RF Design Magazine
About RF Design divider For Advertisers divider Contact Us divider Subscribe to RF Design divider HOME
RSS    Save to Del.icio.us  Digg This


Direct modulation radio hardware architectures for 3G communications systems
Feb 1, 2005 12:00 PM  By Patrick Naraine

The base station transceiver system (BTS) is one of the most expensive network elements in the wireless network infrastructure and directly impacts the cost of the overall network design and deployment. The requirement for increased capacity and higher data rates will inevitably lead to increased cell deployments. To maintain costs at an acceptable level, telecommunications equipment manufacturers (TEMs) will continue to seek more cost-effective solutions to their infrastructure hardware.
 
Resources
Spotlight on Automotive Wireless Connectivity

For the PDF version of this article, click here.

The cost of the BTS is affected by its key functional elements that include antenna, radio transceiver, signal processing subsystems, and support and control hardware and software. An example of a BTS configuration is shown in Figure 1. The BTS uses main branch modules and diversity branch modules. Diversity receiver systems are common in 2G systems, and the 3G specifications[1] also define options for transmitter diversity. As shown in Figure 1, the transmit/receive module (TRM) will contain duplicate transmit and receive sections. For wide area cell sites the receivers will generally be connected to the antennas via a duplexer module (DM) and tower top assembly (TTA). The TTA consists of a low noise amplifier (LNA), which is essential to compensate for cable losses from the antenna to the TRM. The TTA may be equipped with bypass switches (for low noise amplifier protection) and pre-selector filter. On the transmit side, driver power amplifiers (PAs) within the TRM will generally feed high-power amplifiers (HPAs) before passing through the duplexer and antenna.

This article focuses on the receiver and transmitter modules (typically four modules per TRM per sector) and discusses methods of reducing their cost without compromising system performance. The article will focus on the direct conversion receiver (DCR) and direct conversion transmitter (DCT) architecture, since these architectures provide significant costs and size advantages without sacrificing performance. Compared to handset designers, base station designers have been slow in adapting DCR and DCT technologies, but increased pressures for costs and size saving, and continued improved performance will help to accelerate the conversion of base station design.

Table 1. Examples of ACLR1 budget for a WCDMA direct conversion transmitter.
1 -65 -60 -53 -46 -45 Less demands on baseband modulator and driver PA linearity. However, HPA needs more backing off from P1 db point, and could lead to lower power efficiency.
2 -69 -63 -55 -45.5 -45 Medium linearity demands on the modulator and driver PA. Good compromise solution.
3 -77 -72 -60 -45.1 -45 Places more stringent linearity demands on the baseband, modulator and driver PA.

Third-generation systems may use enhanced data for GSM Evolution (EDGE), third-generation code division multiple access (CDMA2000) or wideband code division multiple access (WCDMA) modulation standards. Examples of WCDMA base stations will be given, however, similar analysis could be adopted for the other two 3G systems.

3G direct conversion receiver

The DCR has become increasingly acceptable for applications requiring multiband flexible operation, with size, power and cost constraints. The DCR is able to meet these stringent requirements since it requires fewer components (mixer, filter and amplifiers) leading to lower power, size and cost. The DCR converts the RF signal directly to baseband, eliminating the need for various mixers, image-rejection filters and amplifiers. Figure 2 shows an example of a DCR, which could be adapted for use in 3G base stations. The first element in the receiver chain is the low noise amplifier (LNA). The main function of the LNA is to minimize the overall receiver noise figure. A switched LNA is proposed to handle large input signal dynamic range without driving the rest of the receiver chain into compression. An RF bandpass filter (BPF) follows the LNA. The BPF provides attenuation to any out-of-band interfering signals, and spurious signals generated by the receiver. The filtered input signal is level controlled using a variable gain amplifier (VGA) or a variable attenuator and amplifier combination. The signal is then sent to a demodulator RF integrated circuit (RFIC) where it is demodulated into in-phase (I) and quadrature-phase (Q) baseband signals. The demodulator signal lines are implemented as differential pairs to provide high immunity to noise. I and Q signals from the demodulator are then low-pass filtered (LPF) and amplified by a low-frequency automatic gain control (AGC). The low frequency AGC is critical to maintaining I and Q signal levels within the input operating range of the analog-to-digital converter (ADC).

The receiver (Figure 2) provides a relatively simple architecture with low parts counts. However, the main design challenge, which must be addressed for any DCR, is direct current (dc) offset[2].

The mixing of a local oscillator (LO) signal, which has leaked or coupled to the RF input, with the incoming RF signal, can generate significant dc voltage levels within the receiver. Dc offset can be minimized by maximizing the LO to RF isolation and through harmonic mixers[3]. It is critical to closely match the amplitude and phase response of mixers and amplifiers in the demodulator I-Q paths. A silicon bipolar complementary metal oxide semiconductor (Si-BiCMOS) demodulator, SKY73009 has been designed and tested to achieve amplitude and phase balance within 0.3 dB and 1°, respectively. This demodulator also provides a superior LO-to-RF isolation of 50 dB, essential to maintaining low dc offset signals within the receiver. Systems using lower-performance demodulators, will require more sophisticated and expensive dc-calibration techniques[4].

The DCR shown in Figure 3 uses a switched LNA (part of a front-end receiver RFIC) and the Si-BiCMOS demodulator RFIC described above. With these key components, the RF analysis shows a resulting cascaded noise figure of 3.2 dB and cascaded gain of 70 dB. For most 3G base station implementations (depending on noise figure of TTA and cable losses), a cascaded noise figure of less than 5 dB is acceptable, so the receiver architecture proposed above can provide an additional system margin of 1.9 dB.

3G direct conversion transmitter

Base station transmit diversity is allowed in the 3GPP specifications[1]. With proper alignment (to reduce cross-coupling and to increase branch isolation) and timing synchronization, transmitter diversity can provide improvements to the system signal to noise ratio (SNR) and mitigate multipath and signal fading effects. However, diversity will add to systems implementation cost, size and power consumption. The use of low-cost RFIC components and direct conversion architectures will help mitigate these negative impacts of diversity transmitters.

The direct upconversion transmit architecture shown in Figure 4 is becoming more common due to its low part count and power consumption.

With direct conversion architectures, DACs provide the transmit I-Q signals. I-Q signals from the DACs can be low-pass filtered to remove any aliasing, harmonics and spurs introduced by the DACs. The filtered I-Q signals can then be directly modulated and upconverted to RF frequency using a direct quadrature modulator. The modulator must have adequate amplitude and phase matching between the I-Q branches to minimize corruption of the modulated signal information or error vector magnitude (EVM). The total allowed EVM for a WCDMA base station transmitter is 17.5% using quadrature phase shift keying (QPSK) modulated signals. Systems engineer will allocate most of the EVM systems budget to the HPAs (to improve power efficiency). Therefore, little or no EVM contribution will be expected from the modulator. Direct quadrature modulators are available with amplitude and phase imbalance less than 0.3 dB and 3° respectively, and should contribute less than 5% to the EVM budget.

Signal leakage from LO to RF port needs to be minimized, since in most direct conversion systems the LO and RF input signals are at the same frequency and RF filtering after the modulator will be ineffective in suppressing any LO-RF leakage. Today, direct quadrature modulator RFICs can be obtained with LO to RF isolation of greater than 50 dBc, second-order input intercept point (IIP2) of greater than 60 dBm and NF level of less than -153 dBm/Hz. These RF performance values ensure the modulator will not significantly impact the spurious emissions of the overall transmitter chain.

The upconverted signal can be level controlled to compensate for part-to-part, diversity-to-main branch and temperature gain variations. The variable attenuator will need to handle input power levels of -10 dBm to +10 dBm (typical output levels from modulator), without contributing to the system non-linearity. The AA102-80 variable attenuator will operate in most common wireless bands (0.5 GHz to 2.5 GHz) and has an input third-order product intercept point (IP3) of more than +45 dBm. The settable attenuation range is more than 30 dB with 1 dB step size.

After level control the transmit signal may then be amplified by a linear driver before being fed to the final high-power amplifier (HPA). The driver needs to have sufficient high gain (typically 20 dB to 35 dB) and linearity to suit the overall system requirements. With higher gain driver amplifiers, the number of required stages on the HPA can be reduced, resulting in significant cost and power efficiency savings. High-performance linear drivers are available with RF gain of more than 25 dB and output third-order product intercept point (OIP3) of more than +40 dBm.

A key requirement for any WCDMA transmitter is the adjacent-channel leakage power ratio (ACLR). The 3G specifications[1] call for ACLR1 (one channel or 5 MHz frequency offset) of less than -45 dBc and ACLR2 (two channels or 10 MHz offset) of less than -50 dBc. ACLR measurements are typically conducted with the defined test model 1, which comprises 64 dedicated physical channel (DPCH) signals at 30 ksps and spreading factor of 128. The power levels and timings for the 64 DPCH signals are randomly distributed to simulate a realistic signal environment. The test signal power level timings are defined in the 3G spec document[1].

The ACLR budget must be carefully distributed among the various non-linear sections of the transmitter chain. Examples of how an ACLR budget could be distributed amongst the main transmitter sections are shown in Table 1.

The typical ACLR performance of a high gain linear driver amplifier is shown in Figure 5.

3G transmitted signals will have peak-to-average ratio (PAR) in excess of 10 dB when measured at the 0.01% cumulative complementary distribution (CCD) point. The high PAR of the 3G signals place stringent linearity demands on the power amplifiers. The ACLR Figure 5 shows the amplifier will typically require 8 dB back off from its P1dB point to achieve an ACLR1 of -45 dBc. The ACLR vs. output power response is linear with a slope of approximately 3:1 between -45 dBc and -55 dBc. However, at lower ACLR points the noise floor of the amplifier becomes important and limits the ACLR rejection level. For ACLR budget examples shown in Table 1, the linear driver can provide output power of 22 dBm, 20.5 dBm and 11 dBm for examples 1, 2 and 3 respectively. Example 3 is the least preferred configuration since it places stringent demands on baseband components, modulator and driver PA, and will lower the available output power from these sections, and in turn place a higher demand on the gain requirements of the HPA.

Figure 6 shows an example of a transmitter chain block and level, which could be used to achieve total ACLR1 of greater than the standard specification of -45 dBc[1] with linear output power of +41 dBm. The transmitter ACLR1 budget closely reflects the allocation shown for Example 2 in Table 1.

REFERENCES

  1. 3rd Generation Partnership Project; Technical Spec. Group Radio Access Network; Base Station radio transmission and reception (FDD) (Release 6).

  2. A. Loke et al., “Direct Conversion Radio for digital Mobile Phones-Design Issues, Status and Trends,” IEEE Transactions on Microwave Theory and Techniques, Vol. 50, No. 11, November 2002.

  3. K. Itoh, et al., “Even Harmonic Type Direct Conversion Receiver ICs for Mobile Handsets: Design Challenges and Solutions,” 1999 IEEE Radio Frequency Integrated Circuits Symposium.

  4. C. Holenstein et al., “Adaptive Dual-loop Algorithm for Cancellation of Time-Varying Offsets in Direct Conversion Mixers”, IEEE Radio and Wireless Conference, RAWCON 2000.

ABOUT THE AUTHOR

Patrick Naraine holds a BSEE and MSEE from the McMaster University of Hamilton, Ontario, Canada. He was technical payload manager of Canada's first remote sensing satellite built for the Canadian Space Agency in 1996. He then joined Nortel Networks where he was in charge of RF system designs for AMPS, TDMA and EDGE base station transceivers. From 1999 to 2002 he worked at AT&T Wireless on the system design, testing and deployment of the first U.S. commercial fixed wireless voice and high-speed Internet system. Now he works on system designs for RFICs, MMICs and PAMs at Skyworks Solutions Inc. -Contact him at patrick.naraine@skyworksinc.com.


RSS    Save to Del.icio.us  Digg This

February Defense
 
Back to Top


Contact Us  For Advertisers  For Search Partners  Privacy Policy  Subscribe
© 2012 Penton Media, Inc.

popular searches: zigbee | quadrature modulation | OFDM | WiMAX