Wireless Infrastructure Keeps Pressure On Amplifiers And Oscillators

From cost and ease of integration through critical performance factors, amplifier and oscillator designers are performing seemingly impossible feats to meet next-generation infrastructure needs.

Amplifiers and oscillators have always fueled the "microwave fire." Yet the needs of wireless communications infrastructure in particular are now pushing the limits for both types of components. Designers must provide performance that beats, rather than just meets, the latest wireless specifications. At the same time, costs are sensitive and companies must simplify the design process as much as possible. This trend translates into flexibility with products that can meet the needs of multiple wireless standards. Yet a need remains for specialized, very-high-performing requirements. As all of these varied solutions are pushed to the edge of performance limitsor survive tradeoffs that would once not have been thought possibleoscillator and amplifier manufacturers also must keep in mind the power consumption and other requirements demanded by the greening of base stations.

Of all the evolutionary changes that wireless communications required of amplifiers and oscillators, miniaturization may be the most dramatic. As noted by Kory Stone, VP Sales and Marketing at Anderson Electronics, Inc., "I believe the biggest impact of wireless requirement so far has been to size. The number of handheld or personal devices has explodedespecially for wimaX, cellular, and medical-monitoring devices (Fig. 1). It is the need for portability that has placed the most pressure on the size of the components. That size reduction, in turn, has had the biggest impact in the market and suppliers."

For oscillators in particular, Stone lists size and cost as the two biggest challenges that have been overcome so far: "I believe that frequency products have come a long way in size and cost reduction. Many types of oscillators today are about 1/10 the size and cost they were 10 to 20 years ago."

Impressively, these small devices are now providing broader coverage, allowing customers to target a variety of wireless specifications. "We deal with wireless infrastructure mostly, which is where system specs are derived," says Carissa Sipp, RF Systems Applications Engineer at Texas Instruments. "Phase noise/jitter play a key role in many wireless standards. Our TRF3720 has external loop filters and both a FracN and INT option, which enable superior flexibility for the customer to optimize phase noise/jitter, allow for resolution in the Hz in fractional mode, and over a very wide bandwidth (300 MHz to 4.8 GHz). This is key, as many customers want one device to handle the complete local-oscillator (LO) range needed for their wireless-infrastructure designs."

When it comes to Long Term Evolution (LTE) in particular, Sipp notes that the demands on oscillators become even trickier: "One of the biggest challenges for the PLL/VCO is the MC-GSM spec. The specifications are very stringent and a constant challenge for design to meet phase-noise performance. LTE is also heavily driven by phase noise and the cleaner to source to the modulator, the better the overall EVM performance. This is specific to the LO as, of course, there are many system challenges other than the LO that come into play with all wireless standards."

The TRF372017 upconversion device serves wireless-infrastructure applications for LTE, as well as variants of code-division multiple access (CDMA) and timedivision multiple access (TDMA). At the heart of it are an in-phase/ quadrature (I/Q) modulator and an integer-fractional-N phaselocked loop (PLL)/voltage-controlled oscillator (VCO; Fig. 2). The VCO uses integrated frequency dividers to achieve a continuous tuning range of 300 to 4800 MHz. The LO is available as an output with independent frequency dividers. The device also accepts input from an external LO or VCO.

For wireless infrastructure, there are clear advantages to more integrated approaches like TI's TRF3720. This trend toward integration will certainly continue. In a microchip form factor, for example, OEWaves, Inc. just demonstrated a low-phase-noise, whispering-gallery-mode (WGM), micro-optical- resonator-based opto-electronic oscillator (OEO). This microchip package is based on proprietary, high-quality-factor (Q), crystalline optical resonators and the OEO architecture. It is reportedly capable of microwave frequencies beyond 100 GHz. With significantly lower phase noise/jitter than conventional oscillator technology in a comparable form factor, the company asserts that this microchip allows system spectral purity to be significantly improved without compromising size, weight, and power (SWAP).

Modular approaches also have emerged to ease the design process. For example, a series of six VCXO clock modules from On Semiconductor vows to overcome the cost and performance restrictions related to traditional quartz-based products. As part of the firm's PureEdge family of silicon-based voltage-controlled crystal oscillators (VCXOs), the devices satisfy requirements set by modern 3.3-V, low-voltage-differential signaling (LVDS) clock-generation applications. The series supports frequencies from 74.25 to 160 MHz.

Of course, countless individual components continue to emerge to meet specific infrastructure needs as well. A new debut from Crystek Corp., the CVCO55CC- 3830-3830 VCO, operates at 3830 MHz with a control voltage range of 0.5 ~ 4.5 V. This VCO typically features phase noise of 108 dBc/ Hz at 10 kHz offset with output power of +7 dBm (Fig. 3).

With the "Super Single" oven-controlled crystal oscillator (OCXO), Vectron International promises to find the wireless "Holy Grail" by enabling new services and functionality while increasing performanceand, of course, lowering power consumption. Based on patented microprocessor-correction technology, the MX-041 can achieve temperature stability of 0.6 ppb over 40 to +85C while reducing power consumption by more than 55 percent during warmup and 50 percent during steady-state operation. The "Super Single" construction results in Allan deviation (ADEV) performance of 3 ppb at 1 s along with long-term aging of 10 ppb per year.

Continue on Page 2

Page Title

Many firms are concentrating on the black art of providing increased stability. One example is Frequency Electronics, Inc.'s FEI Series FE-103A, which covers 5 to 20 MHz. This double-oven design uses a fifth-overtone, SC-cut crystal to ensure both long-term and temperature stability. For engineers designing densely packaged systems that demand tight stabilities, Bliley Technologies created the NV47AE OCXO. Targeting VMA slot applications in particular, this OCXO is less than 0.50 in. in height while the majority of OCXOs in its performance category measure 0.75 to 1.0 in. Impressive temperature stability at reasonable cost is available in the DOC series of ovenized oscillators from Connor-Winfield. These surface-mount OCXOs and OCVCXOs provide temperature stabilities in the range of 20 to 100 ppb over commercial, extended commercial, or industrial temperature ranges.

These products are just a small representation of the range of oscillators targeting cellular base stations. In their diversity, however, they reveal a lot about what customers need most and how oscillator makers are answering them. Obviously, some want just the bare-bones product with solid performance while others want a more modular device that they can just plug in. Among other tradeoffs are cost versus performance, stability, and more.

Yet what stands out is that an increasing number of oscillators are starting to simultaneously satisfy multiple demands. With the advancement of next-generation networks like Long Term Evolution-Advanced (LTE-A), balancing design demands will only grow more difficult. "I believe that stability and power will be the next challenge," notes Anderson's Kory Stone. "LTE and WiMAX will need low-cost, high-accuracy frequency standards (Stratum 3E). Combining high stability with low cost is still not a reality. Initially, there will be new implementations of old technology. Unfortunately, these will also drive power consumption. Ultimately, it will require silicon support to overcome these issues in the most effective manner to save both power and cost."

Amping it Up
For amplifiers, many of today's trends mirror those in the oscillator space. For one, wireless-infrastructure providers are often turning to modular approaches to simplify the design process while saving space and cost. Longtime PA maker TriQuint, for example, is now more focused on integrating multiple functions into modules. TriQuint's base-station modules offer designers four levels of RF integration. At the highest level of integration, for instance, the Level 4 modules replace multiple discrete products with one module that incorporates two amplifiers, a digital step attenuator, and all input/output matching circuitry.

Given the many kinds of amplifiers that are available to serve the wireless-infrastructure market, however, much ongoing development focuses on pushing individual amplifiers forward. At this month's ARMMS RF and Microwave meeting and conference, for example, Andrei Grebennikov from Bell Labs, Alcatel-Lucent, Ireland, presented a paper on high-efficiency, multistage Doherty architectures based on inverse class-F PAs for base-station applications. As switching amplifiers, Inverse Class-F PAs have open circuits at the even harmonics with shorts at odd harmonics. They promise very high efficiency. In combination with the Doherty parallel configuration, good linearity can be achieved. The paper explains how these techniques are being applied to make very efficient 20-W base stations.

Essentially, three- and four-stage Doherty architectures can provide lower efficiency peaking points at significant (12-dB) output-power backoffeven for equal device-periphery ratios. In practical implementation, each PA is based on a high-power, gallium-nitride (GaN) HEMT or LDMOS field-effect-transistor (FET) devices. The transmission-line load network corresponds to an inverse Class-F-mode approximation. This theoretical analysis was based on an analytical derivation of the optimum load-network parameters to control the secondand third-harmonic components at the device output (including the device-output parasitic shunt capacitance and series inductance). In a single-carrier, 2.14- GHz WCDMA operating mode, a high drain efficiency of more than 60 percent can be achieved at an average output power of 20 W.

Recent work at California Eastern Laboratories also focused on an inverse class-F power amplifier. The firm's Mouqun Dong has presented the design of a high-efficiency, 5-W PA using an LDMOS transistor from Renesas Electronics, dubbed the NE5511279A. In a break from more traditional approaches, multiple transmission-line sections or resonant circuits were not used to realize the desired harmonic termination conditions for this PA. Instead, the design's output matching circuit simply consists of shunt capacitors and short sections of transmission line.

This circuit topology is commonly used in practical PA designs because of its simplicity and convenience for impedance transformation at fundamental frequency. To a certain extent, however, the same circuit topology also allows the manipulation of second- and third-harmonic impedances to achieve a nearly resistive, large second-harmonic impedance and a small third-harmonic impedance. Dong notes that efficiency of about 85 percent has been achieved at 150 MHz.

Of course, Class-F amplifiers are not the only ones targeting cellular infrastructure. Promising to deliver high linearity performance while consuming low power for cellular and WiMAX infrastructure applications, M/ACOM Technology Solutions recently launched the MAAM-009560 heterojunction bipolar transistor (HBT). This driver amplifier covers 250 to 4000 MHz with a +42- dBm output third-order intercept point over a greater than 20-dB input-power range. It features typical gain of 15 dB.

WiMAX and 4G also are among the applications targeted by the SSPA products from Stealth Microwave, which cover 3.3 to 3.8 GHz. Yet these amplifiers specifically target equipment that requires ultralinear output to meet emission requirements. They can be used in systems ranging from continuous- wave (CW) applications to complex signals, such as those with orthogonal-frequency-division-multiplexing (OFDM) modulation. Power levels range from 2 to 50 W with 20 to 60 dB of gain. Compared to standard solid-state power amplifiers, the modules utilizing pre-distortion technology claim to require less back-off for operation in the linear region.

Also, variable-gain amplifiers (VGAs) are in play. For example, Hittite Microwave Corp.'s HMC926LP5E is a digitally controlled VGA covering 700 to 2700 MHz. At 900 MHz, the gain can be programmed from 6.5 to 38 dB in 0.5-dB steps. The amplifier provides a gain control range of 1 to 32.5 dB at 1900 MHz and from 4 dB to +27.5 dB at 2600 MHz. The HMC926LP5E delivers a noise figure of 4.4 dB in its maximum gain state with an output third-order intercept point to +45 dBm.

To tackle complex digital waveforms in LTE and other applications, Empower RF Systems, Inc. offers correction techniques in its model 7091 amplifier (Fig. 4). This 900-MHz compact, linear PA is the first in a series of communications amplifiers that are being developed for digital waveform requirements in the following bands: 450, 700, 800, 900, 1800, 1900, and 2100 MHz. The 7091 amplifier, based on LDMOS devices, has a built-in predistortion engine with wide instantaneous correction bandwidth. It therefore ensures low distortion and wide-dynamic-range operation. According to the firm, efficiency to 40 percent is possible in this series.

With a series of gain-block amplifiers from GraSen Technology LLC, one part number promises to handle all of the current frequency bands with no change in gain, output power, or third-order-interceptpoint levels (Fig. 5). These amplifiers rely on an indium-gallium-phosphide (InGaP)/gallium-arsenide (GaAs) HBT process. The GSA603-12 InGaP HBT gain block amplifier, for example, covers DC ~ 3500 MHz. At 2 GHz, it provides 18 dB of gain with +16 dBm output power at 1-dB compression and an output third-order intercept point of +28 dBm.

This dizzying array of products offers just a few examples of the types of amplifiers and oscillators that are being developed to serve mobile-infrastructure needs. They take different forms to provide tradeoffs in terms of performance, simplified design, size, or costto name a few. At the same time, they must meet new emission requirements. The need for greener wireless infrastructure will continue to impact development. Most critical, however, is the need to satisfy the newest and emerging wireless specifications.

Over and over again, amplifier and oscillator companies note that it is their customers who push the limits of their designs. Customers do not want to just meet standard specifications; they want to beat them in order to optimize their designs' tradeoff potential. According to TI's Carissa Sipp, "Integration and flexibility demands on the parts and specs are always a tradeoff. The market is constantly challenging designs as the customers themselves place even more stringent specifications over that of the actual wireless standard. Working with the customer in the beginning stages of the definition of parts is critical to understanding the needs and specs they want to meet with the amount of integration they need."

Undoubtedly, customers will continue to push the envelope in amplifier and oscillator developments. Given the vast array of applications within wireless infrastructureand the varied ways in which designers want to satisfy themthe range of amplifier and oscillator solutions will remain diverse. They will be the same, though, in that they will push beyond traditional design to drive wireless infrastructure into the future.

TAGS: Content
Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.