VIEWS: 7 PAGES: 36 POSTED ON: 9/20/2011
IF Receiver for Wideband Digitally Modulated Signals Oren Avraham Direct Instructor: Doctor Ronen Holtzman, Microwave Division, Elisra Electronic Systems Ltd. Supervising Instructor: Professor Raphael Kastner, Department of Physical Electronics, School of Electrical Engineering, Tel Aviv University 1 Presentation Outline Objectives Requirements Block Diagrams Processing Features Research and Reading Subjects Results and Conclusions 2 Primary Objectives I. Design a receiver for wideband digitally modulated signals: Analog processing: full circuit design Digital processing: algorithm principles Software definition (including Gain and NF evaluation for BIT) II. Perform the required research and literary reading on various aspects of the analog and digital processing to assure minimal degradation in signal quality, among which are the following: Analog to Digital Converters and their proper use in the integrated system. Calculation of Sampling Clock Aperture Jitter and its effect on ADC SNR. III. Realize and test a working PCB using SMT components, which performs the analog processing section. 3 System Block Diagram IF Section IF Analog to Digital Input Converter (Analog Processing) FPGAs Optical Optic Digital Transceiver (Digital Processing) Output 4 Main Requirements Input / Output Frequencies: 70 MHz 140 MHz 4 Selectable Bandwidths per Center Frequency SNR degradation (IF Cascade & ADC): 0.1 dB, maximum Maximal Dynamic Range @ 100 KHz Resolution : 2nd Order: 70 dB, minimum 3rd Order: 70 dB, minimum Gain Control Dynamic Range: AGC: 30 dB MGC: 30 dB Output Power Level: Should be chosen to best utilize the ADC’s Dynamic Range. 5 Analog Processing: Block Diagram Pre Selector Filter bank ByPass 70/5 70/10 70/28 IF Input Thermopad/ATT 70/40 AT-117 140/10 20 dB HMC484 DCA RF2360 Power Overload 140/20 Detection Threshold HMC253 140/40 Noise Generator 140/56 IF Comp DET AD8309 μC Control A/D Anti Aliasing Filter RVA2500 IF Thermopad LPF 90 Output VVA 20 dB 140/56 20 dB Integrator AGC Feedback IF Op AD8309 AGC/MGC DET Monitor Amp Output μC D/A A/D μC A/D Control 6 Analog Processing: Main Features Automatic Gain Control Manual Gain Control Sub-Octave Filtration Anti-Aliasing Filtration Noise Injection for BIT Purposes 7 AGC/MGC (1): Circuit RVA2500 20 dB AGC Coupler AGC VVA IF Input IF Output VVA Control Voltage C Analog Operational R IF Amplifier RSSI AGC - Detector (AD8309) + MGC/ Target Digital AGC μC μC D/A A/D A/D PC (Control Software) 8 AGC/MGC (2): IF Detector Log Amp RSSI Versus IF Signal Power at Cascade Output 1.3 1.2 1.1 1 RSSI [Volt] 0.9 0.8 0.7 0.6 0.5 0.4 -40 -35 -30 -25 -20 -15 -10 -5 0 5 Output Power [dBm] 9 AGC/MGC (3): VVA VVA Attenuation Versus Control Voltage 70 60 50 Attenuation [dB] 40 30 20 10 0 0 2 4 6 8 10 12 14 Control Voltage [Volt] 10 Sub-Octave Filtration Performed by the Pre-Selector Filter Bank Improves the effective IP2 by at least 20 dB Composed of Eight Band Pass Filters: 1. 70/5 MHz 2. 70/10 MHz 3. 70/28 MHz 4. 70/40 MHz 5. 140/10 MHz 6. 140/20 MHz 7. 140/40 MHz 8. 140/56 MHz A Bypass channel is included for scenarios in which an extremely low Group Delay Variation is required. 11 Anti-Aliasing Filtration Performed by two filters: 1. LPF 90 MHz 2. BPF 140/56 MHz Designed for a Sampling Frequency of 196.608 MHz (48th multiple of an E1 rate). Replica Rejection for worst case scenario: 1. LPF 90 MHz: 70 dBc 2. BPF 140/56 MHz: 55 dBc 12 Built In Test A Noise Generator (ENR=30 dB) is used for Noise Injection ADC Samples are used as observations and two noise power levels are computed: 1. N1 – Natural Thermal Noise at Cascade Input (-174 dBm/Hz) 2. N2 – Generator Noise at Cascade Input (-144 dBm/Hz) Using the following equations the IF cascade’s Gain and Noise Figure are evaluated: N [W ] N 2 [W ] N 1 [W ] N [W ] 1 NF [dB] ENR 10 log 2 G[dB] 10 log 1 (T T ) k B h o The test scans all signal channels and produces a Pass/Fail report. 13 Digital Processing: Block Diagram 1st FPGA Decimation IQ Distributed Conversion Arithmetic FIRs Downsampling IF Cascade 12-bit ADC ↓D Anti-sinc Filter Output @ 196.608 MHz X2 Frequency Multiplier NCO 98.304 MHz Reference Clock 2nd FPGA P/S RocketIO Optic TX/RX Receiver Interface Output 14 Simplified View of Sampled Spectrum Sampled Spectrum for a 70/40 MHz Analog Input Signal (shown for 0<f<f ) s 1 0.8 Magnitude 0.6 0.4 0.2 0 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 Frequency [MHz] Sampled Spectrum for a 140/56 MHz Analog Input Signal (shown for 0<f<f ) s 1 0.8 Magnitude 0.6 0.4 0.2 0 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 Frequency [MHz] 15 Research and Literary Reading Subjects: ADC-Related System Design Considerations The complete Receiver functions as an Integrated Analog- Digital System. A thorough understanding of the ADC’s effects on the integrated system’s performance is required. This is performed by characterizing the ADC in “RF/IF terms” and designing the IF cascade accordingly. One of the main questions which arise is the following: What is the optimal power of the analog signal at the ADC’s input? 16 Input Power to ADC – Introduction (1) It is a wide spread notion that in order to best utilize the ADC’s Dynamic Range (bits), the analog signal’s power at the ADC’s input should be as high as possible, putting it very close to Full Scale Power (FSP). This yields very demanding Gain and Linearity requirements of the RF cascade preceding the ADC, and causes an inevitable degradation of signal integrity (higher gain means higher intermodulation products, harmonics, likelihood of compression and so forth). Furthermore, the risk of ADC clipping (when the analog signal power exceeds its FSP) runs very high. 17 Input Power to ADC – Introduction (2) Our hypothesis is that it is not the analog signal’s power but rather the amplified Thermal Noise of the RF/IF cascade at the ADC’s input which determines the degradation in system noise performance caused by the ADC. This is based on the fact that ADC noise (Thermal, Quantization and Jitter Induced) can be referred to as white noise. This white noise is added to the amplified thermal noise (which is white is as well) at the ADC’s input. Since both noises are white and statistically independent, the result is their power summation. Therefore, we can calculate the ADC’s Effective Noise Figure. 18 Input Power to ADC: ADC Effective NF ADC FSP [dBm] Typically 1 dB below FSP [-1 dBFS] SNR (Integrated Over Entire Nyquist Bandwidth) [dB] f sampling 10 log 2 ADC Noise Density [dBm/Hz] ADC Effective NF [dB] Thermal Noise Density (kTB=-174 dBm/Hz ) 19 Input Power to ADC – OG (1) In order to perform the power summation of Thermal Noise and ADC Noise in a more intuitive manner, we use the Over Gain approximation: Amplified Thermal ADC Noise: NADC Noise: NRF/IF Natural Thermal RF/IF Combined ADC Noise: -174 dBm/Hz Cascade Noise GRF/IF NFADC NFRF/IF 20 Input Power to ADC – OG (2) The different Noise Power Levels are: OG Δ(OG) N ADC 174 NFADC [dB] [dB] N RF / IF 174 GRF / IF NFRF / IF <-20 ≈ |OG| OG [dB] N RF / IF N ADC GRF / IF NFRF / IF NFADC -20 20.04 N Combined N RF / IF N ADC N RF / IF OG -15 15.1 -10 10.4 OG Summation -6 7 Combined Noise [dBm] 0 3 Δ(OG) [dB] 6 1 Amplified Thermal Noise: NRF/IF [dBm] 10 0.4 15 0.1 Over Gain (OG) [dB] 20 0.04 >20 ≈0 ADC Noise: NADC [dBm] 21 Input Power to ADC – OG (3) SNR Therefore, we can predict the ADC- OG Degradation Induced SNR Degradation based on [dB] [dB] the OG table. <-20 ≈ |OG| We note that for an OG of more than -20 20.04 15 dB, this degradation is negligible. -15 15.1 From this, we would derive the Gain -10 10.4 Requirement of the RF/IF Cascade -6 7 (for a given NF): 0 3 6 1 GRF / IF 15 NFADC NFRF / IF 10 0.4 15 0.1 We verified our hypothesis by 20 0.04 simulation, as presented in the >20 ≈0 following slides. 22 Input Power to ADC – Simulation (1) High OG (20 dB), -50 dBFS Input Sine Wave: Analog Signal at ADC Input Digitized Signal at ADC Output 0 0 Analog Signal ADC Noise Floor Magnitude [dBFS] Magnitude [dBFS] -50 -50 -100 -100 -150 -150 0 10 20 30 40 0 10 20 30 40 Frequency [MHz] Frequency [MHz] ADC Simulation (Single Plot) 0 Analog Signal at ADC Input Digitized Signal at ADC Output Magnitude [dBFS] -50 -100 -150 0 5 10 15 20 25 30 35 40 45 Frequency [MHz] 23 Input Power to ADC – Simulation (2) Low OG (-10 dB), -1 dBFS Input Sine Wave: Analog Signal at ADC Input Digitized Signal at ADC Output 0 0 Analog Signal ADC Noise Floor Magnitude [dBFS] Magnitude [dBFS] -50 -50 -100 -100 -150 -150 0 10 20 30 40 0 10 20 30 40 Frequency [MHz] Frequency [MHz] ADC Simulation (Single Plot) 0 Analog Signal at ADC Input Digitized Signal at ADC Output Magnitude [dBFS] -50 -100 -150 0 5 10 15 20 25 30 35 40 45 Frequency [MHz] 24 Input Power to ADC – Simulation (3) High OG (15 dB), -90 dBFS Input Sine Wave: Analog Signal at ADC Input Digitized Signal at ADC Output 0 0 Analog Signal ADC Noise Floor -50 -50 Magnitude [dBFS] Magnitude [dBFS] -100 -100 -150 -150 0 10 20 30 40 0 10 20 30 40 Frequency [MHz] Frequency [MHz] ADC Simulation (Single Plot) 0 Analog Signal at ADC Input Digitized Signal at ADC Output -50 Magnitude [dBFS] -100 -150 0 5 10 15 20 25 30 35 40 45 Frequency [MHz] 25 Input Power to ADC – Simulation (4) OG Sweep (-30 dB to +30 dB), showing that our hypothesis coincides with the simulation results: SNR Degradation Caused by the ADC Versus Over Gain 30 Theoretic Simulation 25 20 SNR Degradation [dB] 15 10 5 0 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 Over Gain [dB] 26 ADC: Effective IP3 IP3 is the 3rd order Intercept Point, which corresponds to the 3rd order Intermodulation Products of a non-linear analog (RF) component. One of the ADC’s non-linearity parameters is the “Two-Tone Intermodulation Distortion Rejection” (IMDR). We “translate” the ADC’s IMDR to RF/IF terms (IP3) using the following relation: 1 IP 3 ADC [dBm] Pin [dBm] IMDR [dB] 2 Pin [dBm] Pin [dBFS ] FSP [dBm] 27 Effect of Sampling Clock Aperture Jitter on ADC SNR The Effective SNR of an ADC is comprised of several noise sources: ADC DNL RMS Aperture Jitter (Differential Thermal Noise of Sampling Clock Non Linearity) in LSBs 2 2 1 2 VTh , rms 2 SNRADC 20 log 2 f signal rms N N 2 3 2 2 Analog Signal ADC Resolution Frequency (Number of bits) 28 Calculation of Sampling Clock RMS Aperture Jitter (1) The common approach to calculate the RMS Aperture Jitter of a frequency source (such as a sampling clock) is to integrate its Phase Noise as is, and simply translate the result (received in radians) to temporal terms (seconds): fs 2 L f df rms fs 0 29 Calculation of Sampling Clock RMS Aperture Jitter (2) We apply a more modern approach which incorporates a sine function factor, as follows: Frequency Offset Oscillator Phase Noise From Center Frequency (After BPF) fs f 2 2 L f sin df rms 2 f fs s 0 Oscillator (Center) Frequency 30 Calculation of Sampling Clock RMS Aperture Jitter (3) The sine factor attenuates the Phase Noise close to the oscillator center frequency, as depicted in the following figure: sin(/2f/f 0)2 0 -50 Attenuation [dB] -100 -150 -8 -7 -6 -5 -4 -3 -2 -1 0 10 10 10 10 10 10 10 10 10 Normalized Frequency Offset (f/fo) 31 Calculation of Sampling Clock RMS Aperture Jitter (4) At frequencies that are far from the center, the Phase Noise is attenuated by the Clock’s BPF: Filter Response 0 -20 Center Frequency (fs=98.304 MHz) -40 2fs -60 -80 50 100 150 200 Frequency (MHz) 32 Calculation of Sampling Clock RMS Aperture Jitter (5) Phase Noise -60 Before BPF After BPF -80 Magnitude [dBc/Hz] -100 -120 -140 -160 0 1 2 3 4 5 6 7 10 10 10 10 10 10 10 10 Frequency Offset [Hz] 33 Results (1): At 70 MHz Measurement Parameter Requirement 2nd Amp 2nd Amp Enabled Disabled Noise Figure, maximum [dB] 9.5 5.2 8.5 Expected SNR Degradation, 0.1 <10-3 0.085 maximum [dB] DR2 @ 100 KHz 70 76.9 85.25 BW, minimum [dB] DR3 @ 100 KHz 70 72.2 84 BW, minimum [dB] 34 Results (2): At 140 MHz Measurement Parameter Requirement 2nd Amp 2nd Amp Enabled Disabled Noise Figure, maximum [dB] 9.5 5.5 9 Expected SNR Degradation, 0.1 <10-3 0.095 maximum [dB] DR2 @ 100 KHz 70 77 85.25 BW, minimum [dB] DR3 @ 100 KHz 70 72 83.65 BW, minimum [dB] 35 Conclusions I. The Subject of ADC Integration with RF/IF Cascades was explored, reaching the following conclusions: The ADC-Induced SNR Degradation is determined by the level of thermal noise at the ADC’s input. Even sub-LSB signals can be detected by the ADC. The Over Gain approximation proved to be a good method to determine the required Gain of the RF/IF cascade and predict the SNR Degradation. II. The Analog (IF) section of the receiver was designed, built and measured showing electrical performance surpassing the requirements. 36