HDM8513A Users Manual DVB/DSS Compliant Receiver Nov. 2000 Revision 1.0 Electronics Industries Co., Ltd. 1 Direct Broadcast Satellite (DBS) has been one of the most successful new product introductions in the history of consumer electronics. This product represents the first application of digital video compression for broadcast television. Originally intended to provide cable quality television services to remote areas, this product is now offering a competitive replacement to cable services in many urban areas. The first operational systems employ closed proprietary signaling structures. The European Broadcasting Union (EBU) has developed the first open standard (DVB-S) for DBS services. The broadcasting community has embraced this standard which is now being adopted for new systems throughout the world. This widely accepted open standard is essential for DBS to achieve full market potential. The HDM8513ATM is a fully DVB-S&DSS compliant ADC/QPSK demodulator/FEC device which provides an MPEG-2 stream to be processed by the conditional access and video decompression circuits. The demodulator clocked with a fixed frequency is true variable rate over the range of 1 to 55M symbols-per-second. This product achieves the highest performance and flexibility. It minimizes the cost of external circuits, thus reducing overall system cost. 2 Hyundai Electronics Ind. Co., Ltd reserves the right to make changes to its products or specifications to improve performance, reliability, or manufacturability. Information furnished by Hyundai Electronics Ind. Co., Ltd is believed to be accurate and reliable. However, no responsibility is assumed by Hyundai Electronics Ind. Co., Ltd for its use; nor for any infringement of patents or other rights of third parties which may result from its use. No license is granted by its implication or otherwise under any patent rights of Hyundai Electronics Ind. Co., Ltd. For more information contact: Address: Youngdong Bldg. 891, Daechi-dong, Kangnam-gu, Seoul, 135-738, Korea Tel: 82-2-3459-3188 Fax: 82-2-3459-5843 E-mail: [email protected] 3 TABLE OF CONTENTS 1. INTRODUCTION TO THE HDM8513A................................................................................................................6 1.1 FEATURES AND BENEFITS..................................................................................................................................7 2. HARDWARE SPECIFICATION..............................................................................................................................8 3. TECHNICAL OVERVIEW..................................................................................................................................... 18 3.1 DUAL CHANNEL A NALOG TO DIGITAL CONVERTER .................................................................................. 18 3.2 VARIABLE RATE DEMODULATOR.................................................................................................................. 20 3.3 NOISE MEASUREMENT CIRCUIT .....................................................................................................................22 3.4 VITERBI DECODER.............................................................................................................................................24 3.5 A UTONOMOUS A CQUISITION ..........................................................................................................................25 3.6 REED SOLOMON DECODER.............................................................................................................................. 27 3.7 CLOCK GENERATION PLL.................................................................................................................................29 3.8 DBS RECEIVER................................................................................................................................................... 30 4. MECHANICAL SPECIFICATIONS..................................................................................................................... 31 4.1 100 PIN QUAD FLAT PACK ................................................................................................................................31 4.2 64 PIN THIN QUAD FLAT PACK........................................................................................................................33 4.3 RECOMMENDED A NALOG PIN CONNECTION............................................................................................... 35 4.4 RECOMMENDED CLOCK GENERATION CIRCUIT ...........................................................................................35 5. SIGNAL DESCRIPTION....................................................................................................................................... 36 5.1 INPUTS..................................................................................................................................................................36 5.2 OUTPUTS............................................................................................................................................................. 36 5.3 M ONITOR AND CONTROL INTERFACE ...........................................................................................................39 5.4 I2C MODE............................................................................................................................................................. 40 6. REGISTER DEFINITIONS..................................................................................................................................... 42 6.1 W RITE REGISTERS..............................................................................................................................................42 6.2 READ REGISTERS................................................................................................................................................55 APPENDIX.................................................................................................................................................................... 58 A1. A2. A3. A4. LOOP FILTER PROGRAMMING APPLICATION NOTE ................................................................................59 FALSE LOCK ESCAPE A PPLICATION NOTE .................................................................................................62 PERFORMANCE WITH INTERFERENCE .......................................................................................................... 63 NYQUIST CRITERIA CONSIDERATIONS......................................................................................................... 67 LIST OF FIGURES FIGURE 1: FIGURE 2: FIGURE 3: FIGURE 4: FIGURE 5: FIGURE 6: 4 TOP LEVEL BLOCK DIAGRAM....................................................................................................................6 INPUT DATA TIMING DIAGRAM ...............................................................................................................9 INTEL 80C88A READ TIMING DIAGRAM ............................................................................................... 10 INTEL 80C88A WRITE TIMING DIAGRAM............................................................................................. 11 INTEL 8051 READ TIMING DIAGRAM .....................................................................................................12 INTEL 8051 WRITE TIMING DIAGRAM...................................................................................................13 FIGURE 7: MOTOROLA READ TIMING DIAGRAM....................................................................................................14 FIGURE 8: MOTOROLA W RITE TIMING DIAGRAM .................................................................................................15 FIGURE 9: OUTPUT TIMING DIAGRAM FOR NORMAL PARALLEL ....................................................................... 16 FIGURE 10: OUTPUT TIMING DIAGRAM FOR NORMAL SERIAL ...........................................................................16 FIGURE 11: OUTPUT TIMING DIAGRAM FOR REGULATED PARALLEL............................................................... 17 FIGURE 12: OUTPUT TIMING DIAGRAM FOR REGULATED SERIAL MODE1.......................................................17 FIGURE 13: OUTPUT TIMING DIAGRAM FOR REGULATED SERIAL MODE2.......................................................17 FIGURE 14: ADC BLOCK DIAGRAM............................................................................................................................ 19 FIGURE 15: DEMODULATOR BLOCK DIAGRAM....................................................................................................... 20 FIGURE 16: NOISE M EASUREMENT CIRCUIT ...........................................................................................................22 FIGURE 17: NOISE A CCUMULATOR AS A FUNCTION OF SNR AND TIME............................................................ 23 FIGURE 18: VITERBI DECODER...................................................................................................................................24 FIGURE 19: REED SOLOMON DECODER.................................................................................................................... 28 FIGURE 20: CLOCK SIGNAL GENERATION................................................................................................................29 FIGURE 21: TYPICAL SET TOP BOX DEMODULATOR............................................................................................ 30 FIGURE 22: M ECHANICAL CONFIGURATION ...........................................................................................................32 FIGURE 23: M ECHANICAL CONFIGURATION ...........................................................................................................34 FIGURE 24: A NALOG PIN CONNECTION .................................................................................................................... 35 FIGURE 25: CLOCK GENERATION CIRCUIT ..........................................................................................................35 FIGURE 26: I2C WRITE TO THE HDM8513A ...........................................................................................................40 FIGURE 27: I2C READ FROM THE HDM8513A......................................................................................................... 41 FIGURE A1: SYMBOL TIMING RECOVERY TRANSIENT RESPONSE....................................................................... 59 FIGURE A2: CARRIER PHASE RECOVERY TRANSIENT RESPONSE ........................................................................ 60 FIGURE A3: CARRIER PHASE RECOVERY TRANSIENT RESPONSE WITH LOW SNR ..........................................61 FIGURE A4: A DJACENT CHANNEL INTERFERENCE OF 10 DB, 1.35 SPACING.................................................... 64 FIGURE A5: PERFORMANCE WITH INTERFERER AT DIFFERENT CARRIER SPACINGS .....................................65 FIGURE A6: PERFORMANCE WITH +10 DB INTERFERER......................................................................................66 LIST OF TABLES TABLE 1: ABSOLUTE M AXIMUM RATINGS ...............................................................................................................8 TABLE 2: DC CHARACTERISTICS.................................................................................................................................8 TABLE 3: DEMODULATOR SPECIFICATIONS.............................................................................................................9 TABLE 4: AC CHARACTERISTICS.................................................................................................................................9 TABLE 5: INTEL 80C88A READ CYCLE TIMING PARAMETERS (BUSMODE = 1)................................................10 TABLE 6: INTEL 80C88A WRITE CYCLE TIMING PARAMETERS (BUSMODE = 1) .............................................11 TABLE 7: INTEL 8051 READ CYCLE TIMING PARAMETERS (BUSMODE = 1)......................................................12 TABLE 8: INTEL 8051 WRITE CYCLE TIMING PARAMETERS (BUSMODE = 1)................................................... 13 TABLE 9: M OTOROLA READ CYCLE TIMING PARAMETERS (BUSMODE =0).................................................... 14 TABLE 10: M OTOROLA W RITE CYCLE TIMING PARAMETERS (BUSMODE =0).................................................15 TABLE 11: OUTPUT TIMING....................................................................................................................................... 16 TABLE 12: EXAMPLE OF ACQUISITION TIMING .....................................................................................................26 TABLE 13: I2C SLAVE A DDRESS..................................................................................................................................41 5 1. Introduction to the HDM8513A The HDM8513A digital demodulator for direct broadcast satellite receivers is a single chip solution fully compliant with the European Telecommunications Standards Institute (ETSI) specification ETS 300 421. This chip integrates an A/D converter, a variable rate matched filter, a variable rate QPSK demodulator with a Viterbi decoder, a deinterleaver and a Reed Solomon decoder. The HDM8513A, which is implemented in a 0.35 micron CMOS, Triple Layer Metal Process, provides variable rate capability while operating with a fixed frequency sampling clock. Digital samples of baseband I and Q data are generated by an internal A/D converter, then provided to the demodulator at a fixed sample rate. The root raised cosine filter is implemented internally with fully digital techniques. Similarly, the symbol timing recovery and carrier phase tracking functions are performed entirely in the digital domain. This approach provides minimum constraints on external circuits, thus reducing overall system costs. The HDM8513A may be configured by an external processor for a specific symbol rate, and carrier frequency along with loop gain parameters. The HDM8513A provides an external AGC signal which is used to control the gain of the analog signal which is applied to the down-converters. And it also provides a digital AGC internally which controls the gain of the signal out of the matched filters. In addition, the HDM8513A provides fully programmable sweep circuitry to aid in initial acquisition when large frequency offsets may be present. The digital frequency translation capability of the HDM8513A permits this part to be used in frequency multiplexing applications. In this application, an entire transponder bandwidth containing many signals is sampled at a fixed rate. The digital oscillator within the HDM8513A is programmed to the specific desired carrier frequency within that band to permit the selected signal to be passed through the baseband filter and processed by the demodulator circuits. Fixed Rate Sample Clock I Q Symbol Clock 6 A/D Converter 6 Variable Rate QPSK Demodulator 3 3 Data Clock Viterbi Bit Clock Viterbi Decoder Synchronization and Deinterleaving 8 Reed Solomon Decoder 8 Data Frame Sync QPSK Lock Node Sync Viterbi Data FIGURE 1: TOP LEVEL BLOCK DIAGRAM 6 1.1 Features and Benefits * Fully DVB&DSS compliant * Dual 6bit A/D converters * Continuously variable symbol rate from 1Msps to 55Msps (75MHz clock) * Internal digital root raised cosine filter * Less than 0.5 dB implementation loss * Frequency multiplexing capability * Automated frequency search * Internal bias cancellation * Both wideband and narrowband AGC * Noise calibration for antenna steering * Output data rate as high as 68Mbps * Fixed frequency sampling clock * Simple interface with tuner and analog processing * Microcontroller interface * Eight bit parallel or I2C monitor and control interface * I2C by-pass mode * Two package types Part code HDM8513AP HDM8513AT Package 100PQFP 64TQFP 7 2. Hardware Specification Table 1: Absolute Maximum Ratings Rating Ambient Temperature under Bias Storage Temperature Ambient Humidity under Bias Thermal Resistance(Ja) Junction Temperature Voltage on Any Pin VDD, IOVDD Package Material Value -10 to 70 -65 to 150 85(85 c ,500hrs) 45 150 Vss - 0.3V to V DD + 0.5V 5.5 - Compound : CEL-4630SX - Lead Frame : Copper Unit c c % c/W c V V Table 2: DC Characteristics Symbol IDD IOVDD VDD V VIL VIH VOL VOH IIH IIL CIN COUT Parameter Dynamic Current (Power Supply Current) Interface Power Supply Voltage Core Power Supply Voltage ADC Power Supply Voltage Input Low Voltage Input High Voltage Output Low Voltage Output High Voltage Input High Current Input Low Current Input Capacitance (analog pad) Output Capacitance Min. - Max. 395 Units mA 3 3.6 V Test Conditions VDD=3.3, Freq=60Mhz (Typical 367mA) Normal Operation 3 3.6 V Normal Operation 3 3.6 V Normal Operation 0 0.7V DD 0.3V DD VDD+ 0.5 0.4 10 10 10 10 10 V V 2.4 -10 -10 - V V uA uA pF pF IOL = 4 mA IOH = 4 mA VIL = VDD VDD = 3.6, VIL =0.5 Typical 8.76pF (8.52pF) Typical 8.83pF (IOVDD and VDD = 3.3V+ or - 5%, TA = 0 to 70 c, unless otherwise specified) 8 Table 3: Demodulator Specifications Parameter Sampling Clock Frequency Analog Input Full Scale Range Symbol Rate Viterbi Data Rate Reed Solomon Data Rate Implementation Loss Symbol Rate Resolution Carrier Frequency Resolution Acquisition Sweep Range Min. 1MHz 0.9 Vpp 1Msps 20 Clock/(2 ) 20 Clock/(2 ) - Max. 75MHz 1.1 Vpp 55Msps 75Mbps 69Mbps 0.5 dB + or - Clock/2 Table 4: AC Characteristics Symbol tsu1 Parameter Input Data Setup before Clock Min. 6 Max. - Unit ns th1 Input Data Hold after Clock 2 - ns tpw1 Low Pulse Width of Clock 8.7 - ns tpw2 High Pulse Width of Clock 8.1 - ns tpw1 CLOCK tpw2 tsu1 th1 I_IN [5:0] or Q_IN [5:0] FIGURE 2: INPUT DATA TIMING DIAGRAM 9 Table 5: Intel 80C88A Read Cycle Timing Parameters (Busmode = 1) Symbol tsu1 Parameter Input Address and /CE Setup before /RE Inactive Min. 35 Max. - Unit ns 5 - ns th1 Input Address and /CE Hold after /RE Inactive tpw1 td1 /RE Low Duration 200 - ns Delay from /CE to DTACK Active - 35 ns tdoz1 Delay from /RE Inactive to DTACK in Tristate Mode - 10 ns tdoz2 Delay from /RE Inactive to HI_DATA [7:0] Tristate Mode 10 - ns Valid HI_ADDR [4:0] /CE t pw1 /RE td1 th1 Z Z DTACK tdoz1 HI_DATA[7:0] tsu1 tdoz2 FIGURE 3: INTEL 80C88A READ TIMING DIAGRAM Note: HI_ADDR[4:0] is derived from the processor(80C88A) A15-A8 bus and HI_DATA[7:0] is connected to the AD7 - AD0 bus. #This page is only for HDM8513AP. 10 Table 6: Intel 80C88A Write Cycle Timing Parameters (Busmode = 1) Symbol tsu1 th1 Parameter Input Data Setup before /WE Inactive Input Address, Data and /CE Hold after /WE Inactive Min. 20 Max. - Unit ns 8 - ns 200 - ns tpw1 /WE Low Duration td1 tdoz1 Delay from /CE to DTACK Active - 35 ns Delay from /WE Inactive to DTACK in Tristate Mode - 15 ns Valid HI_ADDR [4:0] /CE tpw1 /WE td1 th1 DTACK tdoz1 HI_DATA[7:0] tsu1 FIGURE 4: INTEL 80C88A W RITE TIMING DIAGRAM Note: HI_ADDR[4:0] is derived from the processor(80C88A) A15-A8 bus and HI_DATA[7:0] is connected to the AD7 - AD0 bus. #This page is only for HDM8513AP. 11 Table 7: Intel 8051 Read Cycle Timing Parameters (Busmode = 1) Symbol tsu1 th1 Parameter Input Address Setup before /CE Active Input Address and /CE Hold after /RE Inactive tpw1 /RE Active Duration tpd1 Delay from /RE Active to HI_DATA [7:0] Valid tdoz1 Delay from /RE Inactive to HI_DATA[7:0] Tristate Mode Valid HI_ADDR [4:0] th1 tsu1 /CE tpw1 /RE HI_DATA[7:0] tpd1 tdoz1 FIGURE 5: INTEL 8051 READ TIMING DIAGRAM #This page is only for HDM8513AP. 12 Min. 5 Max. - Unit ns 5 - ns 400 - ns - 40 ns 10 - ns Table 8: Intel 8051 Write Cycle Timing Parameters (Busmode = 1) Symbol tsu1 Parameter Input Address and Data Setup before /WE Active Min. 5 Max. - Unit ns 5 - ns 400 - ns th1 Input Address and Data Hold after /WE Inactive tpw1 tsu2 /WE Active Duration /CE Setup to /WE Active 5 - ns th2 /CE Hold after /WE Inactive 5 - ns Valid HI_ADDR [4:0] tsu1 t h1 tsu2 th2 /CE t pw1 /WE Valid HI_DATA[7:0] FIGURE 6: INTEL 8051 WRITE TIMING DIAGRAM #This page is only for HDM8513AP. 13 Table 9: Motorola Read Cycle Timing Parameters (Busmode =0) Symbol tsu1 Parameter Setup Time of R/W with respect to /CE Active Min. 5 Max. - Unit ns tsu2 td1 Address Setup with respect to /DS Active 5 - ns Delay from DTACK Active to Data Valid - 30 ns th1 R/W Hold with respect to /DS Inactive 5 - ns th2 Address Hold with respect to /DS Inactive 5 - ns th3 Data Hold with respect to /DS Inactive 10 - ns HI_ADDR[4:0] Valid tsu2 th2 /CE /DS th1 tsu1 R/W td1 DTACK th3 HI_DATA[7:0] FIGURE 7: MOTOROLA READ TIMING DIAGRAM Note: External pull-up resistor is required on DTACK. #This page is only for HDM8513AP. 14 Table 10: Motorola Write Cycle Timing Parameters (Busmode =0) Symbol tsu1 Parameter Data Setup to /DS Active Min. 5 Max. - Unit ns tsu2 td1 R/W Setup to /CS and Address 3 - ns /DS Delay from R/W 5 - ns td2 DTACK Delay from /DS Active - 40 ns td3 DTACK Delay from /DS Inactive - 10 ns tpw1 th1 /DS Active Duration 5 - ns Address, /CS and R/W Hold from /DS Inactive 5 - ns th2 Data Hold from /DS Inactive 5 - ns Valid HI_ADDR[4:0] /CS tsu2 tpw1 /DS td1 th1 R/W td2 DTACK HI_DATA[7:0] td3 tsu1 th2 Valid FIGURE 8: MOTOROLA W RITE TIMING DIAGRAM Note: External pull up resistor is required on DTACK. #This page is only for HDM8513AP. 15 Table 11: Output Timing Symbol tsu thd Parameter Output Data Setup before DATA_CLK and DATA_STB Min. 5 Max. - Unit ns 10 - ns Output Data Hold after DATA_CLK and DATA_STB t hd t su DATA_CLK DATA_STB FRAME_SYNC DATA_VALID n-3 DATA n-2 n-1 n xx xx xx xx xx xx xx xx xx 1 2 3 FIGURE 9: OUTPUT TIMING DIAGRAM FOR NORMAL PARALLEL t hd t su DATA_CLK DATA_STB FRAME_SYNC DATA_VALID DATA[0] 8n-8 xx xx 8n-7 8n-6 8n-5 8n-4 xx 8n-3 8n-2 8n-1 8n xx 1 2 3 FIGURE 10: OUTPUT TIMING DIAGRAM FOR NORMAL SERIAL NOTE : In case of DVB, n is 188 In case of DSS, n is 144 16 4 4 t hd t su DATA_CLK DATA_STB FRAME_SYNC DATA_VALID DATA n-3 n-2 n-1 n xx xx xx xx xx xx xx xx xx xx xx xx xx xx xx xx 1 2 3 4 FIGURE 11: OUTPUT TIMING DIAGRAM FOR REGULATED PARALLEL t hd t su DATA_CLK DATA_STB FRAME_SYNC DATA [0] 8n-8 xx xx 8n-7 8n-68n-5 8n-4 xx 8n-3 8n-28n-1 8n xx xx xx xx xx xx xx xx 1 2 3 4 FIGURE 12: OUTPUT TIMING DIAGRAM FOR REGULATED SERIAL MODE1 t hd t su DATA_CLK FRAME_SYNC DATA_STB DATA [0] 8n-8 xx xx 8n-7 8n-6 8n-58n-4 xx 8n-3 8n-28n-1 8n xx xx xx xx xx xx xx xx 1 2 3 4 FIGURE 13: OUTPUT TIMING DIAGRAM FOR REGULATED SERIAL MODE2 NOTE : In case of DVB, n is 188 In case of DSS, n is 144 17 3. Technical Overview 3.1 Dual Channel Analog to Digital Converter The block diagram shown below illustrates internal configuration of the Dual Channel ADC. Baseband signals, in-phase(I) and quadrature phase(Q), which are generated by down converters, are applied to the dual channel ADC and quantized to 6-bit digital codes respectively. The ADC is optimized to allow AC coupled inputs with full scale input range of 1V + or - 10%. An LSB weight is approximately 15.6 mV. The full scale input analog conversion range (Vpp) is determined by the voltages of VTOP and VBOT and simply equal to (VTOP - VBOT). The full scale range is defined as the voltage range that accommodates 63 codes of equally spaced LSBs. Also the ADC supplies its own reference voltages for A/D conversions. The voltages can be monitored by external reference pins. The VTOP, VBOT represent top and bottom reference voltages respectively. REF_I, REF_Q represent middle reference voltages for each channel. All these 4 reference voltage pins should be by-passed to GND via 0.1uF capacitors. The values of internally generated voltage of VTOP and VBOT are 2.0V and 1.0V respectively. Vpp can be adjusted by externally applying voltages to both VTOP and VBOT pins respectively when different conversion ranges are necessary. VTOP can be adjusted as high as 2.3V and VBOT can be as low as 0.5V. A larger input range can be established by taking VTOP higher and VBOT lower than on-chip generated voltages. To supply necessary bias voltages for AC coupled applications, REF_I and REF_Q, which are middle reference voltages for I and Q channel, are connected to the analog input pins (AIN_I and AIN_Q ) respectively through 40 kohm resistors, as shown in the block diagram. For DC coupled applications, these voltages can be used to feed back offset compensation signals. To insure optimum performance, a low impedance analog ground plane is recommended and should be separated from other digital ground planes. The analog power supplies should be by passed at device to analog ground through 0.1uF ceramic capacitors. 18 VTOP AIN_I 6-bit ADC REF_I DI Ref. Voltage Gen. CLOCK AIN_Q 6 6-bit ADC 6 DQ REF_Q VBOT FIGURE 14: ADC BLOCK DIAGRAM 19 3.2 Variable Rate Demodulator The block diagram illustrates the overall configuration of the variable rate QPSK demodulator. Baseband in-phase (I) and quadrature (Q) inputs are applied to the demodulator at a fixed sampling rate. These digital samples are produced by A/D converters which employ AC coupling to minimize DC offset. Symbol Clock Phase Accumulator Is 6 Q Complex Multiplier 6 Symbol Tracking Loop Filter Polyphase Filters Symbol Timing Discriminator Narrowband AGC Io Qo s Sine Cosine Carrier Tracking Loop Filter Digital Oscillator Nominal Carrier Frequency Sweep Control Carrier Phase Discriminator Lock Detector Lock FIGURE 15: DEMODULATOR BLOCK DIAGRAM The carrier frequency error associated with these samples is removed digitally during tracking operations by a complex multiplier and a digitally controlled oscillator, sometimes called a numerically controlled oscillator (NCO). During initial acquisition, coarse frequency error is removed by a combination of the digital AGC within the HDM8513A and external analog tuning circuits. A polyphase filter performs the root raised cosine filtering of the frequency corrected baseband samples. This filter, which implements the function of equation (1), is always configured to have an impulse response duration of 4 symbols regardless of the programmed symbol rate. For low symbol rates, a large number of samples are used, while for high symbol rates a relatively low number of samples are processed for each filter output. The outputs of the polyphase filters are applied to a digital narrowband AGC which insures that the signal is optimally scaled to the Viterbi decoder to an accuracy of + or - 0.5 dB to insure optimum FEC performance. y[k] = Σ h[n] x[k-n] (1) In addition to optimizing performance of the Viterbi decoder, the digital narrowband AGC also insures that the performance of the symbol timing and carrier tracking loops is independent of 20 signal level variations. An analog wideband AGC is also employed to insure that the analog signal applied to the A/D converters is properly scaled. Both the symbol timing and carrier tracking loops are implemented digitally, which eliminates the need for external connections to analog tuning components during steady state operation. This causes the requirements on the analog presampling filter to be relaxed, permitting a lower cost analog front end. For systems which require a narrowband presampling filter, and have the potential for significant frequency error in the LNB (several MHz) the HDM8513A provides a high resolution measure of carrier frequency to permit periodic readjustment of the front end tuner frequency to compensate for drift. The host processor periodically reads the frequency register, then computes appropriate correction to the tuner frequency. The nominal symbol rate and the nominal carrier frequency are programmed into the demodulator to an accuracy provided by 20 bits of resolution, and the system accuracy is equivalent to that of the fixed frequency sampling clock. During initial acquisition, the HDM8513A provides an automated sweep program to facilitate carrier acquisition. The host processor loads a 20 bit register which determines the initial carrier frequency. A 16 bit register is programmed with the number of symbol times the receiver will dwell at each frequency. If the receiver remains at the initial frequency for the programmed number of symbol times without achieving lock, the carrier frequency is incremented by the step frequency value programmed into another 16 bit register. If no lock is achieved, the receiver will continue to increment the frequency until the maximum number of search frequencies, as determined by the value in an eight bit register, is achieved. When the maximum number of search frequencies is reached, the carrier frequency returns to the initial value and the entire process is repeated. Once the host processor determines that lock is achieved by observing the lock flag, it then inhibits the sweep function and programs loop bandwidth parameters which are optimized for steady state performance. 21 3.3 Noise Measurement Circuit When the DBS system is being installed in any place, the most difficult part of the installation is accurate pointing of the antenna toward the satellite. Inaccurate pointing results in loss of margin and greater potential for outages in adverse weather conditions. Existing systems use information from the demodulator forward error correction circuits to provide a measure of antenna pointing. Unfortunately, this method is useful over a range of only several dB above system threshold. The HDM8513A employs a unique circuit for accurate measure of signal strength over a 20 dB range of signal to noise ratio. This method, illustrated in the block diagram, makes use of the fact that the demodulator provides 8 bits of resolution for each of the quadrature output components. This high resolution provides a means of measuring the noise component with great accuracy. The eight bit in-phase demodulator filter output is detected by an absolute value circuit, then passed through an IIR to provide a measure of average signal amplitude. Each sample is then subtracted from this average amplitude to provide an instantaneous noise sample. The absolute value of these noise samples are then averaged by a second IIR to provide a measure of the noise which is roughly proportional to the noise power and inversely proportional to signal to noise ratio. Finally, the Figure 17 illustrates the results of simulations under different noise conditions. This figure illustrates that for signal-to-noise ratio as high as 19 dB, the noise measurement circuit provides a meaningful measure of signal power with worst case resolution of 1 dB. In-Phase Component 8 Absolute Value R 8 Average Magnitude 255 256 16 8 Average Deviation Absolute Value R 8 Instantaneous Deviation 255 256 FIGURE 16: N OISE MEASUREMENT CIRCUIT 22 16 FIGURE 17: NOISE A CCUMULATOR AS A FUNCTION OF SNR AND TIME 23 3.4 Viterbi Decoder The Viterbi decoder accepts 3 bit soft decision samples of the in-phase (I) and quadrature (Q) components of the received signal. Once QPSK lock has been achieved, the decoder searches for the correct code rate, starting with rate 3/4, then proceeding to rate 2/3, 5/6, 7/8 and finally rate 1/2. Each of the possible synchronization phases at each rate is tested as well as the two possible carrier phase ambiguity conditions. Polarity reversal is corrected in the word synchronization logic. Viterbi lock is achieved when the trellis traceback algorithm converges, on the average, within a prescribed number of symbols. Although the algorithm automatically tests for carrier phase ambiguity, there is no provision to automatically correct for phase reversal. Phase reversal can occur if the receiver chain, consisting of an LNB and the tuner, provides an odd number of high side frequency translation operations. A system may be required to operate with different LNBs, some of which provide phase reversal. This condition may be corrected by the host processor, which can set a bit in the down converter to correct for phase reversal. The Viterbi decoder employs the radix two algorithm. The output buffer reserializes the data which is made available, along with the Viterbi data clock as external signals. These signals permit verification of the DVB specification which is referenced to the Viterbi decoder output. ACS Array 128 Trace-back RAM Traceback Memory Controller I Q 3 3 G1 Depuncturing Logic G2 Branch Metric Calculator Decoder Quality Estimate Change Puncture Phase Change Carrier Phase FIGURE 18: VITERBI DECODER 24 Last-In First-Out Buffer Clock Out Data Out Viterbi Lock 3.5 Autonomous Acquisition The HDM8513A provides several features to permit signal acquistion with minimal interaction with the host microcontroller. The host microcontroller must configure the HDM8513A for a specific symbol rate, carrier frequency, carrier sweep conditions, and tracking loop bandwidth. The microcontroller also must monitor lock status to determine when acquisition is achieved. There are many provisions in the HDM8513A to enable the system designer to implement custom algorithms for specific requirements. The microcontroller first must set the lower edge of the carrier search range in the Carrier Frequency registers (04, 05 and 06). Then the processor configures the Carrier Sweep Step Size register (09, 0A) to a value which is less than two times the carrier pull-in range. The number of symbols per dwell is defined in registers (0B,0C), and is typically set to a value of 500 to 1000. The total search range is set by the Number of Search Frequencies as defined in register 0D. The total sweep frequency range is this number times the Carrier Sweep Step Size. The sweep process stops once QPSK carrier lock is detected. If no lock is detected, the sweep process continuously repeats. The QPSK demodulator may lock to any one of four different phase reference states, only one of which produces true I and Q data as it was modulated at the transmitter. If the local phase reference is plus 90 degrees or minus 90 degrees with respect to the true phase, the information provided to the Viterbi decoder will be unintelligible. If the Viterbi decoder is unable to achieve valid lock, it will reattempt lock with a 90 degree phase shift, without external intervention. In the event that the local phase is 180 degrees from the true phase, the data provided to the Viterbi decoder will be inverted, but otherwise valid. The code employed by the Viterbi decoder is transparent, thus the data from the Viterbi decoder will be inverted if the input is inverted. This situation is corrected in the word synchronization circuit. This circuit searches for the unscrambled sync word which occurs once per frame (every 204 bytes at the Viterbi output). Once correlation with the sync word is found, the data is reformatted as a series of bytes with the beginning of each 204 byte frame identified to provide the synchronization information required for the deinterleaver and the Reed Solomon decoder. If the polarity of the sync word is incorrect, the data is inverted before further processing without external interaction. The HDM8513A supports five different code rates, including 1/2, 2/3, 3/4, 5/6 and 7/8. When rate 1/2 is employed, there is a one-to-one correspondence between incoming I and Q samples and G1 and G2 terms required by the Viterbi decoder. The higher rates employ punctured coding techniques which periodically cause either a G1 or G2 term to be deleted. The puncturing pattern can have 6 possible ambiguity states for rate 2/3, 4 states for rate 3/4, 6 states for rate 5/6 and 8 states for rate 7/8. As part of the Viterbi decoding acquisition process, each puncturing state of each code must be tested. Total acquisition requires search of 26 different conditions. The process starts with rate 3/4 coding and proceeds sequentially to rate 2/3, 5/6, 7/8, and finally rate 1/2. In some systems, it may be possible to experience spectral inversion. This might occur when different combinations of LNBs and tuners are employed which implement different frequency translation schemes. Correction of spectral inversion must be corrected with host processor interaction. If the host processor detects that QPSK lock is achieved, but Viterbi lock has not occurred within a specified time, then a bit must be set in the demodulator which reverses the spectrum. 25 The table below illustrates a typical acquisition timing. For this example, the symbol rate is one half of the clock rate. The code rate is set to 5/6, which requires 13 trial and errors before node sync is achieved. The carrier search logic requires 10 dwells at different frequencies (500 symbols per dwell) before demodulator lock is achieved. Table 12: Example of Acquisition Timing Carrier Search Viterbi Node Sync Byte Sync Deinterleaver Flush Reed Solomon Total Timing Bit Times 8,333 2,652 16,000 19,584 1,632 48,201 Symbols 5,000 1,591 9,600 11,750 979 26,950 Clock Cycles 10,000 3,182 19,200 23,500 1,958 57,840 The total time required for acquisition could vary widely, depending upon the carrier search range and the time required for Viterbi node sync. For this example, however, the Byte Sync time and the time required to flush the deinterleaver dominates the total time. If a 60MHz clock were employed, the total acquisition time would be 0.963 milliseconds for this example 26 3.6 Reed Solomon Decoder The serial output from the Viterbi is provided to the Word Sync circuits which searches for the eight bit frame sync word which occurs every 204 bytes. By detecting the polarity of the sync word, this module can correct polarity reversals in the data provided by the Viterbi decoder. Byte serial data is provided to the convolutional deinterleaver, which reorders the received symbols. This process causes errors, which typically occur in bursts from the Viterbi decoder, to be distributed randomly over many blocks. This deinterleaved data is then provided to the Reed Solomon decoder which can reduce an error rate of 2 x10-4 from the Viterbi decoder to less than 1 in 10-10. The Reed Solomon decoder accepts input data in blocks of 204 bytes and produces error corrected blocks of 188 bytes. Maximum 8 bytes per a RS block can be corrected in RS decoder. Reedsolomon block includes on-chip BER calculator at the output of Viterbi to monitor signal quality or estimate the SNR of incoming signal. The calculated value can be read by accessing two read registers via utility bus such as I2C. It represents the number of errors among 220 data bits. The next process is descrambling, not to be confused with the descrambling which is part of conditional access. The purpose of scrambling the transmitted data and performing the inverse in the receiver is to insure that the spectrum of the transmitted waveform is always evenly distributed without significant discrete spectral lines. Without the scrambling/descrambling process, a transmitted sequence of all ones or all zeroes would result in strong spectral components and could interfere with other signals in the same satellite transponder. The final process is data regulation. Viterbi Data and Viterbi Clock occur irregularly according to the code rate. Data clock regulation makes it possible to interface with external common interface devices. To make external bus interface more flexible, interface mode such as parallel or serial can be selected by mode selection register. Parameter Regulate_data_clk Serial_valid Mode_serial Register Bit 5 of 14H register Bit 6 of 14H register Bit 0 of 18H register l NORMAL INTERFACE MODE (parallel/serial) If regulate_data_clk is reset, both parallel interface and serial interface work in normal operation which is same as HDM8513 regardless of serial_valid bit. Parallel interface or serial interface can be alternated by modifying mode_serial bit (Refer to Figure 9 and Figure 10) l REGULATED INTERFACE MODE (parallel) If regulate_data_clk is set,all interfaces are from internal FIFO designed to regulate irregular interface signals. Data clock cycle is a little bit faster than the average of cycle of irregular data clock, so meaningless data can be output in invalid data period. (Refer to Figure 11) l REGULATED INTERFACE MODE (serial) If mode_serial bit is set in the regulate interface mode, regulate interface mode is enabled for serial interface. Regular serial interface mode has two modes for more flexibi lity. The mode selection is controlled by serial_valid bit. If serial_valid is reset, DATA_STB signal alternates when every valid bit is out (mode1). While serial_valid is set, DATA_STB signal sustains high when valid bit is out (mode2). (Refer to Figure 12 and Figure 13) 27 Error Flag Deinterleaver Memory Viterbi Data Viterbi Clock 8 Word Sync. 8 Deinterleaver Memory Control 8 Reed Solomon Decoder Word Clock Frame Clock FIGURE 19: REED SOLOMON DECODER 28 8 Descrambler Data Out Sync. Data Clock 3.7 Clock Generation PLL An integrated VCO is locked to MxN times a reference frequency provided by a external clock. F Ref Reference Divider 1/M F Fb ext_clk PLL Analog Core 1/2 1 int_clk 0 Feedback Divider 1/N enable FIGURE 20: CLOCK SIGNAL GENERATION This programmable PLL consists of a PLL analog core, a reference divider with a divider ratio M, a feedback divider with a divider ratio N, and a divider which askes the duty cycle 1/2. Reference divider and feedback divider are used to synthesize various frequencies from a reference frequency, f ext_clk. Since PLL synchronizes the frequency and phase of two signals, FRef and FFb, f ext_clk M = f int_clk N Internal clock is calculated as follows f int_clk = N f ext_clk M The following two PLL modes are provided to control PLL. (1) PLL Enable mode : The internal clock is connected to the generated clock of the PLL. (2) PLL Disable mode : The PLL is bypassed and the external clock is directly connected to the internal clock. More information can be found on the part of the write register. 29 3.8 DBS Receiver The HDM8513A DVB Demodulator including a dual A/D converter and the MPEG-2 decoder provide the core digital processing technology for a DBS receiver conforming with the DVB standard. DRAM Inte rfa c e 4 80 MHz Loo p Filter Fixed F requency P LL Control Da ta 8 I L-B and Tuner 4 80 MHz IF Do wnconverter Q HDM8513A 8 Conditional Access Clock Video MPEG-2 Demultiplexer Audio BSFC77GV6 SL171 0 AGC 2 L ow Pa ss Filter AGC 1 3 WB AGC Co arse Tuning Step Frequ ency Control M C68306 (M C68340) Host Proce ssor Serial Interfa ce FIGURE 21: TYPICAL SET TOP BOX DEMODULATOR A tuner accepts an L-band RF input from the antenna/LNB assembly located outside the building. A host processor controls the tuner to the nominal center frequency of the target signal. Baseband I and Q outputs from the downconverter are applied to an A/D converter pair which is sampled at a fixed rate, 60MHz as illustrated in this example. The tuner is required to filter the received baseband signal to a bandwidth less than half the sampling rate, but is not required to perform matched filtering. Once the HDM8513A has locked to the target signal, the host processor may read the internal registers to determine the steady state frequency error. This error would be used to make period corrections to the programmed frequency of the tuner PLL. The HDM8513A provides an output which can be used to control the analog AGC in the tuner. This digital signal must be filtered and amplified before applying it to the AGC control element. When the loop is closed, the signal applied to the A/D converters is optimally scaled. 30 4. Mechanical Specifications 4.1 100 Pin Quad Flat Pack 4.1.1 Pin Assignment 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DATA_CLK FRAME_ERROR FRAME_SYNC VDD VSS LNB_TONE SIGMADELTA SYMBOL_CLOCK WB_AGC QPSK_LOCK IOVDD IOVSS TEST15 TEST14 TEST13 TEST12 VDD VSS TEST11 TEST10 TEST9 TEST8 IOVDD IOVSS TEST7 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 TEST6 VDDA VSSA VTOP AIN_I VSSA VDDA REF_I REF_Q AIN_Q VBOT VSS TEST5 TEST4 TEST3 VDDP VSS P TEST2 TEST1 TEST0 XTAL1 CLOCK IOVDD IOVSS RESET 51 HI_ADDR4 76 52 HI_ADDR3 77 53 HI_ADDR2 78 54 HI_ADDR1 79 55 HI_ADDR0 80 56 VDD 81 57 VSS 82 58 HI_DATA7 83 59 HI_DATA6 84 60 HI_DATA5 85 61 HI_DATA4 86 62 IOVDD 87 63 IOVSS 88 64 HI_DATA3 89 65 HI_DATA2 90 66 HI_DATA1 91 67 HI_DATA0 92 68 VDD 93 69 VSS 94 70 VB_NODESYNC 95 71 VB_CLOCK 96 72 VB_DATA 97 73 VDD 98 74 VSS 99 75 BUSMODE 100 DTACK SDA_I2CO SDA_I2C SCL_I2CO SCL_I2C IOVDD IOVSS R/W(/RE) /CE /DS(/WE) VDD VSS DATA7 DATA6 DATA5 LOCK DATA4 DATA3 IOVDD IOVSS DATA2 DATA1 DATA0 DATA_VALID DATA_STB 31 4.1.2 Package Dimensions 23.340 23.090 20.100 19.900 80 51 50 81 17.880 17.908 HDM8513A 14.100 13.900 DVB Demodulator 100 31 1 30 0.380 0.220 0.650 Typ. All Dimensions in mm 3.350 3.000 0.230 0.130 0.7 0.950 0.650 0.500 0.250 1.950 Typ. FIGURE 22: M ECHANICAL CONFIGURATION 32 4.2 64 Pin Thin Quad Flat Pack 4.2.1 Pin Assignment 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 FRAME_ERROR FRAME_SYNC LNB_SYNC WB_AGC IOVDD IOVSS TEST13 TEST12 VDD VSS TEST11 TEST10 TEST9 TEST8 VTOP AIN_I 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 VSSA VDDA REF_I REF_Q AIN_Q VBOT VSSA VDD VSS N/C N/C N/C XTAL1 IOVDD IOVSS RESET 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 VDD VSS I2C_ADD2 IOVDD IOVSS I2C_ADD1 I2C_ADD0 VDD VSS VB_CLOCK VB_DATA BUSMODE SDA_I2C0 SDA_I2C SCL_I2C0 SCL_I2C 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 IOVSS VDD VSS DATA7 DATA6 DATA5 LOCK DATA4 DATA3 IOVDD DATA2 DATA1 DATA0 DATA_VALID DATA_STB DATA_CLK 33 4.2.2 Package Dimensions 12.00 10.00 48 33 32 49 HDM8513AT 10.00 12.00 0.17 Min. 17 64 0.27 Max. 1 16 0.50 All Dimensions in mm 0 Min. 0.95 Min. 1.00 Typ. 1.05 Max. 0-7 0.15 Max. 0.45 Min. 0.08R Min. 1.00 Ref. FIGURE 23: M ECHANICAL CONFIGURATION 34 4.3 Recommended Analog Pin Connection 1.2uH VDD VDD Electrolyte Capacitor 47uF 0.1uF 0.1uF I 0.1uF Down AIN_I Converter AIN_Q Q 0.1uF VTO VBO REF_Q REF_I 0.1uF VSS 0.1uF VSS 0.1uF HDM8513A 0.1uF AGND FIGURE 24: A NALOG PIN CONNECTION 4.4 Recommended Clock Generation Circuit (to HDM8513A) FIGURE 25: CLOCK GENERATION CIRCUIT 35 5. Signal Description 5.1 Inputs XTAL1 XTAL1 can be configured either for sampling clock input or PLL reference clock input . The sampling clock rate must be a minimum of 1.33 times the symbol rate of the signal to be processed and at least equal to the total bandwidth of the signal to be processed. RESET A low on this signal causes the chip to be initialized. I/O registers are not cleared by this signal. This signal is asynchronous with respect to the clock. AIN_I Analog Input Signal for I channel. This should be AC coupled with Analog Input Source via 0.1uF capacitor. AIN_Q Analog Input Signal for I channel. This should be AC coupled with Analog Input Source via a 0.1uF capacitor. 5.2 Outputs VTOP Top Reference Voltage Output of about 2.0V. It should be bypassed to GND by 0.1uF capacitor. External bias voltage can be applied if necessary. VBOT Bottom Reference Voltage Output of 1.0V. It should be bypassed to GND by a 0.1uF capacitor. External bias voltage can be applied if necessary. REF_I Middle Reference Voltage for I Channel. It should be bypassed to GND by a 0.1uF capacitor. REF_Q Middle Reference Voltage for Q Channel. It should be bypassed to GND by a 0.1uF capacitor. DATA [7:0] The eight bit output data is provided in parallel format to be handed to an MPEG decoder for video and audio decompression. 36 DATA_CLK The DATA_CLK is used to latch data and control signal of transport stream. The data and control signals can be programmed to be latched either at positive or negative edge of DATA_CLK. This signal is used in conjunction with DATA_VALID to transfer data from the HDM8513A. The DATA_CLK will continue to toggle during the 16 bytes that the DATA_VALID signal indicates that no data is available (see figure 9 and 10). DATA_VALID When this signal is true, data is valid. This signal is not true during the time the 16 bytes of redundancy information is transmitted for the Reed Solomon decoder. FRAME_SYNC This signal is true at the first byte of a block of 188/144 bytes. DATA_STB This signal is used to transfer data from the HDM8513A to an MPEG decoder. This signal goes from low to high when a new byte of a 188 /144byte MPEG2 data stream block is available. This signal is inactive during the time the 16 redundancy bytes are transferred. FRAME_ERROR This signal goes true when the Reed Solomon decoder detects that an uncorrectable number of errors have occurred. The error flag in the MPEG2 output stream is also set when this flag goes high. WB_AGC This one bit output provides a measure of the external analog gain required for optimizing the signal applied to the analog to digital converters. This signal must be filtered, then applied to the analog gain control. CLOCK This is a buffered clock output signal which may be used to drive other devices with the same clock which drives the HDM8513A. QPSK_LOCK This signal goes true when the QPSK demodulator has achieved phase lock. VB_NODESYNC This signal goes true when the Viterbi decoder has achieved node synchronization. LOCK This signal goes true when the output data is valid and all synchronization functions have been performed. SYMBOL_CLOCK This signal, used for test purposes, goes true for a duration of one clock cycle for each received symbol. For symbol rates equal or greater than half the clock frequency, this signal at times may remain high for two successive clock cycles to indicate that two symbols have occurred. VB_DATA The serial output of the Viterbi Decoder is provided on this pin. The information rate at this point is less than the rate of the input clock ( less than 60Mbps if a 60MHz clock is employed). As long as valid convolutional encoding is employed, there is no constraint that the input signal adheres to MPEG2 format. This data is tapped priod to the polarity correction circuitry, so the data at this point may be inverted. 37 VB_CLOCK The positive edge of this signal indicates that VB_DATA is valid. SIGMADELTA This is an one bit Sigma Delta D/A converter which has 8 bits of resolution. This output must be filtered with an analog low pass filter off the chip. This output may be used for any external analog control. LNB_TONE This is a 22KHz clock output to control the LNB. TEST[15:0] The data provided on the test output signals is defined by data value of register 14H. Refer to register 14H. 38 5.3 Monitor and Control Interface Three different modes are supported for the monitor and control interface. Two of the modes are 8 bit parallel interfaces, one which supports Intel microcontrollers and the other intended for Motorola microcontrollers. The third mode is a serial interface conforming to the I2C standard. The I2C mode is activated by placing BUSMODE high at the same time both /RE and /WE are low simultaneously. When this mode is active, the seven bit I2C slave address of the HDM8513A is configured by the seven least significant bits of the HI_DATA[7:0] bus. HI_DATA [7:0] This bi-directional data bus is used for transferring control parameters to the demodulator and for reading the status registers within the demodulator. /CE Chip enable is an active low input to the demodulator which signifies that the other control signals are active. /RE Read Enable is an active low input to the device which, when active at the same time chip enable is true, permits the device to drive the HI_DATA [7:0] lines. When BUSMODE is 0 (Motorola), this pin is read / not write (see timing diagrams). /WE Write enable is an active low input to the device which, when true at the same time chip enable is true, causes input data on the HI_DATA [7:0] bus to be transferred to the register defined by the HI_ADDR [4:0] bus. When BUSMODE is 0 (Motorola), this pin is not data strobe (see timing diagrams). HI_ADDR [4:0] The address bus defines which location within the device is to be accessed during a read or write operation. BUSMODE BUSMODE selects the type of microcontroller/processor used to setup the chip. When high, an Intel processor/microcontroller interface is used. When low, a Motorola processor interface is used. DTACK Data Acknowledge/Data Ready is a tristate output signal which informs the controlling processor that a data transfer has been acknowledged by the HDM8513A. SCL_I2C This pin provides the clock for the I2C interface when that mode is active. SDA_I2C This pin is the data for the I2C interface and requires an external pull-up resistor as per the I2C standard. SDA_I2CO This pin, which can be by-passed, is the data for the I2C interface. SCL_I2CO This pin, which can be by-passed, provides the clock for the I2C interface. 39 5.4 I2C Mode The HDM8513A utilizes the subaddress technique when the I2C mode is employed. In all cases, the HDM8513A behaves as the slave device (transmitter or receiver), whilst the host behaves as the master device. The seven bit slave address of the HDM8513A is user selectable, being defined by the inputs to HI_DATA[6:0] when the HDM8513A is in I2C mode. Further information on the I2C bus formats and protocols is contained in the Philips Semiconductors I2C specification. In a 100pin configuration, SDA_I2CO and SCL_I2CO are added to provide a by-passing function. When I2C bypass bit is set to zero, SDA_I2CO and SCL_I2CO are disabled. 5.4.1 I2C Write to HDM8513A The master initiates communication with the HDM8513A by generating a start condition and then sending the HDM8513A the slave address defined by the seven bit hardwired address on HI_DATA [6:0]. Per I2C convention, the eighth bit in the address byte is a read/not write bit, and should be set to zero. The HDM8513A will acknowledge the correctly sent slave address, following which the master sends an eight bit word address; this is the address of the first HDM8513A register to be written to. Once the word address has been acknowledged by the HDM8513A, the master can then transmit the byte to be written to the word address. Once this byte is acknowledged by the HDM8513A, the word address is automatically incremented and further data bytes may be transmitted by the master as necessary; one transmission may therefore contain a number of bytes of data to be written to a sequential set of addresses (dummy bytes should be written to addresses not defined in the HDM8513A register set to continue this process). The process is terminated by the master generating a stop condition. Figure 25 depicts this protocol. acknowledgement from slave S 7 Bits SLAVE ADDRESS 0 A acknowledgement from slave acknowledgement from slave 8 Bits WORD ADDRESS A DATA BYTE A P repeat if necessary R/W S - Start Condition A - Acknowledge P - Stop Condition auto increment memory word address FIGURE 26: I2C W RITE TO THE HDM8513A 5.4.2 I2C Read from the HDM8513A To read information from the HDM8513A, the master must first write the desired word address. Hence the master must first generate a start condition and transmit the seven bit HDM8513A slave address defined on HI_DATA[6:0], with the eighth bit (read/not write) set to zero. Once this has been acknowledged by the HDM8513A, the master transmits the first word address from which it wishes to read information. The master must then generate a second start condition and 40 retransmit the HDM8513A slave address, this time with the read/not write bit set to one (read). This will be acknowledged by the HDM8513A, which then assumes the role of slave transmitter and transmits the requested byte. This byte should be acknowledged by the master receiver. If no stop condition is generated by the master, the HDM8513A will increment its word address pointer and transmit the next byte of information. This process is detailed in Figure 26. acknowledgement from slave S 7 Bits SLAVE ADDRESS 0 A acknowledgement acknowledgement from slave from slave 8 Bits WORD ADDRESS acknowledgement from master 7 Bits A S SLAVE ADDRESS 1 A R/W DATA BYTE A R/W auto increment memory word address HDM8513A becomes slave transmitter acknowledgement from master S: Start Condition A: Acknowledge P: Stop Condition DATA BYTE A P last byte auto increment memory word address FIGURE 27: I2C READ FROM THE HDM8513A Table 13: I2C Slave Address I2C_ADD0 0 1 0 1 0 1 0 1 I2C_ADD1 0 0 1 1 0 0 1 1 I2C_ADD2 0 0 0 0 1 1 1 1 I2C Address 0000000 0000011 0001100 0001111 0110000 0110011 0111100 0111111 41 6. Register Definitions 6.1 Write Registers ADDRESS (Hex) 00, 01, 02 Symbol Timing Frequency The 20 bit straight binary number in this field establishes the symbol timing frequency utilized within the demodulator. Bit 7 of address 00 is the MSB and bit 4 of address 02 is the LSB. If Rs is the symbol rate and fc is the clock frequency, the value to be stored in this 20 bit field is the integer portion of Rs (220)/fc. 03 Symbol Timing Loop Gain Control This field establishes the K1 and K2 gain values for the second order loop filter of the symbol tracking loop. Bits 0,1 and 2 determine the straight-through gain, and bits 4,5,6 and 7 determine the integration path gain. The nominal value of this parameter in Hex, is expressed below for different ranges of symbol rate to clock rate ratios: Symbol Rate/Clock 0.75 - 0.40 0.40 - 0.20 0.20 - 0.10 0.10 - 0.05 0.05 - 0.025 0.025 - 0.016 04, 05, 06 Value B6 A5 94 83 72 61 Carrier Frequency The 20 bit, two's complement number in this field establishes the nominal carrier frequency of the demodulator. Bit 7 of address 04 is the MSB and bit 4 of address 06 is the LSB. The number in this 20 bit field multiplied by the clock frequency divided by 220 is the carrier frequency in Hertz. When the carrier sweep function is active, this value defines the starting frequency. 42 07, 08 Carrier Loop Filter Control This field establishes the K1 and K2 gain values for the second order loop filter of the carrier tracking loop. Bits 0,1,2 and 3 determine the straightthrough gain, and bits 4,5, 6 and 7 determine the integration path gain. The nominal value of this parameter in Hex, is expressed below for different ranges of symbol rate to clock rate ratios. Two loop filter configurations are provided at each symbol rate, one for steady state operation(08) and one which is used only for acquisition(07) to permit greater frequency pull-in. Initially the gains are set to acquisition values. When QPSK_LOCK is achieved, they are automatically switched to steady state values. Symbol Rate/Clock Steady State Acqu. 0.75 - 0.40 0.40 - 0.20 0.20 - 0.10 0.10 - 0.05 0.05 - 0.025 0.025 - 0.016 09, 0A 47 47 47 46 45 45 77 77 77 77 77 77 Carrier Sweep Step Size This 16 bit value defines the size of the step of each carrier frequency dwell. Bit 7 of address 09 is the MSB and bit 0 of address 0A is the LSB. The number in this register is divided by 216, and multiplied by the clock frequency to determine the frequency step increment. 0B, 0C Symbols Per Dwell This 16 bit value defines the time, in symbol periods, for which the demodulator will dwell before making the next frequency step in a sweep. Bit 7 of address 0B is the MSB and bit 0 of address 0C is the LSB. 0D Number of Search Frequencies This 8 bit field determines the number of frequency steps which occur during the frequency sweeping process. Combined with the frequency step size, this determines the frequency span of the carrier sweep. 43 0E Narrow Band AGC initial value The six most significant bits of this field establish the initial gain of the AGC. High numbers correspond to low gain associated with low symbol rates. If the narrowband AGC function is enabled, this number is used as a starting point and the closed loop will seek the optimum setting without processor interaction. 44 0F Control Parameters Bit 0. Binary/Two’s Complement When this bit is a zero, the system expects the six bit modulation input samples in two’s complement format, otherwise the input should be in offset binary format. Bit 1. Spectrum Invert When this bit is set to zero, the spectrum of the received signal is inverted. This has the effect of complementing the in-phase channel only. Bit 2. Bias Cancel Enable When this bit is a one, the internal circuit which cancels DC bias on the I and Q inputs is enabled. When this function is enabled, it is assumed that the input signal is scrambled with no significant DC component on either the I or Q. Bit 3. Symbol Track Enable When this bit is set to one, the symbol tracking function is enabled. When this bit is zero the symbol tracking frequency is forced to the nominal 20 bit programmed value. Bit 4. Carrier Track Enable When this bit is set to one, the carrier phase tracking function is enabled. When this bit is zero, the carrier frequency is forced to the 20 bit programmed value. Bit 5. Sweep Hold When this bit is set to one, the sweeping process is inhibited, and the nominal carrier frequency remains at the last value. Bit 6. Narrowband AGC Mode 1 Enable When this bit is set to one and the narrowband AGC is in Mode 1, the narrowband AGC self-adjusts to the optimum gain setting. When the bit is set to zero, the most recent value is held without updating. Bit 7. Automatic Detection of Spectrum Inversion When this bit is set to one, the spectrum inversion is detected automatically. 45 10 Reset Functions Bit 0. Symbol Timing Frequency Accumulator When this bit is set to zero, the frequency accumulator in the symbol tracking loop is cleared to zero. This bit must be set to one in normal tracking operation to implement a second order tracking loop, otherwise the loop is first order. Bit 1. Carrier Phase Tracking Frequency Accumulator When this bit is set to zero, the frequency accumulator in the carrier phase tracking loop is cleared to zero. This bit must be set to one in tracking operation to implement a second order loop filter otherwise the loop is first order. Bit 2. Wideband AGC Accumulator When this bit is set to zero, the accumulator in the wideband AGC is cleared to zero. In normal operation, this bit is set to one. When the wideband AGC is set to Mode 1, this bit has no effect as the integrator must be implemented in the external analog circuits. Bit 3. Narrowband AGC Accumulator When this bit is set to zero, the accumulator in the narrowband AGC is cleared to the initial value defined in location 0E. In normal operation, this bit is set to one. Bit 4. Unused Bit 5. Carrier Sweep Function When this bit is set to zero, the sweep function is disabled and the carrier frequency is forced to the preset value defined in register locations 04, 05 and 06. Bit 6. Viterbi Reset When this bit is set to zero, the accumulator for the signal quality is cleared to zero. In normal operation, this bit is set to one. Bit 7. Reed Solomon Error Counter When this bit is set to zero, the counters for the number of corrected errors and the number of uncorrected code words are cleared to zero. 46 11 Wideband AGC Control Bit 0. Wideband AGC Mode When this bit is set to one (Mode 0), the WB AGC output must be filtered with an external integrating analog filter to implement a first order feedback loop. When this bit is zero (Mode 1), a digital integrator within the HDM8513A performs this function and the only external analog function required is a low pass filter to remove the high frequency components of the sigma delta converter output. Bit 1. WB AGC Invert When this bit is set to zero a high duty factor on the WB AGC output corresponds to too much gain. When the control bit is set to one, high duty factor corresponds to not enough gain. Bit 2. WB AGC Hold During normal tracking operation, this bit is set to one. When this bit is set to zero and the wideband AGC is in Mode 1, the digital integrator is held to the most recent value and loop updates are inhibited. Bit 3. LNB Hold When this bit is set to one, the output of LNB-Tone is held on zero. Bit 4. I2C By-pass When this bit is set to zero, SCL_I2CO and SDA_I2CO are disabled. The default is one and Data/clock can be by -passed. Bits [7:5]. WB AGC Gain This three bit field defines the time constant of the WB AGC in Mode 1. A value of zero corresponds to the shortest time constant and 7 corresponds to the slowest time constant. 12 LNB Tone This eight bit value establishes the control for LNB tone generator. If fL is the desired frequency and fC is the clock frequency, the value to be stored in this 8 bit field is the integer portion of fL(217)/fC. The default value(30H) generates 22KHz tone at 60MHz sampling clock. 13 Sigma Delta This eight bit input value establishes the control for Sigma Delta converter. This function is independent of other demodulator functions and is provided as control for external analog components. 47 14 Test Set-up The eight bit data written to this location defines the data presented on the 16 bit test bus. For configurations where the data is updated once per symbol period, the data changes at the rising edge of SYMBOL_CLOCK (in the case that SYMBOL_CLOCK remains high for consecutive CLOCK cycles, the test port data will also change accordingly during the high period of SYMBOL_CLOCK due to the arrival of another symbol). Bits [2:0]. Test port configuration 00H Output is tristate. 01H Test bits [15:8] provide the I baseband filter output. Test bits [7:0] provide the Q baseband filter output. This information is updated once per symbol period. 02H Test bits [15:0] provide the sixteen most significant bits of the demodulator carrier phase test bits. This information is updated once per symbol period. 03H Test bits [15:0] provide the sixteen most significant bits of the demodulator symbol phase test bits. This information is updated once per symbol period. 04H Test bits [15:8] provide the Reed Solomon output data. Test bits [7:0] provide the deinterleaver output data. This information is updated at the Reed Solomon clock rate; when the transport stream output is configured to parallel output mode, DATA_CLK may be used as an output clock for this data. 05H Test bits [15:10] provide the six bit narrowband AGC accumulator value. Test bits [9:6] provide the four bit value of symbol phase. Test bits [5:4] provide the two bit symbol count value. This information is updated once per symbol period. 06H Test bits [13:8] provide the six bit I-channel data from the ADC. Test bits [5:0] provides the six bit Q-channel data from the ADC. This information is updated at the fixed rate sample clock. 07H In this mode the test pins are used as input pins. The internal ADC is disabled, and the inputs at the test pins are fed directly to the demodulator. Test bits [13:8] are used as I-channel input and test bits [5:0] are used as Q-channel input. This information is updated at the fixed rate sample clock. Bit 3. Transport error Indicator Enable/Disable Enables/Disables the transport error indicator,1 bit indicator in transport header. When this bit is set to 1 and if transport error is internally detected the transport error indicator bit is set to 1. When zero this functionality is disabled. 48 Bit 4. This bit should be fixed to zero Bit 5. Regulated Data Clock Enables/Disables the data and data clock regulator. When this bit is set to 1, data output and data clock are regulated by FIFO operation. When this bit is set to 0, internal data output and internal data clock are bypassed Bit 6. Serial Valid When this bit is set to 1, DATA_STB signal sustains high when valid bit is out (mode 2). When this bit is set to 0, DATA_STB signal alternates when every valid bit is out (mode 1). Refer to figure 12 and figure 13. Bit 7. Clock Polarity This bit is used to select the DATA_CLK polarity either for serial or parallel transport interface. If this bit is set to zero(default value), the transport data and control signals are latched at the positive edge of DATA_CLK. Otherwise, the signals are latched at the negative edge of DATA_CLK. 15 Viterbi Lock Threshold Register 15 to 17 contain control parameters for synchronization in Viterbi decoder. Ordinary users are recommended to use the default value. Bit[7:4] defines the lock threshold for VB_NODESYNC. Viterbi decoder decides that the correct code rate has been found. A large number means it takes longer to find the correct code rate in automatic detection mode. It should be greater than 7. The default value is 12. Bit[3:0] defines the lock fail threshold. Viterbi decoder rejects a code rate and moves on to the next code rate. A small number means Viterbi decoder tries more data before it moves to the next code rate. It should be less than 7. The default value is 2. 16 Viterbi Unlock Threshold This number defines the threshold to maintain the Viterbi lock state. A large number means it needs more bad data to get out of the viterbi lock state and re-start searching the correct code rate. The default value is 1. 49 17 Viterbi Byte-Sync control Once the viterbi lock(VB_NODESYNC) is achieved, the Viterbi decoder tries to find the byte-sync. This 8-bit register is used to set “unlockthreshold” for the byte-sync. Large number means it needs more baddata to get out of the byte-sync state, i.e. less sensitive to noise. The default value is 1. 50 18 Control Parameters for Viterbi and RS Decoders Bit 0. Parallel or Serial Output Controls the transport stream output of the 8513A to serial or parallel mode. “0” (default) means the 8513A MPEG output is parallel. “1” means the 8513A MPEG output is serial. The LSB of the data bus(data[0] - pin 98) is used as the serial output pin. Bit 1. MPEG2 Data “0” (default) means the incoming data is MPEG2 decoded. In this mode a sync byte is expected every 188 bytes. “1” means non-MPEG2 data. The Viterbi decoder doesn’t check the existence of the sync byte. Bit [4:2]. Depuncturing Rate It defines the depuncturing rate of the Viterbi decoder. When vb_autocode is disabled, the depuncturing rate is set to this value. 0 --- 1/2 1 --- 2/3 2 --- 3/4 3 --- 5/6 4 --- 7/8 5 --- 6/7 Bit 5. Viterbi Auto Decoding Mode When this bit is set to 1, the Viterbi decoder automatically finds the correct code rate of the incoming signal. When this bit is set to 0, the code rate is set to the user-defined value at bit[4:2]. The default is 0. Bit 6. DSS Mode When this bit is set to 0, this device operates as DVB mode. When this bit is set to 1, this device operates as DSS mode. In that case, the roll-off factor of the Nyquist filter is set to 0.2. The default is 0 (DVB). Bit 7. BPSK Mode When this bit is set to 0, the demodulator assumes the incoming data is QPSK-modulated. When this bit is set to 1, the demodulator assumes the incoming data is BPSK-modulated. 19 Rate 1/2 Threshold Select This seven bit parameter defines the threshold used in the Viterbi decoder node synchronization process. For rate 1/2, the nominal value is 30 (1EH). 51 1A Rate 2/3 Threshold Select This seven bit parameter defines the threshold used in the node synchronization process. For rate 2/3, the nominal value is 30 (1EH). 1B Rate 3/4 Threshold Select This seven bit parameter defines the threshold used in the node synchronization process. For rate 3/4, the nominal value is 40 (28H). 1C Rate 5/6 Threshold Select This seven bit parameter defines the threshold used in the node synchronization process. For rate 5/6, the nominal value is 60 (3CH). 1D Rate 6/7 Threshold Select This seven bit parameter defines the threshold used in the node synchronization process. For rate 6/7, the nominal value is 60 (3CH). 1E Rate 7/8 Threshold Select This seven bit parameter defines the threshold used in the node synchronization process. For rate 7/8, the nominal value is 60 (3CH). Bit 7. This bit should be fixed to zero. 52 1F Clock Generation PLL Control An integrated VCO is locked to MxN times a reference frequency provided by a external clock Bits [4:0]. N Divider ratio It defines a feedback divider with a divider ratio N. The dafault is 15 (0FH). Bit 5. M Divider ratio It defines a reference divider with a divider ratio M. When this bit is set to 1, the divider ratio M is 4. When this bit is set to 0, the divider ratio M is 1. The default is 1. Bit 6. PLL Enable When this bit is set to 1, the generated clock of the PLL is connected to the internal clock. When this bit is set to 0, the PLL is bypassed and the external clock signal is directly connected to the internal clock. The default is 0. Bit 7. This bit should be fixed to zero. 53 Example for determination of internal clock f int_clk Desired internal clock: f ext_clk External clock supplied: N divider ratio range: 1 – 31 (integer) M divider ratio range: 1 or 4 (integer) N f ext_clk Calculation is as follows:f int_clk = M <Example 1> Desired internal clock: 60MHz External clock supplied: 16MHz N divider ratio range: 1 – 31 (integer) M divider ratio range: 1 or 4 (integer) Calculation is as follows: 60 =16 x N/M Case M =4, N must be 15 Case M =1, N is impossible Only one possible case exists <Example 2> Desired internal clock: 60MHz External clock supplied: 4MHz N divider ratio range: 1 – 31 (integer) M divider ratio range: 1 or 4 (integer) Calculation is as follows: 60 =4 x N/M Case M =1, N must be 15 Case M =4, N is impossible Only one possible case exists. 54 6.2 Read Registers ADDRESS (Hex) 00 Narrowband AGC Accumulator The current value of the six bit AGC accumulator may be read from this location. 01, 02, 03 Symbol Timing Frequency Accumulator The current value of the 20 frequency accumulator in the symbol timing loop filter may be read from these 3 locations. Bit 7 of address 01 is the MSB and bit 4 of address 03 is the LSB. 04, 05, 06 Phase Tracking Frequency Accumulator The current value of the 20 bit frequency accumulator in the carrier phase loop filter may be read from these 3 locations. Bit 7 of address 04 is the MSB and bit 4 of address 06 is the LSB. 07 QPSK Lock Status Bit 0. QPSK Lock Flag When this bit is set to one, the QPSK demodulator is phase locked. 08 Wide Band AGC Accumulator This eight bit value represents the most significant bits of the accumulator in the first order wideband AGC loop. This data only has meaning when the wideband AGC is in Mode 0. 09, 0A Sweep Frequency The 16 bit sweep accumulator is available at this location. Bit 7 of address 09 is the MSB and bit 0 of address 0A is the LSB. Th e receiver frequency is determined by adding the Sweep Frequency with the carrier frequency accumulator (read addresses 04, 05 and 06) and the nominal carrier start frequency (write addresses 04, 05 and 06). 55 0B In-Phase The eight bit output of the In-phase baseband filter is available at this location. This data is updated once per symbol. 0C Quadrature The eight bit output of the quadrature baseband filter is available at this location. This data is updated once per symbol. 0D Noise Power This eight bit output provides a measure of the noise component of the signal when QPSK lock is achieved. Higher numbers correspond to lower signal-to-noise ratio conditions. The quality of this metric is improved if the narrowband AGC is disabled for a minimum of 1000 symbol periods before this parameter is read. 0E, 0F BER Calculator The current value of the 16bit BER is used to monitor the signal quality or estimate the SNR of incoming signal at the output of Viterbi. Bit 7 of address 0E is the MSB and bit 0 of address 0F is the LSB. It represents the number of errors among 220 data bits. 16,17,18 Signal Quality This 24 bit signal provides a measure of quality of the signal processed by the Viterbi decoder. This parameter can be used to infer bit error rate and input signal-to-noise ratio for signals which are within a few dB of threshold. Bit 7 of address 16 is the MSB and bit 0 of address 18 is the LSB. The specific definition of this signal for each coding rate is TBD. 19 Viterbi Rate This three bit number represents the code rate of the Viterbi decoder. Rate 1/2 Rate 2/3 Rate 3/4 Rate 5/6 Rate 7/8 1A 0 1 2 3 4 Reed Solomon Errors The four bit number at this location indicates the number of errors corrected in the most current block of 188 bytes. This number may range from 0 to 8. 56 1B FEC Lock Bit 0. Viterbi Node Sync When this bit is set to one, the Viterbi decoder has successfully established node synchronization. Bit 1. Frame Sync When this bit is set to one, the FEC chip has successfully established word sync and frame sync. Bit 2. Viterbi Byte Sync When this bit is set to one, the Viterbi decoder has successfully established byte-synchronization. Bit 3. Pi Ambiguity When this bit is set to one, the Viterbi decoder has successfully resolved pi ambiguity in the input data. (i.e inverted data) Bit 4. Pi/2 Ambiguity When this bit is set to one, the Viterbi decoder has successfully resolved pi/2 ambiguity in the input data 1C,1D Accumulated Reed Solomon Errors These two registers present a count of corrected errors since it was last reset. Bit 7 of address 1C is the MSB and bit 0 of address 1D is the LSB. These registers are reset by writing value to address 10H. 1E Accumulated Reed Solomon Data This register presents a count of the uncorrected code words since it was last reset. When it reaches its maximum count(255), it rolls back to zero and starts counting again. This register is reset by writing value to address 10H. 1F Device Identifier This register present device identifier. The current value of this register is 93H 57 Appendix 58 A1. Loop Filter Programming Application Note To illustrate that the symbol timing recovery loop and the carrier phase recovery loop are both programmable, several simulations were performed with different loop parameter conditions. These simulations were performed with a symbol rate of two samples per symbol, corresponding to 30M symbols-per-second if a 60MHz clock were utilized. Figure A1 illustrates the transient response of the symbol phase with three different loop conditions (K1=5, K2=10; K1=4, K2=9; and K1=8, K2=7). The vertical scale represents phase over a 360 degree range (524,287 to -524,288). All test cases were run at high signal -to-noise ratio. The highest gain condition could be used for fast acquisition as well as for steady state with high code rate conditions, while the intermediate gain is a suitable steady state setting for rate 1/2 codes (minimum Eb/N0 of 4 dB). The lowest gain setting corresponds to ultra low loop bandwidth and may be considered for maintaining lock without phase jumps during deep signal fades. FIGURE A1: SYMBOL TIMING RECOVERY TRANSIENT RESPONSE 59 Figure A2 illustrates the transient response of the carrier tracking loop with the same loop bandwidth settings at high signal-to-noise ratio. The phase step for this test corresponds to 45 degrees. The actual bandwidth of the carrier loop is greater than that of the symbol loop for the same settings because the carrier loop must cope with greater dynamics (such as frequency offset and drift). Figure A3 illustrates the transient response of the carrier phase tracking loop under the same conditions at minimum signal-to-noise ratio (E b/N0 of 4 dB with rate 1/2 coding). The highest bandwidth case will pull in with a carrier frequency error of + or - .005 of the symbol rate at this minimum signal level. Higher loop bandwidth may be programmed to provide greater pull-in with higher signal-to-noise ratio conditions. FIGURE A2: CARRIER PHASE RECOVERY TRANSIENT RESPONSE 60 FIGURE A3: CARRIER PHASE RECOVERY TRANSIENT RESPONSE WITH LOW SNR 61 A2. False Lock Escape Application Note A QPSK signal will have inherent false lock states at frequency offsets of + or - n/4 of the symbol rate. Most DBS signals which have symbol rates of 20M symbols-per-second or higher will not experience false lock because the carrier frequency uncertainty is less than 1/4 of the symbol rate. The HDM8513A is designed to process low data rate signals which may experience false lock, particularly at high signal-to-noise ratio conditions. The HDM8513A will permit recovery from false lock with some added host processor interaction. Specifically, the processor must initialize the internal carrier frequency search hardware to search over a carrier frequency range of 1/4 of the symbol rate. If QPSK lock is achieved, but no Viterbi lock is achieved, the processor would assume this is a false carrier lock, then program the HDM8513A to search another carrier frequency range covering 1/4 of the symbol rate. When both QPSK lock and Viterbi lock have been achieved, the search is completed. This technique is reliable because the HDM8513A utilizes a fixed frequency clock which is not subject to inaccuracy associated with analog VCOs. This accuracy insures that the multiple search ranges are perfectly continuous with respect to each other with no overlap. 62 A3. Performance with Interference. In order to evaluate the filter employed within the HDM8513A with respect to attenuating out-ofband interference, a test was performed utilizing the COSSAP simulator. The desired signal, at zero frequency, was configured to utilize 16 samples-per-symbol (corresponding to 3.75MHz symbol rate if a 60MHz clock is employed). An interfering signal was added with the same characteristics, except that the amplitude was made to be 10dB higher than that of the desired signal, the data pattern was different and the carrier frequency was offset from that of the desired signal. Several offset frequencies were evaluated for this case. Figure A4 illustrates the spectrum of the test condition when the offset frequency is 1.35 times the symbol rate. Figure A5 illustrates the measured bit error rate for various conditions. The error rate on the I channel was measured separately from that of the Q channel, and the horizontal axis is scaled in dB for one component (I or Q of the signal). For example, the point labeled 1dB corresponds to SNR (noise bandwidth = symbol rate) of 4dB or E b/N0 of 4dB if rate 1/2 coding is employed. The theoretical performance for coherent PSK is shown with the solid line. The curve closest to theoretical is the demodulator performance with no other interferers and corresponds to an implementation loss of about 0.2dB. When the interferer was placed at a frequency of either 2.0 or 1.35 times the symbol rate away from the desired carrier, there is an additional degradation ranging from 0dB to 0.1dB. The worst case occurs when the interferer is placed at only 1.28 times the symbol rate from the carrier of the desired signal. In this case, the performance has degraded with respect to the no interference case by 0.3 to 0.5dB. Figure A6 illustrates performance with an interferer which is 20dB higher than the desired signal and separated in frequency by 2 times the symbol rate. In this case, the performance has degraded by 0.7 to 0.8dB from the case with no interferer. 63 FIGURE A4: A DJACENT CHANNEL INTERFERENCE OF 10 DB, 1.35 SPACING 64 FIGURE A5: PERFORMANCE WITH INTERFERER AT DIFFERENT CARRIER SPACINGS 65 FIGURE A6: PERFORMANCE WITH +10 DB INTERFERER 66 A4. Nyquist Criteria Considerations The HDM8513A is clocked at 60MHz, yet processes signals with symbol rates as high as 45M symbols-per-second. At first thought, this might appear to be violating the Nyquist criteria which states that the sampling rate must be at least twice the highest frequency component. The total bandwidth of the 45Msps signal, with 35% excess bandwidth, is about 60MHz. The samples provided to the HDM8513A are complex samples, which is equivalent to 120M samples-per-second, which does satisfy the Nyquist criteria. Another way of looking at this is to examine the baseband signal. The signal bandwidth covers 60MHz, but the baseband spectrum covers from -30MHz to +30MHz. There are no baseband frequency components greater than 30MHz, and the 60MHz clock is adequate as long as complex samples are taken. 67