Yorgos Palaskas

A 64 GHz LNA With 15.5 dB Gain and 6.5 dB NF in 90 nm CMOS

This paper presents an integrated LNA for millimeter-wave applications implemented in 90 nm CMOS technology. Modeling methodology based solely on electromagnetic simulations, RC parasitic extraction and device measurements up to 20 GHz allows for ldquocorrect-by-constructionrdquo design at mm-wave frequencies and first-pass silicon success. The dual-stage cascode LNA has a peak gain of 15.5 dB at 64 GHz with a NF of 6.5 dB, while drawing 26mA per stage from 1.65 V. Output is 3.8 dBm. At , each stage draws 19 mA, with a peak gain and a NF of 13.5 dB and 6.7 dB, respectively. Measured results are in excellent agreement with simulations, proving the effectiveness of the proposed design methodology. A custom set-up for mm-wave NF measurement is also extensively described in the paper.

A 4-GS/s 4-bit Flash ADC in 0.18-

A 4-bit noninterleaved flash ADC implemented in 0.18-mum digital CMOS achieves a sampling rate of 4 GS/s. A 32 mum by 32 mum, on-chip differential inductor in each comparator extends the sampling rate without an increase in power consumption. A combination of DAC trimming and comparator redundancy reduces the measured DNL and INL to less than 0.15 LSB and 0.24 LSB, respectively. The measured ENOB with a 100 MHz full-power input is 3.84 bits and 3.48 bits, at 3 GS/s and 4GS/s, respectively. The ADC achieves a bit error rate of less than 10-11 at 4 GS/s.

\mu{\hbox {m}}

A class-E power amplifier (PA) utilizes differential switches and a tuned passive output network improves power-added efficiency (PAE) and insensitivity to amplitude variations at its input. A modulator is introduced that takes outphased waveforms as its inputs and generates a pulse-width and pulse-position modulated (PWPM) signal as its output. The PWPM modulator is used in conjunction with a class-E PA to efficiently amplify constant envelope (e.g., GMSK) and non-constant envelope (e.g., QPSK, QAM, OFDM) signals with moderate peak-to-average ratios (PAR). The measured maximum output power of the PA is 28.6 dBm with a PAE of 28.5%, and the measured error vector magnitude (EVM) is 1.2% and 4.6% for GMSK and pi/4-DQPSK (PAR ap 4 dB) modulated signals, respectively.

CMOS

A 2.4 GHz outphasing power amplifier (PA) is implemented in a 32 nm CMOS process. An inverter-based class-D PA topology is utilized to obtain low output impedance and good linearity in the outphasing system. MOS switch non-idealities, such as finite on-resistance and finite rise and fall times are analyzed for their impact on outphasing linearity and efficiency. Outphasing combining is performed via a transformer configured to achieve reduced loss at power backoff. The fabricated class-D outphasing PA delivers 25.3 dBm peak CW power with 35% total system Power Added Efficiency (includes all drivers). Average OFDM power is 19.6 dBm with efficiency 21.8% when transmitting WiFi signals with no linearization required. The PA is packaged in a flip-chip BGA package. Good linearity performance (ACPR and EVM) demonstrates the applicability of inverter-based class-D amplifiers for outphasing configurations.

A Class-E PA With Pulse-Width and Pulse-Position Modulation in 65 nm CMOS

A mm-wave power-harvesting RFID tag is implemented in 90 nm CMOS. Operation at mm-wave reduces antenna size and could allow antenna integration on-chip. This, together with power harvesting that can be used in lieu of a battery, can result in a pinless, CMOS-only tag with no package and no off-chip components whatsoever. The tag harvests energy from the incoming mm-wave continuous wave (CW) signal transmitted by the reader and then uses a 60 GHz free-running oscillator to transmit back pulse-width modulated bursts. An in-depth treatment of the voltage multiplier and associated matching network and implications on tag range are presented. With 2 dBm mm-wave input power, the tag achieves a rate of 5 kb/s. The RFIC size is 1.3 × 0.95 mm2 including pads.

A Flip-Chip-Packaged 25.3 dBm Class-D Outphasing Power Amplifier in 32 nm CMOS for WLAN Application

A digital outphasing transmitter is presented for 2.4-GHz WiFi. The transmitter consists of two delay-based phase modulators and a 26-dBm integrated switching class-D power amplifier. The delay-based phase modulator delays incoming LO edges with a resolution of 1.4 ps (8 bit) required to meet WiFi requirements. A phase MUX architecture is proposed to implement switching between phases once every LO period (2.4 GHz) without generating detrimental glitches at the output. Due to its open-loop nature, the proposed phase modulator is capable of delivering wide OFDM bandwidths up to 40 MHz. The paper analyzes the impact of impairments, e.g., delay mismatch within the delay cells and outphasing mismatches, as well as associated mitigation techniques. The transmitter has been implemented in a 32-nm digital CMOS process and delivers an OFDM average power of 20 dBm with an overall system efficiency of 18.6% when transmitting 54-Mb/s 64QAM signal. The fully digital design is expected to further improve in power dissipation and chip-area with further CMOS scaling.

A mm-Wave Power-Harvesting RFID Tag in 90 nm CMOS

This paper presents an integrated CMOS power amplifier and a technique for correcting AM-PM distortion in power amplifiers. The linearization technique uses a varactor as part of a tuned circuit to introduce a phase shift that counteracts the AM-PM distortion of the power amplifier. The varactor is controlled by the amplitude of the IQ baseband data in a feedforward fashion. The technique has been demonstrated in a 5-GHz class-AB CMOS power amplifier designed for WLAN applications and implemented in a 90-nm CMOS process. The power amplifier delivers 16 dBm of average power while transmitting at 54 Mb/s (64 QAM). The proposed linearization technique is shown to improve the efficiency of the power amplifier by a factor of 2.8

A 2.4-GHz 20–40-MHz Channel WLAN Digital Outphasing Transmitter Utilizing a Delay-Based Wideband Phase Modulator in 32-nm CMOS

A 5-bit flash ADC incorporates 20 mum by 20 mum inductors to improve both comparator preamplification bandwidth and regeneration speed. A switched-cascode scheme reduces comparator kickback. Offset cancellation is achieved by modifying the comparator reference voltages without degrading high-speed performance. The ADC achieves a measured SNDR of 27.5 dB for a 5 MHz input at 4 GS/s, and 23.6 dB for a 1 GHz input at 3.5 GS/s. The power consumption (including clock buffer and ladder) is 227 mW at 3.5 GS/s. The active area is 0.658 mm2

A 5-GHz 20-dBm Power Amplifier With Digitally Assisted AM-PM Correction in a 90-nm CMOS Process

A transformer-combined fully integrated outphasing class-D PA in 45 nm LP CMOS achieves 31.5 dBm peak output power at 2.4 GHz with 27% peak PAE, and supports over 86 dB of output power range. The PA employs dynamic power control (DPC) whereby sections of the PA are turned on or off dynamically according to the instantaneous signal amplitude to reduce power dissipation, especially at back-off. Dynamic on-off switching introduces transients on the power supply that can limit performance. The paper proposes and demonstrates techniques to reduce the impact of such transients. A multi-section slab inductor based transformer combiner is used to allow individual switching of unit PAs. The PA delivers 24.8 dBm average power while meeting 64-QAM WLAN requirements. PAE is 16% when using DPC, which represents a 33% efficiency enhancement compared to the DPC-disabled mode. At lower average power of 20.5 dBm, DPC enables a 140% enhancement in average efficiency, hence increasing battery life.

A 3.5 GS/s 5-b Flash ADC in 90 nm CMOS

This paper presents a fractional frequency divider-by-1.25 and associated all-digital calibration circuitry. The divider can be used in a wireless transceiver to prevent direct or harmonic pulling of the VCO by the power amplifier. Timing errors between the quadrature phases used in the phase-rotating divider introduce fractional spurs at the output. In this design, the timing errors are measured with a stochastic time-to-digital converter with 20 fs resolution, and corrected to suppress output spurs. The fractional divider has been implemented in a 45 nm CMOS LP process and its core dissipates an estimated 17 mA current from a 1.1 V supply. After calibration, fractional spurs are on average below -59 dBc and -50 dBc (¿ ~ 2 dB over 10 samples) with a 2.5 and 3.8 GHz output frequency respectively. Calibration performance has been confirmed for temperatures from -20°C up to 85°C. The low spur level facilitates radio co-existence with no need for additional filtering. This makes this divider a good candidate for WiFi and WiMAX radios up to 3.8 GHz.