Microwave Journal
www.microwavejournal.com/articles/22557-overview-of-rf-switch-technology-and-applications

Overview of RF Switch Technology and Applications

July 15, 2014

Modern RF instrumentation is heavily dependent on switch technology. As both power consumption and space requirements shrink, selecting the appropriate switch solution becomes increasingly important. This article provides an overview of RF switches and considerations for selection based on function within the RF instrument.

OVERVIEW OF TECHNOLOGIES

Solid State Switches

Solid state switches can be divided into two primary categories: diode and field-effect transistor (FET). PIN diode switches are generally realized using discrete implementations and are known for higher power handling and fast switching speed, but the fairly complex biasing schemes and high levels of DC power required are significant disadvantages, especially for battery-operated instrumentation. PIN diodes require a forward current through the device to establish a low series resistance. This direct injection of DC current into the RF channel limits low-frequency operation.

FET-based switches are more commonly found in integrated solutions because the switching behavior is voltage-dependent. The control voltage applied to the gate of the FET “switches” the channel from a low-resistance “on” state to a high-resistance, capacitive “off” state. Three-terminal FET devices feature minimal DC power consumption and separate gate control of the channel. The high impedance of the gate supports a broadband response, but the FET is still frequency limited by finite capacitance between the RF channel and the gate terminal, as well as channel capacitance when the device is in the “off” state.

There are two distinct FET types commonly used in industry today: MESFETs and MOSFETs. MESFETs are fabricated using gallium arsenide (GaAs) and gallium nitride (GaN) processes, while MOSFET devices are commonly silicon-based. A primary distinction between the two types is the asymmetric behavior of the MESFET gate to applied voltage.

Electromechanical and MEMS Switches

Electromechanical (EM) switches use a different mechanism than their solid state counterparts. In an EM switch, a metal contact is actuated to make or break the connection. This technology offers the advantages of low insertion loss, high isolation, and high linearity.1 Emerging microelectromechanical systems (MEMS) switch technology attempts to deliver the advantages of traditional EM switches, but in a small form factor. MEMS switches employ micro-miniaturized mechanical contacts controlled by electrostatic forces to make RF connections.1, 2

PERFORMANCE CATEGORIES

No single switch solution can satisfy all of the various requirements for each application. Frequency range, power handling, linearity, switching speed, insertion loss, isolation, video feedthrough, power consumption, and reliability are important considerations in determining which technology to use.

Frequency Coverage

Operating frequency is often one of the first considerations when selecting an RF switch for a given application. This category is multifaceted in that several other performance metrics, such as power handling and linearity, are dependent on frequency.

Low-frequency performance is another consideration that is equally important. The frequency response of FET and PIN diode switches is influenced by external components, such as RF bias chokes and DC-blocking capacitors. Even though a switch may be specified to operate at very low frequencies, the frequency response of DC-blocking capacitors may limit its ability to do so. Careful selection of these external components is necessary to preserve switch performance.

PIN diode switch performance degrades at low frequencies, where the diode no longer behaves like a variable resistance, but begins to function as a PN junction diode that rectifies the signal, thereby degrading switch linearity. The frequency at which this transition occurs is not clearly defined, but performance degradation can usually be avoided by operating the PIN diode above the transit time frequency of the device (fc), which depends upon its physical size per the equation:2,3,4,5

Math 1

where:

f = transit time frequency (Hz);

t= transit time of device (s).

FET-based switches have the capability of operating down to very low frequencies;2however, power handling degrades due to the impact of the FET’s reactance on the gate.6At the high end, GaAs FET switches have operating frequencies extending beyond 50 GHz. PIN-based switches are available with operating frequencies in excess of 70 GHz.

EM switches can operate all the way down to DC, as their inherent metal-to-metal switch contacts have no fundamental limitation to low-frequency operation or the passing of DC signals. As a miniaturized advancement of EM switch technology, MEMS switches can also operate down to DC; however, more attention must be paid to the maximum current carrying capabilities of the conductor to avoid fusing the switch.

Power Handling and Linearity

Power handling has come a long way from being just a simple maximum current or voltage rating for a switch element. Today’s extremely low energy-per-bit modulation schemes, such as 16,384-QAM of DOCSIS 3.1, and increased potential of self-jamming, as in LTE-A Band 5 and 13, require the RF switch to perform exceptionally well at peak power levels to meet in-band error vector magnitude (EVM) requirements as well as adjacent and out-of-band metrics. Out-of-band harmonics, adjacent channel leakage ratio (ACLR), and spurious levels are critical in today’s communication schemes.

Figure 1

Figure 1 Compression characteristic of a MESFET versus a MOSFET WiFi switch showing the sharper compression behavior of the MOSFET device.

Power handling is determined by the voltage or current limitations of the fundamental device technology. Linearity assesses the ability of the switch to faithfully transfer a signal without distortion. Common metrics are third-order intermodulation products and generated harmonics.

An important consideration is the ability of the technology to scale to power requirements. In MOSFET switches, this is commonly achieved by stacking multiple devices to support the maximum voltage level. A 3 V device requires a stack of four to support a 30 dBm (1 W) 50 ohm referenced RF signal. Multiple devices in series increase the total channel resistance, which is compensated by increasing the gate periphery of each individual device.

Compression characteristics can also affect linearity metrics where AM-AM distortion maps into non-linear behavior and increases EVM. The soft compression behavior of MESFET devices is a drawback versus MOSFET devices, which have a definitive breakdown characteristic. Figure 1 compares a MESFET- and a MOSFET-based WiFi switch showing sharper compression of the MOSFET device.

Switching Speed

For certain systems, the switching speed of an RF switch is critical. “Switching time” is described as the RF signal rise time through the switching elements (typically the 10 to 90 percent transition). The complete turn-on time (tON) includes the delay of the internal bias circuitry and is typically defined as 50 percent of control signal to 90 percent of output signal. Conversely, the turn-off time (tOFF) is defined as 50 percent of control to 10 percent of output. PIN diode and FET switches can switch on the order of microseconds or nanoseconds, whereas EM switches typically take much longer (on the order of milliseconds).

Figure 2

Figure 2 Insertion loss versus switch Ron.

Another important switching time definition, especially in amplitude-sensitive applications, is the time period beyond the 10 to 90 percent switching and is commonly referred to as the “settling time.” This period is defined for amplitude settling within, for instance, 0.1 or 0.05 dB of the final value. GaAs switches have relatively long gate lag periods due to charge trapped in the channel, but progress is being made to reduce this settling delay with improved device structures and biasing schemes.

Related to device switching is repeatability. The insertion loss of mechanical switches can vary from cycle to cycle due to contact surface effects. In comparison, solid state switches are generally more repeatable, but can suffer from variations in insertion loss over temperature, process, and bias levels.

Insertion Loss

Insertion loss defines the amount of signal attenuation that occurs for a given switch path. The insertion loss of a basic single-pole-double-throw (SPDT) switch can be divided among three major factors: (1) resistive loss caused by the finite resistance of the “on” channel; (2) reflective loss due to the resistance RONadded to the load impedance; and (3) signal leakage through any “off” path device capacitance. Figure 2 compares the total insertion loss of a switch to that due to its “on” resistance.

Table 1

Any change to reduce either the “on” resistance or the “off” capacitance while holding all other parameters constant would be beneficial. The switch figure of merit (RON·COFF) provides a means of comparison between different technologies, where a lower figure of merit will result in lower insertion loss. Table 1 shows the figure of merit for common solid state switch technologies. EM and MEMS switches typically offer very low insertion loss and have lower figures of merit than their solid state counterparts. It is important to note that RON·COFF/VMAXis sometimes used instead to capture differences in the maximum device voltage between processes. A drop in RONmay be accompanied by a drop in VMAX, and this may negate the RONimprovement if power handling is a critical parameter.

To achieve broadband insertion loss performance at high frequencies, the challenge is in managing the “off” device capacitance. This is commonly done by reducing device size while accepting degradation in the minimum insertion loss at low frequencies. Sometimes asymmetric switch architectures are used to optimize for unique frequency bands or to optimize insertion loss and isolation for distinct paths, such as minimal loss and power handling on transmit paths and minimal loss and increased isolation on receive paths. Narrowband tuning techniques can be leveraged to minimize capacitive effects and recover much of the insertion loss degradation over a 10 to 20 percent bandwidth.

Isolation

Switch isolation is defined by the transmission of an RF signal to the disengaged throws of the switch. Ideally, the entire input signal flows from the input to the desired output; however, in reality, some of this signal leaks to the undesired outputs.4A PIN diode or FET impedes signal flow through the device when it is not in its conducting state; when reverse biased, the device presents a high resistance. An EM or MEMS-based switch achieves isolation by physically disconnecting the metal-to-metal contact of the conduction path.

The non-conducting states of these devices, however, are not perfect, with non-idealities, such as parasitic capacitances that degrade the ability to impede signal flow. For series PIN diodes, the presence of a parallel junction capacitance degrades isolation between its terminals at higher frequencies. In general, a smaller junction capacitance yields higher signal rejection.3,4For FETs, capacitance between the drain and source introduces similar frequency-dependent effects on isolation.2,6For EM and MEMS-based switches, isolation is determined by the separation distance between switching contacts; greater distance yields higher isolation.

For each switch technology, different topologies can be employed to improve isolation. One primary example is the combination of series and shunt switching elements: when a switch path is disengaged, the series element is biased in the isolation state and the shunt element is biased in the conduction state, which essentially shorts the input signal to ground.2,4,6Both PIN- and FET-based switches are available with low-frequency isolation as high as 80 to 90 dB and high-frequency isolation as high as 40 to 50 dB. EM switches are available with ratings as high as 100 to 120 dB.

Video Feedthrough

Video feedthrough is manifested in the form of voltage transients that appear at the outputs when an RF switch changes path. This unintended signal occurs even without an RF signal passing through the switch. It is generated by the switch’s control signal and the discharge of biasing components.7

Video feedthrough is not commonly specified for RF switches, but can be an important consideration for certain applications. For example, high-gain amplifiers – especially those with automatic gain control – will momentarily compress or even sustain damage due to video feedthrough spikes. Similarly, these voltage spikes could cause damage to or interfere with the operation of instrumentation connected to the switch outputs. Receivers used for covert monitoring for signal intelligence (SIGINT) require very low levels of video leakage to remain invisible to outside observers.

Although this effect can be observed in both PIN diode and FET-based switches, it is generally more prominent in PIN diodes due to the presence of both the DC bias and the RF signal on the same path. The DC-blocking capacitors used to contain the bias within the PIN diode switch will charge and subsequently discharge when the control signal toggles from forward to reverse bias (and vice versa). In FET-based switches, the separation of the control signal from the RF path greatly reduces the amount of video leakage generated during switching.7

Switch Control and Power Consumption

The circuitry required to bias different solid state switch technologies has various degrees of complexity depending on the desired frequency, bandwidth, power levels, and switching speed. PIN diodes may have very fast switching speeds, but the drive circuitry can be complex. Limitations in the ability to tune out parasitic elements result in the need for expensive broadband chokes and more exotic assembly techniques. Bias voltages can also be relatively high and may consume high DC power while requiring large and complex filter networks to combat video feedthrough and other noise from coupling into signal paths.

Control circuits for voltage-controlled FET devices are generally simpler and require lower power, but often have a fixed RC delay that limits switching speed. There is design flexibility in setting the switching speed with possible tradeoffs in low-frequency power handling and insertion loss. Additional components, such as blocking capacitors, may need to be placed in the signal path due to the common-mode level requirements needed to ensure proper channel control.

Many solid state switch solutions are available with integrated bias and control. This reduces system cost and complexity, but can impact performance and limit flexibility. For example, some parts use integrated charge pumps to generate the internal bias levels needed for switching; however, charge pump spurs can be injected into the RF signal.

Reliability and Lifetime

Reliability and lifetime are critical when selecting switches for applications that require repeated switching or expose the device to harsh or uncontrolled environments. EM switches have limited lifetimes due to their mechanical switching mechanisms. Normal wear and tear over time will cause these structures to degrade. Solid-state switches provide the unique advantage of never wearing out by simply being switched under normal operating conditions, since no moving parts are used. For these switches, life expectancy is determined by overall time in operation, not the number of switching cycles. When used within specifications, solid state switches can exceed 500 million cycles, whereas EM switches typically endure 10 million or fewer operations.8Switch lifetime can degrade when operated under conditions that introduce additional stresses, such as elevated operating temperature, thermal fluctuations, or exposure to excessive signal power.

Figure 3

Figure 3 Basic switch architectures that describe the behavior of the unused switch port.

Another potential stress is hot-switching, which occurs when the device is switched while an RF signal is being applied. Although switches are specified for a maximum power handling, this can degrade when hot-switched. PIN diode switches are typically more robust when switched at high powers, whereas switched FETs transition through a region during which high amounts of signal power may be dissipated.9The ability of the FET switch to withstand this increased dissipation depends on its specified power handling. For example, GaN devices exhibit excellent power handling and can thus handle hot-switching at higher power levels.10MEMS switches are typically more sensitive to hot-switching, as micro-welding between contacts can occur, resulting in reduced lifetime.1

RF switches can also be susceptible to electrostatic discharge (ESD), which occurs when static charge is suddenly transferred between surfaces with differing voltage potentials. This can subsequently damage sensitive devices. GaAs switches are generally more sensitive to ESD, with typical ratings of 200 V. MEMS switches are similarly sensitive. PIN diode switches have a moderate sensitivity to ESD, and CMOS technology (including SOI) are relatively robust in this regard, with ratings as high as 4 kV.2

APPLICATIONS

Architecture Considerations

Two basic switch architectures that describe the behavior of the unused switch port are classified as absorptive or reflective (Figure 3). Absorptive switches present a termination (most commonly 50 Ohms) to the unselected arm typically at the expense of increased insertion loss. Reflective switches leave the unused port unterminated.

Reflective switches can be further categorized as either reflective-open or reflective-short. This distinction is critical in understanding the corresponding impact on the circuitry tied to that path. Reflective-open architectures do not have a shunt path to ground in the “off” state; as a result, the loading on the unused port will be minimized. For example, LNA bypass switches are reflective-open in order not to disturb the LNA’s functionality when the switch is in the “off” state. For reflective-short architectures, a shunt path to ground is established. This low impedance renders attached circuitry effectively useless.

Overview

Switching requirements for RF systems vary dramatically, and the performance metrics outlined in this article provide a means by which switch performance can be evaluated and compared. Test and measurement applications use broadband switches extensively, while communications systems are more likely to use narrowband switches to maximize performance. Narrowband designs often use inductance to resonant-tune the switch “off” capacitance in order to minimize insertion loss and maximize isolation over a more restricted bandwidth.

Transmit/receive-based radar applications require very fast switching speeds since this is crucial in setting the minimum close-in range that can be detected by the radar system. Switch too slow and the reflected transmit signal will already be past the receiver by the time the switch settles. Conversely, less time-sensitive applications, such as one-time, “set-and-forget” uses, can target better insertion loss or very low operational frequencies at the expense of a greatly reduced switching speed.

Application Example: RF Receiver Architecture

Covering all possible system considerations is beyond the scope of this article. As such, an example receiver block diagram (see Figure 4) is used to highlight typical switching requirements. The front end of a receiver typically requires an input protection system, which can be based on RF limiters, alone, or combined with a variable attenuator to keep the input power within a predefined range to maintain linear operation. Switches used in a dynamic, power-limiting structure need to switch fast enough to protect the system and prevent damage further down the signal chain. These switches must cover the full frequency range of the receiver without introducing unnecessary distortion as well as handle a wide range of input power levels.

Figure 4

Figure 4 Example receiver block diagram highlighting typical switching requirements.

RF gain control, functionally shown in Figure 4 as a digital step attenuator and a variable-gain LNA, consists of banded gain blocks, that can be bypassed, and various DSAs. The relative input power is detected and the signal can then be amplified or attenuated to ensure optimal power level at the RF filter banks and subsequent mixers. These switches require fast settling times to allow the measurement to reach its final value when gain changes are required.

The next functional block consists of an RF filter bank needed for harmonic and interference suppression. Here, a four-way filter bank is shown, but systems can have more elements depending on filter sharpness and covered bandwidths. These switches require high isolation at the out-of-band frequencies and low insertion loss within the selected frequency bands. This function can be obtained by using a combination of series and parallel switches or by using high-throw-count integrated switches (e.g. SP6T, SP8T).

The first mixer has a tunable local oscillator (LO) with many of the same system elements as the primary signal path. For applications with fast tuning requirements, short settling time switches are required having good isolation in the band of interest. The example shown in Figure 4 employs a direct conversion receiver at its final stage. This provides bandwidth efficiency, but in turn, requires I/Q signals that extend to DC; as such, the switches used in this section need to be DC-capable with low phase distortion at the maximum bandwidth.

CONCLUSION

Selecting the appropriate switch is extremely important for successful RF applications. As highlighted in this article, switch requirements can vary greatly even within a single system, such as an RF receiver. When equipped with the knowledge of available RF switch solutions, application requirements, and performance tradeoffs, engineers can effectively implement their designs and meet today’s advanced systems demands.

References

  1. T.W. Jau, “RF MEMS Switches: High-Frequency Performance and Hot-Switching Reliability,” High Frequency Electronics, February 2013, pp. 32-38.
  2. P. Hindle, “The State of RF/Microwave Switches,” Microwave Journal, Vol. 53, No. 11, November 2010, pp. 20-36.
  3. J.W.E. Doherty and R.D. Joos, “PIN Diode Circuit Designer’s Handbook,” Microsemi Corp., 1998. Available online at www.ieee.li/pdf/pin_diode_handbook.pdf, Accessed May 2014.
  4. R. Cory, “Solid State RF/Microwave Solid State Switches,” Microwave Product Digest, May 2009, pp. 1-7.
  5. Hewlett Packard, “Application Note 922: Applications of PIN Diodes,” Available online at www.qsl.net/n9zia/wireless/pdf/an922.pdf, Accessed May 2014.
  6. R. Cory and D. Fryklund, “Solid State RF/Microwave Switch Technology: Part 2,” Microwave Product Digest, June 2009, pp. 34-38, 60-66.
  7. Agilent Technologies, “Application Note: Video Leakage Effects on Devices in Component Test,” 2007, Available online at http://cp.literature.agilent.com/litweb/pdf/5989-6086EN.pdf, Accessed May 2014.
  8. rydom Inc., “The Life Expectancy of Solid State Relays,” February 2011, Available online at www.crydom.com/en/Tech/Newsletters/Solid%20Statements%20-%20Life%20Expectancy%20of%20SSRs.pdf, Accessed May 2014.
  9. vago Technologies, “Application Note 5376: HSMP-386J 10W Series Switch,” Available online at www.avagotech.com/pages/en/rf_microwave/diodes/pin/hsmp-386j/, Accessed May 2014.
  10. FMD, “GaN Switches Enable Hot Switching at Higher Power,” Microwave Journal, January 2012, pp. 134-136.
  11. . Boles, J. Brogle, D. Hoag and D. Curcio, “AlGaAs PIN Diode Multi-Octave mmW Switches,” 2011 IEEE International Conference on Microwaves, Communications, Antennas and Electronics Systems (COMCAS), 2011, pp. 1-5.
  12. eregrine Semiconductor, “UltraCMOS Process Technology Overview,” Available online at www.psemi.com/content/ultracmos-process/ultracmos-process-tech.php. [Accessed March 2014].

Peter Bacon is director of system integration at Peregrine Semiconductor where he oversees product requirement definition, technology feasibility studies, and application support. Peter has over 30 years of RF product development experience gained at Peregrine, Skyworks, Conexant, IBM, Raytheon, and Harris Microwave. He received his bachelor’s and master’s degrees in electrical engineering from Lehigh University and his master’s degree in business administration from Boston University.

Drew Fischer is an RF/microwave hardware design engineer at National Instruments. He received his bachelor’s and master’s degrees in electrical engineering from California Polytechnic State University San Luis Obispo.

Ruan Lourens is chief analog design engineer at National Instruments, where he designed numerous RF vector signal transceiver and NI FlexRIO adapter modules. Ruan has been granted 11 patents and holds numerous trade secrets in IC design. He obtained his bachelor’s degree in electrical engineering from the University of Pretoria and master’s degree in electrical engineering from NTU School of Engineering.