Edward Takacs,President
Array Wireless
Encinitas, CA
Christopher M. Durso, President
David Dirdo, Chief Technical Officer
Pacific Microwave Research, Inc.
Vista, CA
A Novel High Efficiency Linear PA Design for use in a C-OFDM Telemetry Link
Abstract: A conventionally designed radio frequency amplifier operated in its linear region exhibits
low DC to RF conversion efficiency. Typically, for a power amplifier designed for digital
modulation applications, the amplifier is operated “backed-off” from its P1dB point by a factor of 10
or -10 dB. The typical linear amplifier is biased for either Class A or Class A/B operation
depending on the acceptable design trade-offs between efficiency and linearity between these two
methods. A novel design approach to increasing the efficiency of a linear RF power amplifier using
a modified Odd-Way Doherty technique is presented in this paper. The design was simulated, built
and then tested. The design yields improvements in efficiency and linearity.
Power Amplifier, PA, C-OFDM, Digital Telemetry, Power Added Efficiency, PAE, Drain
Efficiency, Doherty Amplifier, Digital Microwave Transmission, Non –Line-of-Sight Wireless
A radio frequency power amplifier (PA) is used to provide the required signal boost from the output
of the transmitter to meet the design objectives of the link. The most important characteristic of the
amplifier is the high level of linearity required to ensure that no distortion of the COFDM waveform
results from the amplification process. Historically, the power amplifier used with digitally
modulated waveforms is operated approximately 10 dB backed-off (-10dB) from the P1dB point to
maintain operation in the linear range. Operating the amplifier in Class A mode provides an
undistorted amplified replica of the input signal. However, operating the amplifier backed-off from
the P1dB point results in a low DC to RF conversion efficiency.
System linearity and efficiency is critical to optimized operation of multipath resistant waveforms
such as COFDM. Waveform distortion that results from non-linear operation produces high levels
of intermodulation distortion (IMD) which degrades the performance of the transmission system.
When amplifying a typical 2000 carrier COFDM waveform, it is desirable to limit the spectral
regrowth to a level approximately -20 dBc or better. Conventional approaches to amplifier design
(saturated amplifiers) that provide a C/I of >-20 dBc tend to sacrifice efficiency for linearity.
Power added efficiencies (PAE’s) of amplifiers operated in a saturated mode vary from 25% to 70%
depending upon the design specifications and techniques used to achieve the amplified output.
However, a saturated mode power amplifier cannot be used for the transmission of digitally
modulated waveforms because of the resulting waveform distortion. The PAE of an amplifier
operated backed-off by -10 dB is typically on the order of 3 - 5%. Low amplifier efficiency results
in large amounts of DC power consumption for a given RF output power. This is undesirable for a
mobile wireless telemetry link stationed in a man pack, UAV, or other platform because of the
excessive additional weight and size of the batteries needed to generate the extra DC power. In
addition to greater power consumption, additional heat is generated from this inefficient technique.
This problem is a significant system design issue as larger RF outputs are required to transmit digital
signals over greater and greater distances. Improvement in the PAE of the amplifier is critical when
considering the size, weight, and power (SWAP) limitations of a particular platform.
The power amplifier discussed in this paper was designed using the Doherty technique to increase
the DC to RF conversion efficiency, while maintaining the required linearity of the signal to meet
system requirements. The Microwave Doherty Amplifier has been described extensively in
literature (Reference1,2). The technique was first suggested in 1936 as a means to reduce the
distortion in high power broadcast tube amplifiers and is primarily a power conservation technique
(Reference 3). It is currently being re-examined as a means to improve the efficiency and linearity
characteristics of high power microwave RF amplifiers.
The classic Doherty amplifier uses two active devices to achieve the desired output power. The
power transistors are arranged in a balanced configuration, although it is not a fully isolated balanced
structure. The input signal is split using an isolated in-phase divider, however, the output of the
transistors are combined using a non-isolated combining technique. The nature of the output
combining is such that one of the transistors will pull the load of the other transistor in the system as
required during high power operation.
A block diagram of the classic Doherty amplifier is shown in Figure 1. This representation of the
Doherty amplifier is one that is suggested as a reasonable realization of the technique as described in
Reference 3. In this implementation, the input signal to the amplifier passes through a coupler used
to detect the input signal. This signal is sampled and used to drive a variable attenuator in one of the
signal paths. The signal is split into a main path and auxiliary path after the coupler with an in-phase
-3 dB Wilkinson divider. The main path includes a transistor stage that is biased Class A to Class
A/B depending on the desired system performance. This stage is driven in-phase with the input
90 DEG
90 DEG
Figure 1 Block Diagram Realization of the Classic Microwave Doherty Amplifier
The auxiliary stage is biased Class B or Class C and is not conducting for a majority of the signal
cycle. The auxiliary path of the amplifier contains a 90 degree phase shift at the input prior to the
attenuator. This 90 degree phase addition acts to improve the return loss looking into the input of
the in-phase divider. A reasonable return loss looking into the power divider is realized using this
technique even when the auxiliary transistor is cut off. The variable attenuator is used to trim the
signal level amplitude to match the two transistor inputs in magnitude.
The input drive is used to turn the auxiliary stage on and causes the device to conduct. In the
scheme of the Doherty system, the bias points of the main and auxiliary transistors are selected such
that the auxiliary stage starts to conduct just prior to the main stage transistor reaching a drive level
that starts to cause distortion in the device (saturation). The auxiliary device is required to turn on
and reach the same maximum current swing as the main transistor at the maximum power point if
the same drive level is used as input to each transistor. This is graphically shown in Figure 2, where
Imax is the maximum current supplied by the amplifier. Imax will typically take on a value of 105 110% of the device Idss. This slewing action in the Doherty amplifier’s active devices requires the
auxiliary device to possess at least twice the gate periphery (i.e., be at least twice as large) as the
main device for proper operation. (Reference 3).
A 90 degree phase shift is added to the output of the main path transistor. This phase shift serves a
dual purpose in the Doherty amplifier. It is used to phase match the auxiliary and main paths,
accounting for the 90 degrees in-phase added to the auxiliary path in its input circuit. The 90 degree
shift also serves as an impedance inverter, which is a key detail in the successful realization of the
microwave Doherty amplifier. The Doherty amplifier provides reasonably good power added
efficiency as the amplifier is backed off from is P1dB point and maintains a relatively constant
efficiency from the maximum output power point to a point 6 dB backed off from maximum
(Reference 3).
Main Device
Auxiliary Device
FIGURE 2 Doherty Amplifier Device Current Variation vs Input Drive Level
In a conventional power amplifier design, as the input drive signal to the power amplifier is
increased the output signal begins to distort as a result of the non-linear action of the amplifiers
active devices. As the main stage transistor of the Doherty amplifier is driven toward distortion, the
auxiliary begins to conduct. The auxiliary amplifier acts as a current source. The insertion of a
current source in parallel with the output of the main stage pulls the load of the main stage transistor
keeping the main stage in its linear range. The net effect on the amplifier is to increase the linear
operating headroom for the amplifier main stage device.
Depending on the selection of the amplifier bias point, and the I-V curves of the devices used in the
amplifier, different mechanisms will cause distortion. Distortion of the signal will be caused by
voltage limiting as the transistor is overdriven, or it will be caused by current limiting if the device is
biased too close to cutoff. It is possible to have both types of distortion as well. The pulling of the
main device load impedance reduces the distortions due to voltage limiting in the device. The
distortions caused by bias point selection and current limiting are reduced because of the
cancellation of the distortion products between the stages due to the phase shift in the main stage
Recent papers have suggested (Reference 4) that the classic Doherty amplifier may be extended to a
generalized N-way combination of active devices. In particular, previous work has demonstrated
that an increased number of auxiliary paths in the Doherty amplifier can enhance the efficiency of
the amplifier at greater backed-off drive levels while improving the linearity. A schematic diagram
of a generalized N-way Doherty amplifier is shown in Figure 3.
90 DEG,
90 DEG +
(( ( ( ( ( (
90 DEG,
90 DEG +
(( ( ( ( ( (
( (
90 DEG +
(( ( ( ( ( ((
Figure 3 Generalized N-way Doherty Amplifier
Reference 4 suggests that an optimized 3-way (1 main transistor, 2 auxiliary transistors) Doherty
amplifier should out perform a 2-way amplifier in terms of linearity. Both configurations
demonstrate an improvement in PAE relative to an amplifier design that simply uses a backed-off
transistor amplifier design. The comparison of performance metrics between the 2-way and 3-way
Doherty amplifier configuration was done with approximately equal output levels. The
improvement in efficiency was observed at drive levels 6 - 12 dB backed-off from the input drive
level which resulted in maximum output power. A 4-way Doherty configuration was also examined
and found to exhibit reduced linearity relative to the 3-way configuration although it demonstrated
an improvement over the 2-way example. The reference suggests this outcome was likely due to the
sensitivity of IMD3 cancellation among the 3 peaking amps and the main stage.
The 3-way Doherty amplifier demonstrated an improvement in linearity when compared with the
classic (2-way) Doherty amplifier. The linearity of the amplifier was evaluated in terms of Adjacent
Channel Leakage Ratio (ACLR). ACLR is the ratio of the peak signal in any given channel
bandwidth to the noise caused by that signal in the adjacent channels. An improvement of
approximately 13 dB was observed in the ACLR when the 3-way Doherty amp was compared with
the 2-way combination. The improvement in ACLR using the 3-way Doherty amplifier design is
beneficial to the overall system performance and spectral efficiency. When less energy is present
outside the desired transmission channel bandwidth, more operational systems may be placed in any
given amount of spectrum. This is especially important when deploying a group of co-operative
UAV’s or UGV’s.
There are no restrictions on the type and size of device that can be used in each amplifier path.
Devices of different sized gate peripheries than the main path device may be used in the Doherty
amplifier without issue. The devices used in the two auxiliary paths may also be of different sizes,
either from the main path device or from each other. The devices selected do not need to be of the
same technology either (GaAsFET vs PHEMT,for instance). This ability to select any device
presents the power amplifier designer with a wide pallet from which to design an amplifier with the
desired output power and linearity. The designer may choose the devices needed to complete the
design from products that represent either a price or delivery advantage.
A microwave power amplifier using the Doherty design technique was fabricated and tested to verify
its performance and observe the benefits of the Doherty Amplifier. The key RF and DC
performance specifications for the amplifier are listed in Table 1. The power amplifier was designed
for use with a multi-tone, Coded Orthogonal Frequency Division Multiplexing (C-OFDM) signal for
transmission of digital telemetry and video data (Pacific Microwave Research DT-100C1). The
amplifier was designed for use on a mobile platform. The approximate size of the amplifier housing
is 4.5” x 6.5”. It was designed for use with an applied voltage between 11Vdc and 15Vdc.
Frequency of Operation (GHz)
Linear Output Power (dBm)
DC Current (A)
DC Voltage (V)
C/I (dB)
Desired Size
11 to 15
< -15
5.0” x 6.0” x .9”
Table 1 Key Performance Specifications of the 3-Way Microwave Doherty Amplifier
A 3-way microwave Doherty amplifier was built to meet or exceed the performance goals as set
forth in Table 1. A block diagram of the test platform PA is shown in Figure 4. The amplifier that
uses an 8 Watt device in the main path while the first auxiliary path uses an 8 Watt device and the
second auxiliary path uses a 12 Watt device. The power rating of each device is equal to the
anticipated P1dB point of the device. The total amount of device periphery in the amplifier is 28
Watts, or about +7.5 dB higher than the desired output. As a result, the amplifier operates backedoff by -7.5 dB, not the -10 dB of back-off that is the typically used when employing a standard
saturated amplifier in a digital system.
The fabricated power amplifier using the 3-way Doherty technique employs a directional coupler to
sample the input drive. The input signal is detected and used to set the bias on the auxiliary path
devices. The bias levels to the transistors are pulled to approximate the required response according
to the varying drive signal as shown in Figure 2.
A folded 3-way splitter was used to split the signal into three paths. A folded divider has the
advantage of minimizing the physical footprint of the amplifier making it suitable for a greater range
of host vehicles. There is approximately 1 dB of amplitude imbalance among the 3 paths within the
power divider. There is also approximately 60 degrees of phase imbalance among the ports as well.
The phase shifters shown in the amplifier, prior to the active devices, were used to compensate for
the phase imbalance among the three paths in the amplifier. The amplifier traces were tuned using
microwave techniques to compensate for the amplitude imbalance caused by the 3-way divider.
90 DEG
90 DEG
90 DEG
Figure 4 Block Diagram Realization of a 3 Way Microwave Doherty Amplifier
A driver stage was used prior to the folded 3-way in phase divider to provide sufficient level to drive
the output stages in order to attain 5 Watts of output power. The driver device was selected so that
it was about 8 - 10 dB backed-off from its P1dB point when driven at the nominal input drive level,
so it would not contribute distortions to the signal that would be further amplified by the downstream
output stages. Ideally, it is preferable to implement the driver stage as a Doherty amplifier section as
well. However, the size constraints precluded using the Doherty technique on the driver stage
because of the extra devices required.
The output stages were designed using the Doherty technique as indicated previously. An equivalent
section of transmission line was added to the output of each transistor to trim the output impedance
to the desired value. The transmission line has the effect of rotating the impedance looking into the
device output on the Smith chart to a value that improves both linearity and efficiency (Reference 4).
The trim line lengths are selected so that the output impedance of the devices are transformed close
to an open circuit at the combining output junction. This has the effect of canceling the
intermodulation products due to the main and auxiliary devices further improving the linearity. An
isolator was used at the output port in order to present a stable load to the device outputs.
The 3-way microwave Doherty amplifier was tested to verify its performance so that it could be
evaluated against the specifications listed in Table 1. The 3-way Doherty amplifier was also
compared with the estimated performance of a notional power amplifier designed using standard
design methods for use in a digital modulation system. This amplifier is used as the baseline
configuration for the Doherty amplifier.
The standard PA is assumed to operate at a point -8 dB backed-off from its P1dB output power level.
This was done to provide an equivalent output power for comparison against the Doherty amplifier
with respect to the relative conversion efficiency. A driver stage was defined in the both the baseline
design and the Doherty design to ensure the amplifier provided enough gain to meet the specification
minimum. Attenuators were added prior to both stages to trim the drive level to each device,
ensuring that these stages would not be driven into compression causing distortion of the input
signal. Isolators were used on both the amplifier input and output to provide a 50 Ω source and load
impedance to the active devices. A block diagram of the baseline power amplifier design used for
the performance comparison is shown in Figure 5.
Figure 5 Block Diagram of Baseline PA Configuration
The output device of the baseline amplifier was selected to provide equivalent gate periphery to the
aggregate output device gate peripheries found in the 3-way Doherty amplifier. The device selected
for the analysis of the baseline amplifier was the Mitsubishi semiconductor MGFS45V2735. This
device was selected because it is a 30W device that closely matches in size the total gate periphery
built into the 3-way Doherty amplifier. The Mitsubishi transistor is a 30 watt part with a P1dB point
of +45 dBm. The total amount of gate periphery contained in the output stage of the 3-way Doherty
amplifier is 28 Watts providing a net estimated P1dB output power point of +44.5 dBm.
The driver device selected for both the analysis of the baseline power amplifier configuration and the
3-way Doherty amplifier is the MGF0915A. This device exhibits a +36 dBm output P1dB point, and
consumes approximately 9.6 Watts of power while biased at about one-half of rated Idss. The
MGFS45V2735 was analyzed at a bias point that was recommended in the part data sheet. A
breakdown of the estimated DC power consumption for the baseline amplifier based on
recommended operating points from the device data sheet is given in Table 2.
Driver Stage Power Consumption
Output Stage Power Consumption
DC Overhead (Gate voltage, regulation, etc.)
DC Power Total
80 Watts
9.6 Watts
3 Watts
92.6 Watts
Table 2 DC Power Consumption Breakdown for Baseline PA
The 3-way Doherty amplifier was tested and the results were compared with the desired system
specifications and the baseline power amplifier. The Doherty amplifier was driven with C-OFDM
signals and CW signals to determine its operating characteristics. Efficiency and Linearity were
evaluated as a function of output power for the 3-way amp. The Doherty amplifier demonstrated a
minimum gain of +13.8 dB with a net DC power consumption of approximately 50 Watts for an RF
output level of +37 dBm when driven with a multi-tone C-OFDM signal. The C-OFDM signal used
to provide the input drive to the amplifier consists of 2000 individual carriers separated by 4 kHz
spacing. The amplifier operated in this manner over a bandwidth of 225 MHz (8% operating
bandwidth). Table 3 provides a comparison of the measured key operating characteristics and the
desired performance specifications of the amplifier when driven by a C-OFDM signal.
Key Characteristic
Frequency of Operation
Linear Output Power (dBm)
Linear Gain (dB)
DC Current (A)
DC Voltage (V)
C/I (dB)
Desired Size
Measured Performance
Desired Specification
37 min
13.7 min
5.0@ 12V,max
10 to 15
< -18
5.0” x 6.0” x .9”
37 min
13 min
<5 max
11 to 15
< -15
5.0” x 6.0” x .9”
Table 3 Comparison of Measured Performance vs. Key Performance Specifications
The Doherty amplifier under evaluation was driven with a CW signal of varying strength to
determine the characteristics of the amplifier when over-driven by the input signal. The amplifier
was driven by a signal up to +3 dB higher than the +24 dBm input normally used to excite it. The
data was taken at 3 frequency points across the bandwidth of interest. A chart demonstrating the
performance of the amplifier as a function of drive level is shown in Figure 6. The chart shows a
very flat gain characteristic as a function of frequency and drive. The gain remains flat within less
than +/- 0.25dB over the band of interest and up to the desired output level of +37 dBm.
Figure 6 also shows that the amplifier gain starts to compress at an output level of +38 dBm. The
amount of drive level available in the laboratory was not sufficient to drive the device-under-test to
the P1dB compression point. This P1dB compression point may be estimated from the data in the
chart at between +41 - +42 dBm. This level is slightly lower than might be expected from the
amount of device in the amplifier. This is probably due to the amplitude and phase imbalance in the
3-way signal divider (1.5 dB and 60 degrees). Reducing the amount of amplitude and phase
imbalance would improve the combining efficiency of the output stage leading to improved linear
gain characteristics and power added efficiency (compared to those levels that were measured).
Gain vs. Input Level
Gain (dB)
Output Level (dBm)
Figure 6 Doherty 3-Way Amplifier Linear Gain vs Output Power
The performance of the Doherty amplifier at levels backed-off from the desired operating output
power was investigated. Levels backed-off from the output operating point of -3 dB and -6 dB were
observed. The estimated performance of the baseline amplifier was calculated, plotted, and
compared against the Doherty amplifier data at rated output power and backed-off levels. This data
is shown in Figure 7 and demonstrates that the efficiency of the Doherty amplifier at rated output
power is improved by a factor of two when compared with the baseline amplifier. Additionally, the
PAE at levels -3 dB backed-off from are comparable with the estimated performance of the baseline
PAE vs Frequency
Power Added Efficiency (%)
Baseline PA @ 37
Frequency (MHz)
Figure 7 Performance of 3-Way Doherty Amplifier at Backed off Drive Levels
An amplifier was designed and tested using the Doherty design technique to demonstrate the
theorized improvement in power added efficiency and linearity as compared to a conventional
saturated amplifier design when used with a digital modulation waveform. The amplifier was
designed to address the specifications as outlined in Table 1. and used to compare against a power
amplifier using the conventional method of over-sizing the amplifier active devices by +10 dB. The
Doherty amplifier demonstrated an improvement of the power added efficiency by a factor of two
when compared with the baseline amplifier.
1. W.H. Doherty, “ A New High Efficiency Power Amplifier for Modulated Waves”, Proc.
IRE,Vol. 24, No. 9, pp. 1163-1182, 1936
2. F.H. Raab, P. Asbeck, S. Cripps, P.B. Kennington, Z.B. Popovic, N. Pothecary, J.F. Sevic,
N.O. Sokal, “Power Amplifiers and Transmitters for RF and Microwave”, Microwave
Theory and Techniques, IEEE Trans. On, Vol. 50, No. 3, March 2002
3. S. Cripps, “RF Power Amplifiers for Wireless Communications”,Artech House, pp.225-240,
4. Y. Yang, J. Cha, B. Shin, B. Kim, “ A Fully Matched N-Way Doherty Amplifier with
Optimized Linearity”, Microwave Theory and Techniques, IEEE Trans. On, Vol. 51, No. 3,
March 2003
5. Y. Yang, J. Cha, B. Shin, B. Kim,”Experimental Investigation on Efficiency and Linearity of
Microwave Doherty Amplifier”, 2001 MTT-S Digest, pp. 1367-1370
6. M. Iwamoto, A. Williams, P. Chen, A. Metzger, C. Wang, L.E. Larson, P.M. Asbeck,”An
Extended Doherty Amplifier with High Efficiency Over a Wide Power Range”, Microwave
Theory and Techniques, IEEE Transactions on, Volume 49, Issue 12, Date: Dec
2001, Pages: 2472 - 2479
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF