An amplifier With HF Selectivity

Transcription

1 Department of Electrical and Information Technology ETI 041 Radio Project An amplifier With HF Selectivity Author: Jiangfeng Cai Liding Zhao Supervisor: Göran Jönsson Lund 2009

2 c The Department of Electrical and Information Technology Lund University Box 118, S LUND SWEDEN This thesis is set in Computer Modern 10pt, with the L A TEX Documentation System using CJF thesis template. c Jiangfeng Cai & Liding Zhao 2009 Printed in Sweden by E-huset, Lund. November 2009

3 Preface The low-noise amplifier (LNA) is a special type of electronic amplifier used in communication systems such a FM receiver to amplify very weak signals captured by an antenna. As a key component, LNA is placed at the front-end of a radio receiver. By Friis formula, the overall noise figure of the receiver is dominated by the first few stages. For low noise the amplifier needs to have a high amplification in its first stage. To enhance gain, input and output matching circuits are essential. Image rejection is achieved by filtering in this stage. In all, main amplifier specifications include gain, noise figure, bandwidth, compression point and intercept point. A complete design procedure including parameter extraction, design, simulation, implementation and testing of low noise amplifier for FM receiver is described in this report. One feature or challenge of this project to design the image rejection filter that can be implemented by traditional band-pass filter or other structures. This report only focus on the traditional filter due to time constraints. In addition, stability is a big issue. Maybe the easiest way to build an oscillator is to design an amplifier, we hope it never happen however. We would like to express our gratitude to all people who have inspired us and contributed to this project course. First of all, we would like to express our deepest gratitude to our supervisor Göran Jönsson. Mr.Jönsson is much experienced in RF circuit design and teaching. In this project, he gave us many suggestions about layout design and test procedure. Thanks for his reading and commenting of this report. I have to say that it is my pleasure to select his series courses Radio-Radio Electronics-Radio Project. Markus Törmänen gave us a fast tutorial for ADS so that we can easily simulate the circuit. We would like to thanks Lars Hedenstjerna for the construction of the PCB. This is the last course in my course plan. Here, I would like to use opportunity to express my appreciation to Department of Electrical and Information Technology, LTH, Lund University, and all classmates. Two years study in Sweden has become an invaluable experience in my life. Lund, November 2, 2009 Jiangfeng Cai & Liding Zhao

4

5 Contents Preface iii 1 Introduction Superheterodyner Receiver Image Frequency Specifications and Design Platform Amplifier Design Methodology RF Transistor S-Parameter Based Amplifier Design Gain and Noise Figure Matching Network Transistor Biasing Image Rejection Filter Design System Simulation Layout Design and Assembly RF PCB Layout Guide Package and PCB Assembly Skills Measurement Results DC Measurement Amplifier Band-Pass Filter Selective RF Amplifier System Stability Gain and Bandwidth Image Rejection Compression and Intercept Point Noise Figure Conclusions Error Analysis Conclusions Bibliography 19

6

7 Chapter 1 Introduction R F amplifier, as the first stage, is the most critical module in the radio receiver system. Due to Frii s formula 1, the noise figure(nf) is the most significant design specification for any design project. Meanwhile, RF amplifier should have a good selectivity and certain image rejection ratio. 1.1 Superheterodyner Receiver The superheterodyne (usually shortened to supterhet) receiver was invented in 1918 by U.S. Army major Edwin Armstrong in France during World War I. Surprisingly, this design is so superior that it is still today the basic design architecture for all FM and AM radio receivers. The block diagram to a superheterodyne receiver is shown in Figure 1.1 [1]. The essential elements include RF amplifier, local oscillator (LO), mixer, fixed-turned filter and intermediate frequency (IF) amplifier, demodulator and AF amplifier. The advantage to this method is that most of the radio s signal path has to be sensitive to only a narrow range of frequencies [2]. Figure 1.1: Architecture of superheterodyne receiver 1 NR = NR 1 + NR 2 1 A p1 + NR 3 1 A p1 A p2 + + NR n 1 A p1 A p2 A p3 A pn 1

8 2 Introduction Figure 1.2: Frequency conversion in the receiver 1.2 Image Frequency The basic principle of the receiver is frequency mixing to convert the received signal to a constant lower frequency before detection. This constant frequency is called intermediate frequency (IF). The IF system for broadcast FM commonly operates at 10.7 MHz [3]. The output of mixer includes at least four frequencies: f RF, f LO, f LO f RF, f LO +f RF. This frequency conversion is demonstrated in Figure 1.2 [1]. Potentially, two signal could be shifted to same f IF, one at f RF = f LO +f IF and another at f RF = f LO f IF. One or the other of signals, called image frequency, should be filtered out before the mixer to avoid aliasing. When upper one is filtered out, it is called high-side injection. The other case is called low-side injection. The image frequency can be calculated by Equation 1.1. { frf + 2f f image = IF if f LO > f RF (high side injection) (1.1) f RF 2f IF if f LO < f RF (low side injection) The image frequency results in two stations being received at the same time, thus producing interference. Image frequencies can be eliminated by sufficient attenuation on the incoming signal by the RF amplifier filter. 1.3 Specifications and Design Platform Operating Frequency: Gain: Noise Figure: Image Rejection: Quiescent Point: Source/Load Impedance: Power Supply: Matching Network: Network Analyzer: Spectrum Analyzer: Signal Generator: Noise Sources: Digital Multimeter: Circuit Simulation: PCB Layout Design: MHz G t S 21 2 F F opt +2 db 20 db I C =20 ma, V CE =6 V 50 Ω 12 V DC Input and output matching ROHDE & SCHWARZ ZVR ROHDE & SCHWARZ FSEA ROHDE & SCHWARZ SME 03 Noisecom NC346 Precision Noise Sources PHILIPS PM2518X Agilent ADS V2008 Eagle 5.6.0

9 Chapter 2 Amplifier Design Methodology S -parameter based amplifier design method is a classical design methodology for RF circuits. A sufficient stability of amplifier is a compromise between the selection of Γ S and Γ L. Meanwhile, the amplifier should have enough attenuation for image frequency performed by a series band-pass filter. 2.1 RF Transistor BFR520 is the core element of the RF amplifier. BFR520 is a NPN 9 GHz wide band RF transistor from NXP Semiconductors. It features high power gain, low noise figure, high transition frequency and excellent reliability with gold metallization processing technique. Some design parameters are available on the data sheet [4]. The noise figure is F min = 1.35 db with quiescent point I C = 20 ma, V CE = 6 V and operating frequency at 100 MHz as shown in Figure 2.1(b). This point is much close to the design specification. (a) NF and Ga as function of Ic (b) NF and Ga as function of of frequency Figure 2.1: Noise figure and gain of BFR520 transistor

10 4 Amplifier Design Methodology Figure 2.2: Amplifier topology 2.2 S-Parameter Based Amplifier Design A two-port with matching networks is shown in Figure 2.2. Γ S and Γ L refers to the reflection coefficients as seen by the two-port whereas the actual source and load refection coefficients are denoted Γ S and Γ L. The design procedure described below base on S parameter, the values of Γ S, Γ L and matching networks within stable condition to get the desired gain will be determined [5] Gain and Noise Figure The transistor in amplifier is connected as common emitter (CE) configuration. The S-parameters from data sheet were: S 11 = S 21 = S 12 = S 22 = Since = 0.68 < 1 and K = 0.41 < 1 the transistor is conditionally stable and maximum stable gain G MSG = S 21 /S 12 = db, there is some headroom for the design specification G = S 21 2 = db. The gain, noise and stability circles plotted by MATLAB tool DESLIB are shown in the Figure 2.3. Figure 2.3: Stability, gain, NF circles and reflection coefficients

11 2.2 S-Parameter Based Amplifier Design 5 (a) Design of matching networks (b) Input and output matching circuits Figure 2.4: Matching networks Matching Network The selection of Γ S is a compromise between amplifier gain and stability because reflection coefficient Γ L (conjugate of Γ OUT ) is a function of Γ S. From Figure 2.4(a), source reflection (Γ S = ) is very close to gain circle but Γ L = has a relatively safe margin from output stable circle. The input matching is realized by a series capacitor shown in Figure 2.4(b), its value can be calculated by C in = 1 2πf 0 x 1 Z 0 = similarly, for output matching network we have C out = b 2πf 0 1 Z 0 = 1 2π = pf π = 10.4 pf L out = x 1.7 Z 0 = 50 = 138 nh 2πf 0 2π

12 6 Amplifier Design Methodology (a) Shunt feedback I bias (b) Shunt feedback II bias (c) Series feedback bias 2.3 Transistor Biasing Figure 2.5: Various passive biasing circuits Three kinds of passive networks are commonly used in bias circuits, but only the first two configurations are suitable for RF amplifiers as shown in Figure 2.5. Feedback is used in these bias circuits to get stable current and better temperature performance. We select the second configuration due to its stability and good connection to ground. To partition the current gain of the transistor as I C = β 0 I D and I D = β 0 I B with given β 0 = 120, I C = 20 ma and V CC = 12 V I D = I C β0 = = ma and I B = I D β0 = = ma R C = V CC V CE 12 6 = I C + I D + I B = Ω Assume V D = 1.4 V (two times voltage drop of one PN junction) R B1 = V CE V D I D + I B = = 2.31 kω R B2 = V D I D = = Ω R B3 = V D V BE = = 4.2 kω I B Image Rejection Filter Design A band-pass filter is connected in series with RF amplifier to filter out the image frequency. The most common filter approximation is the Butterworth. One advantages of this filter is that the amplitude at -3dB (A(f 0 ) = 1/ 2) is independent of filter order n. According to design specifications, the pass-band is from 88 MHz to 108 MHz. The image frequency at 98 MHz, which is f image = f RF + 2f IF = = MHz from Equation 1.1, has -20dB attenuation. Note the equality between the frequency products ω 1 ω 2 = ω 3 ω 4 = ω I 2 so that all rejection band regions can be calculated as shown in Figure 2.6.

13 2.4 Image Rejection Filter Design 7 Figure 2.6: Amplitude specification of RF filter The filter order n can be calculated by equation A(f) = K 1 + ( ff0 ) 2n (2.1) Theoretically, the order n should be at least 3.7. Given some filter effect of output matching network, the filter order can be truncated to n=3. The detailed calculation result is in Figure 2.7(a) and the transfer function this filter is shown in Figure 2.7(b). (a) Schematic of band-pass filter (b) Simulation of transfer function Figure 2.7: The 3rd order band pass filter

14 8 Amplifier Design Methodology 2.5 System Simulation Figure 2.8: Selective RF amplifier So far we have finished all functional module design for the RF amplifier. But, from the signal isolation point of view, the DC path should isolate the device from the biasing circuits at high frequencies. A radio-frequency-choke (RFC) can be used to comply with our requirements. A RFC is an inductor designed to block a particular frequency in circuit while passing low frequency signal such DC biasing. By placing an RF chock in series with RC, we effectively permit DC component rather than AC component to pass. An RFC coil must be designed such that operating frequency is below the selfresonance frequency. Meanwhile, two decoupling capacitors are used to ensure grounding. The complete schematic of the RF amplifier is shown in Figure 2.8. ADS is an excellent RF circuit design and simulation software developed by Agilent. Figure 2.9 demonstrate the AC analysis of the RF amplifier with two different image rejection filters. Apparently, the 4th order filter has a higher attenuation for image frequency than the 3rd order filter. But db attenuation of the 3rd filter is quite close to specification and the maximum gain is db, previous design has been proved. (a) The 3rd order filter (b) The 4th order filter Figure 2.9: AC simulation

15 Chapter 3 Layout Design and Assembly L ayout design is an art form, and never more so than when you are designing an RF circuit. How to reduce the effects of stray capacitance and inductance introduced by wires is a Gordian knot. A good strategy is to use printed circuit board (PCB) and surface mount devices (SMD). 3.1 RF PCB Layout Guide Below are given some general layout guides that can be used in RF circuit design. Following these rules will help us to avoid some of the most common pitfalls. A solid ground plane should always be used in RF circuits. The purpose is to create an effective 0 V reference node. For SMD PCB, all signal routing is on the top layer and the ground plane will be on the bottom layer. It is a good idea to fill all available space at the signal routing layers with ground plane. Multiple vias could be used to connect the ground planes to the main ground plane. Short connections to ground plane. A via should be placed close to each pad that is to be grounded. Never let two ground pads share one via, that will lead to cross talk between the two pads. The placement of all components should be very compact. Ensure that the RF components are laid out such that all the RF tracks can be kept on the top layer with minimum length and changes of direction. Use mitered bends if signal tracks can t be run in a straight line. Never use right angled bends, which would increase losses and spurious emissions. De-coupling capacitors should be placed as close as possible to the pins that are to be de-coupled. Use one de-coupling capacitor for each node that is to be de-coupled. The value of the capacitors should be chosen so that their series resonance frequency is above the signal frequency they are to de-couple.

16 10 Layout Design and Assembly 3.2 Package and PCB Figure 3.1: PCB layout (Scale 1:1) When planning to layout an RF PCB, the first place to start is to decide the package for all components. To simplify the design, we use only one package (R1206) for all resistors, inductors and capacitors. The transistor BFR520 is encapsulated in a plastic SOT23 envelope. The package of BNC connector is A1944, which can be found in con-coax library. The width of signal traces is inch. The layout area is mm. The PCB layout is shown in Figure Assembly Skills Soldering is an essential technique for every electronic engineer. Proficient skill will improve progress even performance. On the contrary, bad soldering will introduce new interference and even deteriorate the performance. Listed below are some soldering tips and preventative maintenance techniques, detailed information can be found in references [6] [7] [8]. Use appropriate tools. Because of small size of SMD components, the size of soldering iron tip should be small. Keep working surfaces tinned, clean hot tip with damp sponge only before using. Only use rosin core solder and small diameters are most appropriate for small electronics work ( dia. is recommended). The wire such as power supply wire should be twisted and lightly tinned. The terminal of BNC connector also should be tinned before soldering. Solder joints should be inspected when completed to determine if they have been properly made. A good solder joint should has shiny surface and smooth fillet. Use two soldering irons to rearrange or desolder components.

17 Chapter 4 Measurement Results M easurement should be performed step by step. The first thing to do is to test DC bias of transistor so that it has a correct quiescent point. Verify the amplifier and band pass filter respectively before connecting them. The gain and stability of amplifier should satisfy the specification. Check bandwidth and center frequency of the band pass filter. Finally, combine amplifier and filter as a complete system. 4.1 DC Measurement Because S-parameters are dependent on I C and V CE, make sure that the transistor works under the desired quiescent point. Table 4.1 shows some some DC results. Table 4.1: Measurement of DC parameters Quiescent Point Ideal value Measured value I C 20 ma 21.9 ma V CE 6 V V h fe V D 1.4 V V V BE 0.7 V V Measured result shows that the design of DC bias circuit is effective, the result is basically close to the desired specifications. 4.2 Amplifier Gain and stability are the most important specifications for the amplifier. A coupling capacitor is needed to isolate the DC component. Figure 4.1 shows the gain and stability parameters of the pure amplifier. Of great interest in evaluating the stability of an amplifier is stability factors. The amplifier is stable because both input and output coefficient (S 11 and S 22 ) inside the circle. Meanwhile, the K factor is smaller than 1V (0.659 V) at 98 MHz. Gain is db at 98 MHz, which is very close to the design value.

18 12 Measurement Results (a) Input stability (b) Output stability (c) K factor (d) Gain Figure 4.1: Input and output stability without filter 4.3 Band-Pass Filter The measured transfer function of band-pass filter with original value is shown in Figure 4.2(a). Compare with simulation in Figure 2.7, the desired center frequency (98 MHz) is shifted to MHz. One possible reason is the effect of parasitic capacitance. The capacitor C 2 in filter can be adjusted to compensate the effect of stray parameter. Finally, C 2 is fixed at 213 pf, center frequency is 96.5 MHz with 24.1 MHz bandwidth (-2 db) shown in Figure 4.2(b). The amplitude has a -5 db attenuation mainly due to losses in the inductors. (a) C 2 =270 pf (b) C 2 =213 pf Figure 4.2: Butterworth bandpass filter

19 4.4 Selective RF Amplifier 13 (a) Input stability (b) S 12 (c) K factor (d) Output stability Figure 4.3: Input and output stability with filter 4.4 Selective RF Amplifier The selective amplifier specification include gain, bandwidth, image rejection ratio, compression point, intercept point and noise figure. Most of them can be measured by network analyzer. But for IP 3, two-tone test is still an effective method. A precision noise source is necessary for NF test System Stability Compare with stability of the pure amplifier, the bandpass filter deteriorates the input stability as shown in Figure 4.3(a). The input reflection coefficient S 11 is unstable when operating frequency in range between and MHz. K factor becomes a curve rather then a straight line. Output reflection coefficient S 22 is always in stable area as shown in Figure 4.3(d) Gain and Bandwidth The -3 db bandwidth is MHz and center frequency is MHz. Thus the operating frequency range from MHz to MHz, which totally cover the specification requirement. The transducer gain is db at MHz and is almost flat within the band as shown Figure 4.4(a). Compare gain value of pure amplifier (28.31 db), it decreases 6.7 db after using the band pass filter.

20 14 Measurement Results (a) Gain and bandwidth (b) Image rejection Figure 4.4: Gain and image rejection Image Rejection The image frequency of center frequency 98 MHz is MHz, which can be calculate by Equation 1.1 if high side rejection is used. Consequently, the image rejection ratio corresponding these two point is =18.24 db, which is close to the design requirement Compression and Intercept Point The 1 db compression point (CP 1dB ) indicates the power level that causes the gain to drop by 1 db from its small signal value. It is the maximum output signal for linear gain. The SYSTEM MODE-COMPRESS SOI TOI submenu in network analyzer provides automatic CP 1dB measurement [9]. The result is shown in Figure 4.5(a). The worst compression point in-band is 4.56 dbm at 88 MHz. Based on intermodulation products definition, we use two-tone with a small frequency difference method to get red and blue curves as shown in Figure 4.5(b). The 3rd order intercept point is read in the intersection between the extrapolated curves, which can be extracted by linear regression function. From two linear equations, we have IP 3 = dbm. { y1 = ax + b = 0.997x y 2 = ax + b = 3.054x (a) 1 db compression point (b) 3 rd order intercept point Figure 4.5: Compression and intercept point

21 4.4 Selective RF Amplifier 15 Figure 4.6: Noise figure Noise Figure To reduce electromagnetic interference (EMI), the measurement of noise figure should be performaed in RF shielded chamber. With the help of precision noise source, network analyzer and PC data acquisition system, noise figure and gain curves are plotted with room temperature 20 C like Figure 4.6. The noise figure is about 2 db within bandwidth. Some small peaks in noise figure are caused by other device interference such as PC monitoring system. Meanwhile, the gain data is basically above 21 db. The amount of noise added by the noisy front end of a receiver is often expressed in terms of an equivalent temperature. The data acquisition computer also provide equivalent temperature curve as shown with green line in Figure 4.6. The conversion between noise ratio (NR) or noise figure and equivalent noise temperature T eq is T eq = T 0 (NR 1) = T 0 (10 NF 10 1) (4.1) Here, T 0 =290 K is a reference temperature in Kelvins.

22 16 Measurement Results

23 Chapter 5 Conclusions I n this part, the performance comparisons and possible deficiency arising system errors will be analyzed. How to compensate these errors are also mentioned. Simply using passive filter will introduce some problems such gain attenuation and stability variation. 5.1 Error Analysis As a summary, all measured results are listed in Table 5.1 to make a comparison with the desired specifications. Obviously, only the operating frequency and the noise figure totally satisfies the specification. Other parameters have certain errors. Table 5.1: Results comparison Specifications Design value Measured value Operating Frequency MHz MHz Gain db db Noise Figure 3.35 db db Image Rejection 20 db db Quiescent Point I C =20 ma I C =21.59 ma V CE =6 V V CE =5.6 V Two possible quiescent point errors are from transistor gain and bias resistors. The typical DC current gain h FE of transistor is available on the data sheet. But the actual value should be tested with the test PCB for transistor measurement in Lab 2 of radio electronics course [10]. The inaccuracy of nominal bias resistor value will enlarge error further. The gain error is mainly from losses in the band-pass filter. To minimize design error, the measurement of S-parameter may be done before the design. The S-parameters only above 100 MHz are available in the data sheet. The 3rd order band-pass filter used for image rejection is not very competent for the requirement. From previous simulation, at least a 4th order filter is desirable. However, this increase the complexity of adjustment work.

24 18 Conclusions 5.2 Conclusions As it can be seen by previous comparison, most of electrical requirements are fulfilled. There were no big problems during the simulations or testing except insufficient consideration for stray parameters. For one thing, the center frequency of band-pass filter seriously deviate from 98 MHz because of parasitic element, which failed to appear with schematic simulation. An effective solution is to use PCB simulation environment in ADS to evaluate the parasitic effects. For another, the output matching network also is affected by parasitic parameters. In fact the band-pass filter without variable components is not very suitable for selective RF applications. A worth alternative is to use more compact architecture which combine the output matching network and the image rejection filter. Not only does it has a better frequency selectivity, but also it save some components. Actually, Göran Jönsson has introduced this circuit at the beginning of this course. I hope we can try this structure in future.

25 Bibliography [1] Göran Jönsson. Slides for Radio Course. Department of Electrical and Information Technology, Lund University, [2] P.E. Paul H. Young. Electronic Communication Techniques. Prentice Hall, fifth edition, August 17, [3] Joseph J. Carr. Secrets of RF Circuit Design. McGraw-Hill, 3rd edition, [4] BFR520 NPN 9 GHz wideband transistor Data Sheet. Philips semiconductors. download/datasheets/bfr520 3.pdf, September [5] L. Sundström, G. Jönsson, and H. Börjeson. Radio Electronics. Department of Electroscience, Lund University, first edition, [6] Better soldering. solder.htm. [7] Collin J. McKinney. Soldering techniques. UNC Department of Chemistry, fredb/techresource/docs/soldering.pdf, [8] Clyde F. Coombs. Printed Circuits Handbook. McGraw-Hill Professional, 6th edition, [9] ZVR / ZVRE / ZVRL. Vector network analyzer operating manual. ROHDE & SCHWARZIEICE, pages [10] L. Sundström, G. Jönsson, and H. Börjeson. Radio Electronics-Exercises and Laboratory Experiments. Department of Electroscience, Lund University, first edition, 2004.