Broadband Modem Mixed-Signal Front End AD9866 Powerline networking VDSL and HPNA AD9866 PWR DWN MODE TXEN/SYNC TXCLK 2-4X IAMP TxDAC 0 TO –7.5dB IOUT_G+ IOUT_N+ IOUT_N– IOUT_G– 0 TO –12dB 12 CLKOUT_1 CLKOUT_2 CLK SYN. ADIO[11:6]/ Tx[5:0] 2M CLK MULTIPLIER OSCIN XTAL ADIO[5:0]/ Rx[5:0] 12 RXE/SYNC RXCLK AGC[5:0] SPI RX+ ADC 80MSPS 2-POLE LPF 1-POLE LPF RX– 6 4 REGISTER CONTROL 0 TO 6dB ∆ = 1dB – 6 TO 18dB –6 TO 24dB ∆ = 6dB ∆ = 6dB 04560-0-001 APPLICATIONS FUNCTIONAL BLOCK DIAGRAM IOUT_P– Low cost 3.3 V CMOS MxFETM for broadband modems 12-bit D/A converter 2×/4× interpolation filter 200 MSPS DAC update rate Integrated 23 dBm line driver with 19.5 dB gain control 12-bit, 80 MSPS A/D converter −12 dB to +48 dB low noise RxPGA (< 2.5 nV/rtHz) Third order, programmable low-pass filter Flexible digital data path interface Half- and full-duplex operation Backward-compatible with AD9975 and AD9876 Various power-down/reduction modes Internal clock multiplier (PLL) 2 auxiliary programmable clock outputs Available in 64-lead chip scale package or bare die IOUT_P+ FEATURES Figure 1. GENERAL DESCRIPTION The AD9866 is a mixed-signal front end (MxFE) IC for transceiver applications requiring Tx and Rx path functionality with data rates up to 80 MSPS. Its flexible digital interface, power saving modes, and high Tx-to-Rx isolation make it well-suited for half- and full-duplex applications. The digital interface is extremely flexible allowing simple interfaces to digital back ends that support half- or full-duplex data transfers, thus often allowing the AD9866 to replace discrete ADC and DAC solutions. Power saving modes include the ability to reduce power consumption of individual functional blocks or to power down unused blocks in half-duplex applications. A serial port interface (SPI®) allows software programming of the various functional blocks. An on-chip PLL clock multiplier and synthesizer provide all the required internal clocks, as well as two external clocks from a single crystal or clock source. or to an internal low distortion current amplifier. The current amplifier (IAMP) can be configured as a current- or voltagemode line driver (with two external npn transistors) capable of delivering in excess of 23 dBm peak signal power. Tx power can be digitally controlled over a 19.5 dB range in 0.5 dB steps. The Tx signal path consists of a bypassable 2×/4× low-pass interpolation filter, a 12-bit TxDAC, and a line driver. The transmit path signal bandwidth can be as high as 34 MHz at an input data rate of 80 MSPS. The TxDAC provides differential current outputs that can be steered directly to an external load The AD9866 provides a highly integrated solution for many broadband modems. It is available in a space saving, 64-lead lead frame chip scale package (LFCSP), and is specified over the commercial (−40°C to +85°C) temperature range. The receive path consists of a programmable amplifier (RxPGA), a tunable low pass filter (LPF), and a 12-bit ADC. The low noise RxPGA has a programmable gain range of −12 dB to +48 dB in 1 dB steps. Its input referred noise is less than 3.3 nV/rtHz for gain settings beyond 30 dB. The receive path LPF cutoff frequency can be set over a 15 MHz to 35 MHz range or simply bypassed. The 12-bit ADC achieves excellent dynamic performance over a 5 MSPS to 80 MSPS span. Both the RxPGA and the ADC offer scalable power consumption allowing power/performance optimization. Rev. A Information furnished by Analog Devices is believed to be accurate and reliable. However, no responsibility is assumed by Analog Devices for its use, nor for any infringements of patents or other rights of third parties that may result from its use. Specifications subject to change without notice. No license is granted by implication or otherwise under any patent or patent rights of Analog Devices. Trademarks and registered trademarks are the property of their respective owners. One Technology Way, P.O. Box 9106, Norwood, MA 02062-9106, U.S.A. Tel: 781.329.4700 www.analog.com Fax: 781.326.8703 © 2004 Analog Devices, Inc. All rights reserved. AD9866 TABLE OF CONTENTS Specifications..................................................................................... 3 TxDAC and IAMP Architecture .............................................. 28 Tx Path Specifications.................................................................. 3 Tx Programmable Gain Control .............................................. 30 Rx Path Specifications.................................................................. 4 TxDAC Output Operation........................................................ 30 Power Supply Specifications ....................................................... 5 IAMP Current-Mode Operation.............................................. 30 Digital Specifications ................................................................... 6 IAMP Voltage-Mode Operation .............................................. 31 Serial Port Timing Specifications............................................... 7 IAMP Current Consumption Considerations........................ 32 Half-Duplex Data Interface (ADIO Port) Timing Specifications ................................................................................ 7 Receive Path .................................................................................... 33 Full-Duplex Data Interface (Tx and Rx PORT) Timing Specifications ................................................................................ 8 Explanation of Test Levels........................................................... 8 Absolute Maximum Ratings............................................................ 9 Thermal Characteristics .............................................................. 9 ESD Caution.................................................................................. 9 Pin Configuration and Function Descriptions........................... 10 Typical Performance Characteristics ........................................... 12 Rx Path Typical Performance Characteristics ........................ 12 TxDAC Path Typical Performance Characteristics ............... 16 IAMP Path Typical Performance Characteristics .................. 18 Serial Port ........................................................................................ 19 Register Map Description ......................................................... 21 Serial Port Interface (SPI) ......................................................... 21 Digital Interface .............................................................................. 23 Half-Duplex Mode ..................................................................... 23 Full-Duplex Mode ...................................................................... 24 RxPGA Control .......................................................................... 25 TxPGA Control .......................................................................... 27 Transmit Path .................................................................................. 28 Rx Programmable Gain Amplifier........................................... 33 Low-Pass Filter ........................................................................... 34 Analog-to-Digital Converter (ADC)....................................... 35 AGC Timing Considerations.................................................... 36 Clock Synthesizer ........................................................................... 37 Power Control and Dissipation .................................................... 39 Power-Down ............................................................................... 39 Half-Duplex Power Savings ...................................................... 39 Power Reduction Options......................................................... 40 Power Dissipation ...................................................................... 42 Mode Select upon Power-Up and Reset.................................. 42 Analog and Digital Loopback Test Modes.............................. 43 PCB Design Considerations.......................................................... 44 Component Placement.............................................................. 44 Power Planes and Decoupling .................................................. 44 Ground Planes ............................................................................ 44 Signal Routing ............................................................................ 44 Evaluation Board ............................................................................ 46 Outline Dimensions ....................................................................... 47 Ordering Guide .......................................................................... 47 Digital Interpolation Filters ...................................................... 28 REVISION HISTORY 12/04—Data Sheet Changed from Rev. 0 to Rev. A Changes to Specifications Tables .................................................... 3 Changes to Serial Table .................................................................. 19 Changes to Full Duplex Mode section......................................... 24 Changes to Table 14........................................................................ 25 Change to TxDAC and IAMP Architecture section .................. 29 Change to TxDAC Output Operation section............................ 30 Insert equation................................................................................ 37 Change to Figure 84 caption ......................................................... 42 11/03—Revision 0: Initial Version Rev. A | Page 2 of 48 AD9866 SPECIFICATIONS Tx PATH SPECIFICATIONS AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%; fOSCIN = 50 MHz, fDAC = 200 MHz, RSET = 2.0 kΩ, unless otherwise noted. Table 1. Parameter TxDAC DC CHARACTERISTICS Resolution Update Rate Full-Scale Output Current (IOUTP_FS) Gain Error1 Offset Error Voltage Compliance Range TxDAC GAIN CONTROL CHARACTERISTICS Minimum Gain Maximum Gain Gain Step Size Gain Step Accuracy Gain Range Error TxDAC AC CHARACTERISTICS2 Fundamental Signal-to-Noise and Distortion (SINAD) Signal-to-Noise Ratio (SNR) Total Harmonic Distortion (THD) Spurious-Free Dynamic Range (SFDR) IAMP DC CHARACTERISTICS IOUTN Full-Scale Current = IOUTN+ + IOUTN− IOUTG Full-Scale Current = IOUTG+ + IOUTG− AC Voltage Compliance Range IAMPN AC CHARACTERISTICS3 Fundamental IOUTN SFDR (Third Harmonic) IAMP GAIN CONTROL CHARACTERISTICS Minimum Gain Maximum Gain Gain Step Size Gain Step Accuracy IOUTN Gain Range Error REFERENCE Internal Reference Voltage4 Reference Error Reference Drift Tx DIGITAL FILTER CHARACTERISTICS (2× INTERPOLATION) Latency (Relative to 1/FDAC) −0.2 dB Bandwidth −3 dB Bandwidth Stop-Band Rejection (0.289 FDAC to 0.711 FDAC) Tx DIGITAL FILTER CHARACTERISTICS (4× Interpolation) Latency (Relative to 1/FDAC) −0.2 dB Bandwidth Temp Test Level Full Full Full 25°C 25°C Full II IV I V 25°C 25°C 25°C 25°C 25°C V V V IV V −7.5 0 0.5 Monotonic ±2 Full Full Full Full IV IV IV IV 66.6 68.4 0.5 69.2 69.8 −79 81 Full Full Full IV IV IV 2 2 1 25°C Full IV 43.3 25°C 25°C 25°C 25°C 25°C V V V IV V 25°C I 1.23 Full Full V V 0.7 30 Full Full V V 43 0.2187 Cycles fOUT/fDAC Full Full V V 0.2405 50 fOUT/fDAC dB Full Full V V 96 0.1095 Cycles fOUT/fDAC Rev. A | Page 3 of 48 Min Typ Max 12 200 25 2 ±2 2 −1 68.5 +1.5 Unit Bits MSPS mA % FS µA V dB dB dB dB −68.7 105 150 7 13 45.2 dBm dBc dBc dBc dBc mA mA V dBm dBc −19.5 0 0.5 Monotonic 0.5 dB dB dB dB dB V 3.4 % ppm/oC AD9866 Parameter −3 dB Bandwidth Stop Band Rejection (0.289 fOSCIN to 0.711 fOSCIN) PLL CLK MULTIPLIER OSCIN Frequency Range Internal VCO Frequency Range Duty Cycle OSCIN Impedance CLKOUT1 Jitter5 CLKOUT2 Jitter6 CLKOUT1 and CLKOUT2 Duty Cycle7 Temp Full Full Test Level V V Min Full Full Full 25°C 25°C IV IV II V III 5 20 40 25°C Full III III Typ 0.1202 50 Max Unit fOUT/fDAC dB 80 200 60 MHz MHz % ΜΩ//pF ps rms 100//3 12 6 45 55 ps rms % 1 Gain error and gain temperature coefficients are based on the ADC only (with a fixed 1.23 V external reference and a 1 V p-p differential analog input). TxDAC IOUTFS = 20 mA, differential output with 1:1 transformer with source and load termination of 50 Ω, FOUT = 5 MHz, 4× interpolation. IOUN full-scale current = 80 mA, fOSCIN= 80 MHz, fDAC=160 MHz, 2× interpolation. 4 Use external amplifier to drive additional load. 5 Internal VCO operates at 200 MHz , set to divide-by-1. 6 Because CLKOUT2 is a divided down version of OSCIN, its jitter is typically equal to OSCIN. 7 CLKOUT2 is an inverted replica of OSCIN, if set to divide-by-1. 2 3 Rx PATH SPECIFICATIONS AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%; half- or full-duplex operation with CONFIG = 0 default power bias settings, unless otherwise noted. Table 2. Parameter Rx INPUT CHARACTERISTICS Input Voltage Span (RxPGA gain = −10 dB) Input Voltage Span (RxPGA gain = +48 dB) Input Common-Mode Voltage Differential Input Impedance Temp Test Level Full Full 25°C 25°C III III III III Input Bandwidth (with RxLPF Disabled, RxPGA = 0 dB) Input Voltage Noise Density (RxPGA Gain = 36 dB, f−3 dBF = 26 MHz) Input Voltage Noise Density (RxPGA Gain = 48 dB, f−3 dBF = 26 MHz) RxPGA CHARACTERISTICS Minimum Gain Maximum Gain Gain Step Size Gain Step Accuracy Gain Range Error RxLPF CHARACTERISTICS Cutoff Frequency (f−3 dBF ) range Attenuation at 55.2 MHz with f−3 dBF = 21 MHz Pass-Band Ripple Settling Time to 5 dB RxPGA Gain Step @ fADC = 50 MSPS Settling Time to 60 dB RxPGA Gain Step @ fADC = 50 MSPS ADC DC CHARACTERISTICS Resolution Conversion Rate Rx PATH LATENCY1 Full-Duplex Interface Half-Duplex Interface 25°C 25°C 25°C Min Typ Max Unit III III III 6.33 8 1.3 400 4.0 53 2.7 2.4 Ω pF MHz nV/rtHz nV/rtHz 25°C 25°C 25°C 25°C 25°C III III III III III −12 48 1 Monotonic 0.5 dB dB dB dB dB Full 25°C 25°C 25°C 25°C III III III III III NA FULL NA II Full Full V V Rev. A | Page 4 of 48 15 V p-p mV p-p V 35 20 ±1 20 100 12 5 80 10.5 10.0 MHz dB dB ns ns Bits MSPS Cycles Cycles AD9866 Parameter Rx PATH COMPOSITE AC PERFORMANCE @ fADC = 50 MSPS2 RxPGA Gain = 48 dB (Full-Scale = 8.0 mV p-p) Signal-to-Noise (SNR) Total Harmonic Distortion (THD) RxPGA Gain = 24 dB (Full-Scale = 126 mV p-p) Signal-to-Noise (SNR) Total Harmonic Distortion (THD) RxPGA Gain = 0 dB (Full-Scale = 2.0 V p-p) Signal-to-Noise (SNR) Total Harmonic Distortion (THD) Rx PATH COMPOSITE AC PERFORMANCE @ fADC = 80 MSPS3 RxPGA Gain = 48 dB (Full-Scale = 8.0 m V p-p) Signal-to-Noise (SNR) Total Harmonic Distortion (THD) RxPGA Gain = 24 dB (Full-Scale = 126 m V p-p) Signal-to-Noise (SNR) Total Harmonic Distortion (THD) RxPGA Gain = 0 dB (Full-Scale = 2.0 V p-p) Signal-to-Noise (SNR) Total Harmonic Distortion (THD) Rx-to-Tx PATH FULL-DUPLEX ISOLATION (1 V p-p, 10 MHz Sine Wave Tx Output) RxPGA Gain = 40 dB IOUTP± Pins to RX± Pins IOUTG± Pins to RX± Pins RxPGA Gain = 0 dB IOUTP± Pins to RX± Pins IOUTG± Pins to RX± Pins 1 2 3 Temp Test Level Min 25°C 25°C III III 43.7 −71 dBc dBc 25°C 25°C III III 63.1 −67.2 dBc dBc Full Full IV IV 64.3 −67.3 dBc dBc 25°C 25°C III III 41.8 −67 dBc dBc 25°C 25°C III III 58.6 −62.9 dBc dBc 25°C 25°C II II 25°C 25°C III III 83 37 dBc dBc 25°C 25°C III III 123 77 dBc dBc 61.1 Typ Max 62.9 −70.8 −60.8 Unit dBc dBc Includes RxPGA, ADC pipeline, and ADIO bus delay relative to fADC. fIN = 5 MHz, AIN = −1.0 dBFS , LPF cutoff frequency set to 15.5 MHz with Reg. 0x08 = 0x80. fIN = 5 MHz, AIN = −1.0 dBFS , LPF cutoff frequency set to 26 MHz with Reg. 0x08 = 0x80. POWER SUPPLY SPECIFICATIONS AVDD = 3.3 V, DVDD = CLKVDD = DRVDD = 3.3 V; RSET = 2 kΩ, full-duplex operation with fDATA = 80 MSPS,1 unless otherwise noted. Table 3. Parameter SUPPLY VOLTAGES AVDD CLKVDD DVDD DRVDD IS_TOTAL (Total Supply Current) POWER CONSUMPTION IAVDD + ICLKVDD (Analog Supply Current) IDVDD + IDRVDD (Digital Supply Current) POWER CONSUMPTION (Half-Duplex Operation with fDATA = 50 MSPS)1 Tx Mode IAVDD + ICLKVDD IDVDD + IDRVDD Temp Test Level Min Typ Max Unit Full Full Full Full Full V V V V II 3.135 3.0 3.0 3.0 3.3 3.3 3.3 3.3 406 3.465 3.6 3.6 3.6 475 V V V V mA Full IV IV 311 95 342 133 mA mA 25°C 25°C IV IV 112 46 130 49.5 mA mA Rev. A | Page 5 of 48 AD9866 Parameter Rx Mode IAVDD + ICLKVDD IDVDD + IDRVDD POWER CONSUMPTION OF FUNCTIONAL BLOCKS2 (IAVDD + ICLKVDD) RxPGA and LPF ADC TxDAC IAMP (Programmable) Reference CLK PLL and Synthesizer MAXIMUM ALLOWABLE POWER DISSIPATION STANDBY POWER CONSUMPTION IS_TOTAL (Total Supply Current) POWER-DOWN DELAY (USING PWR_DWN PIN) RxPGA and LPF ADC TxDAC IAMP CLK PLL and Synthesizer POWER-UP DELAY (USING PWR_DWN PIN) RxPGA and LPF ADC TxDAC IAMP CLK PLL and Synthesizer 1 2 Temp Test Level Min 25°C 25°C 25°C 25°C 25°C 25°C 25°C 25°C Full III III III III III III IV Typ Max Unit 225 36.5 253 39 mA mA 120 mA mA mA mA mA mA W 87 108 38 10 170 107 1.66 Full 13 mA 25°C 25°C 25°C 25°C 25°C III III III III III 440 12 20 20 27 ns ns ns ns ns 25°C 25°C 25°C 25°C 25°C III III III III III 7.8 88 13 20 20 µs ns µs ns µs Default power-up settings for MODE = LOW and CONFIG = LOW. Default power-up settings for MODE = HIGH and CONFIG = LOW, IOUTP_FS = 20 mA, does not include IAMP’s current consumption, which is application dependent. DIGITAL SPECIFICATIONS AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%; RSET = 2 kΩ, unless otherwise noted. Table 4. Parameter CMOS LOGIC INPUTS High Level Input Voltage Low Level Input Voltage Input Leakage Current Input Capacitance CMOS LOGIC OUTPUTS (CLOAD = 5 pF) High Level Output Voltage (IOH = 1 mA) Low Level Output Voltage (IOH = 1 mA) Output Rise/Fall Time (High Strength Mode and CLOAD = 15 pF) Output Rise/Fall Time (Low Strength Mode and CLOAD = 15 pF) Output Rise/Fall Time (High Strength Mode and CLOAD = 5 pF) Output Rise/Fall Time (Low Strength Mode and CLOAD = 5 pF) RESET Minimum Low Pulse Width (Relative to fADC) Temp Test Level Min Full Full VI VI DRVDD – 0.7 Full VI Full Full Full Full Full Full VI VI VI VI VI VI Rev. A | Page 6 of 48 Typ Max Unit 0.4 12 V V µA pF 3 DRVDD – 0.7 0.4 1.5/2.3 1.9/2.7 0.7/0.7 1.0/1.0 1 V V ns ns ns ns Clock cycles AD9866 SERIAL PORT TIMING SPECIFICATIONS AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%, unless otherwise noted. Table 5. Parameter WRITE OPERATION (See Figure 46) SCLK Clock Rate (fSCLK) SCLK Clock High (tHI) SCLK Clock Low (tLOW) SDIO to SCLK Setup Time (tDS) SCLK to SDIO Hold Time (tDH) SEN to SCLK Setup Time (tS) SCLK to SEN Hold Time (tH) READ OPERATION (See Figure 47 and Figure 48) SCLK Clock Rate (fSCLK) SCLK Clock High (tHI) SCLK Clock Low (tLOW) SDIO to SCLK Setup Time (tDS) SCLK to SDIO Hold Time (tDH) SCLK to SDIO (or SDO) Data Valid Time (tDV) SEN to SDIO Output Valid to Hi-Z (tEZ) Temp Test Level Min Full Full Full Full Full Full Full IV IV IV IV IV IV IV 14 14 14 0 14 0 Full Full Full Full Full Full Full IV IV IV IV IV IV IV Typ Max Unit 32 MHz ns ns ns ns ns ns 32 MHz ns ns ns ns ns ns 14 14 14 0 14 2 HALF-DUPLEX DATA INTERFACE (ADIO PORT) TIMING SPECIFICATIONS AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%, unless otherwise noted. Table 6. Parameter READ OPERATION1 (See Figure 50) Output Data Rate Three-State Output Enable Time (tPZL) Three-State Output Disable Time (tPLZ) Rx Data Valid Time (tVT) Rx Data Output Delay (tOD) WRITE OPERATION (See Figure 49) Input Data Rate (1× Interpolation) Input Data Rate (2× Interpolation) Input Data Rate (4× Interpolation) Tx Data Setup Time (tDS) Tx Data Hold Time (tDH) Latch Enable Time (tEN) Latch Disable Time (tDIS) 1 Temp Test Level Min Full Full Full Full Full II II II II II 5 Full Full Full Full Full Full Full II II II II II II II CLOAD = 5 pF for digital data outputs. Rev. A | Page 7 of 48 Typ Max Unit 80 3 3 MSPS ns 1.5 4 20 10 5 1 2.5 80 80 50 3 3 ns ns ns MSPS MSPS MSPS ns ns ns ns AD9866 FULL-DUPLEX DATA INTERFACE (Tx AND Rx PORT) TIMING SPECIFICATIONS AVDD = 3.3 V ± 5%, DVDD = CLKVDD = DRVDD = 3.3 V ± 10%, unless otherwise noted. Table 7. Parameter Tx PATH INTERFACE (See Figure 53) Input Nibble Rate (2× Interpolation) Input Nibble Rate (4× Interpolation) Tx Data Setup Time (tDS) Tx Data Hold Time (tDH) Rx PATH INTERFACE1 (See Figure 54) Output Nibble Rate Rx Data Valid Time (tDV) Rx Data Hold Time (tDH) 1 Temp Test Level Min Full Full Full Full II II II II Full Full Full II II II Typ Max Unit 20 10 2.5 1.5 160 100 MSPS MSPS ns ns 10 3 0 160 MSPS ns ns CLOAD = 5 pF for digital data outputs. EXPLANATION OF TEST LEVELS I II III IV V VI 100% production tested. 100% production tested at 25°C and guaranteed by design and characterization at specified temperatures. Sample tested only. Parameter is guaranteed by design and characterization testing. Parameter is a typical value only. 100% production tested at 25°C and guaranteed by design and characterization for industrial temperature range. Rev. A | Page 8 of 48 AD9866 ABSOLUTE MAXIMUM RATINGS Table 8. Parameter ELECTRICAL AVDD, CLKVDD Voltage DVDD, DRVDD Voltage RX+, RX−, REFT, REFB IOUTP+, IOUTP− IOUTN+, IOUTN−, IOUTG+, IOUTG− OSCIN, XTAL REFIO, REFADJ Digital Input and Output Voltage Digital Output Current ENVIRONMENTAL Operating Temperature Range (Ambient) Maximum Junction Temperature Lead Temperature (Soldering, 10 sec) Storage Temperature Range (Ambient) Rating 3.9 V maximum 3.9 V maximum −0.3 V to AVDD + 0.3 V −1.5 V to AVDD + 0.3 V −0.3 V to 7 V −0.3 V to CLVDD + 0.3 VS −0.3 V to AVDD + 0.3 V −0.3 V to DRVDD + 0.3 V 5 mA maximum −40°C to +85°C Stresses above those listed under the Absolute Maximum Ratings may cause permanent damage to the device. This is a stress rating only; functional operation of the device at these or any other conditions above those indicated in the operational section of this specification is not implied. Exposure to absolute maximum rating conditions for extended periods may affect device reliability. THERMAL CHARACTERISTICS Thermal Resistance: 64-lead LFCSP (4-layer board). θJA = 24°C/W (paddle soldered to ground plane, 0 LPM air). θJA = 30.8°C/W (paddle not soldered to ground plane, 0 LPM air). 125°C 150°C −65°C to +150°C ESD CAUTION ESD (electrostatic discharge) sensitive device. Electrostatic charges as high as 4000 V readily accumulate on the human body and test equipment and can discharge without detection. Although this product features proprietary ESD protection circuitry, permanent damage may occur on devices subjected to high energy electrostatic discharges. Therefore, proper ESD precautions are recommended to avoid performance degradation or loss of functionality. Rev. A | Page 9 of 48 AD9866 DRVDD DRVSS PWR_DWN CLKOUT2 DVDD DVSS CLKVDD OSCIN XTAL CLKVSS CONFIG MODE IOUT_P+ IOUT_P– IOUT_N+ IOUT_G+ PIN CONFIGURATION AND FUNCTION DESCRIPTIONS 64 63 62 61 60 59 58 57 56 55 54 53 52 51 50 49 ADIO11/Tx[5] 1 48 AVSS ADIO10/Tx[4] 2 47 AVSS 46 IOUT_N– PIN 1 IDENTIFIER 3 ADIO8/Tx[2] 4 45 IOUT_G– ADIO7/Tx[1] 5 44 AVSS ADIO6/Tx[0] 6 43 AVDD ADIO5/Rx[5] 7 42 REFIO ADIO4/Rx[4] 8 41 REFADJ ADIO3/Rx[3] 9 40 AVDD ADIO2/Rx[2] 10 39 AVSS 11 38 RX+ ADIO0/Rx[0] 12 37 RX– RXEN/RXSYNC 13 36 AVSS TXEN/TXSYNC 14 35 AVDD TXCLK/TXQUIET 15 34 AVSS RXCLK 16 33 REFT 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 DRVSS CLKOUT1 SDIO SDO SCLK SEN GAIN/PGA[5] PGA[4] PGA[3] PGA[2] PGA[1] PGA[0] RESET AVSS REFB TOP VIEW (Not to Scale) DRVDD ADIO1/Rx[1] AD9866 Figure 2. Pin Configuration Table 9. Pin Function Descriptions Pin No. 1 2 to 5 6 7 8, 9 10 11 12 13 14 15 Mnemonic ADIO11 Tx[5] ADIO10 to 7 Tx[4 to 1] ADIO6 Tx[0] ADIO5 Rx[5] ADIO4, 3 Rx[4, 3] ADIO2 Rx[2] ADIO1 Rx[1] ADIO0 Rx[0] RXEN RXSYNC TXEN TXSYNC TXCLK TXQUIET Mode1 HD FD HD FD HD FD HD FD HD FD HD FD HD FD HD FD HD FD HD FD HD FD Description MSB of ADIO Buffer MSB of Tx Nibble Input Bits 10 to 7 of ADIO Buffer Bits 4 to 1 of Tx Nibble Input Bit 6 of ADIO Buffer LSB of Tx Nibble Input Bit 5 of ADIO Buffer MSB of Rx Nibble Output Bits 4 to 3 of ADIO Buffer Bits 4 to 3 of Rx Nibble Output Bit 2 of ADIO Buffer Bit 2 of Rx Nibble Output Bit 1 of ADIO Buffer Bit 1 of Rx Nibble Output LSB of ADIO Buffer LSB of Rx Nibble Output ADIO Buffer Control Input Rx Data Synchronization Output Tx Path Enable Input Tx Data Synchronization Input ADIO Sample Clock Input Fast TxDAC/IAMP Power-Down Rev. A | Page 10 of 48 04560-0-002 ADIO9/Tx[3] AD9866 Pin No. 16 Mnemonic RXCLK 17, 64 18, 63 19 20 21 22 23 DRVDD DRVSS CLKOUT1 SDIO SDO SCLK SEN 24 25 to 29 30 GAIN PGA[5] PGA[4 to 0] RESET 31, 34, 36, 39, 44, 47, 48 32, 33 35, 40, 43 37, 38 41 42 45 46 49 50 51 52 53 AVSS REFB, REFT AVDD RX−, RX+ REFADJ REFIO IOUT_G− IOUT_N− IOUT_G+ IOUT_N+ IOUT_P− IOUT_P+ MODE 54 55 56 57 58 59 60 61 62 CONFIG CLKVSS XTAL OSCIN CLKVDD DVSS DVDD CLKOUT2 PWR_DWN 1 Mode1 HD FD Description ADIO Request Clock Input Rx and Tx Clock Output at 2 × fADC Digital Output Driver Supply Input Digital Output Driver Supply Return fDAC/N Clock Output (L = 1, 2, 4, or 8) Serial Port Data Input/Output Serial Port Data Output Serial Port Clock Input Serial Port Enable Input FD HD or FD HD or FD Tx Data Port (Tx[5:0]) Mode Select MSB of PGA Input Data Port Bits 4 to 0 of PGA Input Data Port Reset Input (Active Low) Analog Ground ADC Reference Decoupling Nodes Analog Power Supply Input Receive Path − and + Analog Inputs TxDAC Full-Scale Current Adjust TxDAC Reference Input/Output −Tx Amp Current Output_Sink −Tx Mirror Current Output_Sink +Tx Amp Current Output_Sink +Tx Mirror Current Output_Sink −TxDAC Current Output_Source +TxDAC Current Output_Source Digital Interface Mode Select Input LOW = HD, HIGH = FD Power-Up SPI Register Default Setting Input Clock Oscillator/Synthesizer Supply Return Crystal Oscillator Inverter Output Crystal Oscillator Inverter Input Clock Oscillator/Synthesizer Supply Digital Supply Return Digital Supply Input fOSCIN/L Clock Output, (L = 1, 2, or 4) Power-Down Input HD = half-duplex mode; FD = full-duplex mode. Rev. A | Page 11 of 48 AD9866 TYPICAL PERFORMANCE CHARACTERISTICS Rx PATH TYPICAL PERFORMANCE CHARACTERISTICS AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, fOSCIN = fADC = 50 MSPS, low-pass filter’s f−3 dB = 22 MHz, AIN = −1 dBFS, RIN = 50 Ω, half- or full-duplex interface, default power bias settings. 65 –10 –20 10.5 1MHz 5MHz 10MHz 15MHz 20MHz 62 59 SINAD (dBFS) –30 –40 –50 –60 –70 9.5 56 9.0 53 8.5 50 8.0 47 7.5 44 7.0 –90 0 6.25 12.50 18.75 41 –6 25.00 0 6 12 18 24 30 RxPGA GAIN (dB) FREQUENCY (MHz) 36 04560-0-006 –80 –100 6.5 48 42 Figure 6. SINAD/ENOB vs. RxPGA Gain and Frequency Figure 3. Spectral Plot with 4k FFT of Input Sinusoid with RxPGA = 0 dB and PIN = 9 dBm –55 –30 1MHz 5MHz 10MHz 15MHz 20MHz RBW = 12.2kHz –40 –60 –50 –60 –65 –70 THD (dBc) INPUT REFERRED SPECTRUM (dBm) 10.0 ENOB (Bits) FUND = –1dBFS SINAD = 61.9dBFS ENOB = 10BITS SNR = 64.5dBFS THD = –65.4dBFS SFDR = –64.9dBc (THIRD HARMONIC) RBW = 12.21kHz 0 04560-0-003 REFERRED TO INPUT SPECTRUM (dBm) 10 –80 –90 –70 –75 –100 –110 5 10 15 FREQUENCY (MHz) 20 57 –56 62 –62 59 –68 THD @ 3.14V THD @ 3.3V THD @ 3.46V 54 –74 6 12 18 24 30 RxPGA GAIN (dB) 36 42 48 –45 SINAD @ +25°C SINAD @ +85°C SINAD @ –40°C –50 –55 THD @ +25°C THD @ +85°C THD @ –40°C 56 –60 53 –65 51 –80 50 –70 48 –86 47 –75 45 –21 –18 –15 –12 –9 –6 INPUT AMPLITUDE (dBFS) –3 0 –92 04560-0-005 SINAD (dBFS) 60 65 SINAD (dBFS) SINAD @ 3.14V SINAD @ 3.3V SINAD @ 3.46V –50 THD (dBFS) 63 0 Figure 7. THD vs. RxPGA Gain and Frequency Figure 4. Spectral Plot with 4k FFT of 84-Carrier DMT Signal with PAR = 10.2 dB, PIN = −33.7 dBm, and RxPGA = 36 dB 66 04560-0-007 –85 –6 25 44 –6 0 6 12 18 24 30 RxPGA GAIN (dB) 36 42 (0dBFS = 2V p-p) Figure 8. SINAD/THD Performance vs. RxPGA Gain and Temperature ( fIN = 5 MHz) Figure 5. SINAD and THD vs. Input Amplitude and Supply (fIN = 8 MHz, LPF f−3 dB = 26 MHz; Rx PGA = 0 dB) Rev. A | Page 12 of 48 48 –80 THD (dBc) 0 04560-0-008 –130 –80 04493-0-041 –120 AD9866 Rx PATH TYPICAL PERFORMANCE CHARACTERISTICS AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, fOSCIN = fADC = 80 MSPS, low-pass filter’s f−3 dB = 30 MHz, AIN = −1 dBFS, RIN = 50 Ω, half- or full-duplex interface, default power bias settings. 65 –10 –20 10.5 5MHz 10MHz 15MHz 20MHz 30MHz 62 59 SINAD (dBFS) –30 –40 –50 –60 –70 10.0 9.5 56 9.0 53 8.5 50 8.0 47 7.5 44 7.0 ENOB (Bits) FUND = –1dBFS SINAD = 62.4dBFS ENOB = 10.1BITS SNR = 63.4dBFS THD = –69.3dBFS SFDR = –70.5dBc (THIRD HARMONIC) RBW = 19.53kHz 0 –90 –100 0 10 20 30 41 –6 40 0 6 12 FREQUENCY (MHz) Figure 9. Spectral Plot with 4k FFT of Input Sinusoid with RxPGA = 0 dB and PIN = 9 dBm 18 24 30 RxPGA GAIN (dB) 36 42 04560-0-012 –80 04560-0-009 REFERRED TO INPUT SPECTRUM (dBm) 10 6.5 48 Figure 12. SINAD/ENOB vs. RxPGA Gain and Frequency –30 –55 RBW = 19.53kHz –60 –50 –60 –65 THD (dBc) –70 –80 –90 –70 –75 0 10 20 FREQUENCY (MHz) 30 –85 –6 40 Figure 10. Spectral Plot with 4K FFT of 111-Carrier DMT Signal with PAR = 11 dB, PIN = −33.7 dBm, LPF's f−3 dB = 32 MHz and RxPGA = 36 dB 66 SINAD @ 3.14V SINAD @ 3.3V SINAD @ 3.46V THD @ 3.14V THD @ 3.3V THD @ 3.46V 54 –74 51 –80 48 –86 45 –21 –18 –15 –12 –9 –6 INPUT AMPLITUDE (dBFS) –3 0 –92 SINAD (dBFS) –68 THD (dBFS) 57 18 24 30 RxPGA GAIN (dB) 36 42 48 –40 SINAD @ +25°C SINAD @ +85°C –45 SINAD @ –40°C 61 04560-0-011 –62 12 65 –56 60 6 Figure 13. THD vs. RxPGA Gain and Frequency –50 63 0 59 –50 56 –55 53 –60 50 –65 47 –70 THD @ +25°C THD @ +85°C THD @ –40°C 44 41 –6 0 6 12 18 24 30 RxPGA GAIN (dB) –75 36 42 –80 48 THD (dBc) –130 –80 04560-0-014 –120 SINAD (dBFS) 5MHz 10MHz 15MHz 20MHz 30MHz –110 04560-0-013 –100 04560-0-010 INPUT REFERRED SPECTRUM (dBm) –40 (0dBFS = 2V p-p) Figure 11. SINAD and THD vs. Input Amplitude and Supply (fIN = 8 MHz, LPF f−3 dB = 26 MHz; RxPGA = 0 dB) Figure 14. SINAD/THD Performance vs. RxPGA Gain and Temperature ( fIN = 10 MHz) Rev. A | Page 13 of 48 AD9866 63 –20 –54 62 –25 64.0 –56 61 63.5 –58 60 –60 THD @ 3.14V THD @ 3.3V THD @ 3.47V 62.5 62.0 –62 –64 –30 SNR vs. MSPS @ 3.0VSUP –35 SNR vs. MSPS @ 346VSUP 59 –40 SNR @ 3.13V 58 –45 THD @ 3.13V THD @ 3.46V 57 –50 THD @ 3.3V –66 56 –55 61.0 –68 55 –60 60.5 –70 54 –65 60.0 –72 30 5 10 15 20 INPUT FREQUENCY (MHz) 35 30 14 56.6 12 54.7 10 43.8 8 –40°C 6 +85°C 21.9 4 +25°C 2 30 36 RxPGA GAIN (dB) 0 48 42 43 42 41 40 39 38 04560-0-019 76.6 44 SNR (dB) 16 NOISE SPECTRAL DENSITY (nV/ Hz) 87.5 24 0 3 0.3 GAIN STEP ERROR (dB) 0.4 2 1 0 –1 –2 DEVICE 1 DEVICE 2 DEVICE 3 DEVICE 4 0 6 12 18 24 30 40 50 60 70 80 30 0.2 0.1 0 –0.1 –0.2 –0.3 04560-0-017 DC OFFSET (% of full scale) 0.5 4 –5 –6 20 Figure 19. SNR vs. Filter Cutoff Frequency (50 MSPS; fIN = 5 MHz; AIN = −1 dB; RxPGA = 48 dB) 5 –4 10 CUTOFF FREQUENCY (MHz) Figure 16. Input Referred Integrated Noise and Noise Spectral Density vs. RxPGA Gain (LPF f−3 dB = 26 MHz) –3 –70 80 70 45 04560-0-016 18 INTEGRATED NOISE (µV rms) 20 98.5 0 18 60 Figure 18. SNR and THD vs. Sample Rate and Supply (LPF Disabled; RxPGA = 0 dB; fIN = 8 MHz) 109.4 10.9 50 INPUT FREQUENCY (MHz) Figure 15. SNR and THD vs. Input Frequency and Supply ( LPF f−3 dB = 26 MHz; RxPGA = 0 dB) 32.8 40 36 42 AD9866: GAIN STEP ERROR @ +25°C AD9866: GAIN STEP ERROR @ +85°C AD9866: GAIN STEP ERROR @ –40°C –0.4 –0.5 –6 48 GAIN (dB) 0 6 12 18 24 30 36 42 RxPGA GAIN (dB) Figure 20. RxPGA Gain Step Error vs. Gain (fIN = 10 MHz) Figure 17. Rx DC Offset vs. RxPGA Gain Rev. A | Page 14 of 48 04560-0-020 0 04560-0-015 61.5 53 20 THD (dBc) 63.0 THD (dBc) SNR (dBFS) 64.5 04560-0-018 SNR @ 3.14V SNR @ 3.3V SNR @ 3.47V SNR (dBFS) –52 65.0 48 AD9866 Rx PATH TYPICAL PERFORMANCE CHARACTERISTICS AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, fOSCIN = fADC = 50 MSPS, low-pass filter disabled, RxPGA = 0 dB, AIN = −1 dBFS, RIN = 50 Ω, half- or full-duplex interface, default power bias settings. 2048 1408 1280 1792 1152 1536 1280 896 CODE CODE 1024 1024 768 640 768 04560-0-021 256 0 80 160 240 320 400 480 560 640 04560-0-024 512 512 384 256 720 0 80 160 240 320 TIME (ns) Figure 21. RxPGA Settling Time −12 dB to +48 dB Transition for DC Input (fADC = 50 MSPS, LPF Disabled) 640 720 –6dB GAIN 0dB GAIN +6dB GAIN –2 –4 FUNDAMENTAL (dB) –6 –9 –12 –6 –8 –10 +18dB GAIN +30dB GAIN +42dB GAIN –12 –14 04560-0-022 0 5 10 15 20 25 30 35 40 45 04560-0-025 –16 –15 –18 –20 50 0 5 10 15 INPUT FREQUENCY (MHz) 20 25 30 35 40 45 50 INPUT FREQUENCY (MHz) Figure 22. Rx Low-Pass Filter Amplitude Response vs. Supply (fADC = 50 MSPS, f−3 dB = 33 MHz, RxPGA = 0 dB) Figure 25. Rx Low-Pass Filter Amplitude Response vs. RxPGA Gain (LPF's f−3 dB = 33 MHz) 140 420 10 410 9 400 8 130 TxDAC ISOLATION @ 0dB 7 RESISTANCE (Ω) 390 110 100 90 RIN 380 6 370 5 4 360 CIN 350 3 CAPACITANCE (pF) 120 80 04560-0-023 IAMP ISOLATION @ 0dB 70 60 0 5 10 15 20 25 30 340 2 330 1 320 35 FREQUENCY (MHz) 5 15 25 35 45 55 65 75 85 FREQUENCY (MHz) Figure 23. Rx to Tx Full-Duplex Isolation @ 0 RxPGA Setting (Note: ATTEN @ RxPGA = x dB = ATTEN @ RxPGA = 0 dB − RxPGA Gain) Figure 26. Rx Input Impedance vs. Frequency Rev. A | Page 15 of 48 95 0 105 04493-0-026 AMPLITUDE RESPONSE (dB) 560 0 3.3V 3.0V 3.6V –3 ATTEN @RxPGA = 0dB (dB) 480 Figure 24. RxPGA Settling Time for 0 dB to +5 dB Transition for DC Input (fADC = 50 MSPS, LPF Disabled) 0 –18 400 TIME (ns) AD9866 TxDAC PATH TYPICAL PERFORMANCE CHARACTERISTICS 10 0 0 –10 –10 –20 –20 –30 –30 –40 –40 –50 –50 –60 –60 –70 –80 0 5 10 15 20 04560-0-030 dBm 10 04493-0-072 dBm AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, fOSCIN = 50 MSPS and 80 MSPS, RSET = 1.96 kΩ, 2:1 transformer coupled output (see Figure 63) into 50 Ω load half-or full-duplex interface, default power bias settings. –70 –80 25 0 5 10 FREQUENCY (MHz) –65 –65 –70 –70 IMD (dBFS) –75 4dBm 7dBm –80 10dBm –85 35 40 –75 4dBm 7dBm –80 2.5 5.0 7.5 10.0 12.5 15.0 17.5 –90 20.0 04560-0-031 04560-0-028 0 0 5 15 20 25 30 Figure 31. 2-Tone IMD Frequency Sweep vs. Peak Power with fDATA = 80 MSPS, 2× Interpolation –70 –70 SFDR (dBFS) –65 –75 4dBm 7dBm –80 –85 (RELATIVE TO PEAK POWER) –65 10dBm 10 2-TONE CENTER FREQUENCY (MHz) 10dBm –75 4dBm –80 7dBm 0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 –90 20.0 2-TONE CENTER FREQUENCY (MHz) 04560-0-032 –85 04560-0-029 SFDR (dBFS) 30 –85 Figure 28. 2-Tone IMD Frequency Sweep vs. Peak Power with fDATA = 50 MSPS, 4× Interpolation (RELATIVE TO PEAK POWER) 25 10dBm 2-TONE CENTER FREQUENCY (MHz) –90 20 Figure 30. Dual-Tone Spectral Plot of TxDAC's Output (fDATA = 80 MSPS, 2× Interpolation, 10 dBm Peak Power, F1 = 27.1 MHz, F2 = 28.7 MHz) (RELATIVE TO PEAK POWER) IMD (dBFS) (RELATIVE TO PEAK POWER) Figure 27. Dual-Tone Spectral Plot of TxDAC's Output (fDATA = 50 MSPS, 4× Interpolation, 10 dBm Peak Power, F1 = 17 MHz, F2 = 18 MHz) –90 15 FREQUENCY (MHz) 0 5 10 15 20 25 30 2-TONE CENTER FREQUENCY (MHz) Figure 29. 2-Tone Worst Spur Frequency Sweep vs. Peak Power with fDATA = 50 MSPS, 4× Interpolation Figure 32. 2-Tone Worst Spur Frequency Sweep vs. Peak Power with fDATA = 80 MSPS, 2× Interpolation Rev. A | Page 16 of 48 AD9866 –30 PAR = 11.4 RMS = –1.4dBm –30 –40 –50 –50 dBm –40 –60 –60 –70 –70 –80 –80 04560-0-033 dBm –20 PAR = 11.4 RMS = –1.4 dBm –90 –100 0 5 10 15 20 04493-0-081 –20 –90 –100 25 0 5 10 15 20 25 Figure 33. Spectral Plot of 84-Carrier OFDM Test Vector fDATA = 50 MSPS, 4× Interpolation) PAR = 11.4 RMS = –1.4dBm –40 –40 –50 –50 –60 –60 –70 –70 –80 –80 –90 0 25 50 75 100 125 150 175 04493-0-082 dBm –30 04493-0-079 dBm –30 –90 –100 200 0 20 40 60 80 100 120 140 160 FREQUENCY (MHz) FREQUENCY (MHz) Figure 34. Wideband Spectral Plot of 88-Subcarrier OFDM Test Vector (fDATA = 50 MSPS, 4× Interpolation) Figure 37. Wideband Spectral Plot of 111-Carrier OFDM Test Vector (fDATA = 80 MSPS, 2× Interpolation) 105 100 100 95 2-TONE IMD 2-TONE IMD 85 80 SNR 75 70 04560-0-035 65 60 55 –24 –21 –18 –15 –12 –9 –6 –3 0 AOUT (dBFS) 85 80 75 SNR 70 65 04560-0-038 90 90 (RELATIVE TO PEAK POWER) SNR AND 2-TONE IMD (dBFS) 95 (RELATIVE TO PEAK POWER) 40 –20 PAR = 11.4 RMS = –1.4dBm SNR AND 2-TONE IMD (dBFS) 35 Figure 36. Spectral Plot of 111-Carrier OFDM Test Vector (fDATA = 80 MSPS, 2× Interpolation) –20 –100 30 FREQUENCY (MHz) FREQUENCY (MHz) 60 55 –24 –21 –18 –15 –12 –9 –6 –3 AOUT (dBFS) Figure 35. SNR and SFDR vs. POUT (fOUT = 12.55 MHz, fDATA = 50 MSPS, 4× Interpolation) Figure 38. SNR and SFDR vs. POUT (fOUT = 20 MHz, fDATA = 80 MSPS, 2× Interpolation) Rev. A | Page 17 of 48 0 AD9866 IAMP PATH TYPICAL PERFORMANCE CHARACTERISTICS 48 RBW = 2.3kHz 2.5MHz 46 5MHz 44 42 OIP3 (dBm) 40 10MHz 38 20MHz 15MHz 36 34 0 5 10 15 20 04493-0-087 20 15 10 5 0 –5 –10 –15 –20 –25 –30 –35 –40 –45 –50 –55 –60 04493-0-084 dBm AVDD = CLKVDD = DVDD = DRVDD = 3.3 V, fOSCIN = 50 MSPS, RSET = 1.58 kΩ, 1:1 transformer coupled output (see Figure 64 and Figure 65) into 50 Ω load, half- or full-duplex interface, default power bias settings. 32 30 3.0 25 3.5 4.0 FREQUENCY (MHz) 4.5 5.0 VCM (V) Figure 39. Dual-Tone Spectral Plot of IAMPN Output (IAMP Settings of I = 12.5 mA, N = 4, G = 0, 2:1 Transformer into 75 Ω Loader, VCM = 4.8 V) Figure 42. IOUTN Third-Order Intercept vs. Common-Mode Voltage (IAMP Settings of I = 12.5 mA, N = 4, G = 0, 2:1 Transformer into 75 Ω Load) 0 42 PAR = 11.4 RMS = 10.3dBm –10 40 2.5MHz –20 38 OIP3 (dBm) –40 –50 5MHz 36 10MHz 34 –60 15MHz 04493-0-085 –70 –80 32 0 5 10 15 20 20MHz 30 3.0 25 3.5 4.0 FREQUENCY (MHz) 5.0 Figure 43. IOUTG Third-Order Intercept vs. Common-Mode Voltage (IAMP Settings of I = 4.25 mA, N = 0, G = 6, 2:1 Transformer into 75 Ω Load) 0 0 PAR = 11.4 RMS = 10.4dBm –20 –30 –30 dBm –20 –40 –40 –50 –60 –60 04493-0-086 –50 –70 0 5 10 15 20 PAR = 11.4 RMS = 9.8dBm RBW = 10kHz –10 04493-0-089 –10 dBm 4.5 VCM (V) Figure 40. Spectral Plot of 84-Carrier OFDM Test Vector Using IAMPN in Current-Mode Configuration (IAMP Settings of I = 10 mA, N = 4, G = 0; VCM = 4.8 V) –80 04493-0-088 dBm –30 –70 –80 25 FREQUENCY (MHz) 0 5 10 15 20 25 FREQUENCY (MHz) Figure 41. Spectral Plot of 84-Carrier OFDM Test Vector Using IAMP in Voltage-Mode Configuration with AVDD = 5 V (PBR951 Transistors, IAMP Settings of I = 6 mA, N = 2, G = 6) Figure 44. Spectral Plot of 84-Carrier OFDM Test Vector Using IAMP in Voltage-Mode Configuration with AVDD = 3.3 V (PBR951 Transistors, IAMP Settings of I = 6 mA, N = 2, G = 6) Rev. A | Page 18 of 48 AD9866 SERIAL PORT Table 10. SPI Register Mapping Address (Hex) 1 Bit Breakdown Power-Up Default Value MODE = 0 (Half-Duplex) Description Width CONFIG = 0 SPI PORT CONFIGURATION AND SOFTWARE RESET 0x00 (7) 4-Wire SPI 1 0 (6) LSB First 1 0 (5) S/W Reset 1 0 POWER CONTROL REGISTERS (via PWR_DWN pin) 0x01 (7) 1 0 Clock Syn. 0x02 (6) (5) (4) (3) (2) (1) (0) (7) TxDAC/IAMP Tx Digital REF ADC CML ADC PGA Bias RxPGA CLK Syn. (6) TxDAC/IAMP (5) Tx Digital (4) REF (3) ADC CML (2) ADC (1) PGA Bias (0) RxPGA HALF-DUPLEX POWER CONTROL 0x03 (7:3) Tx OFF Delay (2) Rx _TXEN (1) Tx PWRDN (0) Rx PWRDN CONFIG = 0 CONFIG = 1 Comments 0 0 0 0 0 0 0 0 0 Default SPI configuration is 3-wire, MSB first. 0 0 0 PWR_DWN = 0. Default setting is for all blocks powered on. 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1* 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5 1 1 1 0xFF PLL CLOCK MULTIPLIER/SYNTHESIZER CONTROL 0x04 (5) Duty Cycle Enable 1 0 (4) fADC from PLL 1 0 (3:2) (1:0) 0x05 (2) (1) (0) 0x06 (7:6) (5) (4) (3:2) (1) (0) Rx PATH CONTROL 0x07 (5) (4) (0) MODE = 1 (Full-Duplex) CONFIG = 1 0xFF N/A N/A 0 0 0 0 0 0 PLL Divide-N PLL Multiplier-M OSCIN to RXCLK Invert RXCLK Disabled RXCLK CLKOUT2 Divide CLKOUT2 Invert CLKOUT2 Disable CLKOUT1 Divide CLKOUT1 Invert CLKOUT1 Disable 2 2 1 1 1 2 1 1 2 1 1 00 01 0 0 0 01 0 0 01 0 0 00 10* 0 0 0 01 0 0 01 0 0 00 01 0 0 0 01 0 0 01 0 0 00 01 1* 0 0 01 0 1* 01 0 1* Initiate Offset Cal. Rx Low Power Rx Filter ON 1 1 1 0 0 1 0 1* 1 0 0 1 0 1* 1 Rev. A | Page 19 of 48 PWR_DWN = 1. Default setting* is for all functional blocks powered down except PLL. *MODE = CONFIG = 1. Setting has PLL powered down with OSCIN input routed to RXCLK output. Default setting is for TXEN input to control power on/off of Tx/Rx path. Tx driver delayed by 31 1/fDATA clock cycles. Default setting is Duty Cycle Restore disabled, ADC CLK from OSCIN input, and PLL multiplier × 2 setting. *PLL multiplier × 4 setting. Full-duplex RXCLK normally at nibble rate. *Exception on power-up. Default setting is CLKOUT2 and CLKOUT1 enabled with divide-by-2. *CLKOUT1 and CLKOUT2 disabled. Default setting has LPF ON and Rx path at nominal power bias setting. *Rx path to low power. AD9866 Address (Hex) 1 Bit Breakdown 0x08 (7:0) Power-Up Default Value MODE = 0 (Half-Duplex) Description Width MODE = 1 (Full-Duplex) CONFIG = 0 CONFIG = 1 CONFIG = 0 CONFIG = 1 Comments 0x80 0x61 0x80 0x80 Refer to Low-Pass Filter section. Rx Filter Tuning Cutoff Frequency Tx/Rx PATH GAIN CONTROL 0x09 (6) Use SPI Rx Gain (5:0) Rx Gain Code 8 1 6 0x00 0x00 0x00 0x00 0x0A 1 6 0x7F 0x7F 0x7F 0x7F 1 1 1 1 1 0 1 0 0 0 0 1 0 0 1** 0 1 0 1* 0 0 1 0 1* 0 Default setting is RxPGA control active. *Tx port with GAIN strobe (AD9875/AD9876 compatible). **3-bit RxPGA gain map (AD9975 compatible). 2 01 00 01 01 1 0 0 0 0 1 1 1 N/A 0 0 N/A 0 0 0 0 1 0 0 1 Default setting is 2× interpolation with LPF response. Data format is straight binary for half-duplex and twos complement for full-duplex interface. *Full-duplex only. 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0x00 0 0x00 (6) Use SPI Tx Gain (5:0) Tx Gain Code Tx AND Rx PGA CONTROL 0x0B (6) PGA Code for Tx (5) PGA Code for Rx (3) Force GAIN strobe (2) Rx Gain on Tx Port (1) 3-Bit RxPGA Port Tx DIGITAL FILTER AND INTERFACE 0x0C (7:6) Interpolation Factor (4) Invert TXEN/TXSYNC (2) LS Nibble First* (1) TXCLK neg. edge (0) Twos complement Rx INTERFACE AND ANALOG/DIGITAL LOOPBACK 0x0D (7) Analog Loopback 1 0 0 (6) Digital Loopback* 1 0 0 (5) Rx Port 3-State 1 N/A N/A (4) 1 0 0 Invert RXEN/RXSYNC (2) LS Nibble First* 1 N/A N/A (1) RXCLK neg. edge 1 0 0 (0) Twos complement 1 0 0 DIGITAL OUTPUT DRIVE STRENGTH, TxDAC OUTPUT, AND REV ID 0x0E (7) 1 0 0 Low Drive Strength (0) TxDAC Output 1 0 0 0x0F (3:0) REV ID Number 4 0x00 0x00 Tx IAMP GAIN AND BIAS CONTROL 0x10 (7) Select Tx Gain 1 0x44 0x44 (6:4) G1 3 (2:0) N 3 0x11 (6:4) G2 3 0x62 0x62 (2:0) G3 3 0x12 (6:4) (2:0) Stand_Secondary Stand_Primary 3 3 0x01 0x01 Rev. A | Page 20 of 48 0x44 0x44 0x62 0x62 0x01 0x01 Default setting is for hardware Rx gain code via PGA or Tx data port. Default setting is for Tx gain code via SPI control. Data format is straight binary for half-duplex and twos complement for fullduplex interface. Analog loopback: ADC Rx data fed back to TxDAC. Digital loopback: Tx input data to Rx output port. *Full-duplex only. Default setting is for high drive strength and IAMP enabled. Secondary path G1 = 0, 1, 2, 3, 4. Primary path N = 0, 1, 2, 3, 4. Secondary path stages: G2 = 0 to 1.50 in 0.25 steps and G3 = 0 to 6. Standing current of primary and secondary path. AD9866 Address (Hex) 1 0x13 1 Power-Up Default Value Bit Breakdown Description Width CONFIG = 0 CONFIG = 1 CONFIG = 0 CONFIG = 1 (7:5) (4:3) (2:0) CPGA Bias Adjust SPGA Bias Adjust ADC Bias Adjust 3 2 4 0x00 0x00 0x00 0x00 MODE = 0 (Half-Duplex) MODE = 1 (Full-Duplex) Comments Current bias setting for Rx path’s functional blocks. Refer to Page 41. Bits that are undefined should always be assigned a 0. Table 11. SPI Registers Pertaining to SPI Options REGISTER MAP DESCRIPTION The AD9866 contains a set of programmable registers described in Table 10 that are used to optimize its numerous features, interface options, and performance parameters from its default register settings. Registers pertaining to similar functions have been grouped together and assigned adjacent addresses to minimize the update time when using the multibyte serial port interface (SPI) read/write feature. Bits that are undefined within a register should be assigned a 0 when writing to that register. The default register settings were intended to allow some applications to operate without the use of an SPI. The AD9866 can be configured to support a half- or full-duplex digital interface via the MODE pin, with each interface having two possible default register settings determined by the setting of the CONFIG pin. For instance, applications that need to use only the Tx or Rx path functionality of the AD9866 can configure it for a halfduplex interface (MODE = 0), and use the TXEN pin to select between the Tx or Rx signal path with the unused path remaining in a reduced power state. The CONFIG pin can be used to select the default interpolation ratio of the Tx path and RxPGA gain mapping. SERIAL PORT INTERFACE (SPI) The serial port of the AD9866 has 3- or 4-wire SPI capability allowing read/write access to all registers that configure the device’s internal parameters. Registers pertaining to the SPI are listed in Table 11. The default 3-wire serial communication port consists of a clock (SCLK), serial port enable (SEN), and a bidirectional data (SDIO) signal. SEN is an active low control gating read and write cycle. When SEN is high, SDO and SDIO are three-stated. The inputs to SCLK, SEN, and SDIO contain a Schmitt trigger with a nominal hysteresis of 0.4 V centered about VDDH/2. The SDO pin remains three-stated in a 3-wire SPI interface. Address (Hex) 0x00 Bit (7) (6) Description Enable 4-wire SPI Enable SPI LSB first A 4-wire SPI can be enabled by setting the 4-wire SPI bit high, causing the output data to appear on the SDO pin instead of on the SDIO pin. The SDIO pin serves as an input-only throughout the read operation. Note that the SDO pin is active only during the transmission of data and remains three-stated at any other time. An 8-bit instruction header must accompany each read and write operation. The instruction header is shown in Table 12. The MSB is an R/Windicator bit with logic high indicating a read operation. The next two bits, N1 and N0, specify the number of bytes (one to four bytes) to be transferred during the data transfer cycle. The remaining five bits specify the address bits to be accessed during the data transfer portion. The data bits immediately follow the instruction header for both read and write operations. Table 12. Instruction Header Information MSB 17 R/W 16 N1 15 N0 14 A4 13 A3 12 A2 LSB 11 A1 10 A0 The AD9866 serial port can support both MSB (most significant bit) first and LSB (least significant bit) first data formats. Figure 45 illustrates how the serial port words are built for the MSB first and LSB first modes. The bit order is controlled by the SPI LSB first bit (Register 0, Bit 6). The default value is 0, MSB first. Multibyte data transfers in MSB format can be completed by writing an instruction byte that includes the register address of the last address to be accessed. The AD9866 automatically decrements the address for each successive byte required for the multibyte communication cycle. Rev. A | Page 21 of 48 AD9866 INSTRUCTION CYCLE SEN tS 1/fSCLK DATA TRANSFER CYCLE tLOW tHI SCLK SCLK R/W N1 N2 A4 A3 A2 A1 A0 D71 D61 tDS D1N D0N tDH SDIO R/W N1 N0 A0 D6 D1 D7 D0 04560-0-046 SDATA tH SEN Figure 46. SPI Write Operation Timing INSTRUCTION CYCLE SEN DATA TRANSFER CYCLE SCLK A0 A1 A2 A3 A4 N2 N1 R/W D01 D11 D6N D7N 04560-0-045 SDATA Figure 45. SPI Timing, MSB First (Upper) and LSB First (Lower) When the SPI LSB first bit is set high, the serial port interprets both instruction and data bytes LSB first. Multibyte data transfers in LSB format can be completed by writing an instruction byte that includes the register address of the first address to be accessed. The AD9866 automatically increments the address for each successive byte required for the multibyte communication cycle. Figure 47 illustrates the timing for a 3-wire read operation to the SPI port. After SEN goes low, data (SDIO) pertaining to the instruction header is read on the rising edges of SCLK. A read operation occurs if the read/not-write indicator is set high. After the address bits of the instruction header are read, the eight data bits pertaining to the specified register are shifted out of the SDIO pin on the falling edges of the next eight clock cycles. If a multibyte communication cycle is specified in the instruction header, a similar process as previously described for a multibyte SPI write operation applies. The SDO pin remains three-stated in a 3-wire read operation. tS 1/fSCLK SEN tLOW tHI tDV tDS tDH SDIO R/W tEZ N1 A2 A1 A0 D7 D6 D1 D0 Figure 47. SPI 3-Wire Read Operation Timing Figure 48 illustrates the timing for a 4-wire read operation to the SPI port. The timing is similar to the 3-wire read operation with the exception that data appears at the SDO pin, while the SDIO pin remains high impedance throughout the operation. The SDO pin is an active output only during the data transfer phase and remains three-stated at all other times. tS 1/fSCLK SEN tLOW tHI SCLK tDS SDIO tEZ tDH R/W N1 A2 A1 A0 tEZ tDV SDO D7 D6 D1 D0 Figure 48. SPI 4-Wire Read Operation Timing Rev. A | Page 22 of 48 04560-0-048 Figure 46 illustrates the timing requirements for a write operation to the SPI port. After the serial port enable (SEN) signal goes low, data (SDIO) pertaining to the instruction header is read on the rising edges of the clock (SCLK). To initiate a write operation, the read/not-write bit is set low. After the instruction header is read, the eight data bits pertaining to the specified register are shifted into the SDIO pin on the rising edge of the next eight clock cycles. If a multibyte communication cycle is specified, the destination address is decremented (MSB first) and shifts in another eight bits of data. This process repeats until all the bytes specified in the instruction header (N1, N0 bits) are shifted into the SDIO pin. SEN must remain low during the data transfer operation, only going high after the last bit is shifted into the SDIO pin. 04560-0-047 SCLK AD9866 DIGITAL INTERFACE HALF-DUPLEX MODE The half-duplex mode functions as follows when the MODE pin is tied low. The bidirectional ADIO port is typically shared in burst fashion between the transmit path and receive path. Two control signals, TXEN and RXEN, from a DSP (or digital ASIC) control the bus direction by enabling the ADIO port’s input latch and output driver, respectively. Two clock signals are also used: TXCLK to latch the Tx input data, and RXCLK to clock the Rx output data. The ADIO port can also be disabled by setting TXEN and RXEN low (default setting), thus allowing it to be connected to a shared bus. Internally, the ADIO port consists of an input latch for the Tx path in parallel with an output latch with three-state outputs for the Rx path. TXEN is used to enable the input latch; RXEN is used to three-state the output latch. A five-sample-deep FIFO is used on the Tx and Rx paths to absorb any phase difference between the AD9866’s internal clocks and the externally supplied clocks (TXCLK, RXCLK). The ADIO bus accepts input datawords into the transmit path when the TXEN pin is high, the RXEN pin is low, and a clock is present on the TXCLK pin, as shown in Figure 49. data appears on the bus after a 6-clock-cycle delay due to the internal FIFO delay. Note that Rx data is not latched back into the Tx path, if TXEN is high during this interval with TXCLK present. The ADIO bus becomes three-stated once the RXEN pin returns low. Figure 50 shows the receive path output timing. RXCLK RXEN tOD tPLZ tVT ADIO[9:0] RX0 RX1 RX2 RX3 Figure 50. Receive Data Output Timing Diagram To add flexibility to the digital interface port, several programming options are available in the SPI registers. These options are listed in Table 13. The default Tx and Rx data input formats are straight binary, but can be changed to twos complement. The default TXEN and RXEN settings are active high, but can be set to opposite polarities, thus allowing them to share the same control. In this case, the ADIO port can still be placed onto a shared bus by disabling its input latch via the control signal, and disabling the output driver via the SPI register. The clock timing can be independently changed on the transmit and receive paths by selecting either the rising or falling clock edge as the validating/sampling edge of the clock. Lastly, the output driver’s strength can be reduced for lower data rate applications. Table 13. SPI Registers for Half-Duplex Interface Address (Hex) 0x0C 0x0D 0x0E tDS tPZL 04560-0-050 The digital interface port is configurable for half-duplex or fullduplex operation by pin-strapping the MODE pin low or high, respectively. In half-duplex mode, the digital interface port becomes a 10-bit bidirectional bus called the ADIO port. In full-duplex mode, the digital interface port is divided into two 6-bit ports called Tx[5:0] and Rx[5:0] for simultaneous Tx and Rx operations. In this mode, data is transferred between the ASIC and AD9866 in 6-bit nibbles. The AD9866 also features a flexible digital interface for updating the RxPGA and TxPGA gain registers via a 6-bit PGA port or Tx[5:0] port for fast updates, or via the SPI port for slower updates. See the RXPGA Control section for more information. Bit (4) (1) (0) (5) (4) (1) (0) (7) Description Invert TXEN TXCLK negative edge Twos complement Rx port three-state Invert RXEN RXCLK negative edge Twos complement Low digital drive strength TXCLK ADIO[9:0] tEN TX0 tDIS tDH TX1 TX2 TX3 TX4 04560-0-049 TXEN RXEN Figure 49. Transmit Data Input Timing Diagram The Tx interpolation filter(s) following the ADIO port can be flushed with zeros, if the clock signal into the TXCLK pin is present for 33 clock cycles after TXEN goes low. Note that the data on the ADIO bus is irrelevant over this interval. The half-duplex interface can be configured to act like a slave or a master to the digital ASIC. An example of a slave configuration is shown in Figure 51. In this example, the AD9866 accepts all the clock and control signals from the digital ASIC. Because the sampling clocks for the DAC and ADC are derived internally from the OSCIN signal, it is required that the TXCLK and RXCLK signals be at exactly the same frequency as the OSCIN signal. The phase relationships among the TXCLK, RXCLK, and OSCIN signals can be arbitrary. If the digital ASIC cannot provide a low jitter clock source to OSCIN, use the AD9866 to generate the clock for its DAC and ADC, and pass the desired clock signal to the digital ASIC via CLKOUT1 or CLKOUT2. The output from the receive path is driven onto the ADIO bus when the RXEN pin is high, and a clock is present on the RXCLK pin. While the output latch is enabled by RXEN, valid Rev. A | Page 23 of 48 AD9866 AD9866 12 ADIO [11:0] Tx/Rx Data[11:0] RXEN RXEN TXEN TXEN DAC_CLK TXCLK ADC_CLK RXCLK CLKOUT OSCIN FROM Rx ADC 04560-0-051 12 TO Tx DIGITAL FILTER Figure 51. Example of a Half-Duplex Digital Interface with AD9866 Serving as the Slave Figure 52 shows a half-duplex interface with the AD9866 acting as the master, generating all the required clocks. CLKOUT1 provides a clock equal to the bus data rate that is fed to the ASIC as well as back to the TXCLK and RXCLK inputs. This interface has the advantage of reducing the digital ASIC’s pin count by three. The ASIC needs only to generate a bus control signal that controls the data flow on the bidirectional bus. DIGITAL ASIC AD9866 ADIO [11:0] Tx/Rx Data[11:0] 12 12 TO Tx DIGITAL FILTER nibble rate. Therefore, the 2× or 4× interpolation filter must be used with a full-duplex interface. The AD9866 acts as the master, providing RXCLK as an output clock that is used for the timing of both the Tx[5:0] and Rx[5:0] ports. RXCLK always runs at the nibble rate and can be inverted or disabled via an SPI register. Because RXCLK is derived from the clock synthesizer, it remains active, provided that this functional block remains powered on. A buffered version of the signal appearing at OSCIN can also be directed to RXCLK by setting Bit 2 of Register 0x05. This feature allows the AD9866 to be completely powered down (including the clock synthesizer) while serving as the master. The Tx[5:0] port operates in the following manner with the SPI register default settings. Two consecutive nibbles of the Tx data are multiplexed together to form a 10-bit data-word in twos complement format. The clock appearing on the RXCLK pin is a buffered version of the internal clock used by the Tx[5:0] port’s input latch with a frequency that is always twice the ADC sample rate (2 × fADC). Data from the Tx[5:0] port is read on the rising edge of this sampling clock, as illustrated in the timing diagram shown in Figure 53. Note, TXQUIET must remain high for the reconstructed Tx data to appear as an analog signal at the output of the TxDAC or IAMP. tDS RXCLK FROM Rx ADC tHD TXSYNC RXEN TXEN BUS_CTR Tx[5:0] Tx0LSB Tx1MSB Tx1LSB Tx2MSB Tx3LSB Tx 2 LSB Tx3MSB 04560-0-053 DIGITAL ASIC TXCLK Figure 53. Tx[5:0] Port Full-Duplex Timing Diagram RXCLK CLKIN CLKOUT1 FROM CRYSTAL OR MASTER CLK 04560-0-052 OSCIN Figure 52. Example of a Half-Duplex Digital Interface with AD9866 Serving as the Master FULL-DUPLEX MODE The full-duplex mode interface is selected when the MODE pin is tied high. It can be used for full- or half-duplex applications. The digital interface port is divided into two 6-bit ports called Tx[5:0] and Rx[5:0], allowing simultaneous Tx and Rx operations for full-duplex applications. In half-duplex applications, the Tx[5:0] port can also be used to provide a fast update of the RxPGA (AD9876 backward compatible) during an Rx operation. This feature is enabled by default and can be used to reduce the required pin count of the ASIC (refer to RxPGA Control section for details). In either application, Tx and Rx data are transferred between the ASIC and AD9866 in 6-bit nibbles at twice the internal input/output word rates of the Tx interpolation filter and ADC. Note that the TxDAC update rate must not be less than the The TXSYNC signal is used to indicate to which word a nibble belongs. While TXSYNC is low, the first nibble of every word is read as the most significant nibble. The second nibble of that same word is read on the following TXSYNC high level as the least significant nibble. If TXSYNC is low for more than one clock cycle, the last transmit data is read continuously until TXSYNC is brought high for the second nibble of a new transmit word. This feature can be used to flush the interpolator filters with zeros. Note that the GAIN signal must be kept low during a Tx operation. The Rx[5:0] port operates in the following manner with the SPI register default settings. Two consecutive nibbles of the Rx data are multiplexed together to form a 12-bit data-word in twos complement format. The Rx data is valid on the rising edge of RXCLK, as illustrated in the timing diagram shown in Figure 54. The RXSYNC signal is used to indicate to which word a nibble belongs. While RXSYNC is low, the first nibble of every word is transmitted as the most significant nibble. The second nibble of that same word is transmitted on the following RXSYNC high level as the least significant nibble. Rev. A | Page 24 of 48 AD9866 Rx1MSB Rx1LSB Rx2MSB Rx3LSB Rx3MSB Figure 54. Full-Duplex Rx Port Timing To add flexibility to the full-duplex digital interface port, several programming options are available in the SPI registers. These options are listed in Table 14. The timing for the Tx[5:0] and/or Rx[5:0] ports can be independently changed by selecting either the rising or falling clock edge as the sampling/validating edge of the clock. Inverting RXCLK (via Bit 1 or Register 0x05) affects both the Rx and Tx interface, because they both use RXCLK. DIGITAL ASIC 0x0B 0x0C 0x0D 0x0E Bit (2) (1) (0) (2) (4) (3) Description OSCIN to RXCLK Invert RXCLK Disable RXCLK Rx gain on Tx port Invert TXSYNC (2) (1) (0) (5) (4) (3) (2) (1) (0) (7) LS nibble first TXCLK negative edge Twos complement Rx port three-state Invert RXSYNC 6 GAIN Tx[5:0] Tx Data[5:0] Rx[5:0] Rx Data[5:0] Table 14. SPI Registers for Full-Duplex Interface Address (Hex) 0x05 AD9865/AD9866 OPTIONAL RX_SYNC RXSYNC TX_SYNC TXSYNC 10/12 10/12 TO RxPGA TO Tx DIGITAL FILTER FROM RxADC RXCLK CLKIN CLKOUT1 CLKOUT2 NA OSCIN FROM CRYSTAL OR MASTER CLK 04560-0-055 Rx0LSB DEMUX Rx[5:0] 04560-0-054 tDV RXSYNC Figure 55 shows a possible digital interface between an ASIC and the AD9866. The AD9866 serves as the master generating the required clocks for the ASIC. This interface requires that the ASIC reserve 16 pins for the interface, assuming a 6-bit nibble width and the use of the Tx port for RxPGA gain control. Note that the ASIC pin allocation can be reduced by 3, if a 5-bit nibble width is used and the gain (or gain strobe) of the RxPGA is controlled via the SPI port. MUX tDH RXCLK Figure 55. Example of a Full-Duplex Digital Interface with Optional RxPGA Gain Control via Tx[5:0] NA RxPGA CONTROL LS nibble first RXCLK negative edge Twos complement Low drive strength The default Tx and Rx data input formats are twos complement, but can be changed to straight binary. The default TXSYNC and RXSYNC settings can be changed such that the first nibble of the word appears while TXSYNC, RXSYNC, or both are high. Also, the least significant nibble can be selected as the first nibble of the word (LS nibble first). The output driver strength can also be reduced for lower data rate applications. The AD9866 contains a digital PGA in the Rx path that is used to extend the dynamic range. The RxPGA can be programmed over a −12 dB to +48 dB with 1 dB resolution using a 6-bit word, and with a 0 dB setting corresponding to a 2 V p-p input signal. The 6-bit word is fed into a LUT that is used to distribute the desired gain over three amplification stages within the Rx path. Upon power-up, the RxPGA gain register is set to its minimum gain of −12 dB. The RxPGA gain mapping is shown in Figure 56. Table 15 lists the SPI registers pertaining to the RxPGA. Rev. A | Page 25 of 48 AD9866 48 tSU 42 RXCLK 36 24 Tx [5:0] GAIN 18 12 04560-0-057 GAIN (dB) tHD Tx SYNC 30 GAIN 6 Figure 57. Updating RxPGA via Tx[5:0] in Full-Duplex Mode 04560-0-056 0 –6 –12 0 6 12 18 24 30 36 42 48 54 60 6-BIT DIGITAL WORD-DECIMAL EQUIVALENT 66 Figure 56. Digital Gain Mapping of RxPGA Table 15. SPI Registers RxPGA Control Address (Hex) 0x09 0x0B Bit (6) (5:0) (6) (5) (3) (2) (1) Description Enable RxPGA update via SPI RxPGA gain code Select TxPGA via PGA[5:0] Select RxPGA via PGA[5:0] Enable software GAIN strobe – full-duplex Enable RxPGA update via Tx[5:0] – full-duplex 3-bit RxPGA gain mapping – half-duplex The RxPGA gain register can be updated via the Tx[5:0] port, the PGA[5:0] port, or the SPI port. The first two methods allow fast updates of the RxPGA gain register and should be considered for digital AGC functions requiring a fast closed-loop response. The SPI port allows direct update and readback of the RxPGA gain register via Register 0x09 with an update rate limited to 1.6 MSPS (with SCLK = 32 MHz). Note that Bit 6 of Register 0x09 must be set for a read or write operation. Updating the RxPGA via the Tx[5:0] port is an option only in full-duplex mode.1 In this case, a high level on the GAIN pin2 with TXSYNC low, programs the PGA setting on either the rising edge or falling edge of RXCLK, as shown in Figure 57. The GAIN pin must be held high, TXSYNC must be held low, and GAIN data must be stable for one or more clock cycles to update the RxPGA gain setting. A low level on the GAIN pin enables data to be fed to the digital interpolation filter. This interface should be considered when upgrading existing designs from the AD9876 MxFE product or half-duplex applications trying to minimize an ASIC’s pin count. Updating the RxPGA (or TxPGA) via the PGA[5:0] port is an option for both the half-duplex3 and full-duplex interfaces. The PGA port consists of an input buffer that passes the 6-bit data appearing at its input directly to the RxPGA (or TxPGA) gain register with no gating signal required. Bit 5 or Bit 6 of Register 0x0B is used to select whether the data updates the RxPGA or TxPGA gain register. In applications that switch between RxPGA and TxPGA gain control via PGA[5:0], be careful that the RxPGA (or TxPGA) is not inadvertently loaded with the wrong data during a transition. In the case of an RxPGA to TxPGA transition, first deselect the RxPGA gain register, update the PGA[5:0] port with the desired TxPGA gain setting, and then select the TxPGA gain register. The RxPGA also offers an alternative 3-bit word gain mapping option4 that provides a −12 dB to +36 dB span in 8 dB increments as shown in Table 16. The 3-bit word is directed to PGA[5:3] with PGA[5] being the MSB. This feature is backward-compatible with the AD9975 MxFE, and allows direct interfacing to the CX11647 or INT5130 HomePlug 1.0 PHYs. Table 16. PGA Timing for AD9975 Backward-Compatible Mode Digital Gain Setting PGA[5:3] 000 001 010 011 100 101 110 111 1 Decimal 0 1 2 3 4 5 6 7 Gain (dB) −12 −12 −4 4 12 20 28 36 Default setting for full-duplex mode (MODE = 1). The GAIN strobe can also be set in software via Register 0x0B, Bit 3 for continuous updating. This eliminates the requirement for external GAIN signal, reducing the ASIC pin count by 1. 3 Default setting for half-duplex mode (MODE = 0). 4 Default setting for MODE = 0 and CONFIG =1. 2 Rev. A | Page 26 of 48 AD9866 TXPGA CONTROL 0 –1 –2 –3 –4 –5 –6 –7 –8 –9 –10 –11 –12 –13 –14 –15 –16 –17 –18 –19 –20 Table 17. SPI Registers TxPGA Control Address (Hex) 0x0A 0x0B TxDACs IOUTP OUTPUT HAS 7.5dB RANGE 0x0E IAMPs IOUTN AND IOUTG OUTPUTS HAS 19.5dB RANGE 0 8 16 24 32 40 04560-0-058 Tx ATTENUATION (dBFS) The AD9866 also contains a digital PGA in the Tx path distributed between the TxDAC and IAMP. The TxPGA is used to control the peak current from the TxDAC and IAMP over a 7.5 dB and 19.5 dB span, respectively, with 0.5 dB resolution. A 6-bit word is used to set the TxPGA attenuation according to the mapping shown in Figure 58. The TxDAC gain mapping is applicable only when Bit 0 of Register 0x0E is set, and only the four LSBs of the 6-bit gain word are relevant. The TxPGA register can be updated via the PGA[5:0] port or SPI port. The first method should be considered for fast updates of the TxPGA register. Its operation is similar to the description in the RxPGA Control section. The SPI port allows direct update and readback of the TxPGA register via Register 0x0A with an update rate limited to 1.6 MSPS (SCLK = 32 MHz). Bit 6 of Register 0x0A must be set for a read or write operation. Table 17 lists the SPI registers pertaining to the TxPGA. The TxPGA control register default setting is for minimum attenuation (0 dBFS) with the PGA[5:0] port disabled for Tx gain control. 48 56 64 6-BIT DIGITAL CODE (Decimal Equivalent) Figure 58. Digital Gain Mapping of TxPGA Rev. A | Page 27 of 48 Bit (6) (5:0) (6) (5) (0) Description Enable TxPGA update via SPI TxPGA gain code Select TxPGA via PGA[5:0] Select RxPGA via PGA[5:0] TxDAC output (IAMP disabled) AD9866 TRANSMIT PATH 0 TO –12dB AD9865/AD9866 04560-0-059 TXEN/SYNC TXCLK 2.0 –10 1.5 –20 1.0 0.5 –30 PASS BAND –40 –50 –1.0 –70 –1.5 –80 –2.0 –90 0 DIGITAL INTERPOLATION FILTERS Bits [7:6] 00 01 10 11 0.25 0.50 0.75 1.00 1.25 1.50 1.75 NORMALIZED FREQUENCY (Relative to fDATA) 2.5 10 WIDE BAND Interpolation Factor 4 2 1 (half-duplex only) Do not use 0 2.0 –10 1.5 –20 1.0 0.5 –30 PASS BAND 0 –40 –1.0dB @ 0.45 fDATA –50 –0.5 –60 –1.0 –70 –1.5 –80 –2.0 –90 0 The interpolation filter consists of two cascaded half-band filter stages with each stage providing 2× interpolation. The first stage filter consists of 43 taps. The second stage filter, operating at the higher data rate, consists of 11 taps. The normalized wide band and pass-band filter responses (relative fDATA) for the 2× and 4× low-pass interpolation filters are shown in Figure 60 and Figure 61, respectively. These responses also include the inherent sinc(x) from the TxDAC reconstruction process and can be used to estimate any post analog filtering requirements. –2.5 2.00 Figure 60. Frequency Response of 2× Interpolation Filter (Normalized to fDATA) WIDE BAND RESPONSE (dB) Table 18. Interpolation Factor Set via SPI Register 0x0C –0.5 –1.0dB @ 0.441 fDATA –60 Figure 59. Functional Block Diagram of Tx Path The input data from the Tx port can be fed into a selectable 2×/4× interpolation filter or directly into the TxDAC (for a halfduplex only). The interpolation factor for the digital filter is set via SPI Register 0x0C with the settings shown in Table 18. The maximum input word rate, fDATA, into the interpolation filter is 80 MSPS; the maximum DAC update rate is 200 MSPS. Therefore, applications with input word rates at or below 50 MSPS can benefit from 4× interpolation, while applications with input word rates between 50 MSPS and 80 MSPS can benefit from 2× interpolation. 0 PASS-BAND RESPONSE (dB) 0 TO –7.5dB IAMP 0 04560-0-060 TxDAC WIDE BAND 0.5 1.0 1.5 2.0 2.5 3.0 3.5 NORMALIZED FREQUENCY (Relative to fDATA) –2.5 4.0 PASS-BAND RESPONSE (dB) 2-4X 2.5 10 04560-0-061 ADIO[11:6]/ Rx[5:0] 10 IOUT_G+ IOUT_N+ IOUT_N– IOUT_G– The pipeline delays of the 2× and 4× filter responses are 21.5 and 24 clock cycles, respectively, relative to fDATA. The filter delay is also taken into consideration for applications configured for a half-duplex interface with the half-duplex power-down mode enabled. This feature allows the user to set a programmable delay that powers down the TxDAC and IAMP only after the last Tx input sample has propagated through the digital filter. See the Power Control and Dissipation section for more details. WIDE BAND RESPONSE (dB) ADIO[11:6]/ Tx[5:0] IOUT_P– IOUT_P+ The AD9866 (or AD9865) transmit path consists of a selectable digital 2×/4× interpolation filter, a 12-bit (or 10-bit) TxDAC, and a current-output amplifier (IAMP), as shown in Figure 59. Note that the additional two bits of resolution offered by the AD9866 (vs. the AD9865) result in a 10 dB to 12 dB reduction in the pass-band noise floor. The digital interpolation filter relaxes the Tx analog filtering requirements by simultaneously reducing the images from the DAC reconstruction process while increasing the analog filter’s transition band. The digital interpolation filter can also be bypassed, resulting in lower digital current consumption. Figure 61. Frequency Response of 4× Interpolation Filter (Normalized to fDATA) TxDAC AND IAMP ARCHITECTURE The Tx path contains a TxDAC with a current amplifier, IAMP. The TxDAC reconstructs the output of the interpolation filter and sources a differential current output that can be directed to an external load or fed into the IAMP for further amplification. The TxDAC’s and IAMPS’s peak current outputs are digitally programmable over a 0 to −7.5 dB and 0 to −19.5 dB range, respectively, in 0.5 dB increments. Note that this assumes default register settings for Register 0x10 and Register 0x11. Rev. A | Page 28 of 48 AD9866 Applications demanding the highest spectral performance and/or lowest power consumption can use the TxDAC output directly. The TxDAC is capable of delivering a peak signal power-up to 10 dBm while maintaining respectable linearity performance, as shown in Figure 27 through Figure 38. For power-sensitive applications requiring the highest Tx power efficiency, the TxDAC’s full-scale current output can be reduced to as low as 2 mA, and its load resistors sized to provide a suitable voltage swing that can be amplified by a low-power op amp-based driver. Most applications requiring higher peak signal powers (up to 23 dBm) should consider using the IAMP. The IAMP can be configured as a current source for loads having a well defined impedance (50 Ω or 75 Ω systems), or a voltage source (with the addition of a pair of npn transistors) for poorly defined loads having varying impedance (such as power lines). G × (I–∆I) G × (I+∆I) IOFF2 IOFF2 xN xG IOUTG– I IOUTG+ ±∆IS IOUTN– REFADJ IOUTN+ TxDAC I N × (I–∆I) N × (I+∆I) Figure 62 shows the equivalent schematic of the TxDAC and IAMP. The TxDAC provides a differential current output appearing at IOUTP+ and IOUTP−. It can be modeled as a differential current source generating a signal-dependent ac current, when ∆IS has a peak current of I along with two dc current sources, sourcing a standing current equal to I. The fullscale output current, IOUTFS, is equal to the sum of these standing current sources (IOUTFS = 2 × I). REFIO RSET 0.1µF IOUTP+ I – ∆I IOUTP– IOFF1 IOFF1 xN xG 04560-0-062 I + ∆I IAMP clearing Bit 0 of Register 0x0E. As a result, the IOUTP pins must remain completely open, if the IAMP is to be used. The IAMP contains two sets of current mirrors that are used to replicate the TxDAC’s current output with a selectable gain. The first set of current mirrors is designated as the primary path, providing a gain factor of N that is programmable from 0 to 4 in steps of 1 via Bits 2:0 of Register 0x10 with a default setting of N = 4. Bit 7 of this register must be set to overwrite the default settings of this register. This differential path exhibits the best linearity performance (see Figure 42) and is available at the IOUTN+ and IOUTN− pins. The maximum peak current per output is 100 mA and occurs when the TxDAC’s standing current, I, is set for 12.5 mA (IOUTFS = 25 mA). The second set of current mirrors is designated as the secondary path providing a gain factor of G that is programmable from 0 to 36 via Bits 6:4 of Register 0x10, and Bits 6:0 of Register 0x11 with a default setting of G = 12. This differential path is intended to be used in the voltage mode configuration to bias the external npn transistors, because it exhibits degraded linearity performance (see Figure 43) relative to the primary path. It is capable of sinking up to 180 mA of peak current into either its IOUTG+ or IOUTG− pins. The secondary path actually consists of three gain stages (G1, G2, and G3), which are individually programmable as shown in Table 19. While many permutations may exist to provide a fixed gain of G, the linearity performance of a secondary path remains relatively independent of the various individual gain settings that are possible to achieve a particular overall gain factor. Both sets of mirrors sink current, because they originate from NMOS devices. Therefore, each output pin requires a dc current path to a positive supply. Although the voltage output of each output pin can swing between 0.5 V and 7 V, optimum ac performance is typically achieved by limiting the ac voltage swing with a dc bias voltage set between 4 V to 5 V. Lastly, both the standing current, I, and the ac current, ∆IS, from the TxDAC are amplified by the gain factor (N and G) with the total standing current drawn from the positive supply being equal to 2 × (N + G) × I Figure 62. Equivalent Schematic of TxDAC and IAMP The value of I is determined by the RSET value at the REFADJ pin along with the Tx path’s digital attenuation setting. With 0 dB attenuation, the value of I is I = 16 × (1.23/RSET) (1) For example, an RSET value of 1.96 kΩ results in I equal to 10.0 mA with IOUTFS equal to 20.0 mA. Note that the REFIO pin provides a nominal band gap reference voltage of 1.23 V and should be decoupled to analog ground via a 0.1 µF capacitor. The differential current output of the TxDAC is always connected to the IOUTP pins, but can be directed to the IAMP by Programmable current sources IOFF1 and IOFF2 via Register 0x12 can be used to improve the primary and secondary path mirrors’ linearity performance under certain conditions by increasing their signal-to-standing current ratio. This feature provides a marginal improvement in distortion performance under large signal conditions when the peak ac current of the reconstructed waveform frequently approaches the dc standing current within the TxDAC (0 to −1 dBFS sine wave) causing the internal mirrors to turn off. However, the improvement in distortion performance diminishes as the crest factor (peak-torms ratio) of the ac signal increases. Most applications can disable these current sources (set to 0 mA via Register 0x12) to reduce the IAMP’s current consumption. Rev. A | Page 29 of 48 AD9866 1:1 Table 19. SPI Registers for TxDAC and IAMP 0x11 (7) (6:4) (3) (2:0) 0x12 (6:4) (2:0) Tx PROGRAMMABLE GAIN CONTROL TxPGA functionality is also available to set the peak output current from the TxDAC or IAMP. The TxDAC and IAMP are digitally programmable via the PGA[5:0] port or SPI over a 0 dB to −7.5 dB and 0 dB to −19.5 dB range, respectively, in 0.5 dB increments. The TxPGA can be considered as two cascaded attenuators with the TxDAC providing 7.5 dB range in 0.5 dB increments, and the IAMP providing 12 dB range in 6 dB increments. As a result, the IAMP’s composite 19.5 dB span is valid only if Register 0x10 remains at its default setting of 0x44. Modifying this register setting corrupts the LUT and results in an invalid gain mapping. TxDAC OUTPUT OPERATION The differential current output of the TxDAC is available at the IOUTP+ and IOUTP− pins and the IAMP should be disabled by setting Bit 0 of Register 0x0E. Any load connected to these pins must be ground referenced to provide a dc path for the current sources. Figure 63 shows the outputs of the TxDAC driving a doubly terminated 1:1 transformer with its center-tap tied to ground. The peak-to-peak voltage, V p-p, across RL (and IOUT+ to IOUT−) is equal to 2 × I × (RL//RS). With I = 10 mA and RL = RS = 50 Ω, V p-p is equal to 0.5 V with 1 dBm of peak power being delivered to RL and 1 dBm being dissipated in RS. RL RS 0.1µF TxDAC 0 TO –7.5dB IOUT_P– IOUT_P+ RSET IOUTN+ IOUTG+ IAMP 0 TO –12dB IOUTN– IOUTG– 04560-0-063 (3) (2:0) Description TxDAC output Enable current mirror gain settings Secondary path first stage gain of 0 to 4 with ∆ = 1 Not used Primary path NMOS gain of 0 to 4 with ∆ = 1 Don’t care Secondary path second stage gain of 0 to 1.5 with ∆ = 0.25 Not used Secondary path third stage gain of 0 to 5 with ∆ = 1 IOFF2, secondary path standing current IOFF1, primary path standing current REFIO Bit (0) (7) (6:4) REFADJ Address (Hex) 0x0E 0x10 Figure 63. TxDAC Output Directly via Center-Tap Transformer The TxDAC is capable of delivering up to 10 dBm peak power to a load, RL. To increase the peak power for a fixed standing current, one must increase V p-p across IOUTP+ and IOUTP− by increasing one or more of the following parameters: RS, RL (if possible), and/or the turns ratio, N, of transformer. For example, the removal of RS and the use of a 2:1 impedance ratio transformer in the previous example results in 10 dBm of peak power capabilities to the load. Note that increasing the power output capabilities of the TxDAC reduces the distortion performance due to the higher voltage swings seen at IOUTP+ and IOUTP−. See Figure 27 through Figure 38 for performance plots on the TxDAC’s ac performance. Optimum distortion performance can typically be achieved by: • Limiting the peak positive VIOUTP+ and VIOUTP− to 0.8 V to avoid onset of TxDAC’s output compression. (TxDAC’s voltage compliance is around 1.2 V.) • Limiting V p-p seen at IOUTP+ and IOUTP− to less than 1.6 V. Applications demanding higher output voltage swings and power drive capabilities can benefit from using the IAMP. IAMP CURRENT-MODE OPERATION The IAMP can be configured for the current-mode operation as shown in Figure 64 for loads remaining relatively constant. In this mode, the primary path mirrors should be used to deliver the signal-dependent current to the load via a center-tapped transformer, because it provides the best linearity performance. Because the mirrors exhibit a high output impedance, they can be easily back-terminated (if required). For peak signal currents (IOUTPK up to 50 mA), only the primary path mirror gain should be used for optimum distortion performance and power efficiency. The primary path’s gain should be set to 4, with the secondary path’s gain stages set to 0 (Register 0x10 = 0x84). The TxDAC’s standing current, I, can be set between 2.5 mA and 12.5 mA with the IOUTP outputs left open. The IOUTN outputs should be connected to the transformer, with the IOUTG (and IOUTP) Rev. A | Page 30 of 48 AD9866 outputs left open for optimum linearity performance. The transformer1 should be specified to handle the dc standing current, IBIAS, drawn by the IAMP. Also, because IBIAS remains signal independent, a series resistor (not shown) can be inserted between AVDD and the transformer’s center-tap to reduce the IAMP’s common-mode voltage, VCM, and reduce the power dissipation on the IC. The VCM bias should not exceed 5.0 V and the power dissipated in the IAMP alone is as follows: PIAMP = 2 × (N + G) × I × VCM AVDD 0.1µF IOUT_P– IOUTN+ IOUTG+ IOUTN– 04560-0-064 IOUTG– IOUTPK = (N+G) × 1 P_OUTPK = (IOUTPK)2 × T2 × RL AVDD TxDAC 0 TO –7.5dB IOUTN+ R IOUTPK IAMP 0 TO –12dB AVDD For applications requiring an IOUTPK exceeding 50 mA, set the secondary’s path to deliver the additional current to the load. IOUTG+ and IOUTN+ should be shorted as well as IOUTG− and IOUTN−. If IOUTPK represents the peak current to be delivered to the load, then the current gain in the secondary path, G, can be set by the following equation: G = IOUTPK/12.5 – 4 The B6080 and BX6090 transformers from Pulse Engineering are worthy of consideration for current and voltage modes. RS 0.1µF Figure 65. Voltage-Mode Operation The peak differential voltage signal developed across the npn’s bases is as follows: VOUTPK = R × (N × I) (4) where: N is the gain setting of the primary mirror. I is the standing current of the TxDAC defined in Equation 1. The common-mode bias voltage seen at IOUTN+ and IOUTN− is approximately AVDD − VOUTPK, while the common-mode voltage seen at IOUTG+ and IOUTG− is approximately the npn’s VBE drop below this level (AVDD − VOUTPK − 0.65). In the voltage-mode configuration, the total power dissipated within the IAMP is as follows: PIAMP = 2 × I {(AVDD − VOUTPK) × N + (AVDD − VOUTPK − 0.65) × G} (3) The linearity performance becomes limited by the secondary mirror path’s distortion. TO LOAD IOUTN– IOUTG– A step-down transformer1 with a turn ratio, T, can be used to increase the output power, P_OUT, delivered to the load. This causes the output load, RL, to be reflected back to the IAMP’s differential output by T2, resulting in a larger differential voltage swing seen at the IAMP’s output. For example, the IAMP can deliver 24 dBm of peak power to a 50 Ω load, if a 1.41:1 stepdown transformer is used. This results in 5 V p-p voltage swings appearing at IOUTN+ and IOUTN− pins. Figure 42 shows how the third order intercept point, OIP3, of the IAMP varies as a function of common-mode voltage over a 2.5 MHz to 20.0 MHz span with a 2-tone signal having a peak power of approximately 24 dBm with IOUTPK = 50 mA. DUAL NPN PHILLIPS PBR951 RS 0.1µF IOUTG+ Figure 64. Current-Mode Operation 1 R 04560-0-065 0 TO –12dB IOUTPK RSET RL IOUT_P+ IAMP 0.1µF IOUT_P– T:1 TxDAC 0 TO –7.5dB IBIAS = 2 × (N+G) × 1 REFIO IOUT_P+ REFIO REFADJ RSET The voltage-mode configuration is shown in Figure 65. This configuration is suited for applications having a poorly defined load that can vary over a considerable range. A low impedance voltage driver can be realized with the addition of two external RF bipolar npn transistors (Phillips PBR951) and resistors. In this configuration, the current mirrors in the primary path (IOUTN outputs) feed into scaling resistors, R, generating a differential voltage into the bases of the npn transistors. These transistors are configured as source followers with the secondary path current mirrors appearing at IOUTG+ and IOUTG− providing a signal-dependent bias current. Note that the IOUTP outputs must remain open for proper operation. REFADJ 0.1µF (2) IAMP VOLTAGE-MODE OPERATION (5) The emitters of the npn transistors are ac-coupled to the transformer1 via a 0.1 µF blocking capacitor and series resistor of 1 Ω to 2 Ω. Note that protection diodes are not shown for clarity purposes, but should be considered if interfacing to a power or phone line. The amount of standing and signal-dependent current used to bias the npn transistors depends on the peak current, IOUTPK, required by the load. If the load is variable, determine the worst case, IOUTPK, and add 3 mA of margin to ensure that the npn Rev. A | Page 31 of 48 AD9866 transistors remain in the active region during peak load currents. The gain of the secondary path, G, and the TxDAC’s standing current, I, can be set using the following equation: 100 90 80 (6) IAMPN OUTPUT 50 40 TxDACs AVDD 04560-0-066 30 20 10 1 2 3 4 5 6 7 8 9 10 11 12 13 I (mA) Figure 66. Current Consumption of TxDAC and IAMP in Current-Mode Operation with IOUTN Only (Default IAMP Settings) IAMP CURRENT CONSUMPTION CONSIDERATIONS 150 140 130 IOUTG OUTPUT 120 110 ISUPPLY (mA) The Tx path’s analog current consumption is an important consideration when determining its contribution to the overall on-chip power dissipation. This is especially the case in fullduplex applications, where the power dissipation can exceed the maximum limit of 1.66 W, if the IAMP’s IOUTPK is set to high. The analog current consumption includes the TxDAC’s analog supply (Pin 43) along with the standing current from the IAMP’s outputs. Equation 2 and Equation 5 can be used to calculate the power dissipated in the IAMP for the current and voltage-mode configuration. Figure 66 shows the current consumption for the TxDAC and IAMP as a function of the TxDAC’s standing current, I, when only the IOUTN outputs are used. Figure 67 shows the current consumption for the TxDAC and IAMP as a function of the TxDAC’s standing current, I, when the IOUTN and IOUTG outputs are used. Both figures are with the default current mirror gain settings of N = 4 and G = 12. 60 100 90 80 70 60 50 TxDAC AVDD 40 30 IOUTN OUTPUT 20 10 1.0 04560-0-067 The voltage output driver exhibits a high output impedance if the bias currents for the npn transistors are removed. This feature is advantageous in half-duplex applications (for example, power lines) in which the Tx output driver must go into a high impedance state while in Rx mode. If the AD9866 is configured for the half-duplex mode (MODE = 0), the IAMP, TxDAC, and interpolation filter are automatically powered down after a Tx burst (via TXEN), thus placing the Tx driver into a high impedance state while reducing its power consumption. 70 ISUPPLY (mA) IOUTPK + 3 mA = G × I 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 I (mA) Figure 67. Current Consumption of TxDAC and IAMP in Current-Mode Operation with IOUTN Only (Default IAMP Settings) Rev. A | Page 32 of 48 AD9866 RECEIVE PATH CLKOUT_1 CLKOUT_2 CLK SYN. 2M CLK MULTIPLIER ADIO[11:6]/ Rx[5:0] OSCIN XTAL 10/12 RXEN/SYNC RXCLK ADC 80MSPS SPGA RX+ 2-POLE LPF 1-POLE LPF RX– 0 TO 6dB –6 TO 18dB –6 TO 24dB ∆ = 1dB ∆ = 6dB ∆ = 6dB 4 SPORT GAIN MAPPING LUT 6 REGISTER CONTROL AD9865/AD9866 04560-0-068 PGA[5:0] Figure 68. Functional Block Diagram of Rx Path RX PROGRAMMABLE GAIN AMPLIFIER The RxPGA has a digitally programmable gain range from −12 dB to +48 dB with 1 dB resolution via a 6-bit word. Its purpose is to extend the dynamic range of the Rx path such that the input of the ADC is presented with a signal that scales within its fixed 2 V input span. There are multiple ways of setting the RxPGA’s gain as discussed in the RxPGA Control section, as well as an alternative 3-bit gain mapping having a range of −12 dB to +36 dB with 8 dB resolution. The RxPGA is comprised of two sections: a continuous time PGA (CPGA) for course gain and a switched capacitor PGA (SPGA) for fine gain resolution. The CPGA consists of two cascaded gain stages providing a gain range from −12 dB to +42 dB with 6 dB resolution. The first stage features a low noise preamplifier (< 3.0 nV/rtHz), thereby eliminating the need for an external preamplifier. The SPGA provides a gain range from 0 dB to 6 dB with 1 dB resolution. A look-up table (LUT) is used to select the appropriate gain setting for each stage. To limit the RxPGA’s self-induced input offset, an offset cancellation loop is included. This cancellation loop is automatically performed upon power-up and can also be initiated via SPI. During calibration, the RxPGA’s first stage is internally shorted, and each gain stage set to a high gain setting. A digital servo loop slaves a calibration DAC, which forces the Rx input offset to be within ±32 LSB for this particular high gain setting. Although the offset varies for other gain settings, the offset is typically limited to ±5% of the ADC’s 2 V input span. Note that the offset cancellation circuitry is intended to reduce the voltage offset attributed to only the RxPGA’s input stage, not any dc offsets attributed to an external source. The gain of the RxPGA should be set to minimize clipping of the ADC while utilizing most of its dynamic range. The maximum peak-to-peak differential voltage that does not result in clipping of the ADC is shown in Figure 69. While the graph suggests that maximum input signal for a gain setting of −12 dB is 8.0 V p-p, the maximum input voltage into the PGA should be limited to less than 6 V p-p to prevent turning on ESD protection diodes. For applications having higher maximum input signals, consider adding an external resistive attenuator network. While the input sensitivity of the Rx path is degraded by the amount of attenuation on a dB-to-dB basis, the low noise characteristics of the RxPGA provide some design margin such that the external line noise remains the dominant source. The nominal differential input impedance of the RxPGA input appearing at the device RX+ and RX− input pins is 400 Ω//4 pF (±20%) and remains relatively independent of gain setting. The PGA input is self-biased at a 1.3 V common-mode level allowing maximum input voltage swings of ±1.5 V at RX+ and RX−. AC coupling the input signal to this stage via coupling capacitors (0.1 µF) is recommended to ensure that any external dc offset Rev. A | Page 33 of 48 8.0000 4.0000 2.0000 1.0000 0.5000 0.2500 0.1250 0.0625 0.0312 0.0156 0.0100 –12 04560-0-069 ADIO[11:6]/ Tx[5:0] does not get amplified with high RxPGA gain settings, potentially exceeding the ADC input range. FULL-SCALE PEAK-TO-PEAK INPUT SPAN (V) The receive path block diagram for the AD9866 (or AD9865) is shown in Figure 68. The receive signal path consists of a 3-stage RxPGA, a 3-pole programmable LPF, and a 12-bit (or 10-bit) ADC. Note that the additional 2 bits of resolution offered by the AD9866 (vs. the AD9865) result in a 3 dB to 5 dB lower noise floor depending on the RxPGA gain setting and LPF cutoff frequency. Also working in conjunction with the receive path is an offset correction circuit. These blocks are discussed in detail in the following sections. Note that the power consumption of the RxPGA can be modified via Register 0x13 as discussed in the Power Control and Dissipation section. –6 0 6 12 18 24 30 36 42 48 GAIN (dB) Figure 69. Maximum Peak-to-Peak Input vs. RxPGA Gain Setting that Does Not Result in ADC Clipping AD9866 1.30 0.25 Although the default setting specifies that the LPF be active, it can also be bypassed providing a nominal f−3 dB of 55 MHz. Table 20 shows the SPI registers pertaining to the LPF. Table 20. SPI Registers for Rx Low-Pass Filter Address (Hex) 0x07 0x08 Bit (0) (7:0) Description Enable Rx LPF Target value GAIN (dB) The low-pass filter (LPF) provides a third order response with a cutoff frequency that is typically programmable over a 15 MHz to 35 MHz span. Figure 68 shows that the first real pole is implemented within the first CPGA gain stage, and the complex pole pair is implemented in the second CPGA gain stage. Capacitor arrays are used to vary the different R-C time constants within these two stages in a manner that changes the cutoff frequency while preserving the normalized frequency response. Because absolute resistor and capacitor values are process-dependent, a calibration routine lasting less than 100 µs automatically occurs each time the target cutoff frequency register (Register 0x08) is updated, ensuring a repeatable cutoff frequency from device to device. NORMALIZED GAIN RESPONSE –0.25 1.20 –0.50 1.15 –0.75 1.10 –1.00 1.05 –1.25 1.00 –1.50 0.95 –1.75 0.90 –2.00 0.85 0.80 –2.25 NORMALIZED GROUP DELAY –2.50 0.75 0.70 –2.75 –3.00 NORMALIZED GROUP DELAY TIME RESPONSE (GDT) 1.25 0 04560-0-071 LOW-PASS FILTER 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.65 1.0 0.9 NORMALIZED FREQUENCY Figure 71. LPF’s Normalized Pass-Band Gain and Group Delay Responses The −3 dB cutoff frequency, f−3 dB, is programmable by writing an 8-bit word, referred to as the target, to Register 0x08. The cutoff frequency is a function of the ADC sample rate, fADC, and to a lesser extent, the RxPGA gain setting (in dB). Figure 72 shows how the frequency response, f−3 dB, varies as a function of the RxPGA gain setting. 3 The normalized wideband gain response is shown in Figure 70. The normalized pass-band gain and group delay responses are shown in Figure 71. The normalized cutoff frequency, f−3 dB, results in −3 dB attenuation. Also, the actual group delay time (GDT) response can be calculated given a programmed cutoff frequency using the following equation: FUNDAMENTAL (dB) Actual GDT = Normalized GDT/(2.45 × f−3 dB) –6dB GAIN 0dB GAIN +6dB GAIN +18dB GAIN +30dB GAIN +42dB GAIN 0 (7) –3 –6 –9 –12 04560-0-072 5 –15 0 –18 –5 0 5 10 15 20 25 30 35 40 45 50 INPUT FREQUENCY (MHz) Figure 72. Effects of RxPGA Gain on LPF Frequency Response ( f−3 dB = 32 MHz (@ 0 dB and fADC = 80 MSPS) –15 The following formula1 can be used to estimate f−3 dB for a RxPGA gain setting of 0 dB: –20 –25 04560-0-070 GAIN (dB) –10 –30 –35 0 0.5 1.0 1.5 FREQUENCY 2.0 2.5 3.0 f−3 dB_0 dB = (128/target) × (fADC/80) × (fADC/30 + 23.83) (8) Figure 73 compares the measured and calculated f−3 dB using this formula. Figure 70. LPF’s Normalized Wideband Gain Response 1 Empirically derived for a f−3 dB range of 15 MHz to 35 MHz and fADC of 40 MSPS to 80 MSPS with an RxPGA = 0 dB. Rev. A | Page 34 of 48 AD9866 35 ANALOG-TO-DIGITAL CONVERTER (ADC) 33 The AD9866 features a 12-bit analog-to-digital converter (ADC) capable of up to 80 MSPS. Referring to Figure 68, the ADC is driven by the SPGA stage, which performs both the sample-and-hold and the fine gain adjust functions. A buffer amplifier (not shown) isolates the last CPGA gain stage from the dynamic load presented by the SPGA stage. The full-scale input span of the ADC is 2 V p-p, and depending on the PGA gain setting, the full-scale input span into the SPGA is adjustable from 1 V to 2 V in 1 dB increments. FREQUENCY (MHz) 31 29 27 25 80 MSPS MEASURED 23 80 MSPS CALCULATED 21 17 04560-0-073 19 50 MSPS MEASURED 50 MSPS CALCULATED 15 48 64 80 96 112 128 144 160 176 192 208 224 TARGET-DECIMAL EQUIVALENT Figure 73. Measured and Calculated f−3 dB vs. Target Value for fADC = 50 MSPS and 80 MSPS The following scaling factor can be applied to the previous formula to compensate for the RxPGA gain setting on f−3 dB: Scale Factor = 1 − (RxPGA in dB)/382 (9) This scaling factor reduces the calculated f−3 dB as the RxPGA is increased. Applications that need to maintain a minimum cutoff frequency, f−3 dB_MIN, for all RxPGA gain settings should first determine the scaling factor for the highest RxPGA gain setting to be used. Next, the f−3 dB_MIN should be divided by this scale factor to normalize to the 0 dB RxPGA gain setting (f−3 dB_0 dB). Equation 8 can then be used to calculate the target value. The LPF frequency response shows a slight sensitivity to temperature, as shown in Figure 74. Applications sensitive to temperature drift can recalibrate the LPF by rewriting the target value to Register 0x08. A pipelined multistage ADC architecture is used to achieve high sample rates while consuming low power. The ADC distributes the conversion over several smaller A/D subblocks, refining the conversion with progressively higher accuracy as it passes the results from stage to stage on each clock edge. The ADC typically performs best when driven internally by a 50% duty cycle clock. This is especially the case when operating the ADC at high sample rate (55 MSPS to 80 MSPS) and/or lower internal bias levels, which adversely affect interstage settling time requirements. The ADC sampling clock path also includes a duty cycle restorer circuit, which ensures that the ADC gets a near 50% duty cycle clock even when presented with a clock source with poor symmetry (35/65). This circuit should be enabled, if the ADC sampling clock is a buffered version of the reference signal appearing at OSCIN (see the Clock Synthesizer section) and if this reference signal is derived from an oscillator or crystal whose specified symmetry cannot be guaranteed to be within 45/55 (or 55/45). This circuit can remain disabled, if the ADC sampling clock is derived from a divided down version of the clock synthesizer’s VCO, because this clock is near 50%. 35 The ADC’s power consumption can be reduced by 25 mA, with minimal effect on its performance, by setting Bit 4 of Register 0x07. Alternative power bias settings are also available via Register 0x13, as discussed in the Power Control and Dissipation section. Lastly, the ADC can be completely powered down for halfduplex operation, further reducing the AD9866’s peak power consumption. FREQUENCY (MHz) 30 FOUT ACTUAL 80MHz AND –40°C FOUT ACTUAL 80MHz AND +25°C 25 FOUT ACTUAL 80MHz AND +85°C 15 96 04560-0-074 20 112 128 144 160 176 192 208 224 240 TARGET-DECIMAL EQUIVALENT Figure 74. Temperature Drift of f−3 dB for fADC = 80 MSPS and RxPGA = 0 dB Rev. A | Page 35 of 48 AD9866 AGC TIMING CONSIDERATIONS REFT TO ADCs C1 0.1µF C3 0.1µF When implementing a digital AGC timing loop, it is important to consider the Rx path latency and settling time of the Rx path in response to a change in gain setting. Figure 21 and Figure 24 show the RxPGA’s settling response to a 60 dB and 5 dB change in gain setting when using the Tx[5:0] or PGA[5:0] port. While the RxPGA settling time may also show a slight dependency on the LPF’s cutoff frequency, the ADC’s pipeline delay along with the ADIO bus interface presents a more significant delay. The amount of delay or latency depends on whether a half- or fullduplex is selected. An impulse response at the RxPGA’s input can be observed after 10.0 ADC clock cycles (1/fADC) in the case of a half-duplex interface and 10.5 ADC clock cycles in the case of a full-duplex interface. This latency along with the RxPGA settling time should be considered to ensure stability of the AGC loop. C2 10µF C4 0.1µF REFB 1.0V TOP VIEW C4 C2 04560-0-075 C3 C1 Figure 75. ADC Reference and Decoupling The ADC has an internal voltage reference and reference amplifier as shown in Figure 75. The internal band gap reference generates a stable 1 V reference level that is converted to a differential 1 V reference centered about mid-supply (AVDD/2). The outputs of the differential reference amplifier are available at the REFT and REFB pins and must be properly decoupled for optimum performance. The REFT and REFB pins are conveniently situated at the corners of the CSP package such that C1 (0603 type) can be placed directly across its pins. C3 and C4 can be placed underneath C1, and C2 (10 µF tantalum) can be placed furthest from the package. Table 21. SPI Registers for Rx ADC Address (Hex) 0x04 0x07 0x13 Bit (5) (4) (4) (2:0) Description Duty cycle restore circuit ADC clock from PLL ADC low power mode ADC power bias adjust Rev. A | Page 36 of 48 AD9866 CLOCK SYNTHESIZER The data rate, fDATA, for the Tx and Rx data paths must always be equal. Therefore, the ADC’s sample rate, fADC, is always equal to fDATA, while the TxDAC update rate is a factor of 1, 2, or 4 of fDATA, depending on the interpolation factor selected. The data rate refers to the word rate and should not be confused with the nibble rate in full-duplex interface. XTAL XTAL C1 ÷2N 2M CLK MULTIPLIER OSCIN TO ADC TO TxDAC C2 CLKOUT2 ÷ 2L 04560-0-076 CLKOUT1 ÷ 2R Figure 76. Clock Oscillator and Synthesizer (fDAC). The first option is the default setting and most desirable if fOSCIN is equal to the ADC sample rate, fADC. This option typically results in the best jitter/phase noise performance for the ADC sampling clock. The second option is suitable in cases where fOSCIN is a factor of 2 or 4 less than the fADC. In this case, the divider ratio, N, is chosen such that the divided down VCO output is equal to the ADC sample rate, as shown in the following equation: fADC = fDAC/2N where N = 0, 1, or 2. Figure 77 shows the degradation in phase noise performance imparted onto the ADC’s sampling clock for different VCO output frequencies. In this case, a 25 MHz, 1 V p-p sine wave was used to drive OSCIN and the PLL’s M and N factor were selected to provide an fADC of 50 MHz for a VCO operating frequency of 50, 100, and 200 MHz. The RxPGA input was driven with a near full-scale, 12.5 MHz input signal with a gain setting of 0 dB. Operating the VCO at the highest possible frequency results in the best narrow and wideband phase noise characteristics. For comparison purposes, the clock source for the ADC was taken directly from OSCIN when driven by a 50 MHz square wave. 0 The 2M CLK multiplier contains a PLL (with integrated loop filter) and VCO capable of generating an output frequency that is a multiple of 1, 2, 4, or 8 of its input reference frequency, fOSCIN, appearing at OSCIN. The input frequency range of fOSCIN is between 20 MHz and 80 MHz, while the VCO can operate over a 40 MHz to 200 MHz span. For the best phase noise/jitter characteristics, it is advisable to operate the VCO with a frequency between 100 MHz and 200 MHz. The VCO output drives the TxDAC directly such that its update rate, fDAC, is related to fOSCIN by the following equation: M fDAC = 2 × fOSCIN DIRECT VCO = 50MHz VCO = 100MHz VCO = 200MHz –10 –20 –30 dBFS –40 –50 –60 –70 –80 –90 –100 (10) –110 2.5 where M = 0, 1, 2, or 3. 4.5 6.5 8.5 10.5 12.5 14.5 16.5 18.5 20.5 22.5 FREQUENCY (MHz) M is the PLL’s multiplication factor set in Register 0x04. The value of M is determined by the Tx path’s word rate, fDATA, and digital interpolation factor, F, as shown in the following equation: M = log2 (F × fDATA/fOSCIN) (12) 04560-0-077 The AD9866 generates all its internal sampling clocks, as well as two user-programmable clock outputs appearing at CLKOUT1 and CLKOUT2, from a single reference source as shown in Figure 76. The reference source can be either a fundamental frequency or an overtone quartz crystal connected between OSCIN and XTAL with the parallel resonant load components as specified by the crystal manufacturer. It can also be a TTLlevel clock applied to OSCIN with XTAL left unconnected. (11) Note: if the reference frequency appearing at OSCIN is chosen to be equal to the AD9866’s Tx and Rx path’s word rate, then M is simply equal to log2(F). The clock source for the ADC can be selected in Register 0x04 as a buffered version of the reference frequency appearing at OSCIN (default setting) or a divided version of the VCO output Figure 77. Comparison of Phase Noise Performance when ADC Clock Source is Derived from Different VCO Output Frequencies The CLK synthesizer also has two clock outputs appearing at CLKOUT1 and CLKOUT2. They are programmable via Register 0x06. Both outputs can be inverted or disabled. The voltage levels appearing at these outputs are relative to DRVDD and remain active during a hardware or software reset. Table 22 shows the SPI registers pertaining to the clock synthesizer. CLKOUT1 is a divided version of the VCO output and can be set to be a submultiple integer of fDAC (fDAC/2R, where R = 0, 1, 2, or 3). Because this clock is actually derived from the same set of dividers used within the PLL core, it is phase-locked to them such that its phase relationship relative to the signal appearing Rev. A | Page 37 of 48 AD9866 at OSCIN (or RXCLK) can be determined upon power-up. Also, this clock has near 50% duty cycle, because it is derived from the VCO. As a result, CLKOUT1 should be selected before CLKOUT2 as the primary source for system clock distribution. CLKOUT2 is a divided version of the reference frequency, fOSCIN, and can be set to be a submultiple integer of fOSCIN (fOSCIN/2L, where L = 0, 1, or 2). With L set to 0, the output of CLKOUT2 is a delayed version of the signal appearing at OSCIN, exhibiting the same duty cycle characteristics. With L set to 1 or 2, the output of CLKOUT2 is a divided version of the OSCIN signal, exhibiting a near 50% duty cycle, but without having a deterministic phase relationship relative to CLKOUT1 (or RXCLK). Table 22. SPI Registers for CLK Synthesizer Address (Hex) 0x04 0x06 Rev. A | Page 38 of 48 Bit (4) (3:2) (1:0) (7:6) (5) (4) (3:2) (1) (0) Description ADC CLK from PLL PLL divide factor (P) PLL multiplication factor (M ) CLKOUT2 divide number CLKOUT2 invert CLKOUT2 disable CLKOUT1 divide number CLKOUT1 invert CLKOUT1 disable AD9866 POWER CONTROL AND DISSIPATION POWER-DOWN HALF-DUPLEX POWER SAVINGS The AD9866 provides the ability to control the power-on state of various functional blocks. The state of the PWRDWN pin, along with the contents of Register 0x01 and Register 0x02, allow two user-defined power settings that are pin selectable. The default settings1 are such that Register 0x01 has all blocks powered on (all Bits 0), while Register 0x02 has all blocks powered, down excluding the PLL, such that the clock signal remains available at CLKOUT1 and CLKOUT2. When the PWRDWN pin is low, the functional blocks corresponding to the bits in Register 0x01 are powered down. When the PWRDWN is high, the functional blocks corresponding to the bits in Register 0x02 are powered down. PWRDWN immediately affects the designated functional blocks with minimum digital delay. Significant power savings can be realized in applications having a half-duplex protocol allowing only the Rx or Tx path to be operational at any instance. The power savings method depends on whether the AD9866 is configured for a full- or half-duplex interface. Functional blocks having fast power on/off times for the Tx and Rx path are controlled by the following bits: TxDAC/IAMP, TX Digital, ADC, and RxPGA. Table 23. SPI Registers Associated with Power-Down and Half-Duplex Power Savings Address (Hex) 0x01 0x02 0x03 Bit (7) (6) (5) (4) (3) (2) (1) (0) (7) (6) (5) (4) (3) (2) (1) (0) (7:3) (2) (1) (0) 1 Description PLL TxDAC/IAMP TX Digital REF ADC CML ADC PGA BIAS RxPGA PLL TxDAC/IAMP TX Digital REF ADC CML ADC PGA BIAS RxPGA Tx OFF Delay Rx PWRDWN via TXEN Enable Tx PWRDWN Enable Rx PWRDWN Comments PWRDWN = 0. Default setting is all functional blocks powered on. PWRDWN = 1. Default setting is all functional blocks powered off, excluding PLL. Half-duplex power savings. In the case of a full-duplex digital interface (MODE = 1), one can set Register 0x01 to 0x60 and Register 0x02 to 0x05 (or vice versa) such that the AD9866’s Tx and Rx path are never powered on simultaneously. The PWRDWN pin can then be used to control what path is powered on, depending on the burst type. During a Tx burst, the Rx path’s PGA and ADC blocks can typically be powered down within 100 ns, while the Tx paths DAC, IAMP, and digital filter blocks are powered up within 0.5 µs. For an Rx burst, the Tx path’s can be powered down within 100 ns, while the Rx circuitry is powered up within 2 µs. Setting the TXQUIET pin low allows it to be used with the fullduplex interface to quickly power down the IAMP and disable the interpolation filter. This is meant to maintain backward compatibility with the AD9875/AD9876 MxFEs with the exception that the TxDAC remains powered if its IOUTP outputs are used. In most applications, the interpolation filter needs to be flushed with 0s before or after being powered down. This ensures that, upon power-up, the TxDAC (and IAMP) have a negligible differential dc offset, thus preventing spectral splatter due to an impulse transient. Applications using a half-duplex interface (MODE = 0) can benefit from an additional power savings feature made available in Register 0x03. This register is effective only for a half-duplex interface. Besides providing power savings for half-duplex applications, this feature allows the AD9866 to be used in applications that need only its Rx (or Tx) path functionality through pin-strapping, making a serial port interface (SPI) optional. This feature also allows the PWRDWN pin to retain its default function as a master power control, as defined in Table 10. The default settings for Register 0x03 provide fast power control of the functional blocks in the Tx and Rx signal paths (outlined above) using the TXEN pin. The TxDAC still remains powered on in this mode, while the IAMP is powered down. Significant current savings are typically realized when the IAMP is powered down. With MODE = 1 and CONFIG = 1, Reg. 0x02 default settings are with all blocks powered off, with RXCLK providing a buffered version of the signal appearing at OSCIN. This setting results in the lowest power consumption upon power-up while still allowing AD9865 to generate the system clock via a crystal. For a Tx burst, the falling edge of TXEN is used to generate an internal delayed signal for powering down the Tx circuitry. Upon receipt of this signal, power-down of the Tx circuitry Rev. A | Page 39 of 48 AD9866 55 occurs within 100 ns. The user-programmable delay for the Tx path power-down is meant to match the pipeline delay of the last Tx burst sample such that power-down of the TxDAC and IAMP does not impact its transmission. A 5-bit field in Register 0x03 sets the delay from 0 to 31 TXCLK clock cycles, with the default being 31 (0.62 µs with fTxCLK = 50 MSPS). The digital interpolation filter is automatically flushed with midscale samples prior to power-down, if the clock signal into the TXCLK pin is present for 33 additional clock cycles after TXEN returns low. For an Rx burst, the rising edge of TXEN is used to generate an internal signal (with no delay) that powers up the Tx circuitry within 0.5 µs. 50 IAVDDTxDAC (mA) 45 40 35 30 25 15 10 1 2 3 4 5 6 7 8 9 10 11 12 13 Figure 78. Reduction in TxDAC’s Supply Current vs. Standing Current 65 60 4× INTERPOLATION 55 50 POWER REDUCTION OPTIONS 45 40 2× INTERPOLATION 35 30 1× (HALF-DUPLEX ONLY) 25 04560-0-079 The power consumption of the AD9866 can be significantly reduced from its default setting by optimizing the power consumption vs. performance of the various functional blocks in the Tx and Rx signal path. On the Tx path, minimum power consumption is realized when the TxDAC output is used directly and its standing current, I, is reduced to as low as 1 mA. Although a slight degradation in THD performance results at reduced standing currents, it often remains adequate for most applications, because the op amp driver typically limits the overall linearity performance of the Tx path. The load resistors used at the TxDAC outputs (IOUTP+ and IOUTP−) can be increased to generate an adequate differential voltage that can be further amplified via a power efficient op amp based driver solution. Figure 78 shows how the supply current for the TxDAC (Pin 43) is reduced from 55 mA to 14 mA as the standing current is reduced from 12.5 mA to 1.25 mA. Further Tx power savings can be achieved by bypassing or reducing the interpolation factor of the digital filter as shown in Figure 79. 0 ISTANDING (mA) IDVDD (mA) The Rx path power-on/power-off can be controlled by either TXEN or RXEN by setting Bit 2 of Register 0x03. In the default setting, the falling edge of TXEN powers up the Rx circuitry within 2 µs, while the rising edge of TXEN powers down the Rx circuitry within 0.5 µs. If RXEN is selected as the control signal, then its rising edge powers up the Rx circuitry and the falling edge powers it down. To disable the fast power-down of the Tx and/or Rx circuitry, set Bit 1 and/or Bit 0 to 0. 04560-0-078 20 20 15 20 30 40 50 60 70 80 INPUT DATA RATE (MSPS) Figure 79. Digital Supply Current Consumption vs. Input Data Rate (DVDD = DRVDD = 3.3 V and fOUT = fDATA/10) Power consumption on the Rx path can be achieved by reducing the bias levels of the various amplifiers contained within the RxPGA and ADC. As previously noted, the RxPGA consists of two CPGA amplifiers and one SPGA amplifier. The bias levels of each of these amplifiers along with the ADC can be controlled via Register 0x13 as shown in Table 24. The default setting for Register 0x13 is 0x00. Table 24. SPI Register for RxPGA and ADC Biasing Address (Hex) 0x07 0x13 Rev. A | Page 40 of 48 Bit (4) (7:5) (4:3) (2:0) Description ADC low power CPGA bias adjust SPGA bias adjust ADC power bias adjust AD9866 Bit 5 0 1 0 1 0 1 0 1 ∆ mA 0 −27 −42 −51 −55 27 69 27 01 195 00 190 185 10 180 04560-0-081 11 175 170 20 30 40 50 60 70 80 ADC SAMPLE RATE (MSPS) Figure 81. AVDD Current vs. SPGA Bias Setting and Sample Rate 65 –54 64 –56 63 –58 SNR-00 SNR-01 SNR-10 SNR-11 62 61 60 –66 THD-00 THD-01 THD-10 THD-11 58 –68 –70 56 –20 55 20 –25 60.0 –30 57.5 –35 55.0 –40 SNR_RxPGA = 36dB 52.5 –45 50.0 –50 –60 42.5 –65 THD_RxPGA = 36dB 40.0 000 001 010 011 –70 100 04560-0-080 –55 45.0 30 40 50 60 70 –74 80 Figure 82. SNR and THD Performance vs. fADC and SPGA Bias Setting with RxPGA = 0 dB, fIN = 10 MHz. AIN = −1 dBFS THD_RxPGA = 0dB 47.5 –72 SAMPLE RATE (MSPS) THD (dBc) SNR (dBFS) SNR_RxPGA = 0dB 62.5 –62 –64 59 57 65.0 –60 THD (dBc) Bit 6 0 0 1 1 0 0 1 1 200 SNR (dBc) Bit 7 0 0 0 0 1 1 1 1 205 04560-0-082 Table 25. Analog Supply Current vs. CPGA Bias Settings at fADC = 65 MSPS 210 IAVDD (mA) Because the CPGA processes signals in the continuous time domain, its performance vs. bias setting remains mostly independent of the sample rate. Table 25 shows how the typical current consumption seen at AVDD (Pins 35 and 40) varies as a function of Bits (7:5), while the remaining bits are maintained at their default settings of 0. Only four of the possible settings result in any reduction in current consumption relative to the default setting. Reducing the bias level typically results in a degradation in the THD vs. frequency performance as shown in Figure 80. This is due to a reduction of the amplifier’s unity gain bandwidth, while the SNR performance remains relatively unaffected. CPGA BIAS SETTING-BITS (7:5) Figure 80. THD vs. fIN Performance and RxPGA Bias Settings (000,001,010,100 with RxPGA = 0 and +36 dB and AIN = −1 dBFS, LPF set to 26 MHz and fADC = 50 MSPS) The ADC is based on a pipeline architecture with each stage consisting of a switched capacitor amplifier. Therefore, its performance vs. bias level is mostly dependent on the sample rate. Figure 83 shows how the typical current consumption seen at AVDD (Pin 35 and Pin 40) varies as a function of Bits (2:0) and sample rate, while the remaining bits are maintained at the default setting of 0. Setting Bit 4 or Register 0x07 corresponds to the 011 setting, and the settings of 101 and 111 result in higher current consumption. Figure 84 shows how the SNR and THD performance are affected for a 10 MHz sine wave input for the lower power settings as the ADC sample rate is swept from 20 MHz to 80 MHz. The SPGA is implemented as a switched capacitor amplifier; therefore, its performance vs. bias level is mostly dependent on the sample rate. Figure 81 shows how the typical current consumption seen at AVDD (Pin 35 and Pin 40) varies as a function of Bits (4:3) and sample rate, while the remaining bits are maintained at the default setting of 0. Figure 82 shows how the SNR and THD performance is affected for a 10 MHz sine wave input as the ADC sample rate is swept from 20 MHz to 80 MHz. Rev. A | Page 41 of 48 AD9866 30.8oC/W, if the heat slug remains unsoldered.) If a particular application’s maximum ambient temperature, TA, falls below 85oC, the maximum allowable power dissipation can be determined by the following equation: 220 101 OR 111 210 200 000 190 IAVDD (mA) 001 180 PMAX = 1.66 + (85 − TA)/24 010 170 160 100 150 011 Assuming the IAMP’s common-mode bias voltage is operating off the same analog supply as the AD9866, the following equation can be used to calculate the maximum total current consumption, IMAX, of the IC: 101 04560-0-083 140 130 120 20 30 40 50 60 70 IMAX = (PMAX − PIAMP)/3.47 80 Figure 83. AVDD Current vs. ADC Bias Setting and Sample Rate 65 –54 64 –56 60 –58 SNR-000 SNR-001 SNR-010 SNR-011 SNR-100 SNR-101 –60 –62 –64 59 –66 58 56 55 20 –68 THD-000 THD-001 THD-010 THD-011 THD-100 THD-101 57 30 40 THD (dBc) 61 If the IAMP is operating off a different supply or in the voltage mode configuration, first calculate the power dissipated in the IAMP, PIAMP, using Equation 2 or Equation 5, and then recalculate IMAX, using Equation 14. –70 –72 50 60 70 –74 80 04560-0-084 SNR (dBc) 62 (14) With an ambient temperature of up to 85°C, IMAX is 478 mA. SAMPLE RATE (MSPS) 63 (13) SAMPLE RATE (MSPS) Figure 84. SNR and THD Performance vs. fADC and ADC Bias Setting with RxPGA = 0 dB, fIN = 10 MHz, AIN = −1 dBFS A sine wave input is a standard and convenient method of analyzing the performance of a system. However, the amount of power reduction that is possible is application dependent, based on the nature of the input waveform (such as frequency content, peak-to-rms ratio), the minimum ADC sample, and the minimum acceptable level of performance. Thus, it is advisable that power-sensitive applications optimize the power bias setting of the Rx path using an input waveform that is representative of the application. POWER DISSIPATION The power dissipation of the AD9866 can become quite high in full-duplex applications in which the Tx and Rx paths are simultaneously operating with nominal power bias settings. In fact, some applications that use the IAMP may need to either reduce its peak power capabilities or reduce the power consumption of the Rx path, so that the device’s maximum allowable power consumption, PMAX, is not exceeded. PMAX is specified at 1.66 W to ensure that the die temperature does not exceed 125oC at an ambient temperature of 85oC. This specification is based on the 64-pin LFSCP having a thermal resistance, θJA, of 24oC/W with its heat slug soldered. (The θJA is Figure 78, Figure 79, Figure 81, and Figure 83 can be used to calculate the current consumption of the Rx and Tx paths for a given setting. MODE SELECT UPON POWER-UP AND RESET The AD9866 power-up state is determined by the logic levels appearing at the MODE and CONFIG pins. The MODE pin is used to select a half- or full-duplex interface by pin strapping it low or high, respectively. The CONFIG pin is used in conjunction with the MODE pin to determine the default settings for the SPI registers as outlined in Table 10. The intent of these particular default settings is to allow some applications to avoid using the SPI (disabled by pin-strapping SEN high), thereby reducing implementation costs. For example, setting MODE low and CONFIG high configures the AD9866 to be backward-compatible with the AD9975, while setting MODE high and CONFIG low makes it backwardcompatible with the AD9875. Other applications must use the SPI to configure the device. A hardware (RESET pin) or software (Bit 5 of Register 0x00) reset can be used to place the AD9866 into a known state of operation as determined by the state of the MODE and CONFIG pins. A dc offset calibration and filter tuning routine is also initiated upon a hardware reset, but not with a software reset. Neither reset method flushes the digital interpolation filters in the Tx path. Refer to the Half-Duplex Mode and Full-Duplex Mode sections for information on flushing the digital filters. A hardware reset can be triggered by pulsing the RESET pin low for a minimum of 50 ns. The SPI registers are instantly reset to their default settings upon RESET going low, while the dc offset calibration and filter tuning routine is initiated upon RESET returning high. To ensure sufficient power-on time of the Rev. A | Page 42 of 48 AD9866 various functional blocks, RESET returning high should occur no less than 10 ms upon power-up. If a digital reset signal from a microprocessor reset circuit (such as ADM1818) is not available, a simple R-C network referenced to DVDD can be used to hold RESET low for approximately 10 ms upon powerup. reconstructed output from the TxDAC or IAMP to ensure a minimum level of performance. In this test, the user can exercise the RxPGA as well as validate the attenuation characteristics of the RxLPF. Note that the RxPGA gain setting should be selected such that the input does not result in clipping of the ADC. ANALOG AND DIGITAL LOOPBACK TEST MODES Digital loopback can be used to test the full-duplex digital interface of the AD9866. In this test, data appearing on the Tx[5:0] port is routed back to the Rx[5:0] port, thereby confirming proper bus operation. The Rx port can also be three-stated for half- and full-duplex interfaces. The AD9866 features analog and digital loopback capabilities that can assist in system debug and final test. Analog loopback routes the digital output of the ADC back into the Tx data path prior to the interpolation filters such that the Rx input signal can be monitored at the output of the TxDAC or IAMP. As a result, the analog loopback feature can be used for a half- or full-duplex interface to allow testing of the functionality of the entire IC (excluding the digital data interface). Table 26. SPI Registers for Test Modes Address (Hex) 0x0D For example, the user can configure the AD9866 with similar settings as the target system, inject an input signal (sinusoidal waveform) into the Rx input, and monitor the quality of the Rev. A | Page 43 of 48 Bit (7) (6) (5) Description Analog loopback Digital loopback Rx port three-state AD9866 PCB DESIGN CONSIDERATIONS Although the AD9866 is a mixed-signal device, the part should be treated as an analog component. The on-chip digital circuitry has been specially designed to minimize the impact of its digital switching noise on the MxFE’s analog performance. To achieve the best performance, the power, grounding, and layout recommendations in this section should be followed. Assembly instructions for the micro-lead frame package can be found in an application note from Amkor at: http://www.amkor.com/products/notes_papers/MLF_AppNote _0902.pdf. COMPONENT PLACEMENT If the three following guidelines of component placement are followed, chances for getting the best performance from the MxFE are greatly increased. First, manage the path of return currents flowing in the ground plane so that high frequency switching currents from the digital circuits do not flow on the ground plane under the MxFE or analog circuits. Second, keep noisy digital signal paths and sensitive receive signal paths as short as possible. Third, keep digital (noise generating) and analog (noise susceptible) circuits as far away from each other as possible. To best manage the return currents, pure digital circuits that generate high switching currents should be closest to the power supply entry. This keeps the highest frequency return current paths short and prevents them from traveling over the sensitive MxFE and analog portions of the ground plane. Also, these circuits should be generously bypassed at each device, which further reduces the high frequency ground currents. The MxFE should be placed adjacent to the digital circuits, such that the ground return currents from the digital sections do not flow in the ground plane under the MxFE. The AD9866 has several pins that are used to decouple sensitive internal nodes. These pins are REFIO, REFB, and REFT. The decoupling capacitors connected to these points should have low ESR and ESL. These capacitors should be placed as close to the MxFE as possible (see Figure 75) and be connected directly to the analog ground plane. The resistor connected to the REFADJ pin should also be placed close to the device and connected directly to the analog ground plane. POWER PLANES AND DECOUPLING While the AD9866 evaluation board demonstrates a very good power supply distribution and decoupling strategy, it can be further simplified for many applications. The board has four layers: two signal layers, one ground plane, and one power plane. While the power plane on the evaluation board is split into multiple analog and digital subsections, a permissible alternative would be to have AVDD and CLKVDD share the same analog 3.3 V power plane. A separate analog plane/supply may be allocated to the IAMP, if its supply voltage differs from the 3.3 V required by AVDD and CLKVDD. On the digital side, DVDD and DRVDD can share the same 3.3 V digital power plane. This digital power plane brings the current used to power the digital portion of the MxFE and its output drivers. This digital plane should be kept from going underneath the analog components. The analog and digital power planes allocated to the MxFE may be fed from the same low noise voltage source; however, they should be decoupled from each other to prevent the noise generated in the digital portion of the MxFE from corrupting the AVDD supply. This can be done by using ferrite beads between the voltage source and the respective analog and digital power planes with a low ESR, bulk decoupling capacitor on the MxFE side of the ferrite. Each of the MxFE’s supply pins (AVDD, CLKVDD, DVDD, and DRVDD) should also have dedicated low ESR, ESL decoupling capacitors. The decoupling capacitors should be placed as close to the MxFE supply pins as possible. GROUND PLANES The AD9866 evaluation board uses a single serrated ground plane to help prevent any high frequency digital ground currents from coupling over to the analog portion of the ground plane. The digital currents affiliated with the high speed data bus interface (Pin 1 to Pin 16) have the highest potential of generating problematic high frequency noise. A ground serration that contains these currents should reduce the effects of this potential noise source. The ground plane directly underneath the MxFE should be continuous and uniform. The 64-lead LFCSP package is designed to provide excellent thermal conductivity. This is partly achieved by incorporating an exposed die paddle on the bottom surface of the package. However, to take full advantage of this feature, the PCB must have features to effectively conduct heat away from the package. This can be achieved by incorporating thermal pad and thermal vias on the PCB. While a thermal pad provides a solderable surface on the top surface of the PCB (to solder the package die paddle on the board), thermal vias are needed to provide a thermal path to inner and/or bottom layers of the PCB to remove the heat. Lastly, all ground connections should be made as short as possible. This results in the lowest impedance return paths and the quietest ground connections. SIGNAL ROUTING The digital Rx and Tx signal paths should be kept as short as possible. Also, the impedance of these traces should have a controlled characteristic impedance of about 50 Ω. This prevents poor signal integrity and the high currents that can Rev. A | Page 44 of 48 AD9866 occur during undershoot or overshoot caused by ringing. If the signal traces cannot be kept shorter than about 1.5 inches, series termination resistors (33 Ω to 47 Ω) should be placed close to all digital signal sources. It is a good idea to seriesterminate all clock signals at their source, regardless of trace length. The receive RX+ and RX− signals are the most sensitive signals on the entire board. Careful routing of these signals is essential for good receive path performance. The RX+ and RX− signals form a differential pair and should be routed together as a pair. By keeping the traces adjacent to each other, noise coupled onto the signals appears as common mode and is largely rejected by the MxFE receive input. Keeping the driving point impedance of the receive signal low and placing any low-pass filtering of the signals close to the MxFE further reduces the possibility of noise corrupting these signals. Rev. A | Page 45 of 48 AD9866 EVALUATION BOARD An evaluation board is available for the AD9865 and AD9866. The digital interface to the evaluation board can be configured for a half- or full-duplex interface. Two 40-pin and one 26-pin male right angle headers (0.100 inches) provide easy interfacing to test equipment such as digital data capture boards, pattern generators, or custom digital evaluation boards (FPGA, DSP, or ASIC). The reference clock source can originate from an external generator, crystal oscillator, or crystal. Software and an interface cable are included to allow for programming of the SPI registers via a PC. The analog interface on the evaluation board provides a full analog front-end reference design for power line applications. It includes a power line socket, line transformer, protection diodes, and passive filtering components. An auxiliary path allows independent monitoring of the ac power line. The evaluation board allows complete optimization of power line reference designs based around the AD9865 or AD9866. Alternatively, the evaluation board allows independent evaluation of the TxDAC, IAMP, and Rx paths via SMA connectors. The IAMP can be easily configured for a voltage or current mode interface via jumper settings. The TxDAC’s performance can be evaluated directly or via an optional dual op amp driver stage. The Rx path includes a transformer and termination resistor, allowing for a calibrated differential input signal to be injected into its front end. The Analog Devices, Inc. website offers more information on the AD9865/AD9866 evaluation board. Rev. A | Page 46 of 48 AD9866 OUTLINE DIMENSIONS 9.00 BSC SQ 0.60 MAX 0.60 MAX 0.30 0.25 0.18 49 48 PIN 1 INDICATOR 8.75 BSC SQ TOP VIEW PIN 1 INDICATOR 64 1 *7.25 EXPOSED PAD 7.10 SQ 6.95 (BOTTOM VIEW) 0.45 0.40 0.35 33 32 17 16 0.25 MIN 1.00 0.85 0.80 12° MAX 7.50 REF 0.80 MAX 0.65 TYP 0.05 MAX 0.02 NOM 0.50 BSC SEATING PLANE 0.20 REF *COMPLIANT TO JEDEC STANDARDS MO-220-VMMD EXCEPT FOR EXPOSED PAD DIMENSION Figure 85. 64-Lead Lead Frame Chip Scale Package (LFCSP) [CP-64-3] Dimensions shown in millimeters ORDERING GUIDE Model AD9866BCP AD9866BCPRL AD9866BCPZ1 AD9866BCPZRL1 AD9866CHIPS AD9866-EB 1 Temperature Range −40°C to +85°C −40°C to +85°C −40°C to +85°C −40°C to +85°C Package Description 64-Lead LFCSP 64-Lead LFCSP 64-Lead LFCSP 64-Lead LFCSP DIE Evaluation Board Z = Pb-free part. Rev. A | Page 47 of 48 Package Option CP-64-3 CP-64-3 CP-64-3 CP-64-3 AD9866 NOTES © 2004 Analog Devices, Inc. All rights reserved. Trademarks and registered trademarks are the property of their respective owners. C04560–0–12/04(A) Rev. A | Page 48 of 48