|
Polar
Modulation and Bipolar RF Power Devices
By Earl McCune, Chief Technology Officer and co-founder,
Tropian Inc.
"Polar modulation technology is
experiencing increased interest as the need grows to simultaneously
provide high quality signal modulation of bandwidth efficient
signals while achieving high DC to RF efficiency. Bipolar
transistors have historically been problematic when applied
to the most efficient architectures."
Polar modulation techniques have been used
as early as 1915, to specifically address the problem of
low transmitter power amplifier (PA) efficiency with envelope
varying signals. While polar modulation techniques generally
do increase overall PA efficiency, early implementations
were rudimentary, with reduced output signal quality compared
to contemporary good-practice linear circuit approaches.
However, more recently modern wireless systems have adopted
modulation schemes that provide high bandwidth efficiency,
such as orthogonal frequency division modulation (OFDM),
multi-level PSK and QAM. Use of these signals in battery
powered mobile devices has made RF power amplifier efficiency
a serious concern and polar techniques are experiencing
a renaissance in addressing this problem.
Existing developments of polar modulated transmitters generally
fall within three major categories: Polar Loop, Polar Lite
(sometimes also called "Polar Modulator"), and Direct Polar.
Polar Loop
To address the traditional signal quality problem of a polar
transmitter, feedback control can be used to correct the
output signal. Advantages of such a polar loop includes
improved PA efficiency gained from operating the power amplifier
closer to saturation, a low wideband output noise floor
and a reduction of circuit oscillation tendencies with varying
output load impedance.
Disadvantages include the need for a precision receiver
within the transmitter, control loop bandwidths which are
much greater than the signal bandwidth, a restricted output
power dynamic range, the stability of the non-linear feedback
control loops, and a lack of circuit design techniques when
operating with intentional circuit nonlinearity.
Polar Lite
One way to eliminate the nonlinear loop dynamics difficulties
due to feedback around the PA in the polar loop is to restrict
the polar operation to the signal modulator. In this type
of transmitter, the output signal is amplified from the
output of the modulator using conventional linear amplifier
devices. This approach provides a lower modulator wideband
noise floor, determined by the VCO phase noise, compared
to what is available from a linear design using a quadrature
modulator. Since the polar signals are at a low power level,
it is possible to use digital sigma-delta techniques to
realize the phase modulator allowing improved integration
with digital baseband devices.
The obvious disadvantage of this approach is the use of
a conventional linear amplifier with no improvement in transmitter
efficiency.
Direct Polar
The direct polar approach eliminates the additional receiver
associated with the PA feedback of the polar loop while
still retaining the amplitude modulation on the power amplifier
itself. Direct polar advantages include circuit simplicity
and very high energy efficiency with greatly reduced heat
losses. Furthermore, the circuit bandwidth is typically
only slightly greater than the modulation signal bandwidth,
the broadband noise floor is low with good modulation accuracy,
and circuit oscillation problems are eliminated due to deeply
compressed power amplifier operation.
Disadvantages include the need to fully characterize the
power amplifier in compressed operation across the output
dynamic range, time alignment of the AM/PM signal component
paths, correction (generally through predistortion) of AM-AM
and AM-PM distortion effects, and assurance of thermal and
manufacturing stability.
Direct Polar RF Power
Just as digital power consumption dropped dramatically when
ECL/TTL saturated-amplifier logic circuit structures were
replaced by CMOS switches, direct polar RF power follows
the same principle by replacing the linear circuitry with
an RF switch. The power transistor's role is shifted from
regulating the current through the load, to selecting when
current will flow in the load. The amount of current which
flows through the load impedance is set by the environment
around the power transistor, and is independent of the power
transistor itself. As long as the power transistor presents
an ON resistance negligible to the load resistance on its
output port, the power transistor itself becomes a "negligible"
circuit element, no longer dominating the overall circuit
signal-quality performance and efficiency.
The direct polar RF power "amplifier" is a 3-port structure
(RFin, MagnitudeControlin, RFout). The switching transistor
port impedances are not definable (because the circuit is
not linear), so impedance matching is no longer a valid
design technique. Successful design of such a power stage
requires changes from conventional linear amplifier design
techniques. Instead of setting a load line and biasing to
avoid reaching the endpoints of the load line, the direct
polar power "amplifier" must operate at these load-line
endpoints and avoid the linear region.
The term "amplifier" in its conventional sense is a misnomer
for this device as the output signal is not a proportional
scaling of the RF input signal.
Practical Realization Issues
Despite the advantages, there are still barriers against
widespread adoption of direct polar. These include a lack
of appropriate engineering tools (due to strong circuit
nonlinearities required), dealing with a different set of
AM-AM and AM-PM effects, realization of system-specified
dynamic range for all signals, and maintaining the required
AM to PM time alignment. Finally, the manufacturing community
must gain comfort with this technique, which has properties
and characteristics in general opposition with the linear
RF design.
Key to successful direct polar RF power realization is
the accurate characterization of the power stage, primarily
for output power, AM-AM nonlinearities, and AM-PM distortion.
The direct polar RF power technique is "RF open loop" by
definition; manufacturability is only assured if power stage
characteristics are predictable and consistent. To first
order, the fact that this stage operates as a switch leads
to some confidence - as long as ON is ON and OFF is OFF,
overall performance should be consistent. This premise has
been extensively tested against production variations, temperature
extremes, and accelerated aging. Results have demonstrated
that this characterization is practical for GMSK/EDGE signals
using pHEMT technology. Extensions to WCDMA and WLAN are
also consistent with these results
Bipolar RF Power Challenges
Bipolar RF power devices are designed to be linear controlled
current sources, and not operated as switches. For this
application, full efficiency requires operation as a switch.
Today there is a large bipolar technology manufacturing
base (primarily GaAs HBT) so there is a strong incentive
to apply this technology to polar modulation.
A number of issues tend to make a bipolar device less than
ideal as a switch:
• VCE,SAT - offset voltage reduces the voltage available
across the load resistor RL, reducing available output power
to (VCC-VCE,SAT)2/RL. This lowers efficiency because IC*VCE,SAT
is an additional power dissipation source. The existence
of VCE,SAT redefines the practical zero-envelope voltage.
• RF leakage - When VCC is brought to zero volts,
there is still RF power present at the PA output. This leakage
power adds to the output signal distortion.
• Saturation recovery time - at 2 GHz there are just
a few picoseconds available to shift from the saturated
ON state to the OFF state. Accurate time domain models are
needed to optimize these designs.
• Base current to hold ON state - To hold the bipolar
power transistor in its ON (saturated) state, a continuous
current must flow into the base terminal. This lowers overall
efficiency. Higher beta ratios reduce this requirement.
• Ballast - When a large transistor is comprised of
multiple connected smaller cells, some means of ballast
is necessary to assure even distribution of current among
the constituent cells. It is imperative that the means chosen
does not affect the overall efficiency.
There are also characteristics that make bipolar devices
attractive as switches, including:
• High transconductance - compared to FET technologies,
smaller input voltages are required to alternate the transistor
state between OFF and ON.
• High current density - the nature of bipolar structures
allows for higher current densities than in most FET technologies.
• Small die size - with higher current density, bipolar
technology allows smaller transistor sizes for a given collector
current capability than for FET technologies.
• Low ON resistance - ignoring VCE,SAT and ballast
effects, bipolar transistor ON state resistance is low.
Polar modulation technology is experiencing increased interest
as the need grows to simultaneously provide high quality
signal modulation of bandwidth efficient signals while achieving
high DC to RF efficiency. Bipolar transistors have historically
been problematic when applied to the most efficient architectures.
It is desirable to have the bipolar technology community
develop devices which are intended to operate as saturated
switches, minimizing the problematic effects discussed above
while maximizing the desirable characteristics. Not all
of these design goals are in opposition to objectives in
linear amplifier design, but some are. The application circuitry
is very different in a polar transmitter than in a linear
one. As the acceptance of polar techniques for multimode
transmitter architecture grows, bipolar RF power technology
can maintain its dominant position by adapting to the different
requirements of a polar transmitter.
TROPIAN INC.
www.tropian.com
TXTLINX.COM 74
Email
this article to a friend!
|
