RF CMOS Switches and Filters is the radio frequency switch and filter technology integrated with CMOS for monolithic RF front-end modules โ critical for 5G/mmWave communication enabling compact transceivers with reduced external components.
RF CMOS Switch Architecture:
- Series switch: MOSFET in series with signal path; source drain connected to RF signal line
- Shunt switch: MOSFET connected to ground; allows bypassing signal when activated
- Switch stack: series series MOSFETs for high voltage capability; parallel-series combinations for improved characteristics
- Isolation: off-state isolation >30 dB typical; frequency-dependent; decreases at higher frequencies
- Insertion loss: on-state loss ~0.5-1 dB; loss increases with frequency (resistive loss increases)
RF Switch Figure of Merit (FOM):
- Definition: FOM = f ยท Ron ยท Coff; frequency ร on-resistance ร off-capacitance; tradeoff metric
- Physical interpretation: captures fundamental tradeoff between switch characteristics; lower FOM better
- Frequency scaling: FOM proportional to frequency; higher frequency applications more challenging
- Design tradeoff: reducing Ron increases Coff; reducing Coff increases Ron; optimal design required
SOI CMOS RF Switch:
- Silicon-on-insulator process: thin Si layer on oxide on substrate; eliminates parasitic substrate capacitance
- Parasitic reduction: buried oxide removes substrate coupling; improves isolation and insertion loss
- High-impedance substrate: buried oxide isolates switches from conductive substrate; reduces capacitive coupling
- Scalability: smaller transistor dimensions in advanced CMOS; improved FOM scaling with technology node
- Cost consideration: SOI wafers expensive; justified for performance-critical applications
Bulk Acoustic Wave (BAW) Filters:
- Resonator structure: thin piezoelectric layer (AlN typically) sandwiched between electrodes; thickness determines resonance
- Fundamental mode: mechanical vibration at fundamental frequency determined by thickness resonance condition
- Quality factor Q: high Q (~1000-2000) enables sharp filtering; low insertion loss and sharp passband
- Temperature compensation: temperature coefficient of frequency (TCF) controlled via material composition; stable operation
- Bandwidth: narrow-band filters typical; center frequency and bandwidth set by resonator dimensions
FBAR (Film Bulk Acoustic Resonator):
- Suspended membrane: thin piezoelectric film with electrodes; suspended over cavity or backside etched
- Free boundary conditions: air gap provides acoustic isolation; enables high Q
- Frequency tuning: film thickness determines frequency; very thin (<2 ฮผm) for multi-GHz operation
- Power handling: limited by mechanical stress and piezoelectric breakdown; typically <500 mW
- Manufacturing: requires backside etching or release process; challenging integration with CMOS
RF Front-End Module Integration:
- Transceiver path: transmit path (PA), filter, switch, LNA, receive path integrated monolithically
- PA output: high power limits integration with low-power CMOS; often external or separated in module
- LNA noise figure: critical for receiver sensitivity; high-gain, low-noise requirement
- Switching control: on-chip logic controls transmit/receive paths; eliminates manual switching
- Power consumption: integrated front-end reduces external components and parasitic losses
Co-Integration Challenges:
- Power levels: PA operates at high power (~1-10 W); CMOS transistors limited to lower power
- Thermal management: power dissipation in PA; heat spreads to sensitive analog circuits; thermal isolation needed
- Impedance matching: 50 ฮฉ impedance standard in RF; on-chip impedances higher; matching networks required
- Crosstalk: transmit power couples to receive path; isolation structures (guard rings, shields) prevent degradation
- Substrate coupling: noisy digital circuits affect sensitive analog RF; physical/electrical isolation critical
Insertion Loss and Isolation Characteristics:
- Frequency dependence: insertion loss increases with frequency (skin effect); R_on dominates at higher f
- Bandwidth limitations: switches low-pass characteristics; insertion loss increases above certain frequency
- Isolation improvement: multiple switch stages improve isolation; cascade degradation factor important
- Quality factor (Q): reactive elements improve selectivity; L-match networks provide impedance transformation
- Dynamic behavior: switch transient response; settling time affects switching speed
Switch Stack Design for High Voltage:
- Voltage scaling: series transistors share voltage; each transistor sustains V_dd/N voltage
- Transistor sizing: width/length ratio adjusted for equal voltage distribution; body effect considered
- Body biasing: substrate/well biasing controls threshold voltage; improves voltage distribution
- Breakdown consideration: gate oxide breakdown (V_ox,max ~2-3 MV/cm); limits operating voltage
5G mmWave Applications:
- Frequency range: 28/39/73 GHz bands; higher frequencies enable compact antennas and wider bandwidth
- Integration necessity: external components impractical at mmWave; monolithic integration essential
- Beam steering: phased array antennas require RF switches for beam control; phase shifters and attenuators
- Power efficiency: low insertion loss critical for battery-powered devices; integration reduces parasitic losses
- Module density: higher integration density enables compact transceivers; reduced printed circuit board area
RF CMOS switches and BAW filters provide monolithic RF front-end integration โ enabling compact 5G/mmWave transceivers with minimal external components through advanced process technologies.