Punching through the ether with RF-range extenders
RF-front-end chips make it easier for you to design more distance into radios.
Paul Rako, Technical Editor -- EDN, May 13, 2010
At A Glance
|
| Sidebars: RF-power units and conversions |
| View as PDF |
The
RF (radio-frequency) spectrum is crowded, especially in the unlicensed ISM
(industrial/scientific/medical) bands. The radios in these bands typically use
spread-spectrum frequency modulation to reduce the effects of interference.
Although FHSS (frequency-hopping-spread-spectrum) and DSSS
(direct-sequence-spread-spectrum) techniques can allow multiple radios to share
the same frequency, the interference still exists. It manifests itself as a
reduction in range (Reference 1). As more radios share a
slice of bandwidth, spread-spectrum protocols allow recovery of errors due to
occasional interference. Once a sufficient number of radios all share the same
frequency band, however, each radio's range decreases as the number of clear
channels decreases.You can increase a radio's range by boosting the signal strength of the transmission and improving the sensitivity of the receiver (see the sidebar "RF-power units and conversions"). You can accomplish this task using a directional antenna, which concentrates the RF energy over a smaller angle, or you can boost the power of the transmitter and the gain of the receiver. The problem with directional antennas is that many radios are omnidirectional, especially in portable-system applications, leaving you with only one alternative: boosting the signal with a power amp and a low-noise receiver amplifier.
RF design is tricky, and designing power amplifiers is even more demanding than most other undertakings (Reference 2). RF engineers solve problems using frequency-domain analysis instead of time-domain techniques. RF design also requires specialized software, such as Agilent's ADS, AWR's Microwave Office, and products from Ansoft. Although such software packages have become easier to use, they still require the user to be knowledgeable about the principles of RF design. PCB (printed-circuit-board) design, including materials and impedances, is also critical in RF designs and their performance. You must pay attention to trace length and width to control the impedances of the connections because circuits interact with each other and reduce performance or cause unacceptable RFI (RF interference).
To maintain the robustness of your design and comply with FCC (Federal Communications Commission) regulations, you may need specialized signal-integrity software (Reference 3). This software enables you to predict the performance of the system before you lay out the PCB and then do a postlayout evaluation when you feed the finished layout back into the software. It is also difficult to prototype RF circuits. Using dead-bug and air-ball wiring techniques-so named because of how they look when complete-is strictly verboten (Reference 4). You must make a PCB that is a form, fit, and function equivalent to the production board you intend to sell, and you must specify the PCB in a way that will control its dimensions and thickness and that results in predictable trace impedance over the entire manufacturing run.
Once you have built a prototype, you need high-performance test equipment.
High-quality frequency generators and spectrum analyzers are expensive, and a
good VNA (vector network analyzer) that can measure both the amplitude and phase
of an RF-signal path can easily cost as much as your annual salary. RF-system
designers also need extensive experience. If a designer encounters a design
similar to a previous design, he can often dispense with the expensive design
and layout software and get by with minimal test equipment (Figure 1). An experienced RF engineer can't debug everything with an
antique grid-dip meter but can often get circuits working that inexperienced
engineers cannot despite using the best test equipment available. "We have a lot
of RF customers that have a digital- and firmware-design background," says Mark
Grazier, program manager at Texas Instruments. "Even given a discrete
part-reference design and a BOM [bill of materials], they don't get the
specified range in their designs." Grazier notes that these customers have more
success using TI's integrated ICs, to which they simply add an antenna and a
matching network.
For these reasons, you should realize that undertaking an RF design is not trivial. The needs for specialized design software, PCB-layout software, signal-integrity software, prototyping tools, and high-end test equipment all conspire to make it a challenging project.
Modulation protocols
When facing an RF-range-extender project, you must keep in mind that RF is always analog. Even if the high-level protocol of the radio is digital, such as in a cell phone, a Wi-Fi hotspot, or ZigBee, the radio waves are still in the analog domain. All that comes from the antenna is a sine wave that might wiggle in frequency, as an analog-FM transmission does; hop around, as an FHSS transmission does; or sweep and smear across a frequency, as a DSSS transmission does. The RF-signal path assumes an understanding of analog-design techniques ranging from the voltage divider to the Fourier transform.
One challenging aspect of new radio standards is that they use sophisticated modulation schemes to allow the transmission of more bits over an analog signal, similar to the way that a 56-kbps modem allows the transmission of more bits across a 3-kHz telephone line. Analog-modulation techniques, such as FM (frequency modulation), require no linear-signal amplifiers because the zero crossing of the waveform encodes all the information. The existence of a distorted sine wave doesn't matter, as long as the distortion doesn't affect the zero crossings. Modern modulation schemes pack 4, 8, or even 16 bits of information into each hertz of frequency bandwidth. Like telephone modems, they accomplish this task by relating the analog value of the RF waveform at any instant to a given digital value (Reference 5). Hence, the envelope of the RF signal is important, which implies that the linearity of the transmitter is critical.
Roll your own or buy?
If you want to add range to your RF design, you could try to design discrete RF amplifiers from transistors. Bear in mind, however, that you would need to design a power amplifier for transmission and a low-noise amplifier to improve the reception of the radio. Using this approach entails low-cost parts, but you must have a lot of experience and time, as well as expensive software and test equipment, to develop the design. Worse yet, the variations in the PCB may make each of your products behave differently in the field. You must address the process technology of the transistor you use. The technology you use depends on your power and linearity requirements and your ability. A silicon transistor may serve in a low-power, low-frequency application with a simple modulation, but you must use SiGe (silicon-germanium), GaAs (gallium-arsenide), or GaN (gallium-nitride) transistors if your radio operates at bandwidths higher than 3 GHz.
Designing a radio with discrete transistors rarely pays off. "Fully
integrated products will be more cost-effective than using discrete parts,"
notes James Long, an RF consultant. One problem with discrete designs is the
physical size of the RF-signal chain. You can reduce the size and complexity of
a radio by using an integrated RF power amplifier from Avago, for example
(Figure 2). This 2.4-GHz device can achieve 38.5 dB of gain and includes
three amplifier blocks as well as the gain-switching and amplifier-bias
circuitry, all in a 5×5-mm package. Integrated power amplifiers ensure good
linearity. The Avago MGA-43228, for example, can achieve 29.2-dBm
(decibels-referred-to-milliwatt) power output and transmit 64-QAM
(quadrature-amplitude modulation) or other sophisticated schemes that also
require linearity. Other products in Avago's lineup work in the 2.5- to 2.7-GHz
frequency band.
Maxim Integrated Products makes both low-noise and power amplifiers that you can combine to make your RF front end (Reference 6). Some of these products have 32-, 25-, and 18-dB gain at 800 MHz, 2.4 GHz, and 5.3 GHz, respectively. The company also offers the MAX2642 low-noise amp for 900-MHz cellular- and cordless-phone applications. It uses a SiGe-semiconductor process for better performance than that of discrete silicon transistors or, more surprisingly, GaAs low-noise amps. It has 17-dB gain and sells for 80 cents (1000).
Full-line RF company RFMD provides a wide selection of power amplifiers and
low-noise amps. The company combines transmitting and receiving transceivers
into one chip, such as the ML5805, targeting 5.85-GHz ISM radios (Figure 3). The transceiver integrates a power amp and can attain an output
power of 21 dBm with an input sensitivity of -97 dBm. It also includes a
fractional-frequency synthesizer and comes in a 6×6-mm package. RFMD recently
released the ML2730, an integrated FSK (frequency-shift-keying) transceiver that
includes both power and low-noise amplifiers. The device targets use in the
2.4-GHz ISM band and provides 21-dBm gain at its output amplifier. The unit also
has a frequency-synthesizer subsystem. The low-noise amp achieves -97-dBm input
sensitivity and comes in a 40-pin, 6×6-mm QFN package.
Analog Devices also offers complete transceiver chips, such as the AD9353, but the company incorporates ADCs and DACs into the chips, taking you all the way from RF to digital bits. This device is a complete radio but would be unsuitable if you were just trying to add an RF front end to extend the range of your radio design. Similarly, Semtech makes the XE1205, which can provide 15-dBm output power in the 180- to 1000-MHz range. This part is a complete radio, offering FSK modulation and a digital interface. Silicon Labs also offers a complete ISM transceiver, the Si4421, which can achieve an output power of 7 dBm into 50Ω, which is probably not suitable for a long-range radio. Still, with an input sensitivity of -110 dBm, the device can work with an external power amplifier that you design to get the range you need.
Texas Instruments' RF front-end chips may be ideal for designers who want to
extend the range of a design. TI currently makes three transceivers for this
purpose. The CC1190 targets use in frequency bands lower than 1 GHz, and the
CC2590 and CC2591 work in the 2.4-GHz ISM band. All three parts combine a power
amplifier, a low-noise amplifier, a power-amp preamplifier, a
transmitting/receiving switch, and the requisite logic to let you switch between
transmitting and receiving modes. You need to add only an antenna and a matching
network. The CC1190 operates in the 850- to 950-MHz range and can provide 27-dBm
output power (Figure 4). TI manufactured the device on a cost-effective CMOS process,
and it achieves a 2.9-dB noise figure on the low-noise amp. The CC2590 and
CC2591 extend the range of 2.4-GHz radios and provide 12- and 22-dBm output
power, respectively (Figure 5). Both integrate a balun to interface with the differential
signal from the low-power radio for which you are extending the range. TI also
offers evaluation boards that can speed the development of your system.
Once you boost the output power of your amplifier, you reduce the linearity and deteriorate the EVM (error-vector magnitude) and ACLR (adjacent-channel-leakage ratio) of the communication, especially if your goal is to design an efficient power amplifier. By modulating the RF waveform to levels near the power rails, you compromise the linearity of the output signal. You can look at sophisticated amplifier architectures, such as the Doherty amp, to improve efficiency, but they are also less linear. As a result, you must predistort the RF signal to compensate for the power amplifier's distortion. Normally, you would have to do this task using a complex DSP.
Fortunately, start-up Scintera Networks has come up with a way to perform
predistortion purely in the RF-analog domain (Figure 6). The company's product automatically and continuously
characterizes the linearity of your power amplifier and adaptively updates the
coefficients to correct the linearity. The appealing feature, however, is that
the RF path of the predistortion correction signal is purely analog. This
approach saves both cost and power consumption. Without the need to have a DSP,
you can maximize the performance of the radio and deeply modulate the signal for
lower operating and capital-equipment cost (Figure 7). "Linearization is particularly important to meet EVM for power
amplifiers handling high-order modulation signals, such as LTE [long-term
evolution] and WiMax [worldwide interoperability for microwave access]," says
Roger Merel, vice president of product strategy at Scintera. This linearization
is particularly important when you are using RF-power amplifiers and the
adjacent channel is also licensed spectrum with which you must not
interfere.
Don't despair if your radio needs more range; many system designers these days face this situation. The need to extend the range of your radio will only increase in the future as the radio spectrum becomes more crowded. Now that you understand the scope and trade-offs involved, you can make a practical plan to design, build, and test your new system. You can start with an RF engineer or consultant with extensive experience who uses that experience to design an RF front end from discrete parts, minimizing the risk and reducing the need for expensive software and test equipment. A quicker and simpler alternative is to let the analog- and RF-semiconductor companies apply their extensive experience and expertise to your problem. By buying a ready-made RF front end or assembling one based on a reference design, you remove the risk from your project and can still meet your schedule. Once you have designed a system with more than enough range, you can then reduce the gain in the RF-signal path to save power or increase gain to improve range at the expense of power consumption. Just remember that you should have a working system fairly early in the project because you will need to do testing to ensure that your radio meets FCC and worldwide regulations. Why reinvent the wheel? Use reference designs to get the head start you need in a demanding RF-design project.
You can reach Technical Editor Paul Rako at 1-408-745-1994 and paul.rako@edn.com.
| References |
|
| RF-power units and conversions |
|
Decibels are popular units of measurement among engineers because the units'
logarithmic nature means that you can use them to represent a range of values.
Remember, though, that a decibel is a ratio measurement relative to some
standard. For instance, dBV (decibel voltage) is the measure of a voltage
relative to 1V. The decibel refers to a voltage gain you calculate as
20log(ratio), so a gain of 10 equals a gain of 20 dB. A voltage gain of 1
million is 120 dB. Because power is a multiplication of voltage and current,
decibel power terms are 10log(ratio), so a power gain of 10 is 10 dB. You often
measure RF (radio-frequency) power relative to 1 mW, yielding the
decibels-referred-to-milliwatt scale (Table A). The dBw (decibels-referred-to-watt) scale is power relative to
1W.
|
Talkback
