CDMA RELATED CIRCUIT DESIGN, SIMULATION, MEASUREMENT AND VERIFICATION

Transcription

1 CDMA RELATED CIRCUIT DESIGN, SIMULATION, MEASUREMENT AND VERIFICATION R. MAHMOUDI, P. VALK, J. L. TAURITZ. MICROWAVE COMPONENTS GROUP, DELFT INSTITUTE FOR MICROELECTRONICS AND SUBMICRON TECHNOLOGY (DIMES), LABORATORY ECTM, DELFT UNIVERSITY OF TECHNOLOGY, FELDMANNWEG 17, 2628 CT DELFT, THE NETHERLANDS PHONE , FAX: , Abstract The introduction of the second generation of mobile communication systems and the attendant demands on cost, efficiency and system constraints have altered and complicated design procedures. Complex digital communications signals with relatively large bandwidth and crest factor have rendered inadequate traditional design methods based on small/large signal S-parameters particularly for transmitter power amplifiers. Furthermore, the increasing use of Silicon-based integrated circuits (see [1]) has emphasized the importance of physically based compact transistor models when trading off system demands. To verify the adequacy of these models one can make use of commercially available instruments. The signals generated are, however, not readily accessible to the simulator in use and form a significant bottleneck for the verification. To overcome these constraints, we present an integral strategy for spread spectrum simulation and test, which will be illustrated using the North American Digital Cellular, IS-95 standard. Taking advantage of HP-EEsof's Circuit Envelope Simulator developed for the simulation of RF circuits driven by complex signals, we have implemented the complete IS-95 standard in their Microwave Design System (MDS). This facilitates simulation and test at the hardware and software level using physical CDMA signals as defined in the simulator and generated externally. ACPR verification examples, using the advanced compact transistor model Mextram, are introduced and considered in some depth. Key words: Mobile communication, CDMA, Verification, Virtual loop, Mextram. Introduction The growing use of digital communication systems has complicated RF circuit design. Additional demands such as cost, efficiency and system constraints have complicated the design procedure. Complex digital communications signals with relatively large bandwidth and enormous crest factor have rendered inadequate traditional design methods based on small/large S-parameters particularly for transmitter power amplifiers. Although existing design methods for RF integrated circuits are indicative concerning system performance, the accurate prediction of power amplifier response to complex signals requires the use of advanced compact transistor models enabling the designer to trade off system demands and different circuit topologies such as Class A or AB. Verification of the accuracy with which an advanced model, such as Mextram [4], predicts the system behavior with respect to constraints, e.g. ACPR, becomes an indispensable part of the design procedure.

2 To excite the RF circuit one can make use of commercially available instruments. The signals generated are, however, not readily accessible to the simulator in use and form a significant bottleneck for the verification. To overcome this constraint, we have implemented the complete North American digital cellular, IS-95 CDMA system (see [6]), using the Envelope Circuit Simulator in Hewelett Packard's Microwave Design System which we use in turn to drive directly a pair of arbitrary function generators. This combination enables one to verify the simulated and measured results, when excited by the same stimulus, in one platform, namely MDS (Fig. 1). The adjacent channel power ratio (ACPR) North American CDMA is a Spread Spectrum System which is utilized with the so-called Walsh code (see [3], [5] and [6]), based on the Walsh matrix, a square matrix with binary elements. The baseband of the modulated signal has a bandwidth of MHz, which fixes the number of carriers (channels) to 20. In a fully loaded system, 35 different customers can reuse each channel. Unlike other communication systems such as GSM, each cell of the system is allowed the use of all available channels. In order to confine the influence of the adjacent channels, embodied in the spectral regrowth of the CDMA signal due to the added spurious signals, IS-95 has established the following predetermined criterion, the adjacent channel power ratio, ACPR (see [3]) The ratio of the measured spurious emission power level in a 30 KHz band at an 885 KHz offset to the signal's power measured in MHz band must be less than -42 db. The crest factor and the linearity The crest factor, the ratio of the peak to the root-mean-square value of the signal is particularly important for transmitter designers. In fact, the crest factor relates the ratio of the average power of the modulated signal to the maximum available output power of the amplifier. Since the determination of the crest factor is a function of the observation time and sampling, we have preferred to study the distribution of the ratio of the rectified envelope voltage and its RMS voltage: V REV () t = ( V I () t 2 + V Q () t 2 ) 1 V RMS = -- ( V T I () t 2 + V Q () t 2 ) dt C REV RMS T V 20 REV () t = log V RMS Using MDS we have calculated a signal histogram of the probability density function (PDF) of C REV/RMS for the Reverse Link signal. The result depicted in Fig. 2 indicates a large dynamic range in the power rectified envelope of the signal to its average which is a direct consequence of the signal's bandwidth. These results lead to the following conclusions: Increasing the processed signal's bandwidth can be employed to achieve a larger crest factor. Amplification of the signal requires a linear gain function, independent of input power. Otherwise, different parts of the signal will be amplified non-uniformly impairing the signal's spectrum, introducing the spectral regrowth. Advanced compact transistor models such as Mextram, in the case of bipolar transistors, that accurately model high injection current, quasi saturation and Kirk effect (see [4]), should be used in the design of RF circuits particularly for power transmitters.

3 Measurement system A block diagram of our measurement system is depicted in Fig. 4. The simulated I and Q components are passed to two arbitrary function generators (HP E1445A) as a function of time with a fixed sampling frequency. The generated signals are then up-converted to the desired frequency and power level using the quadratic amplitude modulation technique as implemented in the signal generator (HP ESG-D4000A). Results can be measured in the time or frequency domain using instruments such as an oscilloscope or spectrum analyzer which have been made accessible to MDS. Virtual loop Since the arbitrary function generator has a finite internal memory, generation of a continuously modulated signal implies a repetitive baseband signal (to be supplied by the arbitrary function generator). The cycleless (idiosyncratic) nature of modern communication signals, in general and CDMA in particular, unleashes, however, a chain of spurious spectral components (see Fig. 6). In order to surmount this difficulty, we have introduced a new concept the virtual loop. The virtual loop concept is based on a modification of the digital, analogue conversion technique implemented in communication systems under study using FIR and IIR filters. A low-pass, phase equalized filter According to [2], the baseband of the North American digital cellular, IS-95 system has a bandwidth of MHz and a chip-rate (sample frequency) of 4.9 MHz. This leads to undesired spectral components at a 4.9 MHz offset. A low-pass, phase equalized filter is required to eliminate the spurious signals to the edge of the baseband. By using a quadratic interpolation technique in MDS, we have increased the sample frequency of the processed signal to 9.8 MHz. It is then possible to design a low-pass Butterworth filter, which preserves the signal's relative group delay. Quality factor According to [2], the generated CDMA signal must meet a predetermined norm namely: the quality factor ρ which is in fact the cross correlation coefficient between the reference (simulated) and generated data sets defined as follows: ρ = k y i k 2 y i x i x i k x k and y k are respectively the reference signal (simulated) and the generated data. A quality factor of one means that the data sets are identical and a quality factor of zero indicates that the data sets are orthogonal. IS-95 specifies a quality factor greater than for the generated test signal. Instrumentation for the accurate determination of the overall quality factor should use similar up and down-conversion. In case the last option is not available, one can estimate the overall quality factor of the test system as a product of the quality factors of the arbitrary function generators and the signal generator, assuming that the added instrument's spurious signals are completely uncorrelated (see [8]). Based on this assumption, we can estimate the overall quality factor as the product of the quality factor of the HP ESG-D4000 and HP E1445A which meets the predetermine norm of IS-95. ρ = ρ AFG ρ SFG = = 0.995

4 Verification results Compact transistor models and large signal S-parameters are two methods for predicting the affect of the nonlinear behavior of active devices on CDMA stimuli. The latter method involves determining the transducer power gain of the device as a function of the incident power. This method is occasionally referred to as, AM-AM & AM-PM conversion. In contrast to the complex extraction procedure for compact transistor models, such as Mextram, the required parameter for the last method can be readily characterized using single tone CW measurement. To assess the adequacy of these methods we have used a Philips silicon double poly bipolar transistor (2.4*26 µm 2 effective emitter surface) for which a set of extracted Mextram parameters is available. The required transducer power gain functions are determined for two acquiescent collector currents using a single CW tone. These results along with the related simulation results, using the Mextram model, are illustrated in Fig. 5. Using the simulation and test system described above, we have calculated and measured the device ACPR response as a function of the average power for an incident CDMA signal (see Fig. 6). Although the end results confirm the ability of both methods, the large signal S parameter method tends to predict too optimistic a result. This is clearly a limitation of this method in the monitoring of intermodulation distortion. Conclusion In this paper, we present a new method using MDS for internally generating a proper CDMA signal, which in turn directly drives measurement equipment, enabling one to verify measured and simulated results using the same test signals. The verification results confirm the capability of the Mextram compact transistor model to predict the influence of the non-linear behavior of bipolar devices on CDMA stimuli. Acknowledgment The authors are grateful to Philips Semiconductors, Philips Research, HP-EEsof, Rohde & Schwarz and the Dutch Technology Foundation (STW) for supporting the research described in this paper. Bibliography [1] R. Mahmoudi, J. L. Tauritz, at. al. Impact of CDMA Specifications on Circuit Design. IEEE MTT-S International Microwave Symposium Workshop on LowCost Si-based Technology for Wireless Applications, June 1998, Baltimore, USA. [2] TIA/EIA/IS-95, Mobile Station-Base Station Compatibility Standard for Dual-Mode Wideband Spread Spectrum Cellular System, Global Engineering Documents. [3] TIA/EIA/IS-97, Recommended Minimum Performance Standards for Base Station Supporting Dual-Mode WideBand Spread Spectrum Cellular Mobile Stations, Global Engineering Documents. [4] L. C. N. de Vreede, H. C. de Graaf, J. L. Tauritz, at. al. Advanced Modeling of Distortion Effects in Bipolar Transistor Using The Mextram Model. IEEE Journal on Solid-State Circuits, vol 31, no.1. pp , January [5] D. P. Whipple, The CDMA Standard, Applied Microwave & Wireless. vol. 4,1992, pp [6] R. Mahmoudi, J. L. Tauritz, Performance Testing of The North American CDMA System, Using An Envelope Simulator Wireless Communication Conference. pp 84-88, Aug 1997, Boulder USA. [7] L., W., Couch, Digital and Analog Communication Systems, 4rd ed., Prentice Hall, New York, [8] R. Mahmoudi, J. L. Tauritz, at. al. Characterization and Verification of CDMA Power Amplifiers. 51 th Automatic RF Techniques Group, ARFG, Conference, pp , June 1998, Baltimore, MD, USA.

5 Measurement System DUT I & Q Implemented IS-95 Quadratic Amplitude Modulation Device Model Verification MDS Fig 1: The block diagram of the test and verification system. Fig 2: The signal histogram (PDF) of Reverse signal (OQPSK).Characterization of the measurement system. Added spectral components due to non-periodicity of the signal, Added spectral components caused by sample frequency offset at 4.9 MHz, Added spectral components caused by sample frequency offset at 9.8 MHz Fig 3: Left: The generated spectrum without virtual loop and a sample frequency of 4.9 MHz Right The generated spectrum with virtual loop and a sample frequency of 9.8 MHz

6 MDS Arbitrary Function Generator HP E1445A or HP8770A Low pass filter Function Generator HP ESG-D4000A Oscilloscope HP 54616C Spectrum Analyzer HP 5856E Output Matching Network DUT Input Matching Network Fig 4: The block diagram of the measurement system. Gain [db] Gain [db] Ic Ic acquiescent collector 4.2 ma acquiescent collector 6 ma Fig 5: The simulated and the measured 1.8 GHz of the transducent power gains (left) and the associated collector currents (right) Simulation (AM-AM) Simulation (AM-AM) ACPR ACPR Fig 6: The ACPR verification 1.8 GHz for two acquiescent collector currents 6 ma (left) and 4.2 ma (right).