Radio Design Considerations for EDGE Cell Phones

tmily

Radio Design Considerations for EDGE Cell Phones [Copy link]

EDGE (Enhanced Data Rates for GSM Evolution) is a standard for providing high-speed data services over existing GSM mobile phone networks. As the name implies, EDGE is based on the widely deployed GSM/GPRS networks in North America, Europe and Asia. Many applications are driving the demand for data services over mobile phone networks, including camera phones, online audio, video and text messaging . EDGE's peak data rate can theoretically reach three times that of GPRS, so when designing an EDGE phone, designers must analyze important radio design requirements such as the receiver's AM suppression capability, the transmitter's modulation spectrum, receive band noise, and transmitter current consumption.

EDGE Modulation and Data Rates To provide high data rates, EDGE radios must support a new modulation scheme called 8-PSK in addition to the GMSK modulation used by GSM/GPRS, and they must support this modulation without sacrificing GSM/GPRS performance to ensure backward compatibility. Many different EDGE radio architectures have been proposed to support both GMSK and 8-PSK modulation. Radios using polar loop or polar modulation techniques offer high power-added efficiency, but their calibration requirements and design complexity must be increased to support the power control feedback loop. Other radios add a linear transmitter to the existing GSM/GPRS transceiver to provide the 8-PSK modulation required by EDGE. The advantage of this approach is that it can support 8-PSK modulation while retaining strong GSM/GPRS performance. EDGE Radio Requirements The receiver requirements for GSM/GPRS and EDGE are similar, but the presence of GMSK or 8-PSK modulation blockers complicates the design. Because EDGE supports new modulation schemes on existing GSM/GPRS networks, it introduces new transmitter requirements in addition to the transmitter requirements for GSM/GPRS. Transmitter Requirements The transmitter requirements for mobile EDGE radios are derived from the 3GPP specification and are listed in Table 1. As can be seen from the table, the transmitter requirements for 8-PSK are generally more relaxed, and some are the same as GMSK, such as a 6 dB reduction in output power in the low-band, a 4 dB reduction in the high-band, and a 6 dB reduction in output spectrum requirements for 400 kHz shifts. However, unlike GMSK signals, where the transmitter can compress the signal amplitude, for 8-PSK signals, the transmitter must maintain linear operation during burst transmission in order to preserve the signal amplitude and phase. 　 parameter GMSK 8-PSK Maximum output power * +33 / +30 dBm +27 / +26 dBm Modulation spectrum 400 kHz 600 kHz 1.8 MHz* -60 dBc -60 dBc -63/ -65 dBc -54 dBc -60 dBc -63/ -65 dBc Receive band noise 10 MHz shift ** 20 MHz* -67 dBm -79 / -71 dBm -67 dBm -79 / -71 dBm Modulation accuracy *** RMS value Peak 95th percentile 5 degrees 20 degrees – 9% 30% 15% Origin shift suppression – -30 dBc These requirements are nominal values under normal conditions * Requirements for low band (850, 900) / high band (1800, 1900) ** Requirements for low band (850, 900) only *** Requirements for phase error (GMSK) or EVM (8-PSK) Table 1: EDGE radio receiver requirements 　 Receiver AM Rejection In real-world applications, the target signal is often affected by the noise generated by surrounding mobile phones and other devices. Therefore, according to the requirements of the 3GPP standard, the mobile phone must be able to detect the weak signal to be received in the presence of a large blocking signal under various attenuation conditions representing typical urban or suburban environments. Depending on the receiver architecture, a strong blocking signal may cause nonlinear phenomena, causing the desired signal to be distorted or even completely unreceivable. Figure 1 shows the effect of the blocking signal on the target signal in two receiver architectures. The weak target signal and the strong blocking signal will enter the antenna. If the design adopts a direct conversion receiver, the blocking signal will pass through the switch, SAW and low noise amplifier (LNA), and may leak into the local oscillator of the mixer; if the leakage signal is too strong, the blocking signal may self-mix and generate a DC offset at the output, corrupting the down-converted signal. In a digital low-IF receiver, the mixing process moves the DC offset voltage out of the signal range, which is then removed by the filter. 　 Figure 1: EDGE receiver architecture and AM suppression 　 For direct conversion receivers, it is not easy to eliminate the DC offset at the receiver output because even if the maximum time constant allowed by the bit interval (Tb/2 ~1.8 μs) is used, the resulting filter will still filter out most of the desired signal (fc ~ 88 kHz). Although baseband digital signal processors (DSPs) can use software algorithms to synthesize narrower bandwidth filters on multiple bursts, this approach will result in a loss of about 0.5 dB of the desired signal. In addition, it is not clear how effective the DC algorithm will be in the presence of post-AM blockers, which may be caused by 8-PSK EDGE and W-CDMA, or by blockers appearing in the mid-portion of the burst when performing AM rejection testing. Another consideration is the integration of the synthesizer loop filter. Direct conversion transceivers usually do not integrate them together. In this case, not only does the need for high-precision, low-noise components in the transceiver increase the material consumption of the product, it also creates a coupling mechanism for external noise sources at the board level, causing the phase noise of the local oscillator to become stronger. If the phase noise is too large, the blocking signal may mix with it, causing additional low-frequency distortion and corrupting the received signal. This effect is called reciprocal mixing, which is very common in communication system design. Compared with direct conversion receivers, digital low-IF receivers have better DC bias suppression capabilities, so baseband software no longer needs to provide DC bias calibration functions. Because the synthesizer loop filter and tuning components are integrated together, the reciprocal mixing caused by external phase noise sources will no longer occur. To achieve this more powerful performance, the functional integration of the transceiver must be improved, integrating the dual-channel analog-to-digital converter, digital filtering and dual-channel digital-to-analog converter into the same component. Transmitter Modulation Spectrum and Noise The common transmitter architecture for GSM/GPRS mobile phones is based on an offset phase-locked loop (Offset PLL), which provides strong in-band filtering to meet the stringent requirements of the GMSK spectrum. Out-of-band noise reduction is usually achieved through additional transmitter voltage-controlled oscillator (VCO) phase noise filtering, which allows it to meet the noise requirements of the receive band. However, since the offset phase-locked loop only supports phase modulation and not the amplitude modulation required for 8-PSK modulation, the transmitter architecture must be reconsidered when designing an EDGE radio. Figure 2 shows two possible EDGE transmitter architectures. In a polar loop transmitter, the signal is sent to the power amplifier via different amplitude and phase feedback paths. Polar modulation is a variation of the polar loop that works without feedback circuitry and therefore does not require any coupling circuitry. In a linear up-conversion transmitter, the amplitude and phase of the signal start at the same time. When performing GMSK transmission, the baseband I/Q signal is modulated with a shifted phase-locked loop, and the direct up-conversion mixer and variable gain amplifier (VGA) are bypassed. When performing 8-PSK transmission, the shifted phase-locked loop is not modulated, but acts as a local oscillator in carrier mode to up-convert the baseband I/Q signal. The biggest design challenge for polar loop and polar modulation architectures is to achieve accurate matching of amplitude and phase delays through the feedback loop. Simulation results show that if the delay time mismatch is equal to or greater than 30 ns, the modulation spectrum will not meet the specification requirements at 400 kHz. However, the maximum time constant of 30 ns will limit the filter bandwidth of the transmitter voltage-controlled oscillator to at least 5 MHz, which will limit the phase noise suppression effect of the out-of-band voltage-controlled oscillator. 　 Figure 2: EDGE transmitter architecture and signal path 　 To balance the conflicting requirements of polar architectures, previous approaches have used various PA calibration techniques. However, the amplitude feedback path includes a signal envelope detector that changes with process, supply voltage, and temperature in a different way than the phase discrimination circuit in the phase feedback path. In addition, the PA output phase and amplitude variation patterns are difficult to analyze, making calibration more difficult; the loop may become unstable when the PA gain is large and the output signal is small (i.e., ramping or circuit mismatch). Compared to polar loop or polar modulation architectures, which have not yet been verified in actual products, linear transmitters are the only architecture that has begun to be used in GSM/EDGE mobile phones. CDMA and W-CDMA mobile phones also use this architecture. Linear transmitter architecture allows designers to choose PAs from different manufacturers, which is completely different from polar modulation, which requires special or custom PAs. Highly Integrated EDGE Radios Radios based on highly integrated components offer many advantages to mobile phone designers. For example, compared to discrete solutions, highly integrated solutions require fewer components to purchase and stock, and they provide greater performance because they are free of interference or noise sources external to the components. Engineers also no longer need to perform component-level testing, which ensures higher phone manufacturing yields and fewer product returns. Figure 3 shows one way to design a highly integrated EDGE radio using the Silicon Laboratories Aero? II transceiver. The Aero II transceiver offers the industry's highest functional integration for GSM/GPRS mobile phones. All noise-sensitive circuits are integrated into the chip, including the RF and IF VCOs, VCO tuning components, all loop filters, and all digitally controlled quartz oscillators (DCXOs). They are all integrated into a 5 × 5 cm package. Its companion chip supports linear up-conversion for 8-PSK signaling. 　 Figure 3: EDGE radio using Silicon Laboratories Aero II transceiver in conclusion After years of development, EDGE technology is now widely adopted by GSM/GPRS networks around the world. In order to support EDGE mobile phones, radios based on digital low-IF receivers and linear transmitters also promise to provide strong receiver AM rejection and transmitter performance, as well as transmitter currents similar to other architectures. As EDGE technology gradually becomes mainstream in the market, the increasing functional integration is expected to significantly reduce the number of parts and materials in mobile phone radios to the level of GSM/GPRS mobile phones, which will open a smooth path for highly integrated 3G radios. (The author is the wireless product marketing manager of Silicon Laboratories)