View AD9789BBC_6931717.PDF datasheet online --- IC-ON-LINE

Datasheet File OCR Text:

14-bit, 2400 msps rf dac with 4-channel signal processing ad9789 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: 781.461.3113 ?2009-2011 analog devices, inc. all rights reserved. features docsis 3.0 performance: 4 qam carriers aclr over full band (47 mhz to 1 ghz) ?75 dbc @ f out = 200 mhz ?72 dbc @ f out = 800 mhz (noise) ?67 dbc @ f out = 800 mhz (harmonics) unequalized mer = 42 db on chip and bypassable 4 qam encoders with srrc filters, 16 to 512 interpolation, rate converters, and modulators flexible data interface: 4, 8, 16, or 32 bits wide with parity power: 1.6 w (i fs = 20 ma, f dac = 2.4 ghz, lvds interface) direct to rf synthesis support with f s mix mode built-in self-test (bist) support input connectivity check internal random number generator applications broadband communications systems cmts/dvb cellular infrastructure point-to-point wireless general description the ad9789 is a flexible qam encoder/interpolator/upconverter combined with a high performance, 2400 msps, 14-bit rf digital- to-analog converter (dac). the flexible digital interface can accept up to four channels of complex data. the qam encoder supports constellation sizes of 16, 32, 64, 128, and 256 with srrc filter coefficients for all standards. the on-chip rate converter supports a wide range of baud rates with a fixed dac clock. the digital upconverter can place the channels from 0 to 0.5 f dac . this permits four contiguous channels to be synthesized and placed anywhere from dc to f dac /2. the ad9789 includes a serial peripheral interface (spi) for device configuration and status register readback. the flexible digital interface can be configured for data bus widths of 4, 8, 16, and 32 bits. it can accept real or complex data. the ad9789 operates from 1.5 v, 1.8 v, and 3.3 v supplies for a total power consumption of 1.6 w. it is supplied in a 164-ball chip scale package ball grid array for lower thermal impedance and reduced package parasitics. no special power sequencing is required. the clock receiver powers up muted to prevent start-up noise. product highlights 1. highly integrated and configurable qam mappers, inter- polators, and upconverters for direct synthesis of one to four docsis- or dvb-c-compatible channels in a block. 2. low noise and intermodulation distortion (imd) perfor- mance enable high quality synthesis of signals up to 1 ghz. 3. flexible data interface supports lvds for improved sfdr or cmos input data for less demanding applications. 4. interface is configurable from 4-bit nibbles to 32-bit words and can run at up to 150 mhz cmos or 150 mhz lvds double data rate (ddr). 5. manufactured on a cmos process, the ad9789 uses a proprietary switching technique that enhances dynamic performance. functional block diagram 07852-001 data data data 150mhz lvds/cmos qam/ filter/ nco qam/ filter/ nco qam/ filter/ nco qam/ filter/ nco 32 input pins and 2parity pins dco fs 16 interpolator and bpf + scalars 14-bit 2.4gsps dac data cmos 0to 15 lvds rise cmos 16 to 31 lvds fall spi irq rs retimer data formatter/ assembler figure 1.
ad9789 rev. a | page 2 of 76 table of contents features .............................................................................................. 1 ? applications....................................................................................... 1 ? general description ......................................................................... 1 ? product highlights ........................................................................... 1 ? functional block diagram .............................................................. 1 ? revision history ............................................................................... 2 ? detailed functional block diagrams............................................. 3 ? specifications..................................................................................... 4 ? dc specifications ......................................................................... 4 ? digital specifications ................................................................... 5 ? ac specifications.......................................................................... 6 ? absolute maximum ratings............................................................ 8 ? thermal resistance ...................................................................... 8 ? esd caution.................................................................................. 8 ? pin configurations and function descriptions ........................... 9 ? typical performance characteristics ........................................... 12 ? terminology .................................................................................... 22 ? serial control port.......................................................................... 23 ? serial control port pin descriptions ....................................... 23 ? general operation of serial control port............................... 23 ? instruction word (16 bits) ........................................................ 24 ? msb/lsb first transfers............................................................ 24 ? spi register map............................................................................. 27 ? spi register descriptions.......................................................... 29 ? theory of operation ...................................................................... 39 ? datapath signal processing....................................................... 39 ? digital block upconverter ........................................................ 43 ? digital interface modes ............................................................. 45 ? analog modes of operation ..................................................... 54 ? analog control registers .......................................................... 55 ? voltage reference ....................................................................... 56 ? dac output stages .................................................................... 56 ? clocking the ad9789 ................................................................ 57 ? mu delay controller.................................................................. 58 ? interrupt requests...................................................................... 61 ? recommended start-up sequence .......................................... 62 ? customer bist modes................................................................... 63 ? using the internal prn generator to test qam output ac performance ................................................................................ 63 ? using the internal built-in self-test (bist) to test for digital data input connectivity............................................................. 63 ? qam constellation maps ............................................................. 65 ? channelizer mode pin mapping for cmos and lvds............ 68 ? outline dimensions ....................................................................... 74 ? ordering guide .......................................................................... 74 ? revision history 7/11rev. 0 to rev. a changes to table 2, dac clock input (clkp, clkn): added dac clock rate parameter............................................................. 5 changes to table 3, dynamic performance, dac update rate parameter; added adjusted dac update rate parameter......... 6 changes to captions for figure 42, figure 44, figure 46, figure 49 .......................................................................................... 18 changes to digital 16x tunable band-pass filter section, third paragraph......................................................................................... 44 changes to retimer and latency look-up tables section, second paragraph ........................................................................... 50 changes to captions for figure 122, figure 124, figure 125.... 65 4/09revision 0: initial version
ad9789 rev. a | page 3 of 76 detailed functional block diagrams 07852-002 clk ctl data- path 0 data- path 1 data- path 2 data- path 3 up to 32 bits cmos 0to 15 lvds rise cmos 16 to 31 lvds fall up to 32 bits up to 32 bits up to 32 bits 32 input pins p0 p1 fs dco sum scale bpf f c 4to 32 bits 16 interpolator bpf f c = 0 to f dac /2 retimer lvds/cmos data formatter/ assembler figure 2. digital signal processing functional block diagram bypass qam input scale 24-bit nco 0to f dac /16 07852-003 2 n (n = 0 to 5) (p/q = 0.5 to 1) rate converter p/q 24-bit ch gain 0 to 2 srrc 2 bypass srrc qam mapper figure 3. channel 0 through channel 3 datapath block detail (i and q paths are identical so only one is shown)
ad9789 rev. a | page 4 of 76 specifications dc specifications avdd33 = dvdd33 = 3.3 v, cvdd18 = dvdd18 = 1.8 v, dvdd15 = 1.5 v, f dac = 2.4 ghz, i fs = 20 ma, unless otherwise noted. table 1. parameter min typ max unit dac resolution 14 bits analog outputs offset error 6.5 % fsr gain error (with internal reference) 3.5 % fsr full-scale output current (monotonicity guaranteed) 8.66 20.2 31.66 ma output compliance range ?1.0 +1.0 v output resistance 70 output capacitance 1 pf temperature drift gain 135 ppm/c reference voltage 25 ppm/c reference internal reference voltage 1.2 v output resistance 1 5 k analog supply voltages avdd33 3.14 3.3 3.47 v cvdd18 1.71 1.8 1.89 v digital supply voltages dvdd33 3.14 3.3 3.47 v dvdd18 1.71 1.8 1.89 v dvdd15 1.43 1.5 1.58 v supply currents and power dissipation f dac = 2.4 gsps, f out = 930 mhz, i fs = 25 ma, four channels enabled i avdd33 45 ma i dvdd18 72 ma i cvdd18 180 ma i dvdd33 cmos interface 42 ma lvds interface 16 ma i dvdd15 640 ma f dac = 2.0 gsps, f out = 70 mhz, i fs = 20 ma, cmos interface i avdd33 37.4 38.5 ma i dvdd18 67.3 70.5 ma i cvdd18 155.4 180 ma i dvdd33 40.3 50.7 ma i dvdd15 (four channels enabled, all signal processing enabled) 517 556 ma i dvdd15 (one channel enabled, 16 interpolation only) 365 391 ma power dissipation f dac = 2.4 gsps, f out = 930 mhz, i fs = 25 ma, four channels enabled cmos interface 1.7 w lvds interface 1.63 w 1 use an external amplifier to drive any external load.
ad9789 rev. a | page 5 of 76 digital specifications avdd33 = dvdd33 = 3.3 v, cvdd18 = dvdd18 = 1.8 v, dvdd15 = 1.5 v, f dac = 2.4 ghz, i fs = 20 ma, lvds drivers and receivers are compliant with the ieee std 1596.3-1996 reduced range link, unless otherwise noted. table 2. parameter min typ max unit cmos data inputs (d[31:0], p0, p1) input voltage high, v ih 2.0 3.3 v input voltage low, v il 0 0.8 v input current high, i ih ?10 +10 a input current low, i il ?10 +10 a input capacitance 2 pf setup time, cmos data input to cmos_dco 1 5.3 ns hold time, cmos data input to cmos_dco 1 ?1.4 ns cmos outputs (cmos_fs, cmos_dco) output voltage high, v oh 2.4 3.3 v output voltage low, v ol 0 0.4 v output current high, i oh 12 ma output current low, i ol 12 ma maximum clock rate (cmos_dco) 150 mhz cmos_dco to cmos_fs delay 0.28 0.85 ns lvds data inputs (d[15:0]p, d[15:0]n, parp, parn) input voltage range, v ia or v ib 825 1575 mv input differential threshold, v idth ?100 +100 mv input differential hysteresis, v idthh , v idthl 25 mv input differential input impedance, r in 80 120 maximum lvds input rate 150 msps setup time, lvds differential input data to differential dcox 2 1.41 ns hold time, lvds differential input data to differential dcox 2 0.24 ns lvds outputs (dcop, dcon, fsp, fsn) dcop, fsp = v oa ; dcon, fsn = v ob ; 100 termination output voltage high, v oa or v ob 1375 mv output voltage low, v oa or v ob 1025 mv output differential voltage, |v od | 150 200 250 mv output offset voltage, v os 1150 1250 mv output impedance, single ended, r o 40 140 r o mismatch between a and b, ?r o 10 % change in |v od | between 0 and 1, |?v od | 25 mv change in v os between 0 and 1, ?v os 25 mv output currentdriver shorted to ground, i sa , i sb 20 ma output currentdrivers shorted together, i sab 4 ma power-off output leakage, |i xa |, |i xb | 10 ma maximum clock rate (dcop, dcon) 150 mhz dcox to fsx delay 0.12 0.37 ns dac clock input (clkp, clkn) 3 differential peak voltage 1.4 1.8 v common-mode voltage 900 mv dac clock rate 2400 mhz serial peripheral interface maximum clock rate (f sclk , 1/t sclk ) 25 mhz minimum pulse width high, t pwh 20 ns minimum pulse width low, t pwl 20 ns minimum sdio and cs to sclk setup, t ds 10 ns
ad9789 rev. a | page 6 of 76 parameter min typ max unit minimum sclk to sdio hold, t dh 5 ns maximum sclk to valid sdio and sdo, t dv 20 ns minimum sclk to invalid sdio and sdo, t dnv 5 ns inputs (sdio, sclk, cs ) input voltage high, v ih 2.0 3.3 v input voltage low, v il 0 0.8 v input current high, i ih ?10 +10 a input current low, i il ?10 +10 a outputs (sdo, sdio) output voltage high, v oh 2.4 3.6 v output voltage low, v ol 0 0.4 v output current high, i oh 4 ma output current low, i ol 4 ma 1 see the cmos interface timing se ction for more information. 2 see the lvds interface timing section for more information. 3 see the clock phase noise effects on ac performance section fo r more information. ac specifications avdd33 = dvdd33 = 3.3 v, cvdd18 = dvdd18 = 1.8 v, dvdd15 = 1.5 v, f dac = 2.4 ghz, i fs = 20 ma, digital scale = 0 dbfs, unless otherwise noted. table 3. parameter test conditions/comments min typ max unit dynamic performance dac update rate 2400 msps adjusted dac update rate 1 150 msps output settling time (t st ) to 0.025% 13 ns spurious-free dynamic range (sfdr) f dac = 2000 msps f out = 100 mhz 70 dbc f out = 316 mhz 63 dbc f out = 550 mhz 58 dbc f dac = 2400 msps f out = 100 mhz 70 dbc f out = 316 mhz 70 dbc f out = 550 mhz 60 dbc f out = 850 mhz 60 dbc two-tone intermodulation distortion (imd) f out2 = f out1 + 1.25 mhz f dac = 2000 msps f out = 100 mhz 86 dbc f out = 316 mhz 73 dbc f out = 550 mhz 62 dbc f dac = 2400 msps f out = 100 mhz 86 dbc f out = 316 mhz 74 dbc f out = 550 mhz 66 dbc f out = 850 mhz 66 dbc noise spectral density (nsd) 1-channel qam f dac = 2400 msps f out = 100 mhz p out = ?14.5 dbm ?167 dbm/hz f out = 316 mhz p out = ?15.5 dbm ?166.5 dbm/hz f out = 550 mhz p out = ?18 dbm ?166.5 dbm/hz f out = 850 mhz p out = ?18.5 dbm ?166.5 dbm/hz
ad9789 rev. a | page 7 of 76 parameter test conditions/comments min typ max unit adjacent channel leakage ratio (aclr) f dac = 2293.76 msps measured in 6 mhz channels 1-channel qam f out = 200 mhz (harmonics) ?76 dbc f out = 200 mhz (noise floor) ?82 dbc f out = 500 mhz (harmonics) ?74.5 dbc f out = 500 mhz (noise floor) ?78 dbc f out = 800 mhz (harmonics) ?69 dbc f out = 800 mhz (noise floor) ?78 dbc 2-channel qam f out = 200 mhz (harmonics) ?77.5 dbc f out = 200 mhz (noise floor) ?81 dbc f out = 500 mhz (harmonics) ?68 dbc f out = 500 mhz (noise floor) ?76 dbc f out = 800 mhz (harmonics) ?66 dbc f out = 800 mhz (noise floor) ?76 dbc 4-channel qam f out = 200 mhz (harmonics) ?75 dbc f out = 200 mhz (noise floor) ?76 dbc f out = 500 mhz (harmonics) ?69 dbc f out = 500 mhz (noise floor) ?72 dbc f out = 800 mhz (harmonics) ?67 dbc f out = 800 mhz (noise floor) ?72 dbc wcdma aclr f dac = 2304 msps, mix mode second nyquist zone single carrier f out = 1850 mhz first adjacent channel ?70 dbc second alternate channel ?72.5 dbc third alternate channel ?74 dbc single carrier f out = 2100 mhz first adjacent channel ?68 dbc second alternate channel ?70.4 dbc third alternate channel ?72.7 dbc four carrier f out = 2100 mhz first adjacent channel ?63.5 dbc second alternate channel ?65.1 dbc third alternate channel ?66.9 dbc 1 adjusted dac update rate is calculated as f dac divided by the minimum required interp olation factor. for the ad 9789, the minimum interpolation factor is 16. thus, with f dac = 2400 msps, f dacadj = 2400 msps/16 = 150 msps.
ad9789 rev. a | page 8 of 76 absolute maximum ratings table 4. parameter rating avdd33 to avss ?0.3 v to +3.6 v dvdd18 to dvss ?0.3 v to +1.98 v dvdd33 to dvss ?0.3 v to +3.6 v dvdd15 to dvss ?0.3 v to +1.98 v cvdd18 to avss ?0.3 v to +1.98 v avss to dvss ?0.3 v to +0.3 v clkp, clkn to avss ?0.3 v to cvdd18 + 0.3 v fs, dco to dvss ?0.3 v to dvdd33 + 0.3 v cmos and lvds data inputs to dvss ?0.3 v to dvdd33 + 0.3 v ioutn, ioutp to avss ?1.0 v to avdd33 + 0.3 v i120, vref, iptat to avss ?0.3 v to avdd33 + 0.3 v irq, cs , sclk, sdo, sdio, reset to dvss ?0.3 v to dvdd33 + 0.3 v junction temperature 150c storage temperature range ?65c to +150c stresses above those listed under 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 resistance ja is specified for the worst-case conditions, that is, a device soldered in a circuit board for surface-mount packages. table 5. thermal resistance package type ja jb jc unit notes 25.5 14.4 6.8 c/w 4-layer board, no vias 24.4 c/w 4-layer board, 4 pcb vias 19.0 c/w 8-layer board, 4 pcb vias 164-ball csp_bga 17.2 c/w 8-layer board, 16 pcb vias esd caution
ad9789 rev. a | page 9 of 76 pin configurations and function descriptions a m n p l k b clkn c clkp d nc i120 vref iptat j e f h g 11 1312 11 10 89 765 24 3 4 + + + + ++ ? + + + avss dvdd18 cvdd18 avdd33 + x xx xx xx xx p n 07852-004 cvdd18 dvdd18 a v dd33 ioutn iou t p figure 4. clock and analog pins (top view) a m n p l k b c d cmos_bus cmos_ctrl cmos_fs cmos_dco j e f h g 11 1312 11 10 89 765 24 3 4 07852-005 p1 31 27 23 19 15 11 7 3 p0 30 26 22 18 14 10 6 2 29 25 21 17 13 9 5 1 28 24 20 16 12 8 4 0 bu ct fs dc d[31:0] cmos data inputs parity and control inputs figure 5. cmos mode data input pins (top view) a s clk m sdo n sdio p cs l k b c d j e f h g 11 1312 11 10 89 765 24 3 4 ++++ ++ xxxx 07852-006 ++++ xxxx reset irq sb ck do nc nc r i nc nc nc nc nc nc dvss no connect dvdd15 dvdd33 + x nc io figure 6. digital supply and spi pins (top view) a m parn parp n p l k b c d fsp fsn dcop dcon j e f h g 11 1312 11 10 89 765 24 3 4 07852-007 p+ 15 13 11 9 7 531 p? 15 13 11 9 7 531 14 12 10 8 6 4 2 0 14 12 10 8 6 4 2 0 fs fs dc dc +lvds 14 14 ?lvds figure 7. lvds mode data input pins (top view)
ad9789 rev. a | page 10 of 76 table 6. pin function descriptions pin no. mnemonic description a1, a2, a3, a6, a9, a10, a11, b1, b2, b3, b6, b7, b8, b9, b10, b11, c2, c3, c6, c7, c8, c9, c10, c11, d2, d3, d6, d7, d8, d9, d10, d11, e1, e2, e3, e4, e13, e14, f1, f2, f3, f4, f11, f12, f13, f14 avss analog supply ground. a4, a5, b4, b5, c4, c5, d4, d5 cvdd18 1.8 v clock supply. a7 ioutn dac negative output current. a8 ioutp dac positive output current. a12, a13, b12, b13, c12, c13, d12, d13 avdd33 3.3 v analog supply. a14 nc no connect. leave floating. b14 i120 tie this pin to analog ground with a 10 k resistor to generate a 120 a reference current. c1 clkn negative dac clock input (dacclk). c14 vref band gap voltage reference i/o. decouple to analog ground with a 1 nf capacitor. output impedance is approximately 5 k. d1 clkp positive dac clock input (dacclk). d14 iptat factory test pin. output current, propor tional to absolute temperature, is approximately 10 a at 25c with a slope of approximately 20 na/c. e11, e12 dvdd18 1.8 v digital supply. g1, g2, g3, g4, g7, g8, g11, g12, g13, g14 dvdd15 1.5 v digital supply. h1, h2, h3, h4, h7, h8, h11, h12, h13, h14, j1, j2, j3, j4, j11, j12, j13, j14 dvss digital supply ground. k1, k2, k3, k4, k11, k12, k13, k14 dvdd33 3.3 v digital supply. l1 cs active low chip select for spi. l2, l3, m2, m3, n3, n4, p3, p4 nc not used. leave unconnected. l4 p1/parp cmos/lvds parity bit. l5 d31/d15p cmos/lvds data input. l6 d27/d13p cmos/lvds data input. l7 d23/d11p cmos/lvds data input. l8 d19/d9p cmos/lvds data input. l9 d15/d7p cmos/lvds data input. l10 d11/d5p cmos/lvds data input. l11 d7/d3p cmos/lvds data input. l12 d3/d1p cmos/lvds data input. l13 fsp positive lvds frame sync (fsp) for data bus. l14 cmos_bus active high input. configures data bus for cmos inputs. low input configures data bus to accept lvds inputs. m1 sclk qualifying clock for spi. m4 p0/parn cmos/lvds parity bit. m5 d30/d15n cmos/lvds data input. m6 d26/d13n cmos/lvds data input. m7 d22/d11n cmos/lvds data input. m8 d18/d9n cmos/lvds data input. m9 d14/d7n cmos/lvds data input. m10 d10/d5n cmos/lvds data input. m11 d6/d3n cmos/lvds data input. m12 d2/d1n cmos/lvds data input. m13 fsn negative lvds frame sync (fsn) for data bus.
ad9789 rev. a | page 11 of 76 pin no. mnemonic description m14 cmos_ctrl active high input. enables cmos_dco and cmos_fs signals and disables dcop/dcon and fsp/fsn signals. low input disables cmos_dco and cmos_fs signals and enables dcop/dcon and fsp/fsn signals. n1 sdo serial data output for spi. n2 reset active high input. resets the ad9789. n5 d29/d14p cmos/lvds data input. n6 d25/d12p cmos/lvds data input. n7 d21/d10p cmos/lvds data input. n8 d17/d8p cmos/lvds data input. n9 d13/d6p cmos/lvds data input. n10 d9/d4p cmos/lvds data input. n11 d5/d2p cmos/lvds data input. n12 d1/d0p cmos/lvds data input. n13 dcop positive lvds data clock output (dcop) for data bus. n14 cmos_fs cmos frame sync for data bus. p1 sdio serial data input/output for spi. p2 irq active low, open-drain interrupt request output. pull up to dvdd33 with a 10 k resistor. p5 d28/d14n cmos/lvds data input. p6 d24/d12n cmos/lvds data input. p7 d20/d10n cmos/lvds data input. p8 d16/d8n cmos/lvds data input. p9 d12/d6n cmos/lvds data input. p10 d8/d4n cmos/lvds data input. p11 d4/d2n cmos/lvds data input. p12 d0/d0n cmos/lvds data input. p13 dcon negative lvds data clock output (dcon) for data bus. p14 cmos_dco cmos data clock output for data bus.
ad9789 rev. a | page 12 of 76 typical performance characteristics ? 40 ?45 ?50 ?55 ?60 ?65 ?70 ?75 ?80 ?85 ?90 0 200 400 600 800 1000 1200 f out (mhz) sfdr (dbc) 07852-013 2.4ghz 2.2ghz 2ghz 1.6ghz 1ghz figure 8. sfdr vs. f out over f dac , full-scale current = 20 ma, digital scale = 0 dbfs, temperature = 25c ? 40 ?45 ?50 ?55 ?60 ?65 ?70 ?75 ?80 ?85 ?90 0 200 400 600 800 1000 1200 f out (mhz) harmonic level (dbc) 07852-009 0dbfs ?3dbfs ?6dbfs ?12dbfs figure 9. second-order harmonic vs. f out over digital full scale, f dac = 2.4 ghz, full-scale current = 20 ma, temperature = 25c ? 40 ?45 ?50 ?55 ?60 ?65 ?70 ?75 ?80 ?85 ?90 0 200 400 600 800 1000 1200 f out (mhz) sfdr (dbc) 07852-011 32ma 20ma 8ma figure 10. sfdr vs. f out over full-scale current, f dac = 2.4 ghz, digital scale = 0 dbfs, temperature = 25c ? 40 ?45 ?50 ?55 ?60 ?65 ?70 ?75 ?80 ?85 ?90 0 200 400 600 800 1000 1200 f out (mhz) sfdr (dbc) 07852-010 0dbfs ?3dbfs ?6dbfs ?12dbfs figure 11. sfdr vs. f out over digital full scale, f dac = 2.4 ghz, full-scale current = 20 ma, temperature = 25c ? 40 ?45 ?50 ?55 ?60 ?65 ?70 ?75 ?80 ?85 ?90 0 200 400 600 800 1000 1200 f out (mhz) harmonic level (dbc) 07852-012 0dbfs ?3dbfs ?6dbfs ?12dbfs figure 12. third-order harmonic vs. f out over digital full scale, f dac = 2.4 ghz, full-scale current = 20 ma, temperature = 25c ? 50 ?55 ?60 ?65 ?70 ?75 ?80 ?85 ?90 ?95 ?100 0 200 400 600 800 1000 1200 f out (mhz) sfdr (dbc) 07852-008 +85c +25c ?40c figure 13. sfdr vs. f out over temperature, f dac = 2.4 ghz, full-scale current = 20 ma, digital scale = 0 dbfs
ad9789 rev. a | page 13 of 76 90 80 70 60 50 40 30 0 100 200 300 400 500 600 700 800 900 1000 1100 f out (mhz) imd (dbc) 07852-034 2.4ghz 2.0ghz 1.6ghz 1.0ghz figure 14. third-order imd vs. f out over f dac , full-scale current = 20 ma, digital scale = 0 dbfs, temperature = 25c 90 100 80 70 60 50 40 30 0 100 200 300 400 500 600 700 800 900 1000 1100 f out (mhz) imd (dbc) 07852-038 32ma 20ma 8ma figure 15. third-order imd vs. f out over full-scale current, f dac = 2.4 ghz, digital scale = 0 dbfs, temperature = 25c ? 155 ?157 ?159 ?161 ?163 ?165 ?167 ?169 ?171 ?173 ?175 0 200 400 600 800 1000 1200 f out (mhz) nsd (dbm/hz) 07852-016 2.4ghz 2.0ghz 1.6ghz figure 16. nsd vs. f out over f dac , 1-channel qam, full-scale current = 20 ma 90 100 80 70 60 50 40 30 0 100 200 300 400 500 600 700 800 900 1000 1100 f out (mhz) imd (dbc) 07852-037 0dbfs ?3dbfs ?6dbfs ?12dbfs figure 17. third-order imd vs. f out over digital full scale, f dac = 2.4 ghz, full-scale current = 20 ma, temperature = 25c 90 80 70 60 50 40 30 0 100 200 300 400 500 600 700 800 900 1000 1100 f out (mhz) imd (dbc) 07852-041 +85c +25c ?40c figure 18. third-order imd vs. f out over temperature, f dac = 2.4 ghz, full-scale current = 20 ma, digital scale = 0 dbfs ? 155 ?157 ?159 ?161 ?163 ?165 ?167 ?169 ?171 ?173 ?175 0 200 400 600 800 1000 1200 f out (mhz) nsd (dbm/hz) 07852-019 +85c +25c ?40c figure 19. nsd vs. f out over temperature, 1-channel qam, f dac = 2.4 ghz, full-scale current = 20 ma
ad9789 rev. a | page 14 of 76 ?5 ?15 ?25 ?35 ?45 ?55 ?65 ?75 ?85 50 250 450 650 850 frequency (mhz) aclr (dbc) 07852-015 docsis3 ?40c 0c +25c +85c figure 20. aclr performance over temperature, 1-channel qam, f dac = 2.3 ghz, full-scale current = 20 ma, f out = 200 mhz, sum scale = 48 (docsis spec is ?73 dbc; harmonic exception is ?63 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) harmonic level (dbc) 07852-014 docsis3 25c 65c 85c figure 21. second-order harmonic performance vs. f out over temperature, 1-channel qam, f dac = 2.3 ghz, full-scale current = 20 ma, sum scale = 48 (docsis spec is ?73 dbc; harmonic exception is ?63 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) aclr (dbc) 07852-031 docsis3 25c 65c 85c figure 22. noise floor vs. f out over temperature (aclr measured beyond 30 mhz), 1-channel qam, f dac = 2.3 ghz, full-scale current = 20 ma, sum scale = 48 (docsis spec is ?73 dbc) ?5 ?15 ?25 ?35 ?45 ?55 ?65 ?75 ?85 50 150 250 350 450 550 650 750 850 950 frequency (mhz) aclr (dbc) 07852-018 docsis3 ?40c 0c +25c +85c figure 23. aclr performance over temperature, 1-channel qam, f dac = 2.3 ghz, full-scale current = 20 ma, f out = 800 mhz, sum scale = 48 (docsis spec is ?73 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) harmonic level (dbc) 07852-017 docsis3 25c 65c 85c figure 24. third-order harmonic performance vs. f out over temperature, 1-channel qam, f dac = 2.3 ghz, full-scale current = 20 ma, sum scale = 48 (docsis spec is ?73 dbc; harmonic exception is ?63 dbc) ?5 ?15 ?25 ?35 ?45 ?55 ?65 ?75 ?85 50 250 450 650 850 1050 frequency (mhz) aclr (dbc) 07852-039 docsis3 2.3ghz 2.2ghz 2.4ghz figure 25. aclr performance over f dac , 1-channel qam, f out = 850 mhz, full-scale current = 20 ma, temperature = 25c, sum scale = 48 (docsis spec is ?73 dbc)
ad9789 rev. a | page 15 of 76 0 ?10 ?20 ?30 ?40 ?50 ?60 ?70 ?80 ?90 0 100 200 300 400 500 600 700 800 900 1000 frequency (mhz) aclr (dbc) 07852-040 docsis3 cmos lvds figure 26. aclr performance for cmos and lvds interfaces, 1-channel qam, f out = 840 mhz, f dac = 2.4 ghz, full-scale current = 20 ma, sum scale = 48 (docsis spec is ?73 dbc) ?5 ?15 ?25 ?35 ?45 ?55 ?65 ?75 ?85 50 250 450 650 850 1050 frequency (mhz) aclr (dbc) 07852-042 docsis3 25c 65c 85c figure 27. aclr performance over temperature, 2-channel qam, f out = 800 mhz, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 32 (docsis spec is ?70 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) harmonic level (dbc) 07852-043 docsis3 25c 65c 85c figure 28. third-order harmonic performance vs. f out over temperature, 2-channel qam, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 32 (docsis spec is ?70 dbc; harmonic exception is ?63 dbc) ?5 ?15 ?25 ?35 ?45 ?55 ?65 ?75 ?85 50 250 450 650 850 1050 frequency (mhz) aclr (dbc) 07852-044 docsis3 25c 65c 85c figure 29. aclr performance over temperature, 2-channel qam, f out = 200 mhz, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 32 (docsis spec is ?70 dbc; harmonic exception is ?63 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) harmonic level (dbc) 07852-045 docsis3 25c 65c 85c figure 30. second harmonic performance vs. f out over temperature, 2-channel qam, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 32 (docsis spec is ?70 dbc; harmonic exception is ?63 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) aclr (dbc) 07852-046 docsis3 25c 65c 85c figure 31. noise floor vs. f out over temperature (aclr measured beyond 30 mhz), 2-channel qam, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 32 (docsis spec is ?70 dbc)
ad9789 rev. a | page 16 of 76 0 ?10 ?20 ?30 ?40 ?50 ?60 ?70 ?80 50 250 450 650 850 1050 frequency (mhz) aclr (dbc) 07852-027 docsis3 ?40c 0c +25c +85c figure 32. aclr performance over temperature, 4-channel qam, f out = 200 mhz, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 20 (docsis spec is ?67 dbc; harmonic exception is ?63 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) harmonic level (dbc) 07852-026 docsis3 25c 65c 85c figure 33. second-order harmonic performance vs. f out over temperature, 4-channel qam, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 20 (docsis spec is ?67 dbc; harmonic exception is ?63 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) aclr (dbc) 07852-028 docsis3 25c 65c 85c figure 34. noise floor vs. f out over temperature (aclr measured beyond 30 mhz), 4-channel qam, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 20 (docsis spec is ?67 dbc) 0 ?10 ?20 ?30 ?40 ?50 ?60 ?70 ?80 50 250 450 650 850 1050 frequency (mhz) aclr (dbc) 07852-030 docsis3 ?40c 0c +25c +85c figure 35. aclr performance over temperature, 4-channel qam, f out = 800 mhz, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 20 (docsis spec is ?67 dbc) ? 55 ?60 ?65 ?70 ?75 ?80 ?85 0 100 200 300 400 500 600 700 800 900 1000 f out (mhz) harmonic level (dbc) 07852-029 docsis3 25c 65c 85c figure 36. third-order harmonic performance vs. f out over temperature, 4-channel qam, f dac = 2.3 ghz, full-scale current = 25 ma, sum scale = 20 (docsis spec is ?67 dbc; harmonic exception is ?63 dbc) 0 ?10 ?20 ?30 ?40 ?50 ?60 ?70 ?80 50 250 450 650 850 1050 frequency (mhz) aclr (dbc) 07852-047 docsis3 2.3ghz 2.2ghz 2.4ghz figure 37. aclr performance over f dac , 4-channel qam, f out = 850 mhz, full-scale current = 25 ma, temperature = 25c, sum scale = 20 (docsis spec is ?67 dbc)
ad9789 rev. a | page 17 of 76 07852-020 center 840.00mhz res bw 56khz vbw 560khz span 42mhz sweep 39.12ms (601 pts) a tten 2db ref ?32.76db m freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?18.10dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?65.57 ?83.66 ?68.98 ?87.07 6.375mhz 5.250mhz ?75.01 ?93. 11 ?74.62 ?92.71 12.00mhz 6.000mhz ?76.83 ?94.92 ?76.46 ?94.55 18.00mhz 6.000mhz ?77.17 ?95.26 ?76.56 ?94.66 figure 38. 1-channel qam aclr, f out = 840 mhz, temperature = 25c, sum scale = 48, full-scale current = 20 ma, span = 42 mhz 07852-061 center 840.00mhz res bw 30khz vbw 300khz span 42mhz sweep 136.2ms (601 pts) a tten 2db ref ?35.91db m freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?21.75dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?71.64 ?93.39 ?72.50 ?94.25 6.375mhz 5.250mhz ?73.71 ?95.47 ?66.72 ?88.47 12.00mhz 6.000mhz ?73.58 ?95.33 0.50 ?21.10 18.00mhz 6.000mhz ?73.70 ?95.45 ?66.72 ?88.48 figure 39. 2-channel qam aclr, f out = 840 mhz, sum scale = 32, full-scale current = 25 ma, span = 42 mhz, channel 1 center 840.00mhz res bw 30khz vbw 300khz span 18mhz sweep 58.4ms (601 pts) a tten 2db ref ?32.76db m 0 7852-023 freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?17.98dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?73.99 ?91.97 ?74.93 ?92.91 6.375mhz 5.250mhz ?74.94 ?92.92 ?75.35 ?93.33 figure 40. 1-channel qam aclr, f out = 840 mhz, temperature = 25c, sum scale = 48, full-scale current = 20 ma, span = 18 mhz 07852-066 start 831.00mhz res bw 30khz vbw 300khz stop 873mhz sweep 136.2ms (601 pts) a tten 2db ref ?35.91db m freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?21.29dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?70.07 ?92.16 ?73.20 ?94.49 6.375mhz 5.250mhz ?69.05 ?90.34 ?73.87 ?95.16 12.00mhz 6.000mhz ?0.49 ?21.78 ?73.29 ?94.58 18.00mhz 6.000mhz ?66.61 ?87.90 ?73.98 ?95.27 figure 41. 2-channel qam aclr, f out = 840 mhz, sum scale = 32, full-scale current = 25 ma, span = 42 mhz, channel 2
ad9789 rev. a | page 18 of 76 07852-065 center 840.00mhz res bw 30khz vbw 300khz span 18mhz sweep 58.4ms (601 pts) freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?21.56dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?75.37 ?96.93 ?75.56 ?97. 11 6.375mhz 5.250mhz ?73.85 ?95.41 ?72.54 ?94.10 a tten 2db ref ?35.91db m figure 42. zoomed 2-channel qam aclr, f out = 840 mhz, sum scale = 32, full-scale current = 25 ma, span = 18 mhz, channel 1 center 834.00mhz res bw 30khz vbw 300khz span 42mhz sweep 136.2ms (601 pts) a tten 2db ref ?35.96db m 07852-021 freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?23.63dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?70.33 ?93.96 ?11.07 ?34.70 6.375mhz 5.250mhz ?69.04 ?92.67 ?0.49 ?24.12 12.00mhz 6.000mhz ?70.38 ?94.01 0.00 ?23.63 18.00mhz 6.000mhz ?71.02 ?94.65 0.43 ?23.20 figure 43. 4-channel qam aclr, f out = 840 mhz, temperature = 25c, sum scale = 20, full-scale current = 25 ma, span = 42 mhz, channel 1 07852-067 center 852.00mhz res bw 30khz vbw 300khz span 18mhz sweep 58.4ms (601 pts) a tten 2db ref ?35.91db m freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?21.03dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?75.51 ?96.54 ?75.17 ?96.20 6.375mhz 5.250mhz ?72.55 ?93.58 ?73.90 ?94.93 figure 44. zoomed 2-channel qam aclr, f out = 840 mhz, sum scale = 32, full-scale current = 25 ma, span = 18 mhz, channel 2 center 852.00mhz res bw 30khz vbw 300khz span 42mhz sweep 136.2ms (601 pts) a tten 2db ref ?35.96dbm 07852-022 freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?23.23dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?11.10 ?34.32 ?72.19 ?95.42 6.375mhz 5.250mhz ?0.75 ?23.98 ?68.97 ?92.20 12.00mhz 6.000mhz ?0.59 ?23.81 ?70.32 ?93.55 18.00mhz 6.000mhz ?0.35 ?23.58 ?70.70 ?93.93 figure 45. 4-channel qam aclr, f out = 840 mhz, temperature = 25c, sum scale = 20, full-scale current = 25 ma, span = 42 mhz, channel 4
ad9789 rev. a | page 19 of 76 center 834.00mhz res bw 30khz vbw 300khz span 18mhz sweep 58.4ms (601 pts) a tten 2db ref ?35.96dbm 07852-024 freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?23.62dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?72.95 ?96.56 ?10.86 ?34.48 6.375mhz 5.250mhz ?69.38 ?92.99 ?0.51 ?24.13 figure 46. zoomed 4-channel qam aclr, f out = 840 mhz, temperature = 25c, sum scale = 20, full-scale current = 25 ma, span = 18 mhz, channel 1 50 48 46 44 42 40 38 36 34 32 30 50 150 250 350 450 550 650 750 850 950 f out (mhz) mer (db) 07852-032 +25c +85c ?40c figure 47. modulation error ratio, equalized, 1-channel 256-qam, f dac = 2.29376 ghz, full-scale current = 20 ma, sum scale = 48 (equalization filter from demodulation toolbox on spectrum analyzer used) 50 48 46 44 42 40 38 36 34 32 30 50 150 250 350 450 550 650 750 850 950 f out (mhz) mer (db) 07852-033 +25c +85c ?40c figure 48. modulation error ratio, unequalized, 1-channel 256-qam, f dac = 2.29376 ghz, full-scale current = 20 ma, sum scale = 48 center 852.00mhz res bw 30khz vbw 300khz span 18mhz sweep 58.4ms (601 pts) a tten 2db ref ?35.96dbm 07852-025 freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?23.20dbm/ 6.00000mhz rms results 3.375mhz 750.0khz ?11.20 ?34.40 ?74.44 ?97.64 6.375mhz 5.250mhz ?0.77 ?23.96 ?69.07 ?92.26 figure 49. zoomed 4-channel qam aclr, f out = 840 mhz, temperature = 25c, sum scale = 20, full-scale current = 25 ma, span = 18 mhz, channel 4 50 48 46 44 42 40 38 36 34 32 30 50 150 250 350 450 550 650 750 850 950 f out (mhz) mer (db) 07852-035 +25c +85c ?40c figure 50. modulation error ratio, equalized, 4-channel 256-qam, f dac = 2.29376 ghz, full-scale current = 25 ma, sum scale = 20 (equalization filter from demodulation toolbox on spectrum analyzer used) 50 48 46 44 42 40 38 36 34 32 30 50 150 250 350 450 550 650 750 850 950 f out (mhz) mer (db) 07852-036 +25c +85c ?40c figure 51. modulation error ratio, unequalized, 4-channel 256-qam, f dac = 2.29376 ghz, full-scale current = 25 ma, sum scale = 20
ad9789 rev. a | page 20 of 76 80 75 70 65 60 55 50 45 40 35 30 25 20 15 10 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 f out (mhz) sfdr (dbc) 07852-068 figure 52. sfdr vs. f out in mix mode, f dac = 2.4 ghz, full-scale current = 20 ma (second nyquist zone performance) 80 75 70 65 60 55 50 45 40 35 30 85 90 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 f out (mhz) imd (dbc) 07852-076 figure 53. imd vs. f out in mix mode, f dac = 2.4 ghz, full-scale current = 20 ma (second nyquist zone performance) ? 40 ?45 ?50 ?55 ?60 ?65 ?70 ?75 ?80 ?85 1150 1250 1350 1450 1550 1650 1750 1850 1950 2050 2150 2250 f out (mhz) aclr (dbc) 07852-075 first adjacent channe l second adjacent channe l third adjacent channe l fifth adjacent channe l figure 54. aclr vs. f out in mix mode with one-carrier wcdma, f dac = 2304 mhz, full-scale current = 20 ma (second nyquist zone performance) center 2.100ghz res bw 30khz vbw 300khz span 53.84mhz sweep 174.6ms (601 pts) a tten 0db ref ?32.62db m freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?19.95dbm/ 3.84000mhz rms results 5.000mhz 3.840mhz ?68.93 ?88.88 ?67.99 ?87.94 10.00mhz 3.840mhz ?71.31 ?91.26 ?70.42 ?90.37 15.00mhz 3.840mhz ?73.43 ?93.37 ?72.68 ?92.63 20.00mhz 3.840mhz ?75.12 ?95.07 ?74.89 ?94.84 25.00mhz 3.840mhz ?75.60 ?95.55 ?76.51 ?96.46 07852-092 figure 55. one-carrier wcdma aclr in mix mode, f out = 2.1 ghz, f dac = 2304 mhz, full-scale current = 20 ma center 2.102 50ghz res bw 30khz vbw 300khz span 63.84mhz sweep 207ms (601 pts) freq. lower upper offset ref bw dbc dbm dbc dbm carrier power ?26.06dbm/ 3.84000mhz rms results 5.000mhz 3.840mhz ?0.25 ?26.31 ?0.42 ?26.47 10.00mhz 3.840mhz ?0.42 ?26.48 ?63.50 ?89.56 15.00mhz 3.840mhz ?64.07 ?90.13 ?65.13 ?91.18 20.00mhz 3.840mhz ?65.36 ?91.42 ?66.97 ?93.03 25.00mhz 3.840mhz ?66.86 ?92.92 ?68.70 ?94.76 30.00mhz 3.840mhz ?67.83 ?93.89 ?68.64 ?94.70 07852-093 a tten 2db ref ?38.62db m figure 56. four-carrier wcdma aclr in mix mode, f out = 2.1 ghz, f dac = 2304 mhz, full-scale current = 20 ma
ad9789 rev. a | page 21 of 76 1100 1000 900 800 700 600 500 400 300 200 100 0 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 f dac (ghz) power dissipation (mw) 07852-094 avdd33 dvdd33 (lvds) dvdd33 (cmos) dvdd18 dvdd15 cvdd18 figure 57. power dissipation by supply vs. f dac , 4-channel docsis, f out = 915 mhz, full-scale current = 25 ma (datapath configuration: qam encoder on, srrc filter on, four 2 interpolation filters on) 700 600 500 400 300 200 100 0 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 f dac (ghz) power dissipation (mw) 07852-095 avdd33 dvdd33 (lvds) dvdd33 (cmos) dvdd18 dvdd15 cvdd18 figure 58. power dissipa tion by supply vs. f dac , 16 interpolation, one channel enabled, f out = 70 mhz, full-scale current = 20 ma 200 180 160 140 120 100 80 60 40 20 0 8 101214161820222426283032 full-scale current (ma) power dissipation (mw) 07852-098 avdd33 figure 59. avdd33 power dissipation vs. full-scale current 2000 1800 1600 1400 1200 1000 800 600 400 200 0 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 f dac (ghz) power dissipation (mw) 07852-096 total (cmos) total (lvds) figure 60. total power dissipation vs. f dac , 4-channel docsis, f out = 915 mhz, full-scale current = 25 ma (datapath configuration: qam encoder on, srrc filter on, four 2 interpolation filters on) 1400 1200 1000 800 600 400 200 0 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 f dac (ghz) power dissipation (mw) 07852-097 total (cmos) total (lvds) figure 61. total power dissipation vs. f dac , 16 interpolation, one channel enabled, f out = 70 mhz, full-scale current = 20 ma
ad9789 rev. a | page 22 of 76 terminology monotonicity a dac is monotonic if the output either increases or remains constant as the digital input increases. offset error offset error is the deviation of the output current from the ideal of 0. for ioutp, 0 ma output is expected when all inputs are set to 0. for ioutn, 0 ma output is expected when all inputs are set to 1. gain error gain error is the difference between the actual and ideal output span. the actual span is determined by the output when all inputs are set to 1s minus the output when all inputs are set to 0s. temp er atu re d r i f t temperature drift is specified as the maximum change from the ambient (25c) value to the value at either t min or t max . for offset, gain, and reference drift, the drift is reported in ppm per c. power supply rejection (psr) psr is the maximum change in the full-scale output as the supplies are varied from nominal to minimum and maximum specified voltages. output compliance range the output compliance range is the range of allowable voltage at the output of a current output dac. operation beyond the maximum compliance limits may cause either output stage saturation or breakdown, resulting in nonlinear performance. spurious-free dynamic range (sfdr) sfdr is the difference, in db, between the peak amplitude of the output signal and the peak spurious signal over the specified bandwidth. noise spectral density (nsd) nsd is the converter noise power per unit of bandwidth. nsd is usually specified in dbm/hz in the presence of a 0 dbm full- scale signal. adjacent channel leakage ratio (aclr) the adjacent channel leakage (power) ratio is the ratio, in dbc, between the measured power within a channel relative to its adjacent channels. modulation error ratio (mer) modulated signals create a discrete set of output values referred to as a constellation. each symbol creates an output signal corre- sponding to one point on the constellation. mer is a measure of the discrepancy between the average output symbol magnitude and the rms error magnitude of the individual symbol. intermodulation distortion (imd) imd is the result of two or more signals at different frequencies mixing together. many products are created according to the formula af 1 bf 2 , where a and b are integer values.
ad9789 rev. a | page 23 of 76 serial control port the ad9789 serial control port is a flexible, synchronous serial communications port that allows an easy interface to many industry-standard microcontrollers and microprocessors. the ad9789 serial control port is compatible with most synchronous transfer formats, including both the motorola spi? and intel? ssr protocols. the serial control port allows read/write access to all registers that configure the ad9789. single- or multiple-byte transfers are supported, as well as msb first or lsb first transfer formats. the ad9789 serial control port can be configured for a single bidirectional i/o pin (sdio only) or for two unidirectional i/o pins (sdio/sdo). by default, the ad9789 is in unidirectional long instruction mode (long instruction mode is the only instruction mode supported). serial control port pin descriptions the sclk (serial clock) pin is the serial shift clock. this pin is an input. sclk is used to synchronize serial control port reads and writes. write data bits are registered on the rising edge of this clock, and read data bits are registered on the falling edge. this pin is internally pulled down by a 30 k resistor to ground. sdio (serial data input/output) is a dual-purpose pin that acts as an input only (unidirectional mode) or as both an input and an output (bidirectional mode). the ad9789 defaults to the unidirectional i/o mode (register 0x00[7] = 0). the sdo (serial data output) pin is used only in the unidirectional i/o mode as a separate output pin for reading back data. cs (chip select bar) is an active low control that gates the read and write cycles. when cs is high, sdo and sdio are in a high impedance state. this pin is internally pulled up by a 30 k resistor to dvdd33. m1 l1 n1 p1 ad9789 serial control port 07852-048 sclk cs sdo sdio figure 62. serial control port general operation of serial control port a write or read operation to the ad9789 is initiated by pulling cs low. cs stall high is supported in modes where three or fewer bytes of data (plus the instruction data) are transferred (see ). in these modes, table 7 cs can temporarily return high on any byte boundary, allowing time for the system controller to process the next byte. cs can go high on byte boundaries only and can go high during either part (instruction or data) of the transfer. during cs stall high mode, the serial control port state machine enters a wait state until all data is sent. if the system controller decides to abort the transfer before all of the data is sent, the state machine must be reset by either completing the remaining transfers or by returning cs low for at least one complete sclk cycle (but less than eight sclk cycles). raising cs on a nonbyte boundary terminates the serial transfer and flushes the buffer. in streaming mode (see table 7 ), any number of data bytes can be transferred in a continuous stream. the register address is automatically incremented or decremented (see the msb/lsb first transfers section). cs must be raised at the end of the last byte to be transferred, thereby ending streaming mode. communication cycleinstruction plus data there are two parts to a communication cycle with the ad9789. in the first part, a 16-bit instruction word is written to the ad9789, coincident with the first 16 sclk rising edges. the instruction word provides the ad9789 serial control port with information regarding the data transfer, which is the second part of the communication cycle. the instruction word defines whether the upcoming data transfer is a read or a write, the number of bytes in the data transfer, and the starting register address for the first byte of the data transfer. write if the instruction word is for a write operation, the second part of the communication cycle is the transfer of data into the serial control port buffer of the ad9789. data bits are registered on the rising edge of sclk. the length of the transfer (one, two, or three bytes or streaming mode) is indicated by two bits (n1 and n0) in the instruction byte. when the transfer is one, two, or three bytes (but not streaming mode), cs can be raised after each sequence of eight bits to stall the bus, except after the last byte, where it ends the cycle. when the bus is stalled, the serial transfer resumes when cs is lowered. raising cs on a nonbyte boundary resets the serial control port. during a write, streaming mode does not skip reserved or blank registers; therefore, the user must know what bit pattern to write to the reserved registers to preserve proper operation of the part. it does not matter what data is written to blank registers. most writes to the control registers immediately reconfigure the device. however, register 0x16 through register 0x1d do not directly control device operation. they provide data to internal logic that must perform additional operations on the data before it is downloaded and the device configuration is changed. for any updates to register 0x16 through register 0x1d to take effect, the freqnew bit (register 0x1e[7]) must be set to 1 (this bit is self-clearing). any number of bytes of data can be changed before updating registers. setting the freqnew bit simultaneously updates register 0x16 through register 0x1d. in a similar fashion, any changes to register 0x22 and register 0x23 require parmnew (register 0x24[7]) to be toggled from a low state to a high state before the new values take effect. unlike the freqnew bit, parmnew is not self-clearing.
ad9789 rev. a | page 24 of 76 read if the instruction word is for a read operation, the next n 8 sclk cycles clock out the data from the address specified in the instruction word, where n is 1 to 3 as determined by bits[n1:n0]. if n = 4, the read operation is in streaming mode, continuing until cs is raised. streaming mode does not skip over reserved or blank registers. the readback data is valid on the falling edge of sclk. the default mode of the ad9789 serial control port is the uni- directional mode. in unidirectional mode, the readback data appears on the sdo pin. it is also possible to set the ad9789 to bidirectional mode using the sdio_dir bit (register 0x00[7]). in bidirectional mode, both the sent data and the readback data appear on the sdio pin. a readback request reads the data that is in the serial control port buffer area or the data in the active registers (see figure 63 ). the ad9789 supports only the long instruction mode; therefore, register 0x00[4:3] reads 11 (this register uses mirrored bits). long instruction mode is the default at power-up or reset, and writing to these bits has no effect. the ad9789 uses register addre ss 0x00 to register address 0x55. 07852-049 active registers serial control port sclk sdo sdio cs buffer registers freqnew write register 0x1e = 0x10 to update registers figure 63. relationship between seri al control port buffer registers and active registers of the ad9789 instruction word (16 bits) the msb of the instruction word is r/ w , which indicates whether the instruction is a read or a write. the next two bits, n1 and n0, indicate the length of the transfer in bytes. the final 13 bits (bits[a12:a0]) are the address at which to begin the read or write operation. for a write, the instruction word is followed by the number of bytes of data indicated by bits[n1:n0] (see table 7 ). table 7. byte transfer count n1 n0 bytes to transfer 0 0 1 0 1 2 1 0 3 1 1 streaming mode bits[a12:a0] select the address within the register map that is written to or read from during the data transfer portion of the communication cycle. only bits[a6:a0] are needed to cover the range of the 0x55 registers used by the ad9789. bits[a12:a7] must always be 0. for multibyte transfers, this address is the starting byte address. in msb first mode, subsequent bytes increment the address. msb/lsb first transfers the ad9789 instruction word and byte data can be msb first or lsb first. any data written to register 0x00 must be mirrored, the upper four bits (bits[7:4]) with the lower four bits (bits [3:0]). this makes it irrelevant whether lsb first or msb first is in effect. as an example of this mirroring, the default setting for register 0x00[7:0] is 0x18, which mirrors bit 4 and bit 3. these bits set the long instruction mode (the default and the only mode supported). the default for the ad9789 is msb first. when lsb first is set by register 0x00[1] and register 0x00[6], it takes effect immediately. in multibyte transfers, subsequent bytes reflect any changes in the serial port configuration. when msb first mode is active, the instruction and data bytes must be written from msb to lsb. multibyte data transfers in msb first format start with an instruction byte that includes the register address of the most significant data byte. subsequent data bytes must follow in order from the high address to the low address. in msb first mode, the serial control port internal address generator decrements for each data byte of the multi- byte transfer cycle. when lsb first mode is active, the instruction and data bytes must be written from lsb to msb. multibyte data transfers in lsb first format start with an instruction byte that includes the register address of the least significant data byte followed by multiple data bytes. the internal byte address generator of the serial control port increments for each byte of the multibyte transfer cycle. the ad9789 serial control port register address decrements from the register address just written toward 0x00 for multibyte i/o operations if the msb first mode is active (default). if the lsb first mode is active, the register address of the serial control port increments from the address just written toward 0x55 for multibyte i/o operations. streaming mode always terminates when it reaches address 0x2f. note that unused addresses are not skipped during multibyte i/o operations. table 8. streaming mode (no addresses are skipped) write mode address direction stop sequence lsb first increment 0x02d, 0x02e, 0x02f, stop msb first decrement 0x001, 0x000, 0x02f, stop
ad9789 rev. a | page 25 of 76 table 9. serial control port, 16 -bit instruction word, msb first msb lsb i15 i14 i13 i12 i11 i10 i9 i8 i7 i6 i5 i4 i3 i2 i1 i0 r/ w n1 n0 a12 a11 a10 a9 a8 a7 a6 a5 a4 a3 a2 a1 a0 cs scl k don't care sdio a12 n0 n1 a11 a10 a9 a8 a7 a6 a5 a4 a3 a2 a1 a0 d7 d6 d5 d4 d3 d2 d1 d0 d7 d6 d5 d4 d3 d2 d1 d0 don't care don't care don't care 16-bit instruction header register (n) data register (n ? 1) data r/w 07852-050 figure 64. serial control port writemsb firs t, 16-bit instruction, two bytes of data cs scl k sdio sdo register (n) data 16-bit instruction header register (n ? 1) data register (n ? 2) data register (n ? 3) data a12 n0n1 a11 a10 a9 a8 a7 a6 a5 a4 a3 a2 a1 a0 don't care don't care don't care r/w don't care d7 d6 d5 d4 d3 d2 d1 d0 d7 d6 d5 d4 d3 d2 d1 d0 d7 d6 d5 d4 d3 d2 d1 d0 d7 d6 d5 d4 d3 d2 d1 d0 07852-051 figure 65. serial control port readmsb firs t, 16-bit instruction, four bytes of data t s don't care don't care n1n0a12a11a10a9a8a7a6a5d4d3d2d1d0 don't care don't care r/w t ds t dh t hi t lo t clk t c cs sclk sdio 0 7852-052 figure 66. serial control port writemsb firs t, 16-bit instruction, timing measurements data bit n ? 1 data bit n cs sclk sdio sdo t dv 07852-053 figure 67. timing diagram for serial control port register read sclk don't care don't care 16-bit instruction header register (n) data register (n + 1) data sdio don't care don't care a0 a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 d1 d0 n1n0 d2 d3 d4 d5 d6 d7 d0 d1 d2 d3 d4 d5 d6 d7 r/w cs 0 7852-054 figure 68. serial control port writelsb firs t, 16-bit instruction, two bytes of data
ad9789 rev. a | page 26 of 76 cs sclk sdio t hi t lo t clk t s t ds t dh t c bit n bit n + 1 07852-055 figure 69. serial control port timingwrite table 10. serial control port timing parameter description t ds setup time between data and rising edge of sclk t dh hold time between data and rising edge of sclk t clk period of the clock setup time between cs falling edge and sclk rising edge (start of communication cycle) t s setup time between sclk rising edge and cs rising edge (end of communication cycle) t c t hi minimum period that sclk should be in a logic high state t lo minimum period that sclk should be in a logic low state t dv sclk to valid sdio and sdo (see figure 67 )
ad9789 rev. a | page 27 of 76 spi register map do not write to the following registers unless instructed otherwis e: register 0x34, register 0x35, register 0x37, register 0x3b , register 0x3f, or register 0x40 through register 0x55. table 11. register map addr register name bit 7 bit 6 bit 5 bit 4 bit 3 bit 2 bit 1 bit 0 de- fault 0x00 spi control sdio_dir lsbfirst reset lng_inst 0x18 0x01 saturation counter satcnt[7:0] 0x00 0x02 parity counter parcnt[7:0] 0x00 0x03 interrupt enable parerr bistdone parmset parmclr lockacq locklost saterr reserved 0x00 0x04 interrupt status/clear parerr bistdone parmset parmclr lockacq locklost saterr reserved 0x00 0x05 channel enable reserved chanen[3:0] 0x00 0x06 bypass qam srrc reserved int[4:0] 0x00 0x07 qam/srrc configuration reserved alpha[1:0] reserved mapping[2:0] 0x01 0x08 summing node scalar sumscale[7:0] 0x0d 0x09 input scalar inscale[7:0] 0x20 0x0a ftw0[7:0] 0x00 0x0b ftw0[15:8] 0x00 0x0c nco 0 frequency tuning word ftw0[23:16] 0x00 0x0d ftw1[7:0] 0x00 0x0e ftw1[15:8] 0x00 0x0f nco 1 frequency tuning word ftw1[23:16] 0x00 0x10 ftw2[7:0] 0x00 0x11 ftw2[15:8] 0x00 0x12 nco 2 frequency tuning word ftw2[23:16] 0x00 0x13 ftw3[7:0] 0x00 0x14 ftw3[15:8] 0x00 0x15 nco 3 frequency tuning word ftw3[23:16] 0x00 0x16 q[7:0] 0x00 0x17 q[15:8] 0x00 0x18 rate converter denominator (q) q[23:16] 0x80 0x19 p[7:0] 0x00 0x1a p[15:8] 0x00 0x1b rate converter numerator (p) p[23:16] 0x80 0x1c fc[7:0] 0x00 0x1d interpolating bpf center frequency fc[15:8] 0x00 0x1e frequency update freqnew reserved 0x00 0x1f hardware version reserved ver[3:0] 0x03 0x20 interface configuration cmos_bus cmos_ctrl reserved dco_in v if_mode chanpri par[1:0] 0xc8 0x21 data control bin buswdth[1:0] datwdth cmplx ltncy[2:0] 0x61 0x22 dco frequency reserved dcodiv[2:0] ones[3:0] 0x1f 0x23 internal clock phase adjust dscphz[3:0] sncphz[3:0] 0x85 0x24 parameter update parmnew reserved 0x00 0x25 channel 0 gain chan0gain[7:0] 0x80 0x26 channel 1 gain chan1gain[7:0] 0x80 0x27 channel 2 gain chan2gain[7:0] 0x80 0x28 channel 3 gain chan3gain[7:0] 0x80 0x29 spectrum shaping reserved spec_inv 0x00
ad9789 rev. a | page 28 of 76 addr register name bit 7 bit 6 bit 5 bit 4 bit 3 bit 2 bit 1 bit 0 de- fault 0x2f mu delay control 1 search_ tol search_err track_ err guardband[4:0] 0x0b 0x30 mu control duty cycle duty cycle correct enable inc_dec (factory) manual_adj[5:0] (factory test only) 0x40 0x31 clock receiver 1 clkn_cml[3:0] reserved 0xf0 0x32 clock receiver 2 clk_dis reserv ed psign clkp_cml[3:0] nsign 0x3f 0x33 mu delay control 2 mu_clkdis slope mode[1:0] musamp gain[1:0] mu_en 0x42 0x34 reserved reserved 0x00 0x35 reserved reserved 0xca 0x36 dac bias pdbias reserved msel[1:0] 0x03 0x37 reserved reserved 0x00 0x38 dac decoder reserved dac decoder mode 0x00 0x39 mu delay control 3 mudly[0] search_dir[1:0] muphz[4:0] 0x40 0x3a mu delay control 4 mudly[8:1] 0x00 0x3b reserved reserved 0x00 0x3c full-scale current 1 fsc[7:0] 0x00 0x3d full-scale current 2 reserved fsc[9:8] 0x02 0x3e phase detector control phz_pd reserved cmp_bst auto_cal phz_det_bias[3:0] 0x18 0x3f reserved reserved 0x00 0x40 bist control clkshdn inputsel reserved benable bmode[3:0] 0x00 0x41 bist status bdone bstatus[6:0] 0x00 0x42 padlen[7:0] 0x00 0x43 bist zero padding length padlen[15:8] 0x00 0x44 vectlen[7:0] 0x00 0x45 vectlen[15:8] 0x00 0x46 bist vector length vectlen[23:16] 0x00 0x47 bist clock adjust bclkdiv[3:0] bclkphz[3:0] 0x00 0x48 sign 0 control s0enabl s0rden s0prng s0zero s0neg s0fnlch s0sel[1:0] 0x00 0x49 sign 0 clock adjust s0clkdiv[3:0] s0clkphz[3:0] 0x00 0x4a sign 1 control s1enabl s1rden s1pr ng s1zero s1neg s1fnlch s1sel[1:0] 0x00 0x4b sign 1 clock adjust s1clkdiv[3:0] s1clkphz[3:0] 0x00 0x4c regfnl0freq final rate/ offset control 0 [7:0] 0x00 0x4d regfnl1freq final rate/ offset control 1 [7:0] 0x00 0x50 sgn0[7:0] 0x00 0x51 sgn0[15:8] 0x00 0x52 bist signature 0 sgn0[23:16] 0x00 0x53 sgn1[7:0] 0x00 0x54 sgn1[15:8] 0x00 0x55 bist signature 1 sgn1[23:16] 0x00
ad9789 rev. a | page 29 of 76 spi register descriptions table 12. spi control register (address 0x00) bit bit name description 7 sdio_dir this bit configures the sdio pin as an input-only pin or as a bidirectional input/output pin. both choices conform to the spi standard. 0 = input only. 1 = bidirectional (input/output). 6 lsbfirst this bit configures the spi interface for msb first or lsb first mode. both choices conform to the spi standard. 0 = msb first. 1 = lsb first. 5 reset when set to 1, this bit resets the part. after the pa rt is reset, 0 is written to this bit on the next cycle. 0 = no reset. 1 = software reset. 4 lng_inst this bit sets the spi to long in struction mode; 1 is the only valid value. [3:0] these bits should mirror bits[7:4]. bit 3 should mirror bit 4, bit 2 should mirror bit 5, bit 1 should mirror bit 6, and bit 0 should mirror bit 7. table 13. saturation counter register (address 0x01) bit bit name description [7:0] satcnt[7:0] this read-only register contains the saturation counter. this register reflects the number of samples at the output of the sumscale gain block that overrang e the datapath and are digitally clippe d. the count is cleared by writing a 1 to register 0x04, bit 1. table 14. parity counter register (address 0x02) bit bit name description [7:0] parcnt[7:0] this read-only register contains the input data parity error counter. the count is cleared by writing a 1 to register 0x04, bit 7. table 15. interrupt enable register (address 0x03) bit name description 7 parerr setting this bit to 1 enables a parerr flag to generate an interrupt request. generating an interrupt request results in interrupt bit 7 being set in register 0x04 and the irq pin going low. 6 bistdone setting this bit to 1 enables a bistdone flag to generate an interrupt request. gene rating an interrupt request results in interrupt bit 6 being set in register 0x04 and the irq pin going low. 5 parmset setting this bit to 1 enables a parms_set flag to generate an interrupt request. gene rating an interrupt request results in interrupt bit 5 being set in register 0x04 and the irq pin going low. 4 parmclr setting this bit to 1 enables a parms_clr flag to generate an interrupt request. gene rating an interrupt request results in interrupt bit 4 being set in register 0x04 and the irq pin going low. 3 lockacq setting this bit to 1 enables a lockacq flag to generate an interrupt request. gene rating an interrupt request results in interrupt bit 3 being set in register 0x04 and the irq pin going low. 2 locklost setting this bit to 1 enables a locklost flag to generate an interrupt request. gene rating an interrupt request results in interrupt bit 2 being set in register 0x04 and the irq pin going low. 1 saterr setting this bit to 1 enables a saterr (overflow into 16 interpolator) flag to generate an interrupt request. generating an interrupt request result s in interrupt bit 1 being set in register 0x04 and the irq pin going low. 0 reserved reserved.
ad9789 rev. a | page 30 of 76 table 16. interrupt status/clear register (address 0x04) bit name description 7 parerr if this bit is set to 1, one or more parity errors has occurred. writing a 1 to th is bit clears the interrupt. 6 bistdone if this bit is set to 1, the bist has reached the terminal state. writin g a 1 to this bit clears the interrupt. 5 parmset if this bit is set to 1, the parameter update register (addr ess 0x24) has been updated. writing a 1 to this bit clears the interrupt. 4 parmclr if this bit is set to 1, the parameter update register (addr ess 0x24) has been cleared. writing a 1 to this bit clears the interrupt. 3 lockacq if this bit is set to 1, proper data handoff between the digital engine and the dac core is occurring. 2 locklost if this bit is set to 1, proper data handoff between the digital engine and the dac core has been lost. writing a 1 to this bit clears the interrupt. 1 saterr if this bit is set to 1, one or more saturation errors (o verflow into 16 interpolator) has occurred. writing a 1 to this bit clears the interrupt. 0 reserved reserved. table 17. channel enable register (address 0x05) bit bit name description [7:4] reserved reserved. [3:0] chanen[3:0] a logic 1 in any bit position enables the co rresponding channel; 0000 means that all channels are disabled. setting channels enabled 0000 all channels disabled. 0001 channel 0 enabled. 0010 channel 1 enabled. 0011 channel 0 and channel 1 enabled. 1110 channel 1, channel 2, and channel 3 enabled. 1111 all channels enabled. table 18. bypass register (address 0x06) bit bit name description 7 qam if this bit is set to 1, the qam mappers are bypassed. 6 srrc if this bit is set to 1, the square r oot raised cosine (srrc) filters are bypassed. 5 reserved reserved. [4:0] int[4:0] a logic 1 in any bit position bypasses the corresponding interpolation filter. the preferred order for bypassing interpolation filters is to first bypass filter 0, then filter 1, and so on. setting interpolation filters bypassed 00000 all interpolation filters enabled. 00001 interpolation filter 0 bypassed. 00010 interpolation filter 1 bypassed. 00011 interpolation filter 0 and interpolation filter 1 bypassed. 01111 interpolation filter 0, interpolation filter 1, interpolation filter 2, and interpolation filter 3 bypassed. 11111 all interpolation filters bypassed.
ad9789 rev. a | page 31 of 76 table 19. qam/srrc configuration register (address 0x07) table 20. summing node scalar register (address 0x08) table 21. input scalar register (address 0x09) bit bit name description [7:6] reserved reserved. [5:4] alpha[1:0] these bits set the srrc filter alpha. setting alpha filter 00 0.12 01 0.18 10 0.15 11 0.13 3 reserved reserved. [2:0] mapping[2:0] these bits set the qam encoding. setting qam encoding 000 docsis 64-qam 001 docsis 256-qam 010 dvb-c 16-qam 011 dvb-c 32-qam 100 dvb-c 64-qam 101 dvb-c 128-qam 110 dvb-c 256-qam 111 unused bit bit name description [7:0] sumscale[7:0] this register sets the value of the 2.6 mult iplier that is applied to the output of the channel summing nod e. setting 2.6 multiplier 00000000 0 00000001 0.015625 00000010 0.03125 00001101 0.203125 (default) 11111110 3.96875 11111111 3.984375 bit bit name description [7:0] inscale[7:0] this register sets the value of the 3.5 multiplier that is applied to the input data. this scaling block is in parallel with the qam encoder block and is used when the qam encoder block is bypassed. setting 3.5 multiplier 00000000 0 00000001 0.03125 00000010 0.0625 00100000 1 (default) 11111110 7.9375 11111111 7.96875
ad9789 rev. a | page 32 of 76 the three nco 0 frequency tuning word registers together compose the 24-bit frequency tuning word for nco 0. for more informati on about programming these registers, see the baseband digital upconverter section. table 22. nco 0 frequency tuning word registers (address 0x0a to address 0x0c) address bit name description 0x0a ftw0[7:0] frequency tuning word for nco 0, bits[7:0] 0x0b ftw0[15:8] frequency tuning word for nco 0, bits[15:8] 0x0c ftw0[23:16] frequency tuning word for nco 0, bits[23:16] the three nco 1 frequency tuning word registers together compose the 24-bit frequency tuning word for nco 1. for more informati on about programming these registers, see the baseband digital upconverter section. table 23. nco 1 frequency tuning word registers (address 0x0d to address 0x0f) address bit name description 0x0d ftw1[7:0] frequency tuning word for nco 1, bits[7:0] 0x0e ftw1[15:8] frequency tuning word for nco 1, bits[15:8] 0x0f ftw1[23:16] frequency tuning word for nco 1, bits[23:16] the three nco 2 frequency tuning word registers together compose the 24-bit frequency tuning word for nco 2. for more informati on about programming these registers, see the baseband digital upconverter section. table 24. nco 2 frequency tuning word registers (address 0x10 to address 0x12) address bit name description 0x10 ftw2[7:0] frequency tuning word for nco 2, bits[7:0] 0x11 ftw2[15:8] frequency tuning word for nco 2, bits[15:8] 0x12 ftw2[23:16] frequency tuning word for nco 2, bits[23:16] the three nco 3 frequency tuning word registers together compose the 24-bit frequency tuning word for nco 3. for more informati on about programming these registers, see the baseband digital upconverter section. table 25. nco 3 frequency tuning word registers (address 0x13 to address 0x15) address bit name description 0x13 ftw3[7:0] frequency tuning word for nco 3, bits[7:0] 0x14 ftw3[15:8] frequency tuning word for nco 3, bits[15:8] 0x15 ftw3[23:16] frequency tuning word for nco 3, bits[23:16] the three rate converter denominator (q) registers together compose the 24-bit denominator for the rate converter decimation ra tio. for more information about programming these registers, see the sample rate converter section. table 26. rate converter denominator (q) registers (address 0x16 to address 0x18) address bit name description 0x16 q[7:0] rate converter denominator, bits[7:0] 0x17 q[15:8] rate converter denominator, bits[15:8] 0x18 q[23:16] rate converter denominator, bits[23:16] the three rate converter numerator (p) registers together compose the 24-bit numerator for the rate converter decimation ratio. for more information about programming these registers, see the sample rate converter section. table 27. rate converter numerator (p) registers (address 0x19 to address 0x1b) address bit name description 0x19 p[7:0] rate converter numerator, bits[7:0] 0x1a p[15:8] rate converter numerator, bits[15:8] 0x1b p[23:16] rate converter numerator, bits[23:16]
ad9789 rev. a | page 33 of 76 the two interpolating bpf center frequency registers together compose the 16-bit center frequency of the 16 band-pass interpol ation filter. for more information about programming these registers, see the digital 16 tunable band-pass filter section. table 28. interpolating bpf center frequenc y registers (address 0x1c and address 0x1d) address bit name description 0x1c fc[7:0] center frequency, bits[7:0] 0x1d fc[15:8] center frequency, bits[15:8] table 29. frequency update register (address 0x1e) bit name description 7 freqnew setting this bit to 1 updates the derived registers in the ad9789. this bit must be set for changes to register 0x16 through register 0x1d to take effect. this self-clearing bit is reset to 0 after the derived registers are updated. [6:0] reserved reserved. table 30. hardware version register (address 0x1f) bit name description [7:4] reserved reserved. [3:0] ver[3:0] this read-only register indicates the version of the chip (0011). table 31. interface configuration register (address 0x20) bit bit name description 7 cmos_bus this bit reflects the state of the cmos_bus pin (l14). 6 cmos_ctrl this bit reflects the state of the cmos_ctrl pin (m14). 5 reserved reserved. 4 dco_inv when set to 1, the dco pin is inverted. 3 if_mode this bit sets the data interface mode. 0 = channelizer mode. supports all available interface widths and 8- and 16-bit word widths. supports maximum f baud of f dac /48. 1 = quadrature digital upconverter (qduc) mode. supports 32-bit interface, 16-bit word mode only. supports maximum f baud of f dac /16. 2 chanpri this bit selects the channel prioritization value (used in channelizer mode only). 0 = device expects input samples only for those channels that are enabled. 1 = device expects data for all four channels. data for disabled channels is expected and must be sent, but this data is discarded by the ad9789. [1:0] par[1:0] these bits set the parity checking. for more information, see the parity section. setting parity checking 00 parity checking deactivated 01 iq parity (a value of 0 is expected on the i channel and a value of 1 is expected on the q channel) 10 even parity 11 odd parity
ad9789 rev. a | page 34 of 76 table 32. data control register (address 0x21) bit bit name description 7 bin this bit selects the coding for the device. 0 = twos complement coding. 1 = straight binary coding. [6:5] buswdth[1:0] these bits set the input data bus width for the device. 4 datwdth setting input bus width 00 4 bits 01 8 bits 10 16 bits 11 this bit sets the data-word wi dth that is sent to the datapaths. 0 = 8-bit data-word. 1 = 16-bit data-word. 3 cmplx this bit configures the datapath for real or complex data. 0 = real data. 1 = complex data. 32 bits [2:0] ltncy[2:0] these bits set the turnaround latency from the fs pulse to the internal data sampling time. for more information, see the latency register section. setting latency 000 input data begins to be sampled at approxim ately the first rising edge of dco after fs goes low. 001 input data begins to be sampled at approximately the second rising edge of dco after fs goes low. 111 input data begins to be sampled at approxim ately the eighth rising edge of dco after fs goes low. table 33. dco frequency register (address 0x22) bit bit name description 7 reserved reserved. [6:4] dcodiv[2:0] these bits configure the data clock output (dco) frequency. [3:0] setting dco clock frequency 000 dco clock disabled 001 f dacclk /16 010 f dacclk /32 011 invalid 100 f dacclk /64 101 invalid 11x ones[3:0] these bits always read back 1111. invalid
ad9789 rev. a | page 35 of 76 table 34. internal clock phase adjust register (address 0x23) bit bit name description [7:4] dscphz[3:0] the data sampling clock (dsc) is an internal clock that is used to sample the input data. this clock can occur on 1 of 16 phases to optimize the setup and hold timing of the data interface. setting selected phase 0000 earliest clock phase 0001 second earliest clock phase that occurs 1/16 of a dsc cycle later 1111 last available clock phase [3:0] sncphz[3:0] the synchronization clock (snc) is an internal clock that is used to synchronize the digital datapath clock with the dac clock. this clock can occur on 1 of 16 phases to optimize the dac-to-datapath timing. setting selected phase 0000 earliest clock phase 0001 second earliest clock phase that occurs 1/16 of a dsc cycle later 1111 last available clock phase table 35. parameter update register (address 0x24) bit name description 7 parmnew this bit must transition from 0 to 1 for changes to regi ster 0x22 and register 0x23 to take effect. assuming that this bit was previously set to 0, writing a 1 to this bit ca uses the readback value of the bit to reflect the state of the chip. (the state of the chip is updated very quickly; for this reason, users with slow spi implementations may never read back a 0 after an update.) 0 = values have not been updated. 1 = values have been updated. [6:0] reserved reserved. table 36. channel gain registers (address 0x25 to address 0x28) address register name bit name description 0x25 0x26 0x27 0x28 channel 0 gain channel 1 gain channel 2 gain channel 3 gain chan0gain[7:0] chan1gain[7:0] chan2gain[7:0] chan3gain[7:0] these registers configure a value for the 1.7 multiplier applied to each individual channel just prior to the sumscale block. the range of the channel gain is 0 to 1.9921875 with a step size of 0.0078125. to mute an individual channel, set the scale factor to 0. setting channel gain 00000000 0 00000001 0.0078125 11111111 1.9921875 table 37. spectrum shaping register (address 0x29) bit name description [7:1] reserved reserved. 0 spec_inv setting this bit to 1 spectrally inverts th e signal, effectively multiplying the q data by ?1.
ad9789 rev. a | page 36 of 76 table 38. mu delay control 1 register (address 0x2f) bit bit name description 7 search_tol this bit specifies the accuracy of th e phase search. the optimal value for this bit is 1. 0 = not exact: the search can find a phase within two values of the desired phase. 1 = exact: the search finds the exact phase specified. 6 search_err this bit configures the search behavior when an error is encountered. 0 = stop on error. 1 = retry on error. 5 track_err this bit configures the track behavior if the controller does not find the desired phase. the optimal value for this bit is 0. 0 = continue on error. 1 = reset on error. [4:0] guardband[4:0] these bits set the guard band value. the guard band is defined as follows: guardband[4:0] 8 = number of mu delay codes of guard band from the endpoints if the search mode is alternating, the search proceeds in both directions until the guard band is reached in one direction. when the guard band is reached, the search co ntinues only in the opposite direction. if the desired phase is not found before the guard band is reached in the second direction, the search reverts to the alternating mode and continues looking within the guard band. the search fails if the mu delay reaches the endpoints. for more information, see the mu delay controller section. setting guard band 00000 0 01011 11 (default) 11111 31 table 39. mu control duty cycle register (address 0x30) bit bit name description 7 duty cycle correct enable setting this bit to 1 turns on the mu control duty cycl e correction circuitry. turn on this function before enabling the mu controller. along with the phase comp arator boost (enabled in register 0x3e[5]), this function allows for more robust operation of the mu controller over the entire operating speed of the part. 6 inc_dec reserved (factory use only). [5:0] manual_adj[5:0] reserved (factory use only). table 40. clock receiver 1 register (address 0x31) bit bit name description [7:4] clkn_cml[3:0] these bits adjust the common-mode level at the clkn pin. the recommended value for these bits and the clkp_cml[3:0] bits is 0xf. for more information, see the optimizing the clock common-mode voltage section. [3:0] reserved reserved. table 41. clock receiver 2 register (address 0x32) bit bit name description 7 clk_dis this bit disables or enables the clock receiver. when the ad9789 powers up, this bit is set to 0 to prevent severe output noise that occurs on power-up with no cl ock. when the dac clock is stable, set this bit to 1. 0 = disabled. 1 = enabled. 6 reserved reserved (factory use only; leave at default value). 5 psign this bit specifies the sign for the clkp_cml bits. 0 = negative (recommended). 1 = positive. [4:1] clkp_cml[3:0] these bits adjust the common-mode level at the clkp pin. the recommended value for these bits and the clkn_cml[3:0] bits is 0xf. fo r more information, see the optimizing the clock common-mode voltage section. 0 nsign this bit specifies the sign for the clkn_cml bits. 0 = negative (recommended). 1 = positive.
ad9789 rev. a | page 37 of 76 table 42. mu delay control 2 register (address 0x33) bit bit name description 7 mu_clkdis this bit disables or enables the clock to the mu delay controller. 0 = enabled. 1 = disabled. 6 slope this bit configures the desired slope for the phase meas urement of the mu delay. when the desired phase is measured, the slope of the phase measurement is calculated and compared to the value of this bit. for optimal ac performance, the best setting for the search is a positive slope and a phase value of 14. 0 = negative. 1 = positive. [5:4] mode[1:0] these bits configure the mode of operation for the mu controller. 00 = search and track (recommended). 01 = track only. 10 = search only. 11 = invalid. 3 musamp transitioning this bit from 0 to 1 enables the user to re ad back the mu delay value that the controller locked to (the mudly bits in register 0x39 and re gister 0x3a), as well as the phase that it locked to (the muphz bits in register 0x39). 0 = no action. 1 = transition from 0 to 1 captures the read back of the mu controller phase and delay. [2:1] gain[1:0] these bits set the tracking rate of the mu controller. 00 = slowest tracking. 01 = nominal tracking (recommended). 10 = fastest tracking. 11 = invalid (do not use). 0 mu_en this bit enables or disables the mu controller. before enabling the mu controller, turn on both the phase comparator boost (register 0x3e[5]) and the mu control duty cycle correction circuitry (register 0x30[7]). both of these functions allow for more robust operation of th e mu controller over the entire operating speed of the part. 0 = mu controller off (manual mode). 1 = mu controller on (auto mode). table 43. dac bias register (address 0x36) bit bit name description 7 pdbias setting this bit to 1 powers down the dac circuitry. [6:2] reserved reserved. [1:0] msel[1:0] these bits set the mirror roll-off frequency control, which can be used to adjust the noise contribution of the internal current mirror to optimize the 1/f noise. 00 = bypass the mirror roll-off frequency control. 01 = narrowest bandwidth. 10 = medium bandwidth. 11 = widest bandwidth. table 44. dac decoder register (address 0x38) bit bit name description [7:2] reserved reserved. [1:0] dac decoder mode these bits set the decoder mode for the dac. it is re commended that normal mode (the default) be used. 00 = normal mode. 01 = return to zero mode. 10 = mix mode. 11 = invalid.
ad9789 rev. a | page 38 of 76 table 45. mu delay control 3 register (address 0x39) bit bit name description 7 mudly[0] this bit is the lsb of the mu delay value. along with bits[7:0] in register 0x3a, this bit configures the programmable mu delay; the search algorithm begins at this specified mu delay value. in manual mode, the mudly bits can be written to. in tracking mode, the sampled mudly value can be read back. even though there are 9 bits of resolution for this delay line va lue, the maximum allowable mu delay is 431 (0x1af). the optimal point to begin the search is in the middl e of the delay line, or approximately 216 (0xd8). [6:5] search_dir[1:0] these bits configure the search direction, starting at the selected mu delay value. 00 = search down. 01 = search up. 10 = search up and down (optimal). 11 = invalid. [4:0] muphz[4:0] these bits specify the phase to be measured with the maximum allowable phase being 16 (10000). if a value larger than 16 is loaded, the controller will not lock. when the desired phase is measured, the slope of the phase measurement is calculated and compared to the configured slope, which is specified by the slope bit in register 0x33[6]. for optimal ac performance, the best setting for the search is for a positive slope and a phase value of 14 (01110). table 46. mu delay control 4 register (address 0x3a) bit bit name description [7:0] mudly[8:1] along with bit 7 in register 0x39, these bits config ure the programmable mu delay; the search algorithm begins at this specified mu delay value. in manual mo de, the mudly bits can be written to. in tracking mode, the sampled mudly value can be read back. even though there are 9 bits of resolution for this delay line value, the maximum allowable mu delay is 431 (0x1af). the optimal point to begin the search is in the middle of the delay line, or approximately 216 (0xd8). table 47. full-scale current 1 register (address 0x3c) bit bit name description [7:0] fsc[7:0] along with bits[1:0] in register 0x3d, this register sets the full-scale current for the dac. for more information, see the voltage reference section. setting (includes register 0x3d[1:0]) full-scale current (ma) 0000000000 8.6 1000000000 20 (default) 1011010000 25 1111111111 32.1 table 48. full-scale current 2 register (address 0x3d) bit bit name description [7:2] reserved reserved. [1:0] fsc[9:8] along with the fsc[7:0] bits in register 0x3c, these bi ts set the full-scale current for the dac. for more infor- mation, see table 47 and the voltage reference section. table 49. phase detector control register (address 0x3e) bit bit name description 7 phz_pd powers down the phase detector. this bit is for factory use only; this bit should be set to 0. 6 reserved reserved. 5 cmp_bst comparator boost. this bit is for factor y use only; this bit should always be set to 1. 4 auto_cal this bit is for factory use only; this bit should always be set to 1. [3:0] phz_det_bias[3:0] these bits display the binary weighted current. do not write to these bits (factory use only).
ad9789 rev. a | page 39 of 76 theory of operation qam encoder the ad9789 is a flexible digital signal processing (dsp) engine combined with a high performance, 2400 msps, 14-bit dac ( figure 70 ). the dsp blocks include a qam encoder, a 2 upsampling square root raised cosine (srrc) filter, selectable interpolation from 16 to 512, a rate converter, and a complex modulator. the digital interface can accept up to four channels of complex data. the qam encoder supports constellation sizes of 16, 32, 64, 128, and 256. the on-chip rate converter allows fine resolution of baud rates with a fixed dac sampling clock. the digital upconverters can place the input signals from dc to 0.5 f dac . an analog mix mode extends the output spectrum into the second and third dac nyquist zones. the qam encoder supports seven different standards-compliant mappings. (for illustrations of the supported mappings, see the qam constellation maps section.) the qam encoder receives input data-words of 8 bits in width and maps them into 16, 32, 64, 128, or 256 point constellations. it outputs 5-bit complex qam modulated samples. the mode in which the qam encoder runs is selected via the qam/srrc configuration register (register 0x07[2:0]). qam encoder 5 8 5 i q from input interface 07852-056 control of the ad9789 functions is via a serial peripheral interface (spi). figure 72. qam encoder i/o table 50 lists the available qam mapper modes along with the corresponding input bits and output range. the operation of the qam encoder when configured in docsis 64-qam mode is described in this section. the operation of the qam encoder in the other modes is conceptually the same; only the input data bit encoding and scale factors are different. data data data retimer data formatter/assembler 150mhz lvds/cmos qam/ filter/ nco qam/ filter/ nco qam/ filter/ nco qam/ filter/ nco 32 input pins and 2 parity pins dco fs 16 interpolator and bpf + scalars 14-bit 2.4gsps dac 07852-099 spi irq rs data cmos 0to 15 lvds rise cmos 16 to 31 lvds fall the docsis 64-qam constellation diagram is shown in figure 73 . the constellation diagram shows how the qam encoder input is mapped into the qam constellation. for example, an input data-word of 111111 maps to the constellation point in the upper right corner of the 64-qam constellation. figure 70. top level functional block diagram datapath signal processing 110,111 111,011 010,111 011,011 100,101 101,111 110,101 111,111 110,100 111,000 010,100 011,000 100,000 101,010 110,000 111,010 100,111 101,011 000,111 001,011 000,101 001,111 010,101 011,111 100,100 101,000 000,100 001,000 000,000 001,010 010,000 011,010 010,011 011,001 000,011 001,001 000,001 001,101 100,001 101,101 010,110 011,100 000,110 001,100 000,010 001,110 100,010 101,110 110,011 111,001 100,011 101,001 010,001 011,101 110,001 111,101 110,110 111,100 100,110 101,100 010,010 011,110 110,010 111,110 i q 07852-057 c5 c4 c3, c2 c1 c0 the dsp blocks included on the ad9789 can be grouped into two sections. the first is the datapath signal processing. four identical datapaths, or channels, can be used. a block diagram of a single channel is shown in figure 71 . enabling and disabling each dsp block within the datapath takes effect on all channels. there is independent control of the scaling and the frequency placement of each channel. 2 n (n = 0 to 5) (p/q = 0.5 to 1) rate converter p/q 24-bit ch gain 0 to 2 24-bit nco 0to f dac /16 srrc 2 bypass srrc bypass qam inscale qam mapper 07852-129 figure 71. datapath block diagram the following sections describe each of the dsp blocks included in the datapath. figure 73. docsis 64-qam constellation
ad9789 rev. a | page 40 of 76 table 50. qam mapper input and output range vs. mode itu-t j.83 annex description spi register 0x07, mapping[2:0] bits bit range at output input bits b7 b6 b5 b4 b3 b2 b1 b0 1 b docsis 64-qam 000 ?14 to +14 x x c5 c4 c3 c2 c1 c0 b docsis 256-qam 001 ?15 to +15 c7 c6 c5 c4 c3 c2 c1 c0 a dvb-c 16-qam 010 ?15 to +15 x x x x c3 c2 c1 c0 a dvb-c 32-qam 011 ?15 to +15 x x x c4 c3 c2 c1 c0 a and c dvb-c 64-qam 100 ?14 to +14 x x c5 c4 c3 c2 c1 c0 a and c dvb-c 128-qam 101 ?11 to +11 x c6 c5 c4 c3 c2 c1 c0 a and c dvb-c 256-qam 110 ?15 to +15 c7 c6 c5 c4 c3 c2 c1 c0 unused 111 1 x = dont care. each constellation point corresponds to an i and q coordinate pair, as shown in figure 74. in the figure, two symbols are high- lighted in a 64-qam constellation: i = 14, q = 14 (pair 1) and i = 6, q = ?10 (pair 2). to represent the i and q coordinate points, 5-bit, twos com- plement numbers are used. for example, an input of 011101 into the qam encoder maps to the i = 6, q = ?10 position of the qam-64 constellation and results in output samples of i = 00110, q = 10110. i q 07852-058 14 10 6 2 ?2 ?6 ?10 ?14 ?14 ?10 ?6 ?2 2 6 10 14 symbol i = 14, q = 14 i = 01110, q = 01110 symbol i = 6, q = ?10 i = 00110, q = 10110 figure 74. i and q symbol mapping srrc 2 bypass srrc bypass qam x inscale qam mapper 07852-059 16 16 5 8 16 5 16 16 figure 75. qam mapper and srrc filter detail (i and q paths are identical so only one is shown) input scalar the input scalar block is active only when the qam mapper is bypassed. the value of inscale[7:0] is programmed in register 0x09[7:0]. the scale factor applied to the input data is calculated as follows: 32 0] inscale[7: rscalefacto ? this factor provides a scaling range of the input data from 0 to 7.96875 in steps of 0.03125. the default value of 0x20 provides a scale factor of 1. as shown in figure 76, the output of the input scalar block is rounded to the nearest 16-bit value. if the output exceeds the maximum or minimum value, it is clipped to either positive or negative full scale (0x7fff or 0x8000). round saturate inscale 07852-100 8 figure 76. input scalar block diagram srrc filter the square root raised cosine (srrc) filter performs a 2 interpolation and filtering operation on the input data. the srrc filter has a pass band, transition band, and stop band requirement as per the docsis, euro-docsis, and dvb-c standards. to cover all the standards, the value of alpha can be set to 0.12, 0.13, 0.15, or 0.18. this value is programmed in register 0x07[5:4]. the frequency, f n , is determined by the input data baud rate. the response of the srrc filter is illustrated in figure 77. the srrc filter accepts only five bits at its input and can be bypassed (register 0x06[6]). if the srrc filter is the first block enabled in the datapath, these five bits are the five msbs of the 8-bit data-word.
ad9789 rev. a | page 41 of 76 ? 3.01db 0db 07852-060 frequency <0.4db <0.4db ad9789 rev. a | page 42 of 76 sample rate converter the purpose of the sample rate converter (src) is to provide increased flexibility in the ratio of the input baud rate to the dac update rate. each of the four channelization datapaths contains a sample rate converter (src) that provides a data rate conversion in the range of 0.5 to 1.0 inclusive. the rate conversion factor is set by the ratio of two 24-bit values, p and q. figure 83 is a conceptual block diagram of the src. it can be thought of as an interpolation block, followed by filtering and decimation blocks. p q pq 07852-106 24 24 figure 83. conceptual block diagram of the sample rate converter the values of p and q are set by programming the p[23:0] and q[23:0] registers at address 0x16 through address 0x1b. table 51. register locations for sample rate converter bits numerator (p) denominator (q) [23:16] (byte 2) register 0x1b register 0x18 [15:8] (byte 1) register 0x1a register 0x17 [7:0] (byte 0) register 0x19 register 0x16 the values of p and q should be selected to satisfy the following equation for the desired baud rate (f baud ) and dac clock fre- quency (f dac ). baud dac f q p if 16 = (1) where i is the total interpolation ratio of the srrc filter and the five half-band interpolation filters. if equation 1 is satisfied, the long-term baud rate, f baud , is exactly maintained. no residual frequency offset errors are introduced by the rate conversion process. the values of p and q must be selected within the following constraints: 0.15.0 ? q p (2) q[23] = 1 (3) equation 3 states that the value of q must be shifted so that the msb of q is set. in most systems, the baud rate is a given, and the dac sample rate is selected so that it is high enough to support the signal bandwidth and output frequency requirements. in many cases, it is desirable to set the dac clock rate to a multiple of a system clock rate. the following example shows how p and q can be selected in such a system. example a docsis application has a master system clock that runs at a frequency of f master . several channel baud rates are supported, all of which are fractions of the master clock and can be represented by the following equation: master baud f n m f = (4) equation 1 must be satisfied for f baud to be exactly maintained. to facilitate this, the dac sampling frequency is selected to be a multiple of f master that satisfies the signal bandwidth and output frequency requirements. for f master = 10.24 mhz, a signal band- width requirement of 32 mhz or greater, and a supported output frequency band of up to 1 ghz, the following dac sampling frequency can be selected: mhz76.2293 224 = = master dac f f (5) inserting equation 4 and equation 5 into equation 1 results in equation 6. master master f n m q p if = 16 224 (6) enabling the srrc filter and four of the half-band interpolation filters would result in the total interpolation factor, i, being equal to 32. substituting 32 for i and simplifying equation 6 results in equation 7. 16 7 = m n q p (7) recall that n and m are given by the required baud rate. for example, assume a baud rate of 5.0569 mhz, which results from m = 401 and n = 812. mhz0569.5mhz24.10 812 401 = = baud f (8) p and q can then be calculated from the numerator and denominator of equation 9. 1910x0 1634x0 6416 5684 16 7 401 812 === q p (9) because the value of q must be msb justified, both numbers can be shifted by 11 bits, resulting in the final p and q values of 0xb1a000 and 0xc80000, respectively. baseband digital upconverter the digital upconverter enables each baseband channel to be placed anywhere from dc to f dac /16. the center frequency for each of the four channels is register programmable through the 24-bit frequency tuning words, ftw 0 through ftw 3. for the desired center frequency of each individual channel, the ftw can be calculated as follows: () 12 16 24 ? ? ? ? ? ? ? = dac center f f ftw
ad9789 rev. a | page 43 of 76 the calculated ftw for each channel should be entered into the register locations listed in table 52 . table 52. register locations of ftws for each channel ftw channel 0 channel 1 channel 2 channel 3 [23:16] reg. 0x0c reg. 0x0f reg. 0x12 reg. 0x15 [15:8] reg. 0x0b reg. 0x0e reg. 0x11 reg. 0x14 [7:0] reg 0x0a reg 0x0d reg 0x10 reg 0x13 the ftw sets the frequency of the sine and cosine signals generated by the numerically controlled oscillator (nco). the complex output from the nco is multiplied by the input datapath signal to modulate the signal to the desired output frequency. a conceptual block diagram of the baseband digital upconverter is shown in figure 84 . sin cos ftw nco frequency tuning word 07852-107 24 figure 84. conceptual block diagram of the baseband digital upconverter individual channel scalar the last block in the datapath is an 8-bit scalar (register 0x25 to register 0x28) intended for compensating out any sampling and hardware roll-offs that may be encountered. the scale factor applied to each channel is calculated as follows: 128 : 0]7 chanxgain[ rscalefacto = the range of the channel gain is 0 to 1.9921875 with a step size of 0.0078125. an individual channel can be easily and quickly muted, if desired, by setting the scale factor to 0. table 53. register locations for channel gain scalar chanxgain channel 0 channel 1 channel 2 channel 3 [7:0] reg. 0x25 reg. 0x26 reg. 0x27 reg. 0x28 the default value of the channel gain provides a scale factor of 1. as shown in figure 85 , the output of the input scalar block is rounded to the nearest 16-bit value. if the output exceeds the maximum or minimum value, it is clipped to either positive or negative full scale (0x7fff or 0x8000). round saturate chanxgain[7:0] 07852-108 8 figure 85. individual channel gain control digital block upconverter the second half of the dsp engine on the ad9789 combines the outputs of the four datapaths into one block, scales the block of channels, interpolates by 16 to the full dac rate, and performs a band-pass filter operation allowing the block of channels to be placed anywhere in the nyquist bandwidth of the dac. data- pat h 0 data- pat h 1 data- pat h 2 data- pat h 3 0 7852-109 sum scale bpf f c digital block upconverter 16 interpolator bpf f c = 0 to f dac/2 figure 86. functional block diagram of the digital block upconverter each block of the digital block upconverter is described in more detail in the following sections. summing junction scalar the summing junction scalar block operates on the sum of the four channels. the value of sumscale[7:0] is programmed in register 0x08. the scale factor applied to the data is calculated as follows: 64 : 0] sumscale[7 rscalefacto = this factor provides a scaling range of the input data from 0 to 3.984375 with a step size of 0.015625. the default value of 0x0d provides a scale factor of 0.203125. note that when the channels are summed, they are clipped at the output of the summing junction scalar block if the value exceeds the maximum or minimum full-scale value (0x7fff or 0x8000). if the full 16-bit range of each individual channel is used, the sum scalar should be set to 0x10 (0.25) to avoid the possibility of clipping. round saturate sumscale register 0x08 to s a tur a tion counte r saterr register 0x03[1] 07852-110 8 figure 87. block diagram of the summing junction scalar in practice, the signal-to-noise ratio (snr) of the channel can be improved by increasing the sum scale factor and permitting a small amount of clipping. the larger signal amplitude can improve the snr if the clipping is brief and infrequent.
ad9789 rev. a | page 44 of 76 table 54 shows recommended sum scale values for each qam mapper mode. the criteria used to determine the recommended sum scale values were mer/evm measurements and spectral purity. because clipping results in impulsive noise, it can be observed in the output spectrum as a transient increase in the output noise floor. these sum scale values were chosen such that the transient increases in the noise floor were minimal. these tests were completed for one, two, three, and four carrier outputs at approximately 850 mhz. because clipping can occur in the rf chain following the dac, further verification of these values should be performed at the system level by adding ber tests to the sum scale selection criteria. table 54. recommended sum scale values for all qam mapper modes and channel count qam mode sum scale value (decimal) 1 channel 2 channels 3 channels 4 channels dvb-c 16-qam 48 28 22 16 dvb-c 32-qam 54 34 26 20 dvb-c 64-qam 54 34 26 20 dvb-c 128-qam 80 50 38 30 dvb-c 256-qam 54 34 26 20 docsis 64-qam 54 34 26 20 docsis 256-qam 54 34 26 20 digital 16 tunable band-pass filter the digital band-pass filter works in conjunction with a fixed 16 interpolator (see figure 88). the 16 interpolation filter creates 16 images of the baseband signal in the nyquist band of the dac. the digital band-pass filter must then be tuned to reject the 15 undesired images. the center frequency of the band-pass filter can be placed anywhere from dc to f dac /2. the tuning word for the band-pass filter center frequency can be calculated as follows: ?? 12 __ 16 ??? dac center f f freq center bpf the resulting tuning word is a 16-bit value where the most significant byte is written to register 0x1d[7:0] and the least significant byte is written to register 0x1c[7:0]. 16 f c 07852-111 16 figure 88. conceptual block diagram of 16 tunable band-pass filter the width of the filters stop band is fixed at approximately f dac /64. the effective flat pass band is f dac /64. as can be inferred from figure 89 to figure 91, mistuning of the bpf center frequency can result in unwanted images appearing. care should be taken to appropriately filter the desired signal with the interpolation filters prior to the input of the bpf.. 0 ?20 ?40 ?60 ?80 ?100 0 0.5 1.0 1.5 2.0 frequency (ghz) magnitude (db) 07852-062 figure 89. band-pass filter response at 200 mhz, f dac = 2.4 ghz 0 ?20 ?40 ?60 ?80 ?100 0 0.5 1.0 1.5 2.0 frequency (ghz) magnitude (db) 07852-063 figure 90. band-pass filter response at 1 ghz, f dac = 2.4 ghz 0 ?2 ?4 ?6 ?8 ?10 020406080 frequency (mhz) magnitude (db) 07852-064 figure 91. band-pass filter pass-band detail, f dac = 2.4 ghz
ad9789 rev. a | page 45 of 76 d igital interface modes t he ad9789 can be configured for two main digital interface modes of operation: ? channelizer mode ? quadrature digital upconverter (qduc ) mode in channelizer mode (register 0x20[3] = 0), the interface can be configured for 4- to 32-bit bus widths and can accept up to four channels of complex data. any of the signal processing blocks in the digital datapath can be used. the maximum baud rate supported in channelizer mode is f dac /48. in qduc mode (register 0x20[3] = 1), the interface is fixed at a 32-bit bus width and one channel of complex data. the available signal processing methods are interpolation (16 to 512), rate conversion (0.5 to 1.0), and complex modulation. the maximum baud rate supported in qduc mode is f dac /16. in both channelizer and qduc modes, the input data bus can be configured to accept lvds or cmos data via the cmos_bus pin (l14). if cmos_bus is pulled to 3.3 v, the data bus is configured to accept cmos inputs (d[31:0], p0, and p1). if cmos_bus is pulled to 0 v, the bus is configured to accept lvds inputs (d[15:0]p, d[15:0]n, parp, and parn). two output signals are used to source data into the ad9789. the first is the data clock output signal (dco), which is provided to clock data from the digital data source. dco is a divided-down version of dacclk. the second is the frame sync signal (fs), which is provided to request a new data-word. the average frequency of the fs signal is equal to the symbol rate or baud rate of the data. as with the input data bus, the dco and fs signals can be configured as lvds or cmos outputs via the cmos_ctrl pin (m14). if cmos_ctrl is pulled to 3.3 v, dco and fs are output as cmos signals on the p14 and n14 pins (cmos_dco and cmos_fs), respectively. if cmos_ctrl is pulled to 0 v, dco and fs are output as lvds signals on the n13, p13, l13, and m13 pins (dcop, dcon, fsp, and fsn), respectively. channelizer mode in channelizer mode, the digital interface has programmable bus width, data width, and data format. the bus width, which is the physical width of the digital data bus at the input of the ad9789, can be set to a 4-, 8-, 16-, or 32-bit wide interface. the data width, which is the internal width of the data at the input to the digital datapath, can be set to an 8-bit or 16-bit word. the data format can be programmed for real or complex data. a list of supported interface modes is shown in table 55 . up to 32 bits cmos 0to 15 lvds rise cmos 16 to 31 lvds fall up to 32 bits up to 32 bits up to 32 bits programmable data formatter/assembler retimer 32 input pins p0 p1 fs dco 07852-112 sum scale bpf f c 4 to 32 bits clk ctl f dac 16 interpolator bpf f c = 0 to f dac/2 f dac 16 f dac 16 to 1024 32 data- pat h 0 data- pat h 1 data- pat h 2 data- pat h 3 lvds/cmos figure 92. channelizer mode table 55. interface configurations supported in channelizer mode first input block enabled bus width reg. 0x21[6:5] data width reg. 0x21[4] data format reg. 0x21[3] qam encoder 32 bits 8 bits real 16 bits 8 bits real 8 bits 8 bits real 4 bits 8 bits real srrc filter 32 bits 8 bits complex 16 bits 8 bits complex 8 bits 8 bits complex 4 bits 8 bits complex interpolation 32 bits 16 bits complex filter 16 bits 16 bits complex 8 bits 16 bits complex if the qam encoder is the first block enabled in the datapath, the data width should be set to an 8-bit word and real data format. if the srrc filter is the first block enabled in the datapath, the data width should be set to an 8-bit word and complex data format. if both the qam encoder and the srrc filters are bypassed, the data width should be set to a 16-bit word and complex data format. pin mapping in channelizer mode in cmos mode (cmos_bus and cmos_ctrl pins = 3.3 v), the various interface width options are mapped to the ad9789 input pins as shown in tabl e 56 . table 56. cmos pin assignments for various interface widths interface width pin assignments buswdth[1:0] 4 bits d[3:0] 00 8 bits d[7:0] 01 16 bits d[15:0] 10 32 bits d[31:0] 11
ad9789 rev. a | page 46 of 76 in lvds mode, the various interface width options are mapped to the ad9789 input pins as shown in table 57 . when the inter- face width is set to 32 bits in lvds mode, the interface becomes double data rate (ddr). in ddr mode, the first 16 bits are sampled on the rising edge of the data sampling clock (dsc, which is synchronous to dco), and the second 16 bits are sampled on the falling edge of dsc. all other interface widths are single data rate (sdr), where the input data is sampled on the falling edge of dsc. table 57. lvds pin assignments for various interface widths interface width pin assignments buswdth[1:0] 4 bits d[3:0]p, d[3:0]n 00 8 bits d[7:0]p, d[7:0]n 01 16 bits d[15:0]p, d[15:0]n 10 32 bits d[15:0]p, d[15:0]n rising edge and falling edge 11 in nibble or byte loading, the most significant nibble or byte should be loaded first. data for channel 0 should be loaded first followed by channel 1, channel 2, and channel 3. in complex data format, the in-phase part should be loaded before the quadrature part of the data-word. the data bus is lsb justified when the data for each channel is assembled internally. a few examples of how the interface maps for different configurations follow. for more information on how a particular configuration is mapped, see the channelizer mode pin mapping for cmos and lvds section. example 1 for a cmos interface with a 32-bit bus width, 8-bit data width, real data format, and four channels enabled, the data in table 5 8 is expected on the input port after data is requested. table 58. cmos pin mapping for bus width = 32 bits, data width = 8 bits, data format = real, four channels 1 dco d[31:24] d[23:16] d[15:8] d[7:0] 1 r3 r2 r1 r0 1 r represents the real data loaded to a given channel; the channel number follows r. example 2 for a cmos interface with a 32-bit bus width, 8-bit data width, complex data format, and four channels enabled, the data in table 59 is expected on the input port after data is requested. table 59. cmos pin mapping for bus width = 32 bits, data width = 8 bits, data format = complex, four channels 1 dco d[31:24] d[23:16] d[15:8] d[7:0] 1 q1 i1 q0 i0 2 q3 i3 q2 i2 1 i represents the in-phase term and q represents the quadrature term of the complex data loaded to a given channel; the channel number follows i or q. example 3 for an lvds interface with a 16-bit bus width, 8-bit data width, complex data format, and four channels enabled, the data in table 60 is expected on the input port after data is requested. table 60. lvds pin mapping for bus width = 16 bits, data width = 8 bits, data format = complex, four channels 1 dco d[15:8]p, d[15:8]n d[7:0]p, d[7:0]n 1 q0 i0 2 q1 i1 3 q2 i2 4 q3 i3 1 i represents the in-phase term and q represents the quadrature term of the complex data loaded to a given channel; the channel number follows i or q. example 4 for an lvds interface with a 32-bit bus width, 8-bit data width, complex data format, and four channels enabled, the data in table 61 is expected on the input port after data is requested. table 61. lvds pin mapping for bus width = 32 bits, data width = 8 bits, data format = complex, four channels 1 dco 2 d[15:8]p, d[15:8]n d[7:0]p, d[7:0]n 1 rise q0 i0 1 fall q1 i1 2 rise q2 i2 2 fall q3 i3 1 i represents the in-phase term and q represents the quadrature term of the complex data loaded to a given channel; the channel number follows i or q. 2 rise means that the data is sourced on the rising edge of dcox; fall means that the data is sourced on the falling edge of dcox. dco and fs rates in channelizer mode the dco signal is a data clock output provided to clock data from the digital data source. the dco is a divided version of the dac clock. the fs signal is an output provided to request a new data-word. the average frequency of the fs signal (f fs ) is exactly equal to the symbol rate or baud rate (f baud ) of the data. fs is intended as a request line; timing should be taken from the dco. the frequencies of the dco signal (f dco ), the baud rate (f baud ), and the dac clock (f dac ) are related as shown by the following two equations: baud dac f q p if = 16 (1) f dco = f dac /(16 n ) (2) where: i is the interpolation factor, which can range from 1 to 64. p/q is the rate conversion factor (0.5 to 1.0, inclusive). n is a programmable dco divide factor set using the dcodiv[2:0] bits in register 0x22[6:4]. set dcodiv[2:0] to 1, 2, or 4. a value of 0 disables the dco. a dcodiv value of 3 is not functional. the frequency of the dsc signal is always equal to dco.
ad9789 rev. a | page 47 of 76 before choosing an interface configuration, divide the frequency of dco by the highest frequency baud rate that will be used in the system and truncate it. the result is the number of available dco cycles (cycles avail ) between fs pulses. ? ? ? ? ? ? ? ? = baud dco avail f f floor cycles max each interface configuration requires a particular number of dco cycles between fs pulses to successfully load data into all channels. this number can be calculated using the following formula: b w dw fn cycles interface = where: n is the number of channels enabled (1 to 4). n is always equal to 4 if channel prioritization is set to 1 (see the channel prioritization section). f represents the data format. if the data format is real, f = 1; if the data format is complex, f = 2. dw is the data width in number of bits (8 or 16). bw is the bus width in number of bits (4, 8, 16, or 32). for a successful interface design, the number of dco cycles between fs pulses must be greater than the number of dco cycles required by the interface. design example in this example, a system has the baud rate f fs = 6.4 mhz. if a 4-bit-wide interface is desired for four channels with real data format and a data width of 8 bits, the selected f dco should be at least 8 f fs . first, using equation 1 and equation 2, evaluate the interface speed with n = 1, p/q = 0.7, and i = 32. f dac = 32 0.7 16 6.4 mhz = 2293.76 mhz f dco = 2293.76 mhz/(16 1) = 143.36 mhz the f dco /f baud ratio = 22.4. if a value of n = 2 is selected, the number of available dco cycles is reduced to 11; this option may not be feasible when the latency values are taken into account. see the latency effects on channelizer mode section for more information about latency. channel prioritization when channels are enabled and disabled, the input interface mapping can be affected. if channel prioritization (register 0x20[2]) is set to 0, the device expects input samples for only the channels that are enabled. in this configuration, the physical channel mapping at the dut input can move around based on the number of channels enabled, where channel 0 has highest priority (it never moves location when enabled). if channel prioritization is set to 1, data is expected for all four channels but the data is ignored internally if the channel is disabled. this method is recommended because enabling and disabling channels does not shift the input data bus. if the number of channels enabled is always less than four and the user does not plan to enable and disable channels dynamically, setting channel prioritization to 0 is the best choice because fewer clocks and/or pins are required to transfer the input data. an example of channel prioritization set to 0 is shown in table 62 . in this example, the data interface is configured for cmos with 32-bit bus width, 8-bit data width, and real data format. table 62. input mapping vs. enabled channels, channel prioritization = 0 cmos bit mapping channels [d31:d24] [d23:d16] [d15:d8] [d7:d0] 4 channels enabled channel 3 channel 2 channel 1 channel 0 channel 0 disabled channel 3 channel 2 channel 1 channel 0, channel 2 disabled channel 3 channel 1 the same example behaves differently when channel prioritization is set to 1, as shown in tabl e 63 . table 63. input mapping vs. enabled channels, channel prioritization = 1 cmos bit mapping channels [d31:d24] [d23:d16] [d15:d8] [d7:d0] 4 channels enabled channel 3 channel 2 channel 1 channel 0 channel 0 disabled channel 3 channel 2 channel 1 channel 0, channel 2 disabled channel 3 channel 1 quadrature digital upco nverter (qduc) mode in qduc mode (register 0x20[3] = 1), the data interface is fixed at a 32-bit bus width, 16-bit data width, and complex data format. in qduc mode, only one channel should be enabled. if more than one channel is enabled, identical i and q data is sent to each enabled channel. within the datapath, the qam mapper and the srrc filter must be bypassed (register 0x06[7:6] = 11). iand q 16 bits iand q 16 bits iand q 16 bits iand q 16 bits lvds/cmos 32 input pins p0 p1 fs dco 07852-069 cmos 0to 15 lvds rise cmos 16 to 31 lvds fall i q on off off off bpf f c = 0to f dac /2 16 interpolator bpf f c figure 93. qduc mode
ad9789 rev. a | page 48 of 76 pin mapping in qduc mode in cmos mode, the ad9789 input pins are mapped as shown in table 64 . table 64. pin mapping in qduc mode for cmos interface data bit description pin no. d31 msb of i data l5 d16 lsb of i data p8 d15 msb of q data l9 d0 lsb of q data p12 p1 parity for d[31:16] l4 p0 parity for d[15:0] m4 a m n p l k b c d cmos_bus cmos_ctrl cmos_fs cmos_dco j e f h g 11 1312 11 10 89 765 24 3 4 07852-113 p1 31 27 23 19 15 11 7 3 p0 30 26 22 18 14 10 6 2 29 25 21 17 13 9 5 1 28 24 20 16 12 8 4 0 bu ct fs dc d[31:0] cmos data inputs parity and control inputs a m parn parp n p l k b c d fsp fsn dcop dcon j e f h g 11 1312 11 10 89 765 24 3 4 07852-114 p+ 15 13 11 9 7 531 p? 15 13 11 9 7 531 14 12 10 8 6 4 2 0 14 12 10 8 6 4 2 0 fs fs dc dc +lvds 14 14 ?lvds figure 95. lvds data input pin mapping dco and fs rates in qduc mode in qduc mode, dcodiv should always be set to 1 (register 0x22[6:4] = 001). the clock period of dco is equal to 16 dac clock periods. when only 16 interpolation is required and the rate converter is not used, the data rate of the interface is equal to f dco . if further interpolation or rate conversion is enabled in the datapath, the data rate of the interface is f baud . the average rate of fs, f fs , is equal to the baud rate, f baud . the baud rate can be specified by the following equation: q p f f n dco baud = 2 figure 94. cmos data input pin mapping where: n is the number of 2 interpolation filters enabled. p/q is the rate converter ratio. in lvds mode, the ad9789 input pins are mapped as shown in table 65 . table 65. pin mapping in qduc mode for lvds interface 1 data bit the fs signal becomes a request for data that effectively gates the dco clock and ensures that data is sent at the correct baud rate. if p/q = 1 and n = 0, dco occurs at the baud rate and fs is not required. in this case, fs is inactive (always high). the dco signal can be used as a constant rate clock to request samples from the data source. description pin no. d15p, d15n rising msb of i data l5, m5 d0p, d0n rising lsb of i data n12, p12 d15p, d15n falling msb of q data l5, m5 d0p, d0n falling lsb of q data n12, p12 parp, parn rising parity for d[15:0]p, d[15:0]n rising l4, m4 parp, parn falling l4, m4 parity for d[15:0]p, d[15:0]n falling 1 rising means that the data is sourced on the rising edge of dcox; falling means that the data is sourced on the falling edge of dcox.
ad9789 rev. a | page 49 of 76 7 dco cycles 6 dco cycles 7 dco cycles t pd t pd t pd sample 0 sample 1 sample 2 dco fs d[31:0] dsc 07852-115 figure 96. qduc mode interface timing diagram for design example when fs is active design example in this example, a system has a dac rate of 1600 mhz and a baud rate of 15 mhz. because f dco = f dac /16 = 100 mhz, the ratio of f dco /f fs = 6.667. to satisfy the requirement that p/q be between 0.5 and 1.0, an additional interpolation factor of 8 must be applied, so n = 3. solving for p/q results in 5/6. therefore, three out of every 20 dco clock edges should result in data samples being loaded into the device (the ratio of f fs /f dco = 3/20). figure 96 shows a timing diagram that illustrates the operation of the interface in this example. in the timing diagram, t pd corresponds to the propagation delay between the rising edge of fs and when the first sample in a given transmission is sampled into the ad9789. note that t pd can vary by more than 1 dco cycle. retimer operation the ad9789 uses a three-register retimer. the first two registers are clocked from any one of 16 phases derived from the dac clock. the clock for the last register is fixed to phase 15. the programmable register clocks are the digital sample clock (dsc) and the synchronizer clock (snc). by choosing different phases, fine adjustment of the sampling time can be made to adjust for delays in the data source. register 0x23[7:4] sets the dsc phase (dscphz) and register 0x23[3:0] sets the snc phase (sncphz) to any one of the 16 phases. the last register in the chain is always clocked from phase 15. the parity counters can aid in identifying the edges of the data valid windows. operation in cmos mode is quite similar to oper- ation in lvds mode, as can be seen in figure 97 and figure 98 . q d clk q d clk q d clk 32 32 32 bits 0to 31 cmos data dsc 0to 15 snc 0to 13 phz 15 07852-070 figure 97. cmos re timing registers q d clk q d clk q d clk 16 q d clk 16 16 bits 16 to 31 dsc 0to 15 snc 0to 13 phz 15 q d clk q d clk q d clk 16 16 16 bits 0to 15 lvds data 07852-071 figure 98. lvds rearranges the dsc register register 0x23 and register 0x21[2:0] can provide timing adjust- ments with very low jitter penalty, but they can also be set to the following recommended safe values: x in lvds mode, dscphz = 0, sncphz = 3, ltncy = 0 (see the latency register section) x in cmos mode, dscphz = 0, sncphz = 7, ltncy = 0 (see the latency register section) timing adjustments can then be made in an fpga or other data source. note that selecting phase 14 or phase 15 for sncphz results in a timing violation. in cmos mode, setting dscphz one step behind or at sncphz also results in a timing violation. latency register a latency register, controlled via register 0x21[2:0], follows the three-register retimer and can delay the data up to seven dco clocks in steps of one dco clock. the critical retiming is already done in the first three registers, so an incorrect latency value does not result in a timing violation. the latency value determines which data sample is the first sample in a trans- mission and routes that sample to the appropriate channel. latency is affected by the round-trip delay from when fs goes high to when the first data sample is output from the retimer. if the latency value programmed into the part is incorrect, the input data samples will not be assembled properly.
ad9789 rev. a | page 50 of 76 sample sample sample sample sample sample sample sample sample sample sample sample sample sample sample sample sample sample sample sample 0 1 2 3 4 5 6 7 8 1 62 4 3 24 0 4 85 6 6 47 2 dco fs lvds ddr lvds sdr cmos 07852-116 figure 99. sampling points at delay = 0 retimer and latency look-up tables in practice, the retimer and latency parameters can be reduced to a single verified and guaranteed table that provides delays at optimum sample points from 0 to over 100 dac clocks. the sampling points for lvds ddr, lvds sdr, and cmos inter- face modes are given in figure 99 for delay = 0. the number scale above the dco signal in figure 99 corresponds to the delay value in dac clock cycles in table 66 and table 67 . the delay of the pins should be taken into account. this delay is 800 ps for the output delay and 800 ps for the input delay, for a total of 1.6 ns. this delay is included in the following formulas. see table 66 for a complete set of recommended retimer settings for all delay values. note: for lvds ddr, zero (0) measured delay results in a retimer setting of 20, while for .lvds sdr or cmos, the zero(0) delay corresponds to a retimer setting of 12 at f dac = 2.4 ghz. to us e table 66 and table 67 , probe the fs, dco, and data input signals at the ad9789. while viewing these signals on an oscilloscope, measure the delay between the rising edge of fs and the start of the first data sample and add 1.6 ns from the delay of the pins to this value. normalize this total delay to one dac clock period. the optimum sampling point in number of dac clock cycles, which corresponds to the delay number in table 66 and table 67 , can be found from this measured value for each interface mode. for lvds ddr, 16 16/ ns6.1 + + = dco measured optimal t dela y delay for lvds sdr, 8 16/ ns6.1 + + = dco measured optimal t delay delay for cmos, 8 16/ ns6.1 + + = dco measured optimal t delay delay for a maximum valid sampling window, the sampling point should be fine-tuned based on the data input setup and hold times. if the setup and hold times are symmetric about the dco edge, choosing a sampling point at the center of the data window results in the maximum valid sampling window. for more information on the input data setup and hold times, refer to the cmos interface timing section or the lvds interface timing section. the lat, snc, and dsc values for the optimal sampling point in table 66 or table 67 should be written to the ltncy[2:0] bits in register 0x21[2:0], the sncphz[3:0] bits in register 0x23[3:0], and the dscphz[3:0] bits in register 0x23[7:4], respectively.
ad9789 rev. a | page 51 of 76 table 66. recommended retimer settings for all delay values, lvds mode delay 0 1 2 3 4 5 6 7 lat 0 0 0 0 0 0 1 1 snc 7 8 9 9 10 10 2 3 dsc 8 9 10 11 12 13 14 15 delay 8 9 10 11 12 13 14 15 lat 1 1 1 1 1 1 1 1 snc 3 4 4 5 5 6 6 7 dsc 0 1 2 3 4 5 6 7 delay 16 17 18 19 20 21 22 23 lat 1 1 1 1 1 1 2 2 snc 7 8 9 9 10 10 2 3 dsc 8 9 10 11 12 13 14 15 delay 24 25 26 27 28 29 30 31 lat 2 2 2 2 2 2 2 2 snc 3 4 4 5 5 6 6 7 dsc 0 1 2 3 4 5 6 7 delay 32 33 34 35 36 37 38 39 lat 2 2 2 2 2 2 3 3 snc 7 8 9 9 10 10 2 3 dsc 8 9 10 11 12 13 14 15 delay 40 41 42 43 44 45 46 47 lat 3 3 3 3 3 3 3 3 snc 3 4 4 5 5 6 6 7 dsc 0 1 2 3 4 5 6 7 delay 48 49 50 51 52 53 54 55 lat 3 3 3 3 3 3 4 4 snc 7 8 9 9 10 10 2 3 dsc 8 9 10 11 12 13 14 15 delay 56 57 58 59 60 61 62 63 lat 4 4 4 4 4 4 4 4 snc 3 4 4 5 5 6 6 7 dsc 0 1 2 3 4 5 6 7 delay 64 65 66 67 68 69 70 71 lat 4 4 4 4 4 4 5 5 snc 7 8 9 9 10 10 2 3 dsc 8 9 10 11 12 13 14 15 delay 72 73 74 75 76 77 78 79 lat 5 5 5 5 5 5 5 5 snc 3 4 4 5 5 6 6 7 dsc 0 1 2 3 4 5 6 7 delay 80 81 82 83 84 85 86 87 lat 5 5 5 5 5 5 6 6 snc 7 8 9 9 10 10 2 3 dsc 8 9 10 11 12 13 14 15 delay 88 89 90 91 92 93 94 95 lat 6 6 6 6 6 6 6 6 snc 3 4 4 5 5 6 6 7 dsc 0 1 2 3 4 5 6 7 delay 96 97 98 99 100 101 102 103 lat 6 6 6 6 6 6 7 7 snc 7 8 9 9 10 10 2 3 dsc 8 9 10 11 12 13 14 15 delay 104 105 106 107 108 109 110 111 lat 7 7 7 7 7 7 7 7 snc 3 4 4 5 5 6 6 7 dsc 0 1 2 3 4 5 6 7 delay 112 113 114 115 116 117 x x lat 7 7 7 7 7 7 x x snc 7 8 9 9 10 10 x x dsc 8 9 10 11 12 13 x x table 67. recommended retimer settings for all delay values, cmos mode delay 0 1 2 3 4 5 6 7 lat 0 0 0 0 0 0 1 1 snc 7 8 8 9 9 2 2 3 dsc 0 1 2 3 4 5 6 7 delay 8 9 10 11 12 13 14 15 lat 1 1 1 1 1 1 1 1 snc 3 4 4 5 5 6 6 7 dsc 8 9 10 11 12 13 14 15 delay 16 17 18 19 20 21 22 23 lat 1 1 1 1 1 1 2 2 snc 7 8 8 9 9 2 2 3 dsc 0 1 2 3 4 5 6 7 delay 24 25 26 27 28 29 30 31 lat 2 2 2 2 2 2 2 2 snc 3 4 4 5 5 6 6 7 dsc 8 9 10 11 12 13 14 15 delay 32 33 34 35 36 37 38 39 lat 2 2 2 2 2 2 3 3 snc 7 8 8 9 9 2 2 3 dsc 0 1 2 3 4 5 6 7 delay 40 41 42 43 44 45 46 47 lat 3 3 3 3 3 3 3 3 snc 3 4 4 5 5 6 6 7 dsc 8 9 10 11 12 13 14 15 delay 48 49 50 51 52 53 54 55 lat 3 3 3 3 3 3 4 4 snc 7 8 8 9 9 2 2 3 dsc 0 1 2 3 4 5 6 7 delay 56 57 58 59 60 61 62 63 lat 4 4 4 4 4 4 4 4 snc 3 4 4 5 5 6 6 7 dsc 8 9 10 11 12 13 14 15 delay 64 65 66 67 68 69 70 71 lat 4 4 4 4 4 4 5 5 snc 7 8 8 9 9 2 2 3 dsc 0 1 2 3 4 5 6 7
ad9789 rev. a | page 52 of 76 delay 72 73 74 75 76 77 78 79 lat 5 5 5 5 5 5 5 5 snc 3 4 4 5 5 6 6 7 dsc 8 9 10 11 12 13 14 15 delay 80 81 82 83 84 85 86 87 lat 5 5 5 5 5 5 6 6 snc 7 8 8 9 9 2 2 3 dsc 0 1 2 3 4 5 6 7 delay 88 89 90 91 92 93 94 95 lat 6 6 6 6 6 6 6 6 snc 3 4 4 5 5 6 6 7 dsc 8 9 10 11 12 13 14 15 delay 96 97 98 99 100 101 102 103 lat 6 6 6 6 6 6 7 7 snc 7 8 8 9 9 2 2 3 dsc 0 1 2 3 4 5 6 7 delay 104 105 106 107 108 109 110 111 lat 7 7 7 7 7 7 7 7 snc 3 4 4 5 5 6 6 7 dsc 8 9 10 11 12 13 14 15 delay 112 113 114 115 116 117 x x lat 7 7 7 7 7 7 x x snc 7 8 8 9 9 2 x x dsc 0 1 2 3 4 5 x x latency effects on channelizer mode when selecting an interface configuration in channelizer mode, the number of dco cycles between fs pulses (cycles ava i l ) must be greater than the number of dco cycles required by the inter- face configuration (cycles interface ). latency consumes some of these available dco cycles between fs. this decrease in available dco cycles is a result of the round-trip propagation delay from the fs output of the ad9789 to the respective data sample at the input of the ad9789 (ltncy[2:0]) in addition to the internal latency of the device. for a successful interface design, the following condition must be met: cycles avail cycles interface + ltncy[2:0] + 2 cmos interface timing when the ad9789 is configured with a cmos interface (cmos_ctrl = cmos_bus = 3.3 v), a cmos data clock output signal, dco, is provided to drive data from the data source. the output signal operates at the input data rate, which is equal to f dac /16 when dcodiv = 1. cmos data on the bus is sampled on the rising edge of an internal sampling clock (dsc). note that the frequency of dco is equal to the frequency of dsc and the phase relationship between dco and dsc is determined by dscphz (register 0x23[7:4]). the timing of the input data is referenced to dco for a given phase of dsc. the cmos data input timing over temperature is shown in table 68 for dco_inv = 0 (register 0x20[4]), dscphz = 0 (register 0x23[7:4]), and dcodiv = 1 (register 0x22[6:4]). table 68 also shows the data valid window (dvw). the data valid window is the sum of the setup and hold times of the interface. dvw is the minimum amount of time that valid data must be presented to the device to ensure proper sampling. table 68. cmos data input timing with respect to dco temperature min t s (ns) min t h (ns) min dvw (ns) ?40c 4.9 ?1.4 3.5 +25c 5.1 ?1.6 3.5 +85c 5.3 ?1.7 3.6 ?40c to +85c 5.3 ?1.4 3.9 for any value of dscphz greater than 0, the setup and hold times shift by increments of t dco /16, where t dco is the period of the data clock. t s = 5.3 ns ? (( t dco /16) dscphz ) t h = 0.24 ns + (( t dco /16) dscphz ) dco input data dsc 07852-117 t s t h figure 100. cmos input timing in some interface modes, the delay from the rising edge of dco to the rising edge of fs needs to be known. this delay is summa- rized over temperature in table 69 . dco fs dsc 07852-118 t d figure 101. cmos_dco to cmos_fs delay table 69. timing delay betw een cmos_dco and cmos_fs temperature t d, max dco to fs (ns) t d, min dco to fs (ns) ?40c 0.64 0.28 +25c 0.71 0.4 +85c 0.85 0.49 ?40c to +85c 0.85 0.28
ad9789 rev. a | page 53 of 76 lvds interface timing when the ad9789 is configured with an lvds interface (cmos_ctrl = cmos_bus = 0 v), an lvds data clock out- put signal, dco, is provided to drive data from the data source. the lvds interface may be single data rate (sdr) or double data rate (ddr) depending on the bus width configuration. in sdr, data is sampled into the part only on the falling edge of the internal sampling clock (dsc). note that the frequency of dco is equal to the frequency of dsc, so the effective data rate is equal to the dco frequency. the phase relationship between dco and dsc is determined by dscphz (register 0x23[7:4]). in ddr, data is sampled into the part on both the rising and falling edges of dsc, so the effective data rate is equal to twice the dco frequency. the interface is ddr only when the bus width is equal to 32 bits. the dco frequency is equal to f dac /16 when dcodiv = 1. the timing of the input data is referenced to dco for a given phase of dsc. the lvds input data timing over temperature is shown in table 70 for dco_inv = 0 (register 0x20[4]), dscphz = 0 (register 0x23[7:4]), and dcodiv = 1 (register 0x22[6:4]). table 70. lvds data input timing with respect to dco temperature min t s (ns) min t h (ns) min dvw (ns) ?40c 1.04 0.24 1.28 +25c 1.23 0.16 1.39 +85c 1.41 0.03 1.44 ?40c to +85c 1.41 0.24 1.65 in ddr mode, these setup and hold times must be applied to both edges of dco. in sdr mode, these setup and hold times must be applied to the falling edge of dco. for any value of dscphz greater than 0, the setup and hold times shift by increments of t dco /16, where t dco is the period of the data clock. t s = 1.41 ns ? (( t dco /16) dscphz ) t h = 0.24 ns + (( t dco /16) dscphz ) dco input data dsc t s t h t s t h t s t h t s t h single dat a r a te (sdr) dco input data dsc double data rate (ddr) 0 7852-119 figure 102. lvds input timing, sdr vs. ddr in some interface modes, the delay from the rising edge of dco to the rising edge of fs needs to be known. this delay is summarized over temperature in table 71 . dco fs dsc 07852-120 t d figure 103. lvds dco to fs delay table 71. timing delay between lvds dco and fs temperature t d, max dco to fs (ns) t d, min dco to fs (ns) ?40c 0.37 0.21 +25c 0.35 0.16 +85c 0.32 0.12 ?40c to +85c 0.37 0.12 parity the ad9789 supports parity checking on the input data bus. there are three parity checking modes: even parity, odd parity, and iq parity. in iq parity mode, a value of 0 is always expected on the i channel and a value of 1 is always expected on the q channel. note that iq parity mode is generally useful only when the lvds interface is used. these modes are controlled via register 0x20[1:0]. table 72. parity mode spi settings parity mode register 0x20[1:0] deactivates parity checking 00 iq parity 01 even parity 10 odd parity 11 if parity checking is used, each data-word that is transferred into the ad9789 should have a parity bit accompanying it, regardless of fs. in other words, parity must be valid for every dco edge. the parity bits are located at pin l4 and pin m4. when operating the interface in cmos mode, the input parity bits are referred to as p1 and p0, respectively. when operating the interface in lvds mode, the input parity bits are referred to as parp and parn, respectively. recall that the lvds interface can be single data rate (sdr) or double data rate (ddr), depending on the bus width configu- ration. the interface is ddr only when the bus width is equal to 32 bits.
ad9789 rev. a | page 54 of 76 in qduc mode, where the interface is fixed at a 32-bit bus width, the parity behavior is straightforward (see table 73 ). table 73. parity behavior in qduc mode inter- face bus width even/odd parity iq parity p1 checks d[31:16] p1 = 0 cmos 32 bits p0 checks d[15:0] p0 = 1 [parp, parn] rising checks d[15:0]p, d[15:0]n rising parp rising = 0 parn rising = 1 lvds 1 (ddr) 32 bits [parp, parn] falling checks d[15:0]p, d[15:0]n falling parp falling = 1 parn falling = 0 1 rising corresponds to the data sampled on the rising edge of dsc; falling corresponds to the data sampled on the falling edge of dsc. in channelizer mode, where the interface is configurable for different bus widths, data widths, and data formats, the parity bits check the data-word on the bus. for example, consider a configuration in channelizer mode where the bus width is 4, the data width is 8, and the data format is real. in this case, eight clock cycles are required to transfer all of the baud rate data to represent four channels. in the even parity or odd parity mode, one parity bit and four data bits are sent on each clock; the parity bit checks the four data bits to verify that all of the data was sent over the interface. table 74 summarizes the behavior of the two parity pins and how they interact with the data in all interface modes. table 74. parity behavior in channelizer mode 1 rising corresponds to the data sampled on the rising edge of dsc; falling corresponds to the data sampled on the falling edge of dsc. if a parity error occurs, the parity counter (register 0x02[7:0]) is incremented. the parity counter continues to accumulate until it is cleared or until it reaches a maximum value of 255. the count can be cleared by writing a 1 to register 0x04[7]. an irq can be enabled to trigger when a parity error occurs by writing a 1 to register 0x03[7]. the status of irq can be meas- ured via register 0x04[7] or by using the irq pin (pin p2). if using the irq pin and more than one irq is enabled, the user must check register 0x04 when an irq event occurs to determine whether the irq was caused by a parity error. the irq can also be cleared by writing a 1 to register 0x04[7]. analog modes of operation the ad9789 uses a quad-switch architecture that can be config- ured to operate in one of three modes via the serial peripheral interface: normal mode, rz mode, and mix mode. the quad-switch architecture masks the code-dependent glitches that occur in a conventional two-switch dac. figure 104 shows the waveforms for a conventional dac and the quad-switch dac. in the two-switch architecture with d1 and d2 in different states, a switch transition results in a glitch. however, if d1 and d2 are at the same state, the switch does not create a glitch. this code- dependent glitching causes an increased amount of distortion in the dac. in the quad-switch architecture, two switches are always transitioning at each half clock cycle, regardless of the code; therefore, code-dependent glitches are eliminated, but a constant glitch at 2 f dac is created. input data dacclk 2-switch dac output 4-switch dac output (normal mode) d1 d2 d3 d4 d5 d1 d2 d3 d4 d5 d1 d2 d3 d4 d5 d6 d7 d8 d9 d10 d6 d7 d8 d9 d10 d6 d7 d8 d9 d10 t t 07852-072 figure 104. two-switch and quad-switch dac waveforms the quad-switch architecture can also be easily configured to perform an analog mix or return-to-zero (rz) function. in mix mode, the output is effectively chopped at the dac sample rate. the rz mode is similar to mix mode, except that the inter- mediate data samples are replaced with midscale values instead of inverting values. figure 105 shows the dac waveforms for both mix mode and rz mode. inter- face bus width even/odd parity iq parity p1 ignored p1 = 0 cmos 4 bits p0 checks d[3:0] p0 = 1 p1 ignored p1 = 0 cmos 8 bits p0 checks d[7:0] p0 = 1 p1 ignored p1 = 0 cmos 16 bits p0 checks d[15:0] p0 = 1 p1 checks d[31:16] p1 = 0 cmos 32 bits p0 checks d[15:0] p0 = 1 lvds (sdr) 1 4 bits [parp, parn] falling checks d[3:0]p, d[3:0]n falling not supported lvds (sdr) 1 8 bits [parp, parn] falling checks d[7:0]p, d[7:0]n falling not supported lvds (sdr) 1 16 bits [parp, parn] falling checks d[15:0]p, d[15:0]n falling not supported [parp, parn] rising checks d[15:0]p, d[15:0]n rising parp rising = 0 parn rising = 1 lvds (ddr) 1 32 bits [parp, parn] falling checks d[15:0]p, d[15:0]n falling parp falling = 1 parn falling = 0
ad9789 rev. a | page 55 of 76 input data dacclk 4-switch dac output ( f s mix mode) 4-switch dac output (return-to- zero mode) d1 d2 d3 d4 d5 d1 d2 d3 d4 d5 d1 d2 d3 d4 d5 d6 d7 d8 d9 d10 d6 d6 d7 d7 d8 d8 d9 d9 d10 d10 t 07852-073 ?d10 ?d1 ?d6 ?d5 ?d2 ?d3 ?d4 ?d9 ?d8 ?d7 t figure 105. mix mode and rz mode dac waveforms switching between analog modes reshapes the sinc roll-off inherent at the dac output. the performance and maximum amplitude in all three nyquist zones is affected by this sinc roll-off depending on where the carrier is placed, as shown in figure 106 . 0 ?10 ?5 ?15 ?20 ?25 ?30 ?35 0 0.5 1.0 1.5 2.0 2.5 3.0 frequency (hz) amplitude (dbm) 07852-074 first nyquist zone second nyquist zone third nyquist zone normal mode rz mode mix mode figure 106. sinc roll-off for each analog operating mode (f s = 2 dacclk) the rz mode, with its lower but flat response, can be quite useful for quick checks of system frequency response. analog control registers the ad9789 includes registers for optimizing its analog performance. these registers include noise reduction in the output current mirror and output current mirror headroom adjustments. mirror roll-off frequency control using the msel[1:0] bits (register 0x36[1:0]), the user can adjust the noise contribution of the internal current mirror to optimize the 1/f noise. figure 107 shows msel vs. the 1/f noise with 20 ma full-scale current into a 50 resistor. frequency (khz) noise (dbm/hz) ? 110 ?115 ?120 ?125 ?130 ?140 ?135 11 0 07852-083 1 0 0 msel = 11 msel = 10 msel = 01 msel = 00 figure 107. 1/f noise with respect to msel bits
ad9789 rev. a | page 56 of 76 voltage reference the ad9789 output current is set by a combination of digital control bits and the i120 reference current, as shown in figure 108 . current scaling fsc[9:0] ad9789 dac full-scale current 10k? 1nf vref i120 avss i120 v bg 1.2v + ? 07852-084 figure 108. voltage reference circuit the reference current is obtained by forcing the band gap voltage across an external 10 k resistor from i120 (pin b14) to ground. the 1.2 v nominal band gap voltage, vref (pin c14), generates a 120 a reference current in the 10 k resistor. this current is adjusted digitally by fsc[7:0] (register 0x3c[7:0]) and fsc[9:8] (register 0x3d[1:0]) to set the output full-scale current, i fs , in milliamperes. i fs = 0.023 fsc[9:0] + 8.58 the full-scale output current range is approximately 8.6 ma to 32.1 ma for register values from 0x000 to 0x3ff. the default value of 0x200 generates 20 ma full scale. the typical range is shown in figure 109 . 35 30 25 20 i fs (ma) 15 10 5 0 0 200 400 600 800 dac gain code 1000 07852-085 figure 109. full-scale current vs. dac gain code always connect a 10 k resistor from the i120 pin to ground and use the digital controls to adjust the full-scale current. the ad9789 is not a multiplying dac. applying an analog signal to i120 is not supported. vref (pin c14) must be bypassed to ground with a 1 nf capacitor. the band gap voltage is present on this pin and can be buffered for use in external circuitry. the typical output impedance is near 5 k. if desired, an external reference can be used to overdrive the internal reference by connecting it to the vref pin. iptat (pin d14) is used for factory testing and can simply be left floating. iptat is an output current that is proportional to absolute temperature. at 25c, the output current is approx- imately 10 a and follows a slope of approximately 20 na/c. for optimal docsis 3.0 aclr performance, the full-scale output current settings provided in table 75 are recommended. table 75. recommended full-scale current settings vs. number of qam channels number of qam channels recommended i fs (ma) fsc[9:0] 1 20 512 2 25 720 3 25 720 4 25 720 dac output stages to properly evaluate the ad9789 in the lab, three distinct output coupling circuits were used. figure 110 shows the optimal output network when measuring traditional dac performance specifications such as sfdr and imd performance with sine waves. 90? 70? ioutp ioutn 90? jtx-2-10t 07852-121 figure 110. recommended transformer output stage for single-tone/multitone measurements figure 111 shows the optimal output network when measuring signals in mix mode (second or third nyquist zone). the band- width of the center tap transformer is not sufficient to support mix mode outputs, so the best solution is to use a wideband balun by itself. 90 ? 70? ioutp ioutn mabact0039 90? 07852-122 figure 111. recommended transformer output stage for mix mode
ad9789 rev. a | page 57 of 76 finally, when measuring performance for cmts and other digital tv applications, it is advantageous to insert a 1 db, 1.2 ghz chebyshev low-pass filter between the dac and the transformer to better control the impedance seen at the dac core. this helps to decrease the folded back harmonics for higher frequency outputs. the optimal transformer for cmts measurements is the jtx-2-10t, which consists of a balun and center-tapped transformer in a single package. this output stage is shown in figure 112 . the buffer, in turn, can be easily driven from lower level signals such as cml or attenuated pecl that might be encountered on a pcb. this buffer also provides very low, 100 fs added random jitter, which is important to obtain the optimal ac performance from the ad9789. a functional block diagram of the adclk914 is shown in figure 113 . figure 114 shows the recommended schematic for the adclk914 /ad9789 interface. refer to the adclk914 data sheet for more information. any time that the noise floor from the dac cannot meet the specifications in this data sheet, the clock should be examined. 90 ? 90? 70? ioutp 5.6nh 5.6nh ioutn 4.7pf 2.2pf 4.7pf jtx-2-10t 07852-123 50 ? v ref v cc v ee v t d d 50? 50? 50 ? q q adclk914 07852-124 figure 112. recommended transformer output stage for cmts measurements traces from the dac to the transformer should be 50 imped- ance to ground each in figure 110 and figure 112 and 25 to ground each in figure 111 to avoid unnecessary parasitics. figure 113. adclk914 fu nctional block diagram the internal 50 resistors shown at the adclk914 inputs are rated to carry currents from pecl or cml drivers. the v t pin can be connected to v cc , a pecl current sink, or the internal v ref , or it can be left floating depending on the source. the common-mode input range of the adclk914 does not include lvds voltage levels, so ac coupling is required in that case. clocking the ad9789 to provide the required signal swing for the internal clock receiver of the ad9789, it is necessary to use an external clock buffer chip to drive the clkp and clkn inputs. these high level, high slew rate signals should not be routed any distance on a pcb. the recommended clock buffer for this application is the adclk914 . this ultrafast clock buffer is capable of providing 1.9 v out of each side into a 50 load terminated to v cc (3.3 v) for a total differential swing of 3.8 v. 5432 1 gnd nc nc v ee v cc v t v ref v ee v cc q nc nc q d nc nc d 1 16 15 14 13 2 3 4 5678 12 11 10 9 c31 0.1f c0402 gnd c32 0.01f c0402 gnd c33 0.1f c0402 gnd c34 0.01f c0402 gnd adclk914 supp ly decoupling vcc33 vcc33 vcc33 vcc33 clkp clkn r13 49.9 ? r15 49.9 ? r14 49.9 ? r17 100 ? r0402 c99 2400pf c0803h50 c102 2400pf c0803h50 u3 adclk914 gnd gnd gnd gnd c83 0.01f c81 0.01f c82 0.01f j3 pstrnkpe4117 07852-125 figure 114. adclk914/ad9789 interface circuit for use with a lab generator
ad9789 rev. a | page 58 of 76 optimizing the clock common-mode voltage in addition to the system that optimizes the handoff timing, an additional system sets the common-mode voltage of the clock. this system can be used to prop erly align the crossing point of the clkp and clkn signals to ensure that the duty cycle of the clock is set properly. figure 115 shows how the common-mode voltage of clkp and clkn is set. there are eight switches controlled by the clkp_cml bits (register 0x32[4:1]) and the clkn_cml bits (register 0x31[7:4]) for both the clkp and clkn signals. the direction of the adjustment is determined by the psign and nsign bits (register 0x32, bit 5 and bit 0). if psign and nsign are low, the common-mode voltage decreases with clkp_cml/clkn_cml values. if psign and nsign are high, the common-mode voltage increases with clkp_cml/ clkn_cml values, as shown in figure 116 . with both clkp_cml and clkn_cml set to 0, the feedback path forces the common-mode voltage to be set to approximately 0.9 v. the optimal ac performance occurs at a setting of ?15 on both the clkp and clkn offset bits. clkp/clkn cvdd18 clkx_cml sign = 0 clkx_cml sign = 1 07852-081 figure 115. clock common-mode control 1.10 1.05 1.00 0.95 0.90 0.85 0.80 0.75 0.70 ?15 ?13 ?11 ?9 ?7 ?5 ?3 ?1 1 3 5 7 9 11 13 15 offset code common-mode voltage (v) 07852-082 clkp clkn figure 116. common-mode voltage with respect to clkp_cml/clkn_cml and psign/nsign clock phase noise effect s on ac performance the quality of the clock source driving the adclk914 deter- mines the achievable aclr performance of the ad9789. table 76 summarizes the close-in aclr for a four-carrier docsis signal at 900 mhz with respect to various phase noise profiles. (all aclr values are specified in dbc.) table 76. four-carrier docsis close-in aclr performance at 900 mhz for various phase noise profiles phase noise (dbc) band profile 1 profile 2 profile 3 profile 4 spec 750 khz to 6 mhz ?71 ?67.2 ?62.4 ?59.1 ?60 6 mhz to 12 mhz ?70.9 ?70.3 ?67 ?63.8 ?63 12 mhz to 18 mhz ?71 ?70.8 ?70.8 ?70.8 ?65 table 77 shows the phase noise at various offsets for each profile. (all phase noise numbers are specified in dbc/hz.) table 77. phase noise summary for each profile phase noise (dbc/hz) offset 1 profile 1 profile 2 profile 3 profile 4 2 khz ?114.8 ?112.8 ?111.7 ?111.2 20 khz ?117.8 ?115.5 ?114.6 ?113.8 200 khz ?128.3 ?118.9 ?118.3 ?116.8 2 mhz ?148.5 ?127.9 ?122.2 ?117.9 20 mhz ?152.5 ?149.9 ?148 ?145.7 1 at offsets less than 500 khz, the measurement instrume nt dominates the phase noise performance. to meet the close-in aclr requirements for four-carrier docsis, the phase noise found in profile 3 is the minimum requirement necessary. mu delay controller the mu delay adjusts timing between the digital and analog blocks. the mu delay controller receives phase relational information between the digital and analog clock domains. the control system continuously adjusts the mu delay to maintain the desired phase relationship between the digital and analog sections. a top level diagram of the mu delay within the dac is shown in figure 117 . 14-bit 2.4gsps dac digital circuitry 14 16 mu control mu det mu delay 16-bit data dac clock 07852-077 figure 117. mu delay controller block diagram the mu controller has two modes of operation: initial phase search and phase tracking. in the phase search mode, the con- troller looks for the initial mu delay value to use before going into tracking mode. in tracking mode, the controller makes adjustments to the initial mu delay value to keep the phase at the desired value. the initial phase search is required because multiple mu delay settings may result in the desired phase, but the device may not operate correctly at all of those mu delay values.
ad9789 rev. a | page 59 of 76 o perating the mu controller in auto mode t he mu controller is enabled via register 0x33[0]. enabling the controller sets in motion the phase search mode. before enabling the controller, it is important to turn on both the phase comparator boost (register 0x3e[5]) and the mu control duty cycle correction circuitry (register 0x30[7]). both of these functions allow for more robust operation of the mu controller over the entire operating speed of the part. the three modes of operation for the mu con- troller are specified by the mode[1:0] bits in register 0x33[5:4] as follows: ? search and track (00) (optimal setting) ? track only (01) ? search only (10) t he search algorithm begins at a specified mu delay value set using the mudly[8:0] bits, where the lsb is located in register 0x39[7] and the msbs are located in register 0x3a[7:0]. even though there are nine bits of resolution for this delay line value, the maximum allowable mu delay is 431 (decimal). the optimal point to begin the search is in the middle of the delay line, or approximately 216. the initial search algorithm works by sweep- ing through different mu delay values until the desired phase is measured; this phase is specified using the muphz[4:0] bits in register 0x39[4:0], with the maximum allowable phase being 16. if values larger than 16 are loaded, the controller will not lock. when the desired phase is measured, the slope of the phase measurement is calculated and compared to the desired slope, which is specified by the slope bit in register 0x33[6]. for optimal ac performance, the best setting for the search is a positive slope and a phase value of 14. if the phase and slope match the configured values, the search algorithm is finished. the search_tol bit (register 0x2f[7]) can be used to specify the accuracy of the search as follows: ? not exact (0): can find a phase within two values of the desired phase ? exact (1): finds the exact phase specified (optimal setting) figure 118 shows a typical plot of mu phase vs. mu delay line value at 2.4 gsps. starting at the selected mu delay value, the search direction can be specified via the search_dir[1:0] bits in register 0x39[6:5]. the three possible choices for the search are as follows: ? down only (00) ? up only (01) ? alternating up and down (10) (optimal setting) i f the search direction is alternating, the search proceeds in both directions until a programmable guard band is reached in one of the directions, specified by the guardband[4:0] bits in register 0x2f[4:0]. when the guard band is reached, the search continues only in the opposite direction. if the desired phase is not found before the guard band is reached in the second direc- tion, the search reverts to the alternating mode and continues looking within the guard band. t he search fails if the mu delay reaches the endpoints. if the controller does not find the desired phase during the search, the track_err bit (register 0x2f[5]) determines the corrective action as follows: ? continue (0): continues to search (optimal setting) ? reset (1) 18 16 14 12 10 8 6 4 2 0 0 40 80 120 160 200 240 280 320 360 400 440 mu delay mu phase 07852-078 guard band guard band search starting location desired phase and slope f igure 118. typical mu phase characteristics @ 2.4 gsps t o determine whether the search is on the correct slope, the controller measures the slope by first incrementing and then decrementing the mu delay value until any of the following events happens: ? the phase changes by 2. ? the phase is equal to 16 (the maximum value). ? the phase is equal to 0 (the minimum value). ? the mu delay is 431 (the maximum value). ? the mu delay is 0 (the minimum value). after incrementing and then decrementing the mu delay value, the values of the measured phases are compared to determine whether the slope matches the desired slope. to consider the slope valid, the positive direction phase and the negative direction phase must be on opposite sides of the desired phase. examples of valid and invalid phase choices are shown in figure 119 and figure 120 . desired desired positive slope negative slope 10 11 12 13 14 15 9 8 7 6 5 4 07852-079 figure 119. valid positive and negative slope phase examples
ad9789 rev. a | page 60 of 76 desired desired 12 13 13 14 14 15 15 4 3 3 2 2 11 07852-080 figure 120. invalid slope phase examples when the initial mu delay value has been found by the search algorithm, the tracking mode is enabled. in tracking mode, a simple control loop is used to increment by 1, decrement by 1, or not change the mu delay value depending on the measured phase. the control loop uses the desired slope to determine whether the mu delay should be incremented or decremented. no attempt is made to determine whether the actual slope has changed or is still valid. two status bits, lockacq (register 0x04[3]) and locklost (register 0x04[2]) are available to signal proper operation of the control loop. if the current phase is more than five steps away from the desired phase and the lockacq bit was previously set, the lockacq bit is cleared and the locklost interrupt bit is set. furthermore, if lock is lost, the controller can remain in the tracking loop, or it can be reset to start the search again. by setting the musamp bit high (register 0x33[3]) from a low state, the user can read back the mu delay value that the controller locked to by reading the mudly bits (register 0x39[7] and register 0x3a[7:0]), as well as the phase it locked to by reading back the muphz[4:0] bits (register 0x39[4:0]). these bits will no longer read back the value that the search started at or the desired phase, but instead will read back the mu delay line value and phase that the controller is locked to. table 78 lists register writes and reads to lock to the controller. the program assumes that the clock receiver is already enabled and that a clean lock is provided. the typical locking time for the mu controller is approximately 180,000 dac cycles (at 2 gsps, ~75 s). table 78. ad9789 mu delay controller routine address data r/w description 0x30 0x80 write enable duty cycle correction. 0x31 0xf0 write set common-mode level of clkn: clkn_cml = 0xf. 0x32 0x9e write set common-mode level of clkp: clkp_cml = 0xf. set direction of clkp_cml and clkn_cml: psign = 0; nsign = 0. enable clock receiver: clk_dis = 1. 0x3e 0x38 write set phase comparator boost (auto_cal must be set to its default value, 1). 0x24 0x00 write 0x24 0x80 write enable digital clocks. 0x2f 0xce write search for exact phase with a guard band of 98 codes from endpoints. 0x33 0x42 write set search slope to positive. 0x39 0x4e write set search phase to 14, search up and down. 0x3a 0x6c write set start point of search to mid- point of mu delay line (code 216). 0x03 0x00 write disable lock and lock lost indicators. 0x04 0xfe write clear lock and lock lost indicators. 0x03 0x0c write enable lock and lock lost indicators. 0x33 0x43 write enable mu delay controller and start search/track routine. 0x33 0x4b write set mu phase read bit high. 0x33 0x43 write set mu phase read bit low. 0x04 read check lock and lock lost bits: lockacq should be on. locklost should be off. 0x39 read check phase readback (should be equal to 14).
ad9789 rev. a | page 61 of 76 interrupt requests operating the mu controller in manual mode t he following interrupt (irq) requests can be used for additional information and verification of the status of various functional blocks: in manual mode, the user must sweep through all the mu delay values and record the phase value at each value of mudly as shown in figure 118 . every time that the mudly value is stepped, the musamp bit must be toggled from low to high to read the corresponding phase for the specified mu delay line value. it is not possible to keep read high and continuously read back the phase value. as with auto mode, the optimal ac perfor- mance occurs at a positive slope and a phase of 14; therefore, when the curve is complete, choose the mudly value that corresponds to this condition and write that value to the mudly[8:0] bits (register 0x39[7] and register 0x3a). ? parerrtriggered when one or more parity errors occurs on the data bus ? parmsettriggered when parmnew is set and internally registered ? parmclrtriggered when parmnew is cleared and internally registered ? lockacqtriggered when the mu controller is locked to the user-defined phase calculating mu delay line step size ? locklosttriggered when the mu controller loses lock (if the lockacq bit was previously set) stepping through all of the mu delay line values and plotting mu phase vs. mu delay not only allows the user to find the optimal mu delay value, but can also allow the user to determine the mu delay line step size. to calculate the step size, take one full cycle of the mu phase curve and divide the period of the dac clock by this delta. from figure 118 , the two transition points are approximately 56 and 270, providing a delta of approx- imately 214 steps. therefore, the mu delay line step size would be approximately 2 ps/step, as shown in the following equation: ? saterrtriggered when one or more saturation errors occurs each irq is enabled using the enable bits in the interrupt enable register, register 0x03. the status of the irq can be measured in one of the following ways: via the spi bits found in the interrupt status/clear register (register 0x04) or using the irq pin (pin p2). if the pin is used to determine that an interrupt has occurred, it is necessary to check register 0x04 to determine which bit caused the interrupt because the pin indicates only that an interrupt has occurred. to clear an irq, it is necessary to write a 1 to the bit in register 0x04 that corresponds to the interrupt. ps95.1 214 ghz4.2 1 = ? ? ? ? ? ? ? ? if the mu controller is enabled, this value allows the user to calculate (in picoseconds) how much drift is in their system with respect to the dac clock period over temperature.
ad9789 rev. a | page 62 of 76 recommended start-up sequence the steps necessary to optimize the performance of the part and generate an output waveform are listed in table 79 . table 79. recommended system start-up sequence step description register data 0 power up the ad9789. 0 apply the clock. 1 enable the clock receiver an d set the clock cml. 0x32 0x9e 1 enable duty cycle correction. 0x30 0x80 2 0x24 0x00 3 enable digital clocks. 0x24 0x80 4 0x2f 0xce 4 0x33 0x42 4 0x39 0x4e 4 set up mu controller. 0x3a 0x6c 5 disable all interrupts. 0x03 0x00 6 clear all interrupts. 0x04 0xfe 7 enable mu control interrupts. 0x03 0x0c 8 enable mu delay controller. 0x33 0x43 9 set up digital datapath. 0x06 to 0x15 9 set up rate converter. 0x16 to 0x1b 9 set up bpf center frequency. 0x1c to 0x1d 9 set up interface. 0x20 to 0x23 9 set up channel gains. 0x25 to 0x28 9 set up spectral invert. 0x29 9 set up full-scale current. 0x3c to 0x3d 10 wait until mu delay controller is locked (spi read) 1 . 0x04 0x08 11 update rate converter and bpf. 0x1e 0x80 12 0x24 0x00 13 update interface clocks. 0x24 0x80 14 enable channels. 0x05 15 enable other interrupts if desired. 0x03 1 typical lock time of the mu controller is approximately 180,000 dac cycles (at 2 gsps, ~75 s).
ad9789 rev. a | page 63 of 76 customer bist modes using the internal prn generator to test qam output ac performance the ad9789 can be configured to enable an on-chip pseudo- random number (prn) generator. the prn output is connected to the front end of the datapath and disconnects the datapath from the input pins. in this way, the prn generator can be used in conjunction with the on-chip qam encoder to generate a qam output. the prn generator allows the user to measure the ac performance of a qam signal at the dac output without an external data source. to enable the internal prn generator via the serial port, follow these steps. 1. ensure that the clock is enabled and that the clock common-mode level is set to its optimal value by setting the registers in table 80 to the values shown in the table. table 80. register settings to configure the clock register data description 0x30 0x80 enable duty cycle correction. 0x31 0xf0 set the common-mode level of clkn: clkn_cml = 0xf. 0x32 0x9e set the common-mode level of clkp: clkp_cml = 0xf. set psign = 0, nsign = 0. enable clock receiver (clk_dis = 1). 2. configure bist mode for prn generation and disconnect the inputs by setting the registers in tabl e 81 to the values shown in the table. table 81. register settings to configure prn generation register setting 0x42 0x10 0x43 0x00 0x44 0x10 0x45 0x00 0x46 0x00 0x47 0x10 0x49 0x16 0x4b 0x17 0x4c 0x4e 0x4d 0x1f 0x05 0x0f 3. cycle the parmnew bit to ensure that the digital clocks are active by first setting register 0x24 to 0x00, and then setting register 0x24 to 0x80. 4. start prn generation by setting the registers in table 82 to the values shown in the table. table 82. register settings to start prn generation register setting 0x48 0xab 0x4a 0xab 0x40 0x56 after the prn generator is started, users can freely configure the datapath for their desired test configuration as long as register 0x40 to register 0x55 are not modified. to disable the prn generator, write 0x00 to register 0x40. using the internal built-in self-test (bist) to test for digital data input connectivity the ad9789 includes an internal built-in self-test (bist) engine that processes incoming data and creates a signature that can be read back via the serial port. this bist feature can be configured to observe the static state of the digital data input pins (l4 to l12, m4 to m12, n5 to n12, and p5 to p12) and to reflect the state of these pins via the signature registers (register 0x50 to register 0x55). in this way, the user can verify digital data input connectivity. testing connectivity for lvds interface mode to test the connectivity of the digital data input pins in lvds interface mode, follow these steps. 1. ensure that the clock is enabled and that the clock common-mode level is set to its optimal value by setting the registers in table 83 to the values shown in the table. table 83. register settings to configure the clock register data description 0x30 0x80 enable duty cycle correction. 0x31 0xf0 set the common-mode level of clkn: clkn_cml = 0xf. 0x32 0x9e set the common-mode level of clkp: clkp_cml = 0xf. set psign = 0, nsign = 0. enable clock receiver (clk_dis = 1). 2. cycle the parmnew bit to ensure that the digital clocks are active by first setting register 0x24 to 0x00, and then setting register 0x24 to 0x80. 3. configure the lvds interface for high speed, 16-bit bus width, 16-bit data width operation by setting the registers in table 84 to the values shown in the table. table 84. register settings for lvds interface register setting 0x20 0x08 0x21 0x41 0x22 0x1f 0x23 0x87
ad9789 rev. a | page 64 of 76 4. configure pin mode by setting the registers in table 85 to the values shown in the table. table 85. register settings to configure pin modes register setting 0x42 0x00 0x43 0x08 0x44 0x00 0x45 0x08 0x46 0x00 0x47 0x10 0x49 0x1c 0x4b 0x1c 0x4c 0x00 0x4d 0x00 5. cycle the parmnew bit to ensure that the interface configuration was updated by first setting register 0x24 to 0x00, and then setting register 0x24 to 0x80. 6. apply static lvds data to the input ports. 7. enable the bist pin test by setting the registers in table 86 to the values shown in the table. table 86. register settings for bist pin test register setting 0x48 0x80 0x4a 0x80 0x40 0x55 8. read back the signature registers (register 0x50 to register 0x55) to determine the pin states (see table 87 ). table 87. signature register settings register associated lvds pairs 0x50 data bits d[7:0] 0x51 data bits d[15:8] 0x52 parity par 0x53 data bits d[7:0] (repeated) 0x54 data bits d[15:8] (repeated) 0x55 parity par (repeated) testing connectivity for cmos interface mode to test the connectivity of the digital data input pins in cmos interface mode, follow these steps. 1. ensure that the clock is enabled and that the clock common-mode level is set to its optimal value by setting the registers in table 88 to the values shown in the table. table 88. register settings to configure the clock register data description 0x30 0x80 enable duty cycle correction. 0x31 0xf0 set the common-mode level of clkn: clkn_cml = 0xf. 0x32 0x9e se t the common-mode level of clkp: clkp_cml = 0xf. set psign = 0, nsign = 0. enable clock receiver (clk_dis = 1). 2. cycle the parmnew bit to ensure that the digital clocks are active by first setting register 0x24 to 0x00, and then setting register 0x24 to 0x80. 3. configure the cmos interface for high speed, 32-bit bus width, 16-bit data width operation by setting the registers in table 89 to the values shown in the table. table 89. register settings for cmos interface register setting 0x20 0x08 0x21 0x61 0x22 0x1f 0x23 0x87 4. configure pin mode by setting the registers in table 90 to the values shown in the table. table 90. register settings to configure pin modes register setting 0x42 0x00 0x43 0x08 0x44 0x00 0x45 0x08 0x46 0x00 0x47 0x10 0x49 0x1c 0x4b 0x1c 0x4c 0x00 0x4d 0x00 5. cycle the parmnew bit to ensure that the interface configuration was updated by first setting register 0x24 to 0x00, and then setting register 0x24 to 0x80. 6. apply static cmos data to the input ports. 7. enable the bist pin test by setting the registers in table 91 to the values shown in the table. table 91. register settings for bist pin test register setting 0x48 0x80 0x4a 0x80 0x40 0x55 8. read back the signature registers (register 0x50 to register 0x55) to determine the pin states (see table 92 ). table 92. signature register settings register associated cmos pairs 0x50 data bits d[23:16] 0x51 data bits d[31:24] 0x52 parity p1 0x53 data bits d[7:0] 0x54 data bits [d15:8] 0x55 parity p0
ad9789 rev. a | page 65 of 76 qam constellation maps 07852-086 q i i k q k = 10 i k q k =11 i k q k = 00 i k q k = 01 1011 1001 1010 1000 0010 0011 0000 0001 1101 1100 1111 1110 0100 0110 0101 0111 i k q k are the two msbs in each quadrant. figure 121. dvb-c 16-qam constellation 0 7 8 5 2 - 0 8 8 q i i k q k = 10 i k q k are the two msbs in each quadrant. i k q k =11 i k q k = 00 i k q k = 01 101100 101110 100110 100100 101101 101111 100111 100101 101001 101011 100011 100001 101000 101010 100010 100000 001000 001001 001101 001100 001010 001011 001111 001110 000010 000011 000111 000110 000000 000001 000101 000100 110100 110101 110001 110000 110110 110111 110011 110010 111110 111111 111011 111010 111100 111101 111001 111000 010000 010010 011010 011000 010001 010011 011011 011001 010101 010111 011111 011101 010100 010110 011110 011100 figure 122. dvb-c 64-qam constellation 07852-087 q i i k q k = 10 i k q k =11 i k q k = 00 i k q k = 01 10111 10011 10010 10101 10001 10110 10100 10000 00110 00010 00100 00101 00111 00000 00001 00011 11011 11001 11000 11111 11101 11100 11010 11110 01000 01100 01110 01001 01101 01010 01011 01111 i k q k are the two msbs in each quadrant. figure 123. dvb-c 32-qam constellation 0 7 8 5 2 - 0 8 9 i k q k = 10 /2 rotation i k q k =11 rotation i k q k = 00 i k q k = 01 3 /2 rotation 11010 11011 01011 01010 11000 11001 01001 01000 10000 10001 10101 10100 11100 11101 10010 10011 10111 10110 11110 11111 00010 00011 00111 00110 01110 0 1111 00000 00001 00101 00100 01100 01101 11 1 1 357911 9 3 5 7 i k q k are the two msbs in each quadrant. figure 124. dvb -c128-qam constellation3
ad9789 rev. a | page 66 of 76 0 7 8 5 2 - 0 9 0 i k q k = 10 /2 rotation i k q k =11 rotation i k q k = 01 3 /2 rotation 0000 0001 0101 0100 0100 0101 0001 0000 0010 0011 0111 0110 0110 0111 0011 0010 1010 1011 1111 1110 1110 1111 1011 1010 1000 1001 1101 1100 1100 1101 1001 1000 1000 1001 1101 1100 1100 1101 1001 1000 1010 1011 1111 1110 1110 1111 1011 1010 0010 0011 0111 0110 0110 0111 0011 0010 0000 0001 0101 0100 0100 0101 0001 0000 1 13579111315 3 5 7 9 11 13 15 00 10 01 11 10 11 00 01 i k q k = 00 i k q k are the two msbs in each quadrant. figure 125. dvb-c 256-qam constellation 110,111 111,011 010,111 011,011 100,101 101,111 110,101 111,111 110,100 111,000 010,100 011,000 100,000 101,010 110,000 111,010 100,111 101,011 000,111 001,011 000,101 001,111 010,101 011,111 100,100 101,000 000,100 001,000 000,000 001,010 010,000 011,010 010,011 011,001 000,011 001,001 000,001 001,101 100,001 101,101 010,110 011,100 000,110 001,100 000,010 001,110 100,010 101,110 110,011 111,001 100,011 101,001 010,001 011,101 110,001 111,101 110,110 111,100 100,110 101,100 010,010 011,110 110,010 111,110 i q 07852-091 c5 c4 c3, c2 c1 c0 figure 126. docsis 64-qam constellation
ad9789 rev. a | page 67 of 76 1110, 1111 1111, 1101 1110, 1011 1111, 1001 1110, 0111 1111, 0101 1110, 0011 1111, 0001 0000, 1111 0011, 1111 0100, 1111 0111, 1111 1000, 1111 1011, 1111 1100, 1111 1111, 1111 1100, 1110 1101, 1100 1100, 1010 1101, 1000 1100, 0110 1101, 0100 1100, 0010 1101, 0000 0000, 1100 0011, 1100 0100, 1100 0111, 1100 1000, 1100 1011, 1100 1100, 1100 1111, 1100 1010, 1111 1011, 1101 1010, 1011 1011, 1001 1010, 0111 1011, 0101 1010, 0011 1011, 0001 0000, 1011 0011, 1011 0100, 1011 0111, 1011 1000, 1011 1011, 1011 1100, 1011 1111, 1011 1000, 1110 1001, 1100 1000, 1010 1001, 1000 1000, 0110 1001, 0100 1000, 0010 1001, 0000 0000, 1000 0011, 1000 0100, 1000 0111, 1000 1000, 1000 1011, 1000 1100, 1000 1111, 1000 0110, 1111 0111, 1101 0110, 1011 0111, 1001 0110, 0111 0111, 0101 0110, 0011 0111, 0001 0000, 0111 0011, 0111 0100, 0111 0111, 0111 1000, 0111 1011, 0111 1100, 0111 1111, 0111 0100, 1110 0101, 1100 0100, 1010 0101, 1000 0100, 0110 0101, 0100 0100, 0010 0101, 0000 0000, 0100 0011, 0100 0100, 0100 0111, 0100 1000, 0100 1011, 0100 1100, 0100 1111, 0100 0010, 1111 0011, 1101 0010, 1011 0011, 1001 0010, 0111 0011, 0101 0010, 0011 0011, 0001 0000, 0011 0011, 0011 0100, 0011 0111, 0011 1000, 0011 1011, 0011 1100, 0011 1111, 0011 0000, 1110 0001, 1100 0000, 1010 0001, 1000 0000, 0110 0001, 0100 0000, 0010 0001, 0000 0000, 0000 0011, 0000 0100, 0000 0111, 0000 1000, 0000 1011, 0000 1100, 0000 1111, 0000 1110, 0001 1101, 0001 1010, 0001 1001, 0001 0110, 0001 0101, 0001 0010, 0001 0001, 0001 0000, 0001 0001, 0011 0000, 0101 0001, 0111 0000, 1001 0001, 1011 0000, 1101 0001, 1111 1110, 0010 1101, 0010 1010, 0010 1001, 0010 0110, 0010 0101, 0010 0010, 0010 0001, 0010 0010, 0000 0011, 0010 0010, 0100 0011, 0110 0010, 1000 0011, 1010 0010, 1100 0011, 1110 1110, 0101 1101, 0101 1010, 0101 1001, 0101 0110, 0101 0101, 0101 0010, 0101 0001, 0101 0100, 0001 0101, 0011 0100, 0101 0101, 0111 0100, 1001 0101, 1011 0100, 1101 0101, 1111 1110, 0110 1101, 0110 1010, 0110 1001, 0110 0110, 0110 0101, 0110 0010, 0110 0001, 0110 0110, 0000 0111, 0010 0110, 0100 0111, 0110 0110, 1000 0111, 1010 0110, 1100 0111, 1110 1110, 1001 1101, 1001 1010, 1001 1001, 1001 0110, 1001 0101, 1001 0010, 1001 0001, 1001 1000, 0001 1001, 0011 1000, 0101 1001, 0111 1000, 1001 1001, 1011 1000, 1101 1001, 1111 1110, 1010 1101, 1010 1010, 1010 1001, 1010 0110, 1010 0101, 1010 0010, 1010 0001, 1010 1010, 0000 1011, 0010 1010, 0100 1011, 0110 1010, 1000 1011, 1010 1010, 1100 1011, 1110 1110, 1101 1101, 1101 1010, 1101 1001, 1101 0110, 1101 0101, 1101 0010, 1101 0001, 1101 1100, 0001 1101, 0011 1100, 0101 1101, 0111 1100, 1001 1101, 1011 1100, 1101 1101, 1111 1110, 1110 1101, 1110 1010, 1110 1001, 1110 0110, 1110 0101, 1110 0010, 1110 0001, 1110 1110, 0000 1111, 0010 1110, 0100 1111, 0110 1110, 1000 1111, 1010 1110, 1100 1111, 1110 i q 07852-126 c7 c6 c5 c4, c3 c2 c1 c0 figure 127. docsis 256-qam constellation
ad9789 rev. a | page 68 of 76 channelizer mode pin mapping for cmos and lvds table 93 lists the available combinations of data input configu- ration parameters when the ad9789 is in channelizer mode. many of these configurations require multiple clocks to load all channels. all of these configurations are described in detail in table 96 and table 97 . table 94 and table 95 , along with figure 128 and figure 129 , describe cmos and lvds data input pin mapping. cmos mode is always single data rate and samples on the rising edge of dsc. lvds mode is single data rate (sdr) for bus widths of 4 bits through 16 bits and double data rate (ddr) for a bus width of 32 bits. table 93. data input configurations for channelizer mode bus width data width data format 4 8 real 4 8 complex 8 8 real 8 8 complex 8 16 complex 16 8 real 16 8 complex 16 16 complex 32 8 real 32 8 complex 32 16 complex table 94. cmos pin assignments for various interface widths interface width pin assignments buswdth[1:0] 4 bits d[3:0] 00 8 bits d[7:0] 01 16 bits d[15:0] 10 32 bits d[31:0] 11 table 95. lvds pin assignments for various interface widths interface width pin assignments buswdth[1:0] 4 bits d[3:0]p, d[3:0]n 00 8 bits d[7:0]p, d[7:0]n 01 16 bits d[15:0]p, d[15:0]n 10 32 bits d[15:0]p, d[15:0]n rising edge and falling edge 11 a m n p l k b c d cmos_bus cmos_ctrl cmos_fs cmos_dco j e f h g 11 1312 11 10 89 765 24 3 4 07852-127 p1 31 27 23 19 15 11 7 3 p0 30 26 22 18 14 10 6 2 29 25 21 17 13 9 5 1 28 24 20 16 12 8 4 0 bu ct fs dc d[31:0] cmos data inputs parity and control inputs figure 128. cmos da ta input pin mapping a m parn parp n p l k b c d fsp fsn dcop dcon j e f h g 11 1312 11 10 89 765 24 3 4 07852-128 p+ 15 13 11 9 7 531 p? 15 13 11 9 7 531 14 12 10 8 6 4 2 0 14 12 10 8 6 4 2 0 fs fs dc dc +lvds 14 14 ?lvds figure 129. lvds data input pin mapping
ad9789 rev. a | page 69 of 76 in table 96 , r represents real data loaded to a given channel, i represents the in-phase term, and q represents the quadrature term of complex data. the channel number follows r, i, or q. table 96. channelizer mode configurations and channel construction: cmos interface, channel prioritization = 1 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 r 0 2 r 0 3 r 1 4 r 1 5 r 2 6 r 2 7 r 3 4 8 real 8 r 3 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 i 0 2 i 0 3 q 0 4 q 0 5 i 1 6 i 1 7 q 1 8 q 1 9 i 2 1 0 i 2 1 1 q 2 1 2 q 2 1 3 i 3 1 4 i 3 1 5 q 3 4 8 complex 1 6 q 3 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 r 0 2 r 1 3 r 2 8 8 real 4 r 3 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 i 0 2 q 0 3 i 1 4 q 1 5 i 2 6 q 2 7 i 3 8 8 complex 8 q 3
ad9789 rev. a | page 70 of 76 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 i0 2 i0 3 q0 4 q0 5 i1 6 i1 7 q1 8 q1 9 i2 10 i2 11 q2 12 q2 13 i3 14 i3 15 q3 8 16 complex 16 q3 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 r1 r0 16 8 real 2 r3 r2 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 q0 i0 2 q1 i1 3 q2 i2 16 8 complex 4 q3 i3 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 i0 2 q0 3 i1 4 q1 5 i2 6 q2 7 i3 16 16 complex 8 q3 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 32 8 real 1 r3 r2 r1 r0 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 q1 i1 q0 i0 32 8 complex 2 q3 i3 q2 i2 datapath configuration cmos pin mapping bw dw format dco [d31:d28] [d27:d24] [d23:d20] [d19:d16] [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 q0 i0 2 q1 i1 3 q2 i2 32 16 complex 4 q3 i3
ad9789 rev. a | page 71 of 76 in ddr mode, rise corresponds to data sampled on the rising edge of dsc; fall corresponds to data sampled on the falling ed ge of dsc. table 97. channelizer mode configurations and channel construction: lvds interface, channel prioritization = 1 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 r0 2 r0 3 r1 4 r1 5 r2 6 r2 7 r3 4 8 real 8 r3 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 i0 2 i0 3 q0 4 q0 5 i1 6 i1 7 q1 8 q1 9 i2 10 i2 11 q2 12 q2 13 i3 14 i3 15 q3 4 8 complex 16 q3 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 r0 2 r1 3 r2 8 8 real 4 r3 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 i0 2 q0 3 i1 4 q1 5 i2 6 q2 7 i3 8 8 complex 8 q3
ad9789 rev. a | page 72 of 76 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 i0 2 i0 3 q0 4 q0 5 i1 6 i1 7 q1 8 q1 9 i2 10 i2 11 q2 12 q2 13 i3 14 i3 15 q3 8 16 complex 16 q3 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 r1 r0 16 8 real 2 r3 r2 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 q0 i0 2 q1 i1 3 q2 i2 16 8 complex 4 q3 i3 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 i0 2 q0 3 i1 4 q1 5 i2 6 q2 7 i3 16 16 complex 8 q3 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 rise r1 r0 32 8 real 1 fall r3 r2 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 rise q0 i0 1 fall q1 i1 2 rise q2 i2 32 8 complex 2 fall q3 i3
ad9789 rev. a | page 73 of 76 datapath configuration lvds pin mapping bw dw format dco [d15:d12] [d11:d8] [d7:d4] [d3:d0] 1 rise i0 1 fall q0 2 rise i1 2 fall q1 3 rise i2 3 fall q2 4 rise i3 32 16 complex 4 fall q3
ad9789 rev. a | page 74 of 76 outline dimensions 111808-a * compliant to jedec standards mo-219 with the exception to package height. 0.80 bsc 0.80 ref a b c d e f g h j k l m n p 14 13 12 11 10 9 8 7 6 5 4 3 2 1 bottom view 10.40 bsc sq detail a top view detail a coplanarity 0.08 0.53 0.48 0.43 0.38 0.33 0.28 ball diameter seating plane 12.00 bsc sq a1 ball corner * 1.30 1.22 1.14 0.96 0.89 0.82 0.24 ref 0.65 ref figure 130. 164-ball chip scale package ball grid array [csp_bga] (bc-164-1) dimensions shown in millimeters ordering guide model 1 temperature range package description package option AD9789BBCz ?40c to +85c 164-ball chip scale package ball grid array (csp_bga) bc-164-1 AD9789BBCzrl ?40c to +85c 164-ball chip scale package ball grid array (csp_bga) bc-164-1 AD9789BBC ?40c to +85c 164-ball chip scale package ball grid array (csp_bga) bc-164-1 AD9789BBCrl ?40c to +85c 164-ball chip scale package ball grid array (csp_bga) bc-164-1 ad9789-ebz evaluation board for cmts and normal mode evaluation ad9789-mix-ebz evaluation board for mix mode evaluation 1 z = rohs compliant part.
ad9789 rev. a | page 75 of 76 notes
ad9789 rev. a | page 76 of 76 notes ?2009-2011 analog devices, inc. all rights reserved. trademarks and registered trademarks are the prop erty of their respective owners. d07852-0-7/11(a)

▲Up To Search▲

Price & Availability of AD9789BBC

	To Download AD9789BBC Datasheet File
If you can't view the Datasheet, Please click here to try to view without PDF Reader .