probability, for the multi-
carrier scenarios the PAPR varies about a 10 dB
value.
4 CDMA BACKOFF DESIGN
REQUIREMENTS
In designing a CDMA power amplifier, here are a
few factors to consider. These are:
• What is the IMD level resulting from
maximum input power at the input of
amplifier?
• What is the amplifier technology choice?
WINSYS 2007 - International Conference on Wireless Information Networks and Systems
136

• What is amplifier correction technology
choice?
• What is the PAPR probability threshold
choice?
• Is the FCC spectral mask complied with?
• Is there a hard clipping requirement in the
design?
4.1 Amplifier and Correction
Technology Choices
4.1.1 Amplifier Technology Choices
Here are popular choices for amplifiers where gains
are normalized to 1GHz bandwidth:
• Bipolar : IMD= -30 dBc, Gain = 8 dB
• LDMOS: IMD= -40 dBc, Gain = 11 dB
• GaAsFET: IMD= -45 dBc, Gain = 14 dB
4.1.2 Amplifier Correction Technologies
The following are linearization techniques of
amplifiers where correction capacity means the
range where the worst IMD (relatively largest
magnitude) products reside, below the fundamental
carrier, after linearization or correction takes place.
Effective bandwidth of correction and relative cost
are also indicated:
• Feedforward: Correction capacity from 30 to
35 dB, BW > 25 MHz, relative cost is high.
• Envelope Feedback: Correction capacity from
15 to 20 dB, BW< 5 MHz, relative cost is
medium.
• Predistortion: Correction capacity from 3 to 7
dB, BW > 25 MHz, relative cost is low.
• Adaptive Predistortion: Correction capacity
from 10 to 20 dB, BW= 10 to 15 MHz,
relative cost is medium.
4.2 PAPR Probability Tolerance
A 10
-4
probability that a signal would exceed its
PAPR average value is a reasonable one that can be
tolerated in the design of a power amplifier. In
concept , when hard clipping an amplifier, a -20 dBc
IMD is produced, with a 10
-4
PAPR probability
occurrence. This probability means a fraction of 1 in
10,000 signals. The IMD power works out to be
10log
10
(10
-4
) = -40 dB, below this -20 dBc level. A
total -60dBc of IMD power level, below the
fundamental power, would comply with the FCC
spectral mask of the CDMA signal.
When examining Figure 6 for a relationship
between probability of exceeding PAPR value and
PAPR in dB for a realistic multicarrier CDMA
signal, here is what can be observed:
• For 10
-4
probability, PAPR ~ 10 dB
• For 10
-3
probability, PAPR ~ 8 dB
• For 10
-2
probability, PAPR ~ 6 dB
• For 10
-1
probability, PAPR ~ 4 dB.
5 DESIGN CHOICES
The CDMA Emission mask (FCC Part 24 rules)
calls out for better than -57 dBc at greater than 1.25
MHz offset.
Figure 7: FCC CDMA Emission Mask.
The conventional design choice of amplifiers
utilizes bipolar transistor technology with -30 dBc of
IMD level and feedforward linearization with a
minimum of 30 dB correction capability. This is -60
dBc of IMD level from the 1 dB compression point,
which looks 3dB lower than the minimum required -
57 dBc in the FCC CDMA Emission mask. A 10
-4
probability of occurrence matches approximately a
10 dB PAPR value. This means going 10 dB below
the -57 dBc level to meet the 10
-4
probability of not
exceeding the PAPR value. In other words this is
only 7 dB below the calculated -60 dBc of IMD
level which implies a 7 dB power backoff from the
1dB compression point.
More recently, the choice of amplifier
technology has developed to GaAsFET with -45 dBc
of IMD level and correction technology of adaptive
predistortion linearization with a minimum 15 dB of
correction capability at the 1 dB compression point.
Again this is -60 dBc at the 1 dB compression point.
For a change, a 10
-3
distribution probability might be
chosen to match a PAPR occurrence of 8 dB. This
means going 8 dB below the -57 dBc level to meet
the 10
-3
probability of not exceeding the PAPR
value. In other words this is only 5 dB below the
calculated -60 dBc of IMD level which implies a 5
dB power backoff from the 1dB compression point.
DESIGN CONSIDERATIONS FOR MULTI-CARRIER CDMA BTS POWER AMPLIFIERS
137

6 COMMERCIAL BASESTATION
AMPLIFIERS
Commercial basestation power amplifiers (PA) need
be evaluated against the analyzed technology
choices and design criteria, if cellular operators are
to optimize the number of power amplifier units
required in their multicarrier basestation network
and save on capital and operational expenditure.
Here are typical specifications of basestation power
amplifiers, supplied by US PA vendors to system
vendors of cellular operators:
• PA rated at 50W (47 dBm) maximum average
power output per sector.
• Power available after insertion and cable
losses equals 35.4 W.
• Input power backoff claimed to be 10 to 12 dB
• Nominal PA Gain = 58 dB.
• Conventional bipolar feedforward technology
• Maximum input average power = - 5 dBm.
• PA unit automatically reduces gain above -5
dBm to prevent it from being overdriven.
• Two cascaded PA units are required for two
carriers and a maximum of three carriers (16-
20 W per carrier).
6.1 Design Specs Implications
Considering a 47 dBm average power output per
sector and the minimum claimed 10 dB input power
backoff, this means that the average output power
before backoff would work out 57 dBm. If now
instead we consider the researched 7 dB backoff
requirement, then the maximum average output
power after backoff would be 57-7= 50 dBm ( 100
W).
Taking out the insertion/cable losses as
referenced by the cellular operator vendor, the
maximum available output power is now 70.8% x
100W = 70.8 W.
6.2 Lessons Learned
• Cellular Operator Vendors claim a maximum
available transmission power of 35.4 W per
power amplifier unit. Hence two cascaded
power amplifier units would be required to
accommodate up to three carriers, where each
carrier is rated at 16 – 20 W.
• Our researched analysis shows that, based on
a combination of a sufficiently conservative 7
dB backoff power consideration, the
insertion/cable losses, Part 24 FCC spectral
mask and a maximum transmission power
/carrier not exceeding 20 W, a single amplifier
unit of 70.8 W should accommodate up to
three carriers.
• The backoff power is not a hardware design
specification. It should be possible, through
the power amplifier unit software control to
relax it from 10 or 12 dB down to a sufficient
7 dB value.
• A 10
-4
probability (1 in 10,000 signal
occurrences), that a multicarrier CDMA
PAPR threshold would not be exceeded is a
conservative one that can be met by a 7 dB
power backoff and still be under the -57 dBc
level dictated by the FCC emission mask.
• Estimated savings in sparing a 2
nd
power
amplifier unit for a mere second carrier in a
multicarrier deployment could well exceed $
10 Million for less than 200 tri-sector cells.
6.3 Assessment of Risks
It is fair to assess some of the risks involved in
optimizing the number of basestation amplifier units
in a multi carrier cellular infrastructure. Life
expectancy of the power amplifier would be an issue
to some degree, depending on the output percentage
power. Cooling the power amplifier, on the other
hand, is a prime factor in maintaining its higher
output power. The Mean Time Between Failures
(MTBF) is impacted by the those conditions.
7 CONCULSIONS
• The analysis, in this article, is vendor specific,
given the maximum information we could
secure, at the time.
• Further analysis from other vendors would
highly depend on obtaining the information,
necessary for the analysis.
• Resolving the risk assessment factors would
translate the value of implementing two and
up to three carriers into a single 50 W Power
Amplifier.
• There would be a definite cost saving in
optimizing the number of amplifier units
needed in a two and three carrier
basestation/sector.
• We realized that upfront savings in a US
market of approximately 550 multi carrier
basestation sectors would exceed $ 10 M.
WINSYS 2007 - International Conference on Wireless Information Networks and Systems
138

REFERENCES
Zekavat, S.A. Nassar, C.R., "Achieving high-capacity
wireless by merging multicarrier CDMA systems and
oscillating-beam smart antenna arrays", IEEE
Transactions on Vehicular Technology, vol. 52, no. 4,
July 2003, pp 772-778.
Hongxi, Xue; Davies, R.; Beach, M.; McGeehan, J.,
"Application of linearised amplifiers in adaptive
antennas", MTT-S Symposium on Technologies for
Wireless Applications, Feb 1995, pp 51-56.
Yang, S., "The application of low noise amplifiers in
CDMA cellular and PCSsystems for coverage and
capacity enhancements", IEEE Radio and Wireless
Conference, Aug 1998, pp. 181-184.
Wright, Andrew and Nesper, Oliver, " Multi-Carrier
WCDMA Basestation Design Considerations -
Amplifier Linearization and Crest Factor Control",
Technology White Paper, PMC-Sierra, Inc., Issue 1,
Aug 2002, pp. 1-35.
Liberti, Joseph and Rappaport, Theodore, Smart Antennas
for Wireless Communications: IS-95 and Third
Generation CDMA Applications, Prentice Hall, Upper
Saddle River, New Jersey, 1999.
Ata, Osama W., “Two-tone and Nine-Tone Excitations in
Future Adaptive Predistorted Linearized Basestation
Amplifiers of Cellular Radio.”, Wireless Personal
Communications, Intr. Jr., Kluwer Academic
Publishers, Vol. 16, No. 1, Jan 2001, pp.1-19.
Kenny, J .S. and Luke, A., “Design Considerations for
Multicarrier CDMA Basestation Power Amplifiers.”,
Microwave Journal, Vol 42, No. 2, Feb 1999, pp. 76-86.
CDMA Emission Mask: Federal Communications
Commission (FCC) Part 24 rules.
DESIGN CONSIDERATIONS FOR MULTI-CARRIER CDMA BTS POWER AMPLIFIERS
139