by Manish Shah, Vice President of Engineering, Tagore Technology
The deployment of 5G base stations has been ramping up in the last few years, with carriers working with base station equipment suppliers to meet the demand for increasing network capacity and higher data rates. These deployments are targeted between 2 to 5 GHz. Most of the deployment is targeted for the n77, n78, and n79 bands, where a much bigger bandwidth spectrum is allocated for 5G cellular services. The operators and base station equipment providers are confronted with different architecture choices in terms of spectrum efficiency, total power consumption, complexity, total size, and weight of implementation.
A massive MIMO (mMIMO) or an Active Antenna System (AAS) is the architecture of choice for many of the 5G base station remote radio head (RRH) designs. This architecture groups together many antennas to provide high spectral efficiency. This system requires many RF front ends (RFFEs) to be implemented in limited board space. RRH units are typically mounted on poles, which adds additional constraints in terms of size and weight to ease the installation and maintenance. The RFFE total loss budget is also critical to managing total thermal dissipation and improving receiver sensitivity for higher throughput. These constraints necessitate the right technology choice, which provides the best overall solution in terms of size, weight, power, and cost (SWaP-C).
Figure 1 shows the configuration of a typical 5G sub-6 GHz mMIMO radio unit in which active antennas work together to create multiple beams to improve spectral efficiency. Each transmit and receive line-up consists of a transceiver feeding a driver and Doherty final amplifier in the transmitter and a fail-safe switch and LNA gain stages feeding the receive chain.
The dedicated transmit and receive chain is the most flexible in terms of implementation, as it allows most of the signal processing to be performed in the digital domain. However, it is not the most efficient in terms of power, size, and cost efficiency, as it requires a dedicated line-up, especially if each line-up has low power RF output. This is because the power consumed by the baseband ADC and DAC do not scale with output power but depend on bandwidth requirements. Thus, it is critical that each line-up is designed for maximum RF output power to achieve the desired EIRP to meet throughput requirements in the coverage area of the base station.
Requirements for 5G mMIMO RFFE
Figure 2 depicts the typical implementation of an RRH RFFE. On the transmitter side, the signal from the transceiver is amplified by several stages of pre-driver and drivers before it is fed into the Doherty’s final amplifier. Doherty architecture is used to achieve higher back-off efficiency and meet linearity requirements with the help of digital pre-distortion (DPD) algorithms. The receive path includes the high-power fail-safe switch followed by two stages of LNAs. The second stage of the LNA has a bypass mode to improve the dynamic range of the receiver. The function of the fail-safe switch is to protect the receiver in the event of a high VSWR condition caused by damage or a transmission line discontinuity. In such cases, the transmitted power is reflected to the radio.
To protect the receiver from high reflected power, a fail-safe switch is included in the receiver line-up that switches to the receive port during the receive time slot and to the transmit port during the transmit time slot. If a high VSWR occurs during transmission, the fail-safe switch routes the reflected power from the antenna to a 50 ohm load through a circulator at the transmit port, preventing damage to the receiver. For reliable fail-safe operation, the RF switch must have low insertion loss in the receive path to maximize receiver sensitivity.
It should also handle maximum transmit power and provide high isolation from the receive port to protect the receiver from damage. The switch should withstand high power levels until the system detects fault conditions. High VSWR is detected through a coupler, as shown in Figure 2. A typical system implementation can detect fault conditions within 10 s of the fault, which means that the fail-safe switch must handle the full transmit power for this time period. A fail-safe switch is also required to have isolation of 25 to 35 dB in the transmit mode, the value of which depends on the transmitter’s maximum peak power and the LNA’s power handling capability.
For example, if the LNA can handle a peak power of 25 dBm and the maximum transmit power from the DPA is 600 W to achieve 60 W of average power with a 10 dB PAPR, isolation of at least 33 dB is needed to protect the receiver chain. The system typically requires a switching speed of 1 us to meet overall TDD timing. Two stages of LNAs are necessary before the transceiver to achieve a low receive noise figure.
The requirements for the LNA in massive MIMO base station receiver front-end applications are as follows:
- High gain to provide high gain to amplify weak received signals without introducing significant noise and distortion
- Low noise figure, which is a measure of the noise added by the amplifier to the input signal. A low noise figure is crucial to maintain the system’s sensitivity and achieve a high Signal-to-Noise Ratio (SNR).
- High linearity reduces inter-modulation distortion and spurious emissions, which is important in an mMIMO system where there are multiple antennas and signals being transmitted and received simultaneously
- Wideband operation is necessary to cover multiple frequency bands and support different wireless standards, which is essential in an mMIMO system
- Low power consumption reduces the overall power consumption of the base station and increases energy efficiency
- Small size is increasingly important, especially in smaller base stations, so integrating multiple LNAs on a single chip or module is highly desirable
Gallium Nitride (GaN) technology is advantageous in designing high-performance RF switches that offer high power handling capability, low insertion loss, and high linearity. There are two requirements for RF devices used in high power RF switches. The “on” arm of the switch must handle very high RF current, while the “off” arm devices need to handle very high voltages. For example, 100 W of RF power in a 50 ohm system produces 100 VDC peak voltage and peak current of 2 A.
There are two other switching technologies: SOI and pin diodes. An SOI-based switch device has a very low breakdown voltage, in the range of 3 VDC, so it’s not very well suited for high power switches. In addition, SOI-based switches require the stacking of many FETs to realize higher power handling, which makes them less rugged. Pin diode-based high power switches also require a high blocking voltage to keep the diode off and a high bias current to lower the on-resistance of the diode. It also requires many passive components for biasing purposes, which require a significant PC board area that makes it difficult to integrate with an LNA.
Space and power consumption are at a premium for mMIMO architectures as there are many RFFEs that must fit in small spaces, so the pin diode-based solution is not ideal. Figure 3 compares a GaN-based fail-safe switch to a pin diode-based switch implementation with 100 W average power handling and an 8 dB PAPR in fail-safe operation. The 32×32 mMIMO system would have more than 27 W of power dissipation in a pin-diode-based solution versus 19 mW for a GaN-based solution, and the pin-diode-based solution would require almost five times more components and 10 times more board area.
GaAs pHEMTs have been a technology of choice for LNA design. This wide band gap, high electron mobility technology offers high gain, low noise, and high linearity performance needed for a high-performance LNA. The dielectric constant and semi-insulating properties of GaAs are also well suited for achieving the high Q needed for amplifier input and output matching circuits. Most GaAs foundries also offer MIM capacitors, resistors, and pHEMT devices to realize small logic, bias, and low power switching, which is necessary for LNA with a bypass function. Overall, the combination of GaAs material properties, low-noise transistors, and optimized matching networks allow GaAs LNAs to achieve low noise figures, high linearity, and high performance in a compact and cost-effective design.
An Integrated Fail-safe Switch/LNA Module
GaN and GaAs technologies can be integrated into a single package while maintaining their key features. This combination allows the design of GaAs LNA and GaN HEMT RF switches to be realized in a single package without sacrificing their electrical performance. Figure 4 shows a 20 W average power, 8 dB PAPR, dual-channel fail-safe switch/LNA module in a single package.
Figure 5 shows 100 W average power, 8 dB PAPR single-channel fail-safe switch/LNA in a 5 x 5-mm package. Both devices harness the benefit of GaN technology to realize the switch function and GaAs technology to realize the LNA function. This is an attractive solution when amplifying received signals with low distortion, maximizing receiver sensitivity, and adding extra fail-safe protection. GaN enables a much smaller size, low power, and rugged switch function, and GaAs enables very low noise figure, high gain, and dual-stage amplifier function.
For small-signal performance, the figure of merit for a switch is the insertion loss. To achieve low insertion loss, the fail-safe RF switch stack FETs should have a very small on-resistance (Ron), and the switch must also be designed to provide the best isolation in the off path so the on path is not affected. The fail-safe RF switch in Tagore’s TSL8329M and TSL8029M is designed so that the receive on-state has minimal impact on the overall noise figure. For large signals, in case of a failure, the fail-safe RF switch must handle the high peak voltage swing from the high peak-to-average power ratio (PAPR) signals during the transmission mode.
It also must be able to manage high power to guarantee that all the power is delivered to the 50 ohm load in Figure 2. The receive off-state has sufficient isolation so that the maximum power seen by the first stage of the LNA is below its limit. Both designs can be tuned for different cellular bands from 2 to 5 GHz. Figure 6 shows the peak power handling of TSL8329M of 100 W, and Figure 7 shows the peak power handling of TSL8029M, which has a peak power handling ability of 600 W.
The LNA operates in three modes: bypass/low-gain mode, high-gain mode, and power-down mode. In bypass mode, one LNA is bypassed using the switch, and in bypass/low-gain mode, the second stage is bypassed using switches. In the high-gain mode, both amplifiers are cascaded to provide high gain. The power-down mode is designed to pull down all active devices, resulting in low power consumption and very low shutdown current. The transmit GaN switch and internal dies are arranged so that the device can deliver very good channel-to-channel isolation.
For dual front-end receivers in mMIMO systems, the required minimum channel-to-channel isolation depends on the modulation scheme and the system design. The TSL8329M offers 40 dB of channel-to-channel isolation in receive mode and 55 dB of isolation in transmit mode. The board design of the TSL8329M is optimized for improved isolation and reduced coupling between channels.
Figures 8 and 9 show the noise figure performance of TSL8329M and TSL8029M, respectively. The TSL8329M achieves a noise figure below 1 dB, whereas the TS8029N achieves a noise figure of 1.2 dB. The TSL8329M has 32 dB of gain in high-gain mode and an OIP3 of 35 dBm at 3.6 GHz when tuned for the n77 and n78 bands. The TSL8029M has 34 dB of gain in high-gain mode and an OIIP3 of 32 dBm when tuned for the n40 and n41 bands. Figure 10 shows the total size of the TSL8029M implementation, which only requires 5 VDC.
The high-power fail-safe switch and LNA module harness the unique properties of GaN and GaAs technology to realize high power switch and LNA functions, respectively, in a small package. The design eliminates the complexity of pin diode-based fail-safe switch solutions, and further advancements in packaging technology should allow the integration of all functions of the RFFE, Doherty power amplifier, switch and LNA in a single package as well.