GSM/ EDGE Mobile Technology Power Amplifier Characterization. Matthew Angert Masters Thesis First Draft March 5, I. Introduction...

Pages 54
Views 13
of 54
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Description
GSM/ EDGE Mobile Technology Power Amplifier Characterization Matthew Angert Masters Thesis First Draft March 5, 2004 I. Introduction... 3 II. Background...4 A. Power Amplifier Characteristics... 4 B. I-Q
Transcript
GSM/ EDGE Mobile Technology Power Amplifier Characterization Matthew Angert Masters Thesis First Draft March 5, 2004 I. Introduction... 3 II. Background...4 A. Power Amplifier Characteristics... 4 B. I-Q Modulation... 5 C. Encoding Data in GSM and EDGE: Encoding Data in GSM Encoding Data in EDGE... 6 D. GSM/ EDGE Signal Structure... 9 III. Measurement Theory A. Power Added Efficiency B. Power versus Time Measurement C. Error Magnitude Vector Measurement D. Output RF Spectrum - Adjacent Channel Power Output RF Spectrum due to Modulation and Wideband Noise Output RF Spectrum due to Switching Transients IV. Measurement Implementation A. Overview of Testing B. Power Amplifier Module Evaluation Boards C. Implementation - Power Added Efficiency D. Implementation of Tests Using Vector Signal Analyzer E. Implementation of Tests Using PSA Spectrum Analyzer F. Biasing the PA G. PSA Testing - Automated Testing Codes written in Agilent Pro VEE V. Results for VSA Measurements A. Power Added Efficiency Results B. Power versus Time C. EVM Measurements D. ORFS - Spectrum due to Modulation... 43 E. ORFS - Spectrum due to Switching G. Results of VSA Showing Dynamic Range as an Issue: VI. Results for PSA Measurements A. Power versus Time Results B. EVM Measurements C. ORFS - Spectrum due to Modulation D. ORFS - Spectrum due to Switching VII. Conclusions VIII. References... 56 I. Introduction In United States Code Division Multiple Access (CDMA) technology is primarily used for cellular phones, but worldwide Global System for Mobile Communications (GSM) (or Groupe Speciale Mobile) is the most widely deployed wireless network. Worldwide GSM Mobile Phone Users Soon to Reach 1 Billion, GSM Group Says The GSM Association, an industry association promoting GSM mobile phones, announced that worldwide users of GSM mobile phones have reached million. 1 GSM is only second generation (2G) meaning it older and has limited data rate and capability. The Enhanced Datarates for GSM (also abbreviated with Global 2 (EDGE) is third generation (3G) (some call it 2.5G indicating it is more advanced than 2G technology but not quite as high data rate as a 3G system). EDGE uses much of the same network as GSM. EDGE is viewed as a stepping-stone that more of the infrastructure can be reused and thus cost is diminished. To utilize the enhanced features of EDGE, better components (ie. power amplifiers) need to be developed. Along with the better components new measurement techniques need to be developed and implemented to ensure the quality of the new EDGE components. II. Background A. Power Amplifier Characteristics The final stage in a RF front end before the antenna is usually a power amplifier (PA). The PA creates a large output power from the small input power signal to be sent out to the antenna. Many issues arise in power amplifiers due to their distortion, large power consumption, and high cost of these devices. Because of these issues, characterizing and minimizing the negative effects of power amplifiers are of utmost concern in mobile telephone design. The PA is usually integrated into a power amplifier module, which consists of filters to use multiple bands, some sensing loops, matching components, etc. 3 As more and more components are being pushed to become integrated on a single chip, the PA cannot be integrated with the rest of the transceiver because of the high amounts of power needed to run these devices. Linearity and harmonics - Ideally we want the transfer function of a power amplifier to be linear over the entire input range for all desired outputs. All realistic amplifiers have a limited linear range and outside this range output saturation and distortions occur as shown in Figure 1. Many times a larger issue is that the non-linear characteristics cause harmonics of the input signal to be created at the output. These harmonics can cause interference in other frequency channels and could also mix in the receiver to create distortion at the frequency of interest. 4 Figure 1. Output Power versus Input Power showing distortion and compression characteristics of transfer function of PA. B. I-Q Modulation A signal that has been modulated by a carrier can be represented by At ()cos( ωct+ θ ()) t. To make a signal have more efficient bandwidth usage, quadrature amplitude modulation (QAM) can be employed. If two signals are to be transmitted, m 1 (t) and m 2 (t), they can be sent by having the carrier frequency 90 degrees out of phase of each other. Using the same local oscillator, the signal is delay by 90 degrees, the cos( ω t c ) is modulated with m 1(t) and sin( ω t c ) is modulated with m 2(t). The advantage of doing quadrature modulation comes from the fact that sine and cosine are orthogonal to each other. Even though both signals have the same center frequency and bandwidth, they do not interfere because of this orthogonality. Therefore two signals can occupy the same bandwidth that one signal would occupy if it did not use quadrature modulation. 5 Quadrature amplitude modulation is used in GSM and EDGE. All QAM transmitted signals can be represented as: x() t = m ()cos( t ω t) + m ()sin( t ω t) 1 c 2 c where m 1 (t) and m 2 (t) are two separate message signals and wc is 2πfc The in-phase portion of the signal is the one associated with the cosine and is denoted I(t). The quadrature portion of the signal, which is the one 90 degrees out of phase with the cosine signal, is denoted as Q(t). Therefore the signal can be written as: x() t = I()cos( t ω t) + Q()sin( t ω t) c c The I(t) and the Q(t) can be plotted against each other to get a picture of the signal. This is known as an I-Q Diagram or constellation diagram. For example the EDGE I-Q diagram is shown in Figure 2. Figure 2. I-Q Constellation Diagram of an EDGE Signal 6 C. Encoding Data in GSM and EDGE: 1. Encoding Data in GSM To understand EDGE modulation, first look at the simpler GSM modulation. The modulation scheme used in GSM is Gaussian Minimum Shift Keying (GMSK). GMSK signals are generated by sending the signal through a Gaussian prefilter, which reduces the side lobes. 7 Then the signal is encoded in the following manner. Four options are created: positive in-phase and quadrature components, positive in-phase component and negative quadrature component, negative in-phase component and positive quadrature component, and negative in-phase and quadrature components. As the signal shifts phase, the I-Q position changes from one position to the next and this determine what bit is encoded. For example if the I and Q components are both positive at a certain point, by changing the signal to having negative in-phase component and positive quadrature component would encode a bit 1 as shown in Figure 3. As all the bits are encoded the signal shifts clockwise or counterclockwise to encode all the bits. As can be seen from Figure 3 at every time the signal is a constant distance from the origin, which signifies that it always has constant amplitude. Because the GSM signal has constant amplitude, the design for components is simplified. Figure 3. Illustration of MSK 8 2. Encoding Data in EDGE The modulation scheme used in EDGE is a variation on 8-Phase Shift Keying (8- PSK) called 3π/8 8-PSK. 8-PSK Modulation In Phase Shift Keying the I-Q plane has many more positions that can allow the encoding of more bits at a time point. Having one or more bits being indicated by one sample of a signal creates symbols. By using 8-PSK the eight positions are used so that each sample indicates three bits. In this way the data rate to transmit the information is increased from 1 bit per sample to 3 bits per sample comparing the GMSK to the 8-PSK modulation schemes. Again by changing position relative to the current position, the bits are encoded onto the carrier as shown in Figure 4. An issue with 8-PSK is that in certain transition the signal crosses the origin of the I-Q diagram, which indicates that the amplitude is becoming zero. The amplitude becoming zero at any time causes problems in the amplifiers used because distortions occur. Also zero crossings result in discontinuities in envelope and phase of the waveform. Amplifiers have only a certain linear range and amplifiers can cause distortions when the signal is very small and is close to zero. In EDGE, to avoid the amplitude become zero, 3π/8 8-PSK modulation scheme is used. In this scheme there are still eight possible destinations starting at each sample point, but they are each shifted so that at no time does the signal cross zero and shown in Figure 5. 3π/8 equals 67.5º, which is amount of shift, that occurs between the present symbol and the adjacent symbol. Effectively the targets generated are like the 16-PSK modulation scheme. Looking at Figure 5, the targets switch from white to gray with each sample taken to encode the data onto the carrier. The bits are mapped into symbols using Grey Code, which means each adjacent symbol differs only by one bit. Figure 4. 8-PSK Modulation 9 Figure 5. 3π/8 8-PSK Modulation 10 Therefore EDGE modulation scheme of 3π/8 8-PSK has the advantage of GMSK that it has no zero crossings and therefore does not cause distortions to the amplifier. The main challenge of 3π /8 8-PSK is that the signal no longer has a constant envelope. Because the signal no longer has a constant envelope, provides major issues for components in particular, the power amplifier. With a constant envelope, the power amplifier s linear range essentially needs to be zero because the signal s amplitude does not change, but with EDGE s 3π /8 8-PSK, the linear range needs to be large. As shown in Table 1, EDGE s required linear range is quite significant. GSM (GMSK Modulation) EDGE (3π/8 8-PSK Modulation) Peak-to-average ratio 0 db 3.2 db Peak-to-minimum 0 db 17 db Table 1. GSM and EDGE peak to average and peak to minimum ratios 11 D. GSM/ EDGE Signal Structure GSM and EDGE both use Time Division Multiple Access (TDMA) and Frequency Division Multiple Access (FDMA). The TDMA signal is broken up into eight equal time slots, which repeat in what is called a frame as shown in Figure 6. Figure 6. TDMA signal with 8 repeating timeslots. GSM and EDGE use a signal with a 200 khz bandwidth for each carrier. The frequency bands used for GSM EDGE include: Frame Structure: The frame and timeslot structure is shown in Figure 7. Each timeslot is us and therefore a frame is ms. Within each timeslot tail bits are at the beginning and end of each slot. These are included to tell when the signal starts and stops. Then the actual data is contained next, which included two control bits and a midamble in between. The midamble contains information about the burst such as what data rate is being used, etc. At the end of each timeslot there are guard bits, which simply add a buffer between this slot and the next. Because the signal is on only during its intended time and off in the others it is known as a burst. In this one timeslot is where one user transmits, or receives, the signal. During the timeslot other types of bursts can be transmitted including a frequency correction burst, synchronization burst, and dummy burst. The frequency correction burst corrects for frequency errors during transmission. Synchronization bursts make the time slots begin and end at the correct time as time delays can occur during transmission. A dummy burst is a burst that is just blank. 12 Figure 7. GSM and EDGE Burst Structure 13 The Figure 7, which is GSM s burst structure, shows the timeslot of 577 us containing bits. In EDGE s case the same structure can be used but consider each bit as a symbol. Only in the Data portion of the timeslot does the symbols equal 3 bits per symbol; in the rest of the burst each symbol is still 1 bit. Therefore the number of symbols in the timeslot would be symbols. Not counting the guard bits, gives 148 symbols, which is known as the entire burst. Not counting half a symbol of the front tail and half of a symbol of the end tail, gives 147 symbols, which is known as the useful part of the burst. Some refer to the burst as containing 142 symbols, which indicates that both tails are not included in the definition of the burst. Because GSM and EDGE have the same time duration for a slot and have the same occupied bandwidth in frequency, GSM and EDGE signals can exist in the same frame. This compatibility provides designers with more freedom and flexibility in that systems can use EDGE, GSM, or both and change between the two depending on conditions. Figure 8 shows a frame with EDGE at two different power levels, GSM at two different power levels, and empty slot. Figure 8. GSM and EDGE versus time showing both EDGE and GSM coexisting 14 GSM and EDGE both have a symbol rate of khz. GSM using GMSK modulation has 1 bit per symbol as shown previously. EDGE using 8-PSK has 3 bits per symbol and as shown in Table 2, higher data rate can be reached by using EDGE. GSM/ GPRS Modulation GMSK Bits/symbol 1 Symbol rate ksym/s Modulation Bit Rate kbps Ratio Data Rate per time slot 22.8 kbps Data Rate per time slot (User Rate) 14.4 kbps Theoretical Max User Rate 20 kbps Data Rate per eight time slots (User Rate) kbps Theoretical Max User Rate Eight Slots 160 kbps Table 2. Summary of GSM and EDGE characteristics 15 EDGE 3π /8 8PSK ksym/s kbps 69.2 kbps 48 kbps 59.2 kbps 384 kbps kbps III. Measurement Theory 1. Power Added Efficiency - Determines how well the power amplifier converts DC power into RF power. 2. Error Vector Magnitude Determines the quality of the modulation and how much modulation error has occurred. 3. Power versus Time Determines the distortion of the waveform shape and the interference into other time slots. Output RF Spectrum Adjacent Channel Power Determines the amount of interference into other frequency channels. 4. Spectrum due to modulation Determines the adjacent channel power that comes from the steady state component of the signal. 5. Spectrum due to switching Determines the adjacent channel power that comes from the transient component of the signal. A. Power Added Efficiency The end goal of a power amplifier is to take a weak RF input power and using the DC biasing of the circuit, create a strong RF output power. A good measure of performance of a power amplifier is therefore how efficiently this conversion is done. Power Added Efficiency is defined by the below equation: Pout - Pin PAE(%)= 100% Vsupply Isupply (Equ 1) where Pout is RF output power of PA in Watts, Pin is RF input power in Watts, and Vsupply and Isupply are the DC voltage and current of the source. The Vector Signal Analyzer was used to measure the RF output power and the Electronic Signal Generator created the RF input power. Both of these quantities are in dbm so the correct equation becomes: ( ) Pin( dbm) Pout dbm PAE(%)= % Vsupply Isupply 10 (Equ 2) where Poutm(dBm) is RF output power of PA in dbm, Pin(dBm) is RF input power in dbm, and Vsupply and Isupply are the DC voltage and current of the source. PAE Calibration: To protect the VSA a 20 db attenuator had to be added after the output of the power amplifier. This attenuator is not precisely 20 db and the two adapters and two cables used for measurement provide some attenuation so this should be calibrated out to provide an actual measure of PAE. Running the setup over the maximum power levels of the ESG at the frequency ranges of interest, the attenuation of the 20 db attenuator, cords, and adapters totaled 22.0 db. Therefore the final equation used to calculate PAE is: ( ) Pin( dbm) Poutm dbm PAE(%)= % Vsupply Isupply 10 (Equ 3) where Poutm(dBm) is RF output power of PA in dbm, Pin(dBm) is RF input power in dbm, and Vsupply and Isupply are the DC voltage and current of the source. B. Power versus Time Measurement Because the EDGE signal is a TDMA signal, measuring the power amplitude versus time can tell a great deal about the quality of the signal. The power versus time measurement ensures the signal has the proper shape in amplitude and because EDGE uses TDMA, it ensures the burst is not shifted in time into other timeslots. What an EDGE signal that is only using one slot looks like in time domain can be seen in Figure 9. When amplifier distortion occurs the burst will become distorted and appear in other timeslots. This measurement determines the acceptable limits of the distorted shape and interference into the other timeslots. Figure 9. EDGE Signal in the Time Domain A mask is used to determine the proper shape of the EDGE waveform in time by providing an upper and lower limit of the waveform. The shape is determined from the data structure of the EDGE signal as discussed previously. The mask of a normal burst EDGE signal can be found in Figure The mean of the useful part of the burst (the center 147 symbols) is used as the 0 db reference point. If trace is between upper and lower mask, this indicates a passing of the test. db +4 +2, (**) (***) (147 symbols) 7056/13 (542,8) µ s (*) t ( µ s) Figure 10. Time mask for normal duration bursts (NB) at 8-PSK modulation 17 C. Error Magnitude Vector Measurement The I-Q Diagram is a useful tool when Error Vector Magnitude (EVM) is used for analysis. After each measured I-Q position is determined, the difference between the measured signal and the ideal reference signal can be determined. Ideally the difference in the measured and the ideal is zero and any difference greater than zero indicates error. Because EDGE signals are amplitude and phase modulated signals, analyzing the EVM provides a good analysis of the modulation accuracy of the signal because both possible errors are taken into account. Figure 11 illustrates the measured, ideal, and error signals on the I-Q plane. Figure 11. Measured, ideal, and error signals on the I-Q plane used in EVM 18 From the geometry of the above signals, the Error Vector Magnitude can be found from: EVM = ( I I ) + ( Q Q ) (Equ 4) 2 2 meas ref meas ref To create the ideal reference signal the measurement device, whether a Vector Signal Analyzer or Spectrum Analyzer, would follow a procedure shown in Figure 12. The input is demodulated and then modulated, which then determines the ideal reference signal. Figure 12. Illustrates how reference and measured signal is created from measured signal. 19 When measuring the EVM for an entire burst for all symbols in the burst, a few particular EVM analyses tell a great deal about the error of the modulation and these include RMS EVM, Peak EVM, and 95 Percentile EVM. RMS EVM The root mean square EVM (RMS EVM) over all samples is an indicator of the extent of errors occurring throughout all the samples and is calculated in Equation 5. RMS EVM is the ratio of the RMS error vector value and the RMS of the ideal vector value for all symbols measured. Doing this normalizes each error vector to the reference signal to provide a useful measure of the error. 20 E(k) = Error Vector at symbol k S(k) = Ideal Vector at symbol k EVM rms = k k Ek ( ) Sk ( ) % (Equ 5) Peak EVM If large errors are occurring at some symbols, but small errors are occurring at others, the RMS EVM will not give a high value for errors. Therefore using peak EVM gives a value, which will determine if any symbol is have very high errors. Ek ( ) EVM ( k) = 100% 1 2 Sk ( ) N k 2 (Equ 6) where N is number of samples EVM peak = max[ EVM ( k)] (Equ 7) The 95 th Percentile EVM is the EVM value that 95% of all the symbols measured are at or below. When the 95 th Percentile value is determined, this would indicate that only 5% of the symbols measured have an EVM value higher than the 95 th Percentile value. For example if the 95 th Percentile value is 10% and there are 147 symbols, this would indicate that for 140 symbols the EVM value is 10% or less. 21 For EDGE signals in mobile cellular devices, the EVM specifications can be found in Table 3. Normal Conditions Extreme Conditions RMS EVM 9% 10% Peak EVM 30% 30% 95
Advertisements
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x