4590 Patrick Henry Drive Santa Clara, CA 95054-1817 Telephone: (408) 988-3500 Facsimile: (408) 988-0279 CMOS RFIC Switches: Simple and inexpensive, the latest 2.5GHz versions pose a legitimate challenge to GaAs switches. By Joe Grimm, Business Development Manager, RFIC Switches, California Eastern Laboratories There are a number of reasons why CMOS switches have caught the attention of RF engineers. First of all, they re inexpensive. They re also simple they don t have the dual voltage control issues or the need for complex peripheral circuitry other technologies require. But certain limitations have narrowed the range of application for CMOS switches. Their P 1dB power handling capability is lower relative to GaAs devices. Plus, until recently, they ve been limited to applications under 1GHz. But advancements in CMOS processing technology, like NEC s Silicon-on-Insulator (SOI) process, have nearly doubled the F T of these parts. The SOI process eliminates the depletion regions of the source and drain on the substrate, resulting in reduced capacitance. PMOS and NMOS are also placed closer together, allowing for a denser design and reduced metal capacitance. The results are higher speeds and lower power consumption and the ability to deliver insertion loss, isolation, and switching speed performance competitive with other switch technologies to 2.5GHz. Old School RF Switching The original RF switch, the PIN Diode, employs a number of discrete components to switch RF signals (Figure 1). A typical SPDT application requires a pair of PIN diodes, two decoupling capacitors, three bias inductors, and three DC blocking capacitors, plus an external driver to control switching speed. While not exactly a simple, compact solution, PIN Diode switches do have their positive traits. They re highly linear, and they can handle high levels of input power.
PIN Diodes are also tunable to high frequencies, but not without trade-offs. To improve isolation, two or more diodes can be lined up in series but not without an accompanying increase in insertion loss. Basically, PIN Diodes are current-controlled resistors. To reduce insertion loss, they require the application of large amounts of DC power to reduce resistivity in the I (Intrinsic) region. This obviously affects battery life. This characteristic, coupled with the number of components a PIN Diode solution requires, makes the technology a poor choice for portable handheld products. GaAs The Current Favorite Due to their low DC power consumption, small size, and relative simplicity compared to PIN Diodes, GaAs switches have become the device-of-choice in the wideband world. They re available for a broad range of applications from 50MHz to 6GHz, and they re offered by a number of manufacturers in a variety of different packages. GaAs switches are basically voltage-controlled resistors and generally require two complementary control voltages to operate. As these voltages are applied, an internal array of transistors acts as either a low or high impedance resistor, opening and closing the switch. But there s an intrinsic problem. GaAs switches are not designed to be used
with positive, low voltage power supplies. A common work-around is to place DC blocking capacitors (Figure 2) at the input and output of the switch. While these additional components allow the switch to be controlled by positive control voltages, they also limit its bandwidth and add to the insertion loss. Plus they eat up real estate. CMOS The Challenger There are a number of reasons CMOS is gaining favor with designers of RF products. First and foremost is cost. The CMOS process is inherently less expensive than GaAs, so the devices themselves are less expensive. The CMOS process also enables the driver and control circuitry to be combined on the same die with the switch, and to be operated via a single control pin. This simplifies the design process, and reduces the parts count. Unlike GaAs, CMOS switches use a single, positive voltage control, so no DC blocking capacitors are needed (Figure 3). This further reduces the parts count, shrinks the overall size, and lowers the manufacturing costs. Plus the problems that accompany blocking caps reduced bandwidth, more insertion loss, and increased noise are eliminated.
CMOS also feature extremely low power consumption. This, combined with their small size and the fact that they have no need for additional peripheral components, makes them an ideal choice for small, portable, battery-powered designs. Until recently, CMOS switches have been somewhat limited in application by their frequency range and power handling capability. With the introduction of new CMOS devices like NEC s UPD5710TK (Figure 4), this range of applications is rapidly expanding. The UPD5710TK operates at frequencies from DC to 2.5GHz, making it a good candidate for mobile communications handsets, set top boxes, instrumentation, and other
RF applications. It promises to be the beginning of a whole new generation of low cost, high frequency, high performance CMOS switches.