Tactical military radio requirements have seen significant changes over the last several years. In order to meet the communication needs of Future Combat Systems, programs like the Joint Tactical Radio System demand that radios be software reconfigurable. The RF hardware associated with software-defined radios must be capable of transmitting over a variety of waveforms with varying channel bandwidths over frequency bands ranging from 2 MHZ to 2.8 GHZ and beyond. In addition, the handheld and small form factor versions of these radios require reduced parts count to optimize size, weight and cost. They must also provide greater power efficiency to increase battery life.
These requirements pose great challenges for radio designers. Traditional analog inphase/quadrature (I/Q) radio architectures are an excellent solution for radios designed for band-specific operation. However, they are not well suited to multimode multiband operation due to their band-specific nature. This article introduces design of a digital polar transmitter architecture to overcome the limitations of classical analog transmitter architectures in multiband and multimode radios.
I/Q Transmitter Limitations
I/Q-based transmitter design can be problematic when multiband operation is required. Baseband mixing produces unwanted spurious products requiring use of surface acoustic wave (SAW) bandpass filters to reject these adverse signals and minimize wideband noise. The use of bandpass filters forces designers to diplex the signal path or build individual transmitter chains for each band of interest. This design means additional circuitry is required in order to achieve multiband operation. Additional circuitry increases the size, weight and, in most cases, power consumption of the transmitter chain, which is counter to some design objectives.
To achieve multimode operation, analog transmitters architectures must be able to accommodate constant envelope signals (phase and frequency modulations), as well as non-constant envelope signals (amplitude modulated) simultaneously. To avoid distortion of non-constant envelopes, analog transmitters must employ linear (class A) or use predistortion techniques to “linearize” slightly saturated (class A/B) amplifiers. Both of these implementations sacrifice power efficiencies and result in decreased battery life for the user when the transmitter is operating in constant envelope mode. Table 1 summarizes the disadvantages of current design techniques.
|I/Q modulators||Lower efficiency/noise|
|SAW filters||Lower efficiency/higher cost/size|
|Linear PAs||Lower efficiency|
|No “true” multimode capability||Performance capability/higher cost/size|
Digital Polar Transmitter
True multimode multiband operation can be achieved in a single transmit chain through digital polar design. Fig. 1 depicts a high-level architecture abstraction of a digital polar transmitter. The digital I and Q datastreams generated by the baseband IC (modem) are input to a CORDIC processor. The CORDIC processor transforms the Cartesian coordinates (sine and cosine) into polar coordinates (amplitude and phase). The amplitude (A) and phase information are separated and sent down separate paths until they are recombined in the digital power amplifier.
The phase information extracted from the original signal (either constant envelope or non-constant envelope) is transformed into a constant envelope signal. This is achieved by phase modulating a phase lock loop designed to output the desired transmit frequencies. The resulting signal may now be amplified by compressed amplifiers without concern of distorting amplitude information.
The extracted amplitude information is quantized into control bits. These bits are used to modulate a digital power amplifier (DPA). Each bit is a digital representation of the amplitude envelope. The control bits are used to switch amplifier elements of the DPA into on or off states. The examples use a 7-bit control word offering 128 unique amplitude-modulation states. Fewer quantization states can be implemented if decreased amplitude-modulation resolution is acceptable. More quantization states also can be implemented for greater resolution.
The digitized amplitude envelope and the phase-modulated RF carrier are synchronized and recombined within the DPA to produce linear and efficient RF transmission. The result is a precise and repeatable phase and amplitude-modulated signal. Fig. 2 illustrates how the phase and digital amplitude are combined within the DPA to produce a linear output signal. These results were taken under p/4-DQPSK modulation. Fig. 2a depicts the constant envelope phase signal that is used to drive the DPA. When the digital envelope is activated, the RF signal is amplitude modulated to produce an amplified RF output signal containing phase and amplitude information (Fig. 2b).
Since the transmitted signal quality does not depend on the linearity of the amplifying devices, the transmitter maintains efficiency by consuming less current across the range of the operating output powers. The DPA can provide improved efficiency for constant envelope and non-constant envelope waveforms simultaneously. Fig. 3 shows the current consumption of a linear power amplifier in a typical CDMA transmitter, compared to the current consumption of a digital power amplifier. Clearly, high dynamic range (>50 dB) can be provided through the control of the base bias voltages of the DPA while retaining superior efficiency.
Digital Transmitter Linearity
All amplifiers are subject to some level of AM/AM and AM/PM signal distortions when driven into saturation. These distortions can degrade the quality of transmitted signal. The digital transmitter architecture has the advantage of providing a method for correcting nonlinearities at the digital base-band. To correct for these nonlinear effects, the DPA is characterized in each of its digital states. This information is then used to select the correct digital state to achieve the desired modulation characteristic.
This digital approach offers several advantages over analog implementations of polar modulation-based transmitters. Due to the intrinsically digital nature of the digital power amplifier, the AM/AM and AM/PM characteristics are exclusively defined by the behavior of this GaAs HBT-based integrated circuit. The other analog components on the transmitter transfer characteristics, including linear regulators, have no influence on the output signal quality. This characteristic has significant implications for the repeatability and simplicity of look-up tables, which are used to implement digital correction. The overall effect demonstrates the fundamental advantage of digital over analog predistortion systems that typically require complex and expensive adaptive feedback circuits to achieve equal levels of performance.
Fig. 4 shows the measured AM/AM and AM/PM characteristics of the DPA for an applied ramp signal. Ideally, as the digital state (the amplitude of the envelope) increases, the RF output voltage should increase linearly and there should be no phase deviation. This linear behavior generally holds true at low power levels, but more distortion exists at high power levels as the DPA moves toward compression. At high power levels, digital optimization is easily applied to compensate for any amplitude and phase nonlinearities. Fig. 4a shows the performance before and after digital optimization is applied. The improvements in the digital amplifier characteristics are readily apparent.
The DPA behaves as a digital-to-analog converter and is subject to the quantization noise associated with digital signal generation. The typical solution (such as implementing a lowpass reconstruction filter after the DPA to remove quantization noise and sampling replicas) will introduce excessive loss and degrade efficiency. To overcome this effect, baseband signal processing is used to provide out-of-band noise suppression. Fig. 5 shows an example of the impact of this signal processing on the noise performance.
In this example, the baseband signal processing on the amplitude path provides more than 35 dB of noise suppression. When coupled with the appropriate digital filtering of the phase path, the digital transmitter architecture eliminates the need for SAW filters ahead of the DPA. This design provides a significant architectural advantage and demonstrates the unique attributes of implementing digital polar modulators.
Full digital control and programmability, coupled with the wideband-amplitude modulation, enables digital polar transmitters to generate true multimode operation. Furthermore, given that this architecture can eliminate “band-specific” hardware (such as filters and narrow-band amplifiers) in the transmitter, it can be tasked for different frequency bands and allow for different modes of operation by simply reconfiguring frequency, clocks and digital filtering coefficients that support the desired standards.
Significant advantages are available to multimode, multiband radio designers who choose to use digital polar transmitter design techniques. These advantages include:
Elimination of I/Q modulators that create mixer spurs and signal imperfections due to mismatch.
Elimination of SAW filters through the use of digital signal processing algorithms.
Efficient power operation in both constant envelope and non-constant envelope modes.
True multimode/multiband capability.
These advantages all add up to smaller size, lower weight and less power consumption enabling multiple waveforms to be ported onto smaller form factor platforms then were ever before possible.
About the Author
Steven Hurwitz is manager of strategic initiatives for M/A-COM Inc.'s Aerospace & Defense business unit.