null  null
Application Note AN-003
Wideband Doherty Amplifier for WiMAX
Introduction
Doherty Theory of Operation
This application note describes the operational theory
and characterization of a Doherty amplifier reference
circuit that incorporates two Nitronex NPT25015 GaN
transistors. The amplifier is designed for operation over
the 2.5 - 2.7 GHz WiMAX band.
The Doherty configuration employs two transistors in a
quadrature-like configuration, as shown in Figure 1.
WiMAX orthogonal frequency-division multiplexed
(OFDM) and orthogonal frequency-division multiple
access (OFDMA) signals typically have peak-to-average
envelope excursions of 8-12 dB or more. To avoid
clipping, power amplifiers must be significantly backed
off saturation. Unfortunately, backoff is accompanied by a
dramatic drop in drain efficiency. To boost efficiency in
complex modulation environments such as WCDMA and
WiMAX, load modulation techniques such as the Doherty
amplifier configuration are becoming more widely
adopted.
The Nitronex family of GaN devices is well suited for
implementing the Doherty amplifier configuration. The
NPT25015 is a 15W peak power HFET with a saturated
efficiency >50% and is inherently broadband since it
contains no internal matching structures. Within the
Doherty circuit the NPT25015 devices operating at 10 dB
back-off typically exhibit >30% drain efficiency over a
200 MHz operating bandwidth. This level of performance
may be difficult or impossible to achieve with other
transistor technologies such as LDMOS and GaAs.
The Doherty amplifier circuit is more complex than a
two stage parallel amplifier. However, the increased
complexity is offset by a more flexible amplifier
architecture. That is, Doherty circuit operation can be
tailored by adjusting the transition point to trade off
efficiency, gain and linearity. For example, the transition
point can be increased to enhance efficiency at the
expense of linearity. The ability to adapt the amplifier
operating point is an important feature of the Doherty
configuration.
Figure 1: Doherty amplifier functional diagram.
The upper device is termed the Carrier or Main
amplifier. It is biased in the linear class AB mode. The
amplifier shown beneath the carrier amplifier is termed
the Peaking or Auxiliary amplifier. It is biased in the nonlinear class C mode. High efficiency results from 1)
dynamically adapting the load presented to the main and
peaking amplifiers over their dynamic range and 2)
operating the peaking amplifier in the highly efficient
class C mode.
The dynamic load adaption is produced by an impedance
inverter, used in conjunction with an impedance
transformer. A 50 ohm quarter wave transmission line
impedance inverter is used in classical Doherty designs,
accompanied by a 50/ 2 (35 ohm) quarter wave
transformer. The transformer rotates the 50 ohm system
characteristic impedance down to 25 ohms at point “K” in
Figure 1. At low power levels, the main amplifier acts as
a current source while the class C biased peaking
amplifier is pinched off. Ideally, the peaking amplifier
“off-state” impedance (looking into the peaking amplifier
output match) is infinite, effectively appearing as an open
September 2007
Application Note AN-003
circuit to the carrier amplifier. The carrier amplifier
voltage swing looking into a higher transformed load
increases, thereby producing higher gain and efficiency.
The higher gain compensates somewhat for the gain lost
from the as yet non-contributing auxiliary amplifier.
The point at which the peaking amplifier begins to turn
on is the Doherty amplifier transition point. At the
transition point the carrier amplifier loaded such that it is
near saturation. As the amplifier is driven harder the load
at point “K” appears larger. The impedance reflected back
from point “K” to the carrier amplifier decreases. The
carrier amplifier remains near saturation while load adapts
dynamically, maintaining a high efficiency as power
output increases. At rated power both 50 ohm amplifiers
are in parallel, so 25 ohm net impedance is seen at “K”.
The output transformer rotates the 25 ohm impedance to
50 ohms, ideally matched to the system characteristic
impedance. At maximum power output both amplifiers
are operating near saturation and Doherty efficiency
reaches a maximum.
In a large peak-to-average signal environment like
WiMAX, the peaking amplifier contributes power to the
load only during short bursts when the instantaneous
power exceeds the average power. Otherwise the peaking
amplifier is inactive. With proper phasing, the Doherty
amplifier produces the same peak power as two class AB
amplifiers in parallel, but with higher efficiency under
high backoff conditions.
GaN Doherty Circuit Architecture
The carrier and peaking amplifiers utilize the Nitronex
NPT25015 device. The NPT25015 is a 15W, unmatched,
8 mm gate periphery plastic overmolded GaN HFET. The
matching circuit for both the main and auxiliary amplifier
is shown in Figure 2.
Figure 2: GaN amplifier circuit diagram
A 3 dB quadrature splitter divides the input signal
equally between the carrier and peaking amplifiers. At the
output a microstrip quarter wave impedance inverter
combines the carrier and peaking amplifier signal paths.
The combined output signal is applied to a quarter wave
microstrip impedance transformer. An RF/DC crossover
is used to bridge the drain feed connection before reaching
the output RF connector.
Powering up the GaN Doherty Circuit
When working with GaN HEMT devices, it is important
to follow the proper biasing sequence. Please refer to
Nitronex Application note AN-1 for the correct method to
bias GaN devices.
The Doherty reference fixture is shown in Figure 3. Gate
bias for the main and auxiliary amplifiers is supplied via
two BNC connectors. The drain voltage is supplied
through a single BNC connector. A two position jumper
connects the drain supply to both amplifiers. A second
BNC connector can be added to the fixture to facilitate
independent main and auxiliary drain voltages.
If a
second drain voltage supply is used, the two position
jumper must be removed. The main amplifier bias voltage
should be set for approximately -1.3V to establish the
main amplifier IDQ at 180 mA. The auxiliary amplifier
gate bias should be set to -2.75 V for highest peak power
and best efficiency. Auxiliary amplifier bias adjustment
changes the Doherty amplifier transition point, which can
be useful in customizing Doherty circuit performance.
September 2007
Application Note AN-003
Figure 3: GaN Doherty Reference Circuit
Once the bias points are established, the drain voltage
can be safely applied. The carrier and peaking amplifier
use the same drain voltage (24V). There is a two position
jumper on the fixture board which should be set so a
common drain voltage can be applied to the drains. With
the jumper set, increase the drain voltage to 24V. IDQ
should be 180 mA at 24V. The carrier amplifier gate bias
voltage may require some adjustment to achieve 180 mA
IDQ. Refer to the “Measured Results” section for typical
performance with a WiMAX signal.
Figure 4: Doherty Power Sweep Platform
The AM-AM and AM-PM responses are captured with
the network analyzer in the power sweep mode at 2.5, 2.6
and 2.7 GHz. The power gain should be about 10 dB at
low drive levels, and then falls off with compression. A
typical swept measured response at 2.6 GHz is shown in
Figure 5.
Once the amplifier has been exercised over a range of
bias and drain voltages, the designer can choose the best
tradeoff between gain, linearity and efficiency for a
particular application.
Characterizing the Doherty Circuit
It is recommended that the GaN Doherty be initially
characterized for AM-AM and AM-PM responses as well
as return loss. A preamplifier is required to boost the
network analyzer drive to put the Doherty circuit into full
compression. The test setup is shown in Figure 4.
The Doherty fixture is supplied with SMA type RF input
and output connectors. The reference design test fixture
also supports optional N-type flange connectors. With the
input and output connected to a 50 ohm generator and
load, set the carrier and peaking amplifier bias point,
following the bias setting procedure described in AN-1.
Figure 5: 2.6 GHz Doherty AM-AM and AM-PM.
Experiment with the relationship between drain voltage
value and main and auxiliary amplifier bias settings and
the Doherty 1 dB compression point to observe
performance tradeoffs.
A high peak to average signal such as WCDMA or
OFDM/OFDMA should be used to characterize Doherty
linearity. The linearity test setup is shown in Figure 6.
The network analyzer in Figure 5 is replaced by a dual
channel power meter and a wide bandwidth capable
spectrum analyzer.
September 2007
Application Note AN-003
Table 2 summarizes the typical Doherty configuration
responses at 2.6 GHz using operating point parameters
from Table 1. Doherty testing was performed with a 10
MHz bandwidth 64QAM 3/4 OFDMA signal with a 9.5
dB PAR @0.01% CCDF during the TX period of a 50%
duty cycle TDD waveform.
Table 2: Summarized Doherty Amplifier Performance
Figure 6: Doherty Linearity Testing Platform
The signal generator must be able to generate high peak
to average waveforms The preamplifier should be capable
of delivering 40 dBm peak power to the Doherty. The
setup is used to capture linearity metrics such ACP, EVM
and drain efficiency. Again, note how linearity is
influenced by adjusting the supply voltage and main and
auxiliary amplifier bias points. Increasing the drain
supply voltage will typically improve ACP and EVM at
high power levels, but at the expense of efficiency. Do
not exceed 32V drain voltage.
Gain and drain efficiency can be traded off by adjusting
the peaking amplifier bias point (VGSAUX). Figure 7
characterizes Doherty efficiency and gain against peaking
amplifier bias point (VGS) adjustments.
Measured Results
Using the nominal Doherty operating point parameters
shown in Table 1, the NPT25015 Doherty reference
design amplifier delivers 46.5 dBm typical peak power
from 2.5 to 2.7 GHz. WiMAX systems have demanding
linearity requirements, defined by Error Vector Magnitude
(EVM) and adjacent channel leakage ratio (ACLR).
Because the peaking amplifier is a highly non-linear
amplifier, Doherty EVM and ACLR are generally poorer
than conventional class AB amplifiers. Fortunately,
linearity can be substantially enhanced with the
application of Digital Pre-Distortion (DPD) techniques.
Predistorted measured results were captured with an
open loop Matlab based predistortion platform without
applying memory mitigation.
Table 1: Recommended bias point settings
Drain voltage
24V
Main amp IDQ
180mA
Figure 7: Doherty VGSAUX response curves
Peaking amp VGSQ
-2.75V
September 2007
Application Note AN-003
Figures 8 and 9 illustrate Doherty performance using
DPD linearization. Figure 8 shows Doherty single carrier
spectral performance at 32 dBm average power (35 dBm
during transmit) with a 10 MHz bandwidth 64QAM 3/4
OFDM signal with a 9.5 dB PAR @0.01% CCDF at 2.6
GHz. The bias points and supply voltages are provided in
Table 1. Conventional DPD correction is shown in blue.
Memory mitigation DPD is shown in purple. The GaN
based Doherty reference design exhibits negligible
memory effects, so memory mitigation provides little to
no benefit.
Summary
The Doherty reference design incorporates two state of
the art NPT25015 GaN transistors from Nitronex to
demonstrate a versatile load modulation technique. The
reference design achieves broadband performance and
high saturated efficiency with a complex modulated TDD
waveform. The reference design enables designers to
quickly become familiar with tailoring the Doherty
amplifier to meet the demanding efficiency, gain,
emissions and EVM requirements in the 2.5-2.7 GHz
WiMAX band.
Figure 8: Doherty spectrum with/ without DPD
Figure 9 illustrates Doherty EVM with and without
DPD, gain, and efficiency performance with a TDD
OFDMA waveform.
Figure 9: Doherty performance with/without DPD.
September 2007
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

advertisement