View GF9105A_1092097.PDF datasheet online --- IC-ON-LINE

Datasheet File OCR Text:

gennum corporation p.o. box 489, stn. a, burlington, ontario, canada l7r 3y3 tel. +1 (905) 632-2996 fax. +1 (905) 632-5946 e-mail: info@gennum.com www.gennum.com revision date: march 2000 document no. 521 - 88 - 03 data sheet features ? drop in replacement for the gf9105 with lower power and increased functionality new mode for hvf output new mode for using low frequency clocks with non- multiplexed i/o data optimized host if control signals for ensured shared bus compatibility multiple format conversions from one device 4:2:2:4 <-> 4:4:4:4 4:2:2:4 <-> r/g/b/key 4:2:2:4 <-> y/u/v/key y/u/v/key <-> r/g/b/key 4:4:4:4 <-> r/g/b/key 4:4:4:4 <-> y/u/v/key itu-r-601 compliant interpolation/decimation filters supports both single link 4:4:4:4 (smpte rp174) and dual link 4:4:4:4 (smpte rp175) compliant i/o transparent conversions between y/u/v and r/g/b color spaces. fully programmable 3x3 color space converter (csc) 13 bit color space converter coefficients 13 bit key channel scaling coefficient multiplexed and non-multiplexed i/o data bi-directional i/o data ports with tri-stating parallel host if for reading and writing multiplier coefficients and device configuration words single +5v power supply. device overview the GF9105A is a drop in replacement for the gf9105 with lower power and increased functionality. this increased functionality gives the user the option of having hvf output signals and the option of using a low frequency clock when operating with non-multiplexed input and output data. the GF9105A is a flexible vdsp engine capable of performing a variety of format conversions. the flexible architecture of the GF9105A also allows the user to perform a wide range of dsp functions that require a general 3x3 multiplier structure and/or high performance 1:2 interpolation and 2:1 decimation filters. device configuration is selected by writing configuration words through an asynchronous parallel interface (host if). the GF9105A accepts either multiplexed or non- multiplexed input data and may produce either multiplexed or non-multiplexed output data. external h, v and f inputs allow for the insertion of trs words into multiplexed output data streams. all interpolation and decimation filtering required for itu-r- 601 compliant 4:2:2:4 <-> 4:4:4:4 sample rate conversions has been integrated into the GF9105A. in addition, all input and output offset adjustments required for transparent conversions between the y/u/v and r/g/b color spaces have been included within the GF9105A. the color space converter within the GF9105A has 13 bit multiplier coefficients, has 13 bit output resolution, maintains full precision throughout the 3x3 calculation and has a true unity gain by-pass mode. sufficient resolution is maintained within the color space converter to ensure that truly transparent y/u/v <-> r/g/b conversions may be achieved. a user programmable output clipper allows the GF9105A to output a variety of word lengths to meet specific system requirements. the GF9105A is packaged in a 160 pin mqfp package, operates from a single +5v supply. general functionality of GF9105A core ordering information part number package GF9105Acqq 160 pin mqfp output offset adjust key y/g, cb/b, cr/r, key or y/g, cb/b, cr/r, or y/g key, cb/b, cr/r or key xx or cr/r xx or cb/b y/g, cb/b, cr/r, key or y/g, cb/b, cr/r, or y/g cr/r or xx key or key, cb/b, cr/r cb/b or xx 3 x 3 matrix multiplier key scaler demux 4:4:4:4 or 4:2:2:4 h_blank and input offset adjust y/g y/g y/g y/g y/g y/g y/g cb/b cr/r key cb/b cr/r key cb/b cr/r key key cb/b cr/r 13 13 13 11 cb/b cr/r 13 13 13 11 int int output clip output multiplexer key key cb/b cr/r cb/b cr/r dec dec multi gen tm GF9105A component digital transcoder
521 - 88 - 03 2 pin description pin no. symbol description 11, 20, 51, 60, 80, 101, 121, 141, 150 v dd +5 v 5% power supply. 4, 7, 10, 14, 21, 43, 52, 61, 63, 72, 81, 88, 96, 100, 105, 113, 120, 129, 138, 140, 149, 158 gnd ground. 147, 148, 151-157, 159, 160, 1, 2 p1 12..0 data port no. 1: depending on device configuration, p1 12..0 may operate as an input data port or an output data port. note: when hvf output is enabled h is always presented on p1 12 regardless of the state of input/ output . 131-137, 139, 142- 146 p2 12..0 data port no. 2: depending on device configuration, p2 12..0 may operate as an input data port or an output data port. note: when hvf output is enabled v is always presented on p2 12 regardless of the state of input/ output . 115-119, 122-128, 130 p3 12..0 data port no. 3: depending on device configuration, p3 12..0 may operate as an input data port or an output data port. note: when hvf output is enabled f is always presented on p3 12 regardless of the state of input/ output . 102-104, 106-112, 114 p4 10..0 data port no. 4: depending on device configuration, p4 10..0 may operate as an input data port or an output data port. 54, 53, 50-44, 42-39 p5 12..0 data port no. 5: depending on device configuration, p5 12..0 may operate as an input data port or an output data port. 70-64, 62, 59-55 p6 12..0 data port no. 6: depending on device configuration, p6 12..0 may operate as an input data port or an output data port. 86-82, 79-73, 71 p7 12..0 data port no. 7: depending on device configuration, p7 12..0 may operate as an input data port or an output data port. 99-97, 95-89, 87 p8 10..0 data port no. 8: depending on device configuration, p8 10..0 may operate as an input data port or an output data port. 22 sync_cb synchronization: control signal input. sync_cb is used to synchronize the GF9105A to the incoming data stream. 24 h_blank horizontal blanking: control signal input. h_blank is used to replace portions of the input data with a user selectable set of blanking levels. 25 dp_en data port enable : control signal input. dp_en is used to enable and disable data ports p1 - p8. 17 h horizontal: control signal input. h identifies the horizontal blanking interval for the output multiplexer. 16 v vertical: control signal input. v identifies the vertical blanking interval for the output multiplexer. 18 f field: control signal input. f is used to identify field information for the output multiplexer. 26 cs chip select : host interface control signal input. 23 r/w read/write : host interface control signal input. 27-31 addr 4..0 coefficient address: input port to identify which GF9105A device address shall be written to/read from. 3, 5, 6, 8, 9, 12, 13, 15 coeff_port 7..0 coefficient port: host interface bi-directional data port for color space converter coefficients, key scaler coefficient and device configuration words. 19 clk system clock: all timing information is relative to the rising edge of clk. 32 tck jtag test clock input: independent clock signal for jtag.
521 - 88 - 03 3 fig. 1 GF9105A data pin designations 33 tdi jtag test data input: serial input for jtag test data. 34 tms jtag test mode select: serial input for selecting jtag test mode. 35 trst jtag test reset : connect to gnd for normal operation. 36 tdo jtag test data output: serial output for jtag test data. 37 tn_in connect to v dd . 38 pto no connect. pin description pin no. symbol description 147 148 151 152 153 154 155 156 157 159 160 1 2 131 132 133 134 135 136 137 139 142 143 144 145 146 115 116 117 118 119 122 123 124 125 126 127 128 130 102 103 104 106 107 108 109 110 111 112 114 27 28 29 30 31 3 5 6 8 9 12 13 15 54 53 50 49 48 47 46 45 44 42 41 40 39 70 69 68 67 66 65 64 62 59 58 57 56 55 86 85 84 83 82 79 78 77 76 75 74 73 71 99 98 97 95 94 93 92 91 90 89 87 GF9105A 22 24 25 17 16 18 26 23 19 32 33 34 35 36 37 38 p1 12 /h out p1 11 p1 10 p1 9 p1 8 p1 7 p1 6 p1 5 p1 4 p1 3 p1 2 p1 1 p1 0 p2 12 /v out p2 11 p2 10 p2 9 p2 8 p2 7 p2 6 p2 5 p2 4 p2 3 p2 2 p2 1 p2 0 p3 12 /f out p3 11 p3 10 p3 9 p3 8 p3 7 p3 6 p3 5 p3 4 p3 3 p3 2 p3 1 p3 0 p4 10 p4 9 p4 8 p4 7 p4 6 p4 5 p4 4 p4 3 p4 2 p4 1 p4 0 p8 10 p8 9 p8 8 p8 7 p8 6 p8 5 p8 4 p8 3 p8 2 p8 1 p8 0 p7 12 p7 11 p7 10 p7 9 p7 8 p7 7 p7 6 p7 5 p7 4 p7 3 p7 2 p7 1 p7 0 p6 12 p6 11 p6 10 p6 9 p6 8 p6 7 p6 6 p6 5 p6 4 p6 3 p6 2 p6 1 p6 0 p5 12 p5 11 p5 10 p5 9 p5 8 p5 7 p5 6 p5 5 p5 4 p5 3 p5 2 p5 1 p5 0 addr 4 addr 3 addr 2 addr 1 addr 0 coeff_port 7 coeff_port 6 coeff_port 5 coeff_port 4 coeff_port 3 coeff_port 2 coeff_port 1 coeff_port 0 sync_cb h_blank dp_en h v f cs r/w clk tck tdi tms trst tdo tn_in pto p1 12..0 p2 12..0 p3 12..0 p4 10..0 p8 10..0 p7 12..0 p6 12..0 p5 12..0 v dd n.c jtag
521 - 88 - 03 4 fig. 2 GF9105A pin connections fig. 3a GF9105A equivalent input circuit fig. 3b GF9105A equivalent output circuit v dd p3 7 p3 6 p3 5 p3 4 p3 3 p3 2 p3 1 gnd p3 0 p2 12 /v out p2 11 p2 10 p2 9 p2 8 p2 7 p2 6 gnd p2 5 gnd v dd p2 4 p2 3 p2 2 p2 1 p2 0 p1 12 /h out p1 11 gnd v dd p1 10 p1 9 p1 8 p1 7 p1 6 p1 5 p1 4 gnd p1 3 p1 2 GF9105A top view 80 79 78 77 76 75 74 73 72 71 70 69 68 67 66 65 64 63 62 61 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 99 98 97 96 95 94 93 92 91 90 89 88 87 86 85 84 83 82 81 v dd p7 7 p7 6 p7 5 p7 4 p7 3 p7 2 p7 1 gnd p7 0 p6 12 p6 11 p6 10 p6 9 p6 8 p6 7 p6 6 gnd p6 5 gnd v dd p6 4 p6 3 p6 2 p6 1 p6 0 p5 12 p5 11 gnd v dd p5 10 p5 9 p5 8 p5 7 p5 6 p5 5 p5 4 gnd p5 3 p5 2 p1 1 p1 0 coeff 7 gnd coeff 6 coeff 5 gnd coeff 4 coeff 3 gnd v dd coeff 2 coeff 1 gnd coeff 0 v h f clk v dd gnd sync_cb r/w hblank dp_en cs addr 4 addr 3 addr 2 addr 1 addr 0 tck tdi tms trst tdo tn_in pt0 p5 0 p5 1 gnd p3 8 p3 9 p3 10 p3 11 p3 12 /f ou t p4 0 gnd p4 1 p4 2 p4 3 p4 4 p4 5 p4 6 p4 7 gnd p4 8 p4 9 p4 10 v dd gnd p8 10 p8 9 p8 8 gnd p8 7 p8 6 p8 5 p8 4 p8 3 p8 2 p8 1 gnd p8 0 p7 12 p7 11 p7 10 p7 9 p7 8 gnd 116 115 114 113 112 111 110 109 108 107 106 105 104 103 102 101 100 120 119 118 117 input n substrate p well gnd n p v dd d1 d2 p+ n+ n substrate p well d1 d2 gnd output n p p+ n+ v dd
521 - 88 - 03 5 fig. 4a functional block diagram of GF9105A (output/input = 0, hvf_out = 0) fig. 4c functional block diagram of GF9105A (output/input = 0, hvf_out = 1) fig. 4b functional block diagram of GF9105A (output/input = 1, hvf_out = 0) fig. 4d functional block diagram of GF9105A (output/input = 1, hvf_out = 1) GF9105A signal processing core bi-directional data ports processing core output processing core input output/input = 0 p1 12..0 p2 12..0 p3 12..0 p4 10..0 p5 12..0 p6 12..0 p7 12..0 p8 10..0 13 13 13 13 13 13 13 13 11 c1 c2 c3 c4 13 13 13 13 11 c5 c6 c7 c8 11 11 GF9105A signal processing core bi-directional data ports processing core output processing core input output/input = 0 p1 12 /h out p1 11..0 p2 12 /v out p2 11..0 p3 12 /f out p3 11..0 p4 10..0 p5 12 p5 11..0 p6 12 p6 11..0 p7 12 p7 11..0 p8 10..0 12 12 12 11 } } } } c1 c2 c3 c4 12 12 12 11 } } } } c5 c6 c7 c8 12 12 12 11 12 12 12 11 p1 12..0 p2 12..0 p3 12..0 p4 10..0 processing core output processing core input p5 12..0 p6 12..0 p7 12..0 p8 10..0 output/input = 1 GF9105A signal processing core bi-directional data ports c1 c2 c3 c4 c5 c6 c7 c8 13 13 13 13 13 13 13 13 13 13 13 13 11 11 11 11 p1 12 /h out p1 11..0 p2 12 /v out p2 11..0 p3 12 /f out p3 11..0 p4 10..0 processing core output processing core input p5 12 p5 11..0 p6 12 p6 11..0 p7 12 p7 11..0 p8 10..0 output/input = 1 GF9105A signal processing core bi-directional data ports } } } } c1 c2 c3 c4 } } } } c5 c6 c7 c8 12 12 12 12 12 12 12 12 12 12 12 12 11 11 11 11
521 - 88 - 03 6 fig. 5a functionality of GF9105A processing core when int/dec = 1, hvf_out = 0 fig. 5b functionality of GF9105A processing core when int/dec = 0, hvf_out = 0 hb[1:0] ioa[1:0] ooa[1:0] clp_d[1:0] muxed out 4:4:4:4/4:2:2:4_out matrix & key scaler coefficients h v f 10 10 10 10 output offset adjust output/input key y/g, cb/b, cr/r, key xx or cr/r key, cb/b, cr/r or key xx or cb/b y/g, cb/b, cr/r,key or key or key, cb/b, cr/r cb/b or xx cr/r or xx c1 c2 c3 c4 c5 c6 c7 c8 3 x 3 matrix multiplier gs9001 gs9001 demux 4:4:4:4 or 4:2:2:4 h_blank and input offset adjust y/g y/g cb/b cr/r key cb/b cr/r key cb/b cr/r key key y/g y/g y/g cb/b cr/r sync_cb h_blank 13 2 2 2 2 s 13 13 11 y/g cb/b cr/r 13 13 13 11 clock rnd8/10 rnd8/10 rnd8/10 dp_en lowf int int bypass_f output clipping output multiplexer key cb/b cr/r muxed_in fil_rnd key scaler or y/g, cb/b, cr/r,or y/g y/g, cb/b, cr/r, or y/g sl/dl_out sl/dl_in hb [1:0] ioa [1:0] 13 13 11 13 key y/g, cb/b, cr/r, key or cb/b or xx cr/r or xx c1 c2 c3 c4 3 x 3 matrix multiplier key scaler demux 4:4:4:4 or 4:2:2:4 y/g cb/b cr/r key cb/b cr/r key cb/b cr/r key key y/g cb/b cr/r key y/g y/g y/g cb/b cr/r sync_cb h_blank 13 2 2 2 2 13 13 11 y/g cb/b cr/r 13 13 13 11 output clipping output multiplexer dec dec bypass_f clp_d [1:0] muxed out 4:4:4:4/4:2:2:4_out matrix & key scaler coefficients h v f output offset adjust output/input y/g, cb/b, cr/r, key or xx or cb/b xx or cr/r key, cb/b, cr/r or key c5 c6 c7 c8 gs9001 gs9001 h_blank and input offset adjust clock rnd8/10 rnd8/10 rnd8/10 dp_en muxed_in key or key, cb/b, cr/r s y/g, cb/b, cr/r, or y/g y/g, cb/b, cr/r, or y/g sl/dl_out sl/dl_in lowf ooa [1:0] fil_rnd
521 - 88 - 03 7 fig. 5c functionality of GF9105A processing core when int/dec = 1, hvf_out = 1 fig. 5d functionality of GF9105A processing core when int/dec = 0, hvf_out = 1 hb[1:0] ioa[1:0] ooa[1:0] clp_d[1:0] muxed out 4:4:4:4/4:2:2:4_out matrix & key scaler coefficients h v f 10 10 10 10 output offset adjust output/input key y/g, cb/b, cr/r, key or key or key, cb/b, cr/r cb/b or xx cr/r or xx c5 c6 c7 c8 3 x 3 matrix multiplier gs9001 gs9001 demux 4:4:4:4 or 4:2:2:4 h_blank and input offset adjust y/g y/g cb/b cr/r key cb/b cr/r key cb/b cr/r key key y/g y/g y/g cb/b cr/r sync_cb h_blank 2 2 2 2 s y/g cb/b cr/r 13 13 13 11 clock rnd8/10 rnd8/10 rnd8/10 dp_en int int bypass_f output clipping output multiplexer key cb/b cr/r muxed_in fil_rnd key scaler y/g, cb/b, cr/r, key or y/g, cb/b, cr/r,or y/g y/g, cb/b, cr/r, or y/g sl/dl_out sl/dl_in 12 12 12 11 c2 c4 c3 c1 h v f xx or cb/b xx or cr/r key, cb/b, cr/r or key lowf hb [1:0] ioa [1:0] 12 12 11 12 key y/g, cb/b, cr/r, key or cb/b or xx cr/r or xx 3 x 3 matrix multiplier key scaler demux 4:4:4:4 or 4:2:2:4 y/g cb/b cr/r key cb/b cr/r key cb/b cr/r key key y/g cb/b cr/r key y/g y/g y/g cb/b cr/r sync_cb h_blank 12 2 2 2 12 12 11 c2 c4 c3 c1 h v f y/g cb/b cr/r 13 13 13 11 output clipping output multiplexer dec dec clp_d [1:0] muxed out 4:4:4:4/4:2:2:4_out matrix & key scaler coefficients h v f output offset adjust y/g, cb/b, cr/r, key or xx or cb/b xx or cr/r key, cb/b, cr/r or key c5 c6 c7 c8 gs9001 gs9001 h_blank and input offset adjust clock rnd8/10 rnd8/10 dp_en muxed_in key or key, cb/b, cr/r s y/g, cb/b, cr/r, or y/g y/g, cb/b, cr/r, or y/g sl/dl_out sl/dl_in 2 bypass_f output/input rnd8/10 lowf ooa [1:0] fil_rnd
521 - 88 - 03 8 GF9105A detailed device description input/output data ports the GF9105A has 8 bi-directional data ports, labelled p1 to p8. p1 to p3 and p5 to p7 are 13-bit data ports while p4 and p8 are 11-bit data ports. the output/input control bit and the hvf_out control bit ( see host programming section and figures 4a - 4d) control how p1 to p8 are configured. when output/input is set low and when hvf_out is set low, p1 12..0 , p2 12..0 , p3 12..0 , p4 10..0 are configured as input video data ports and p5 12..0 , p6 12..0 , p7 12..0 , p8 10..0 are configured as output video data ports (refer to figure 4a). when output/input is set low and when hvf_out is set high, p1 11..0 , p2 11..0 , p3 11..0 , p4 10..0 are configured as input video data ports and p5 12..0 , p6 12..0 , p7 12..0 , p8 10..0 are configured as output video data ports. in this mode, p1 12 , p2 12 , p3 12 are configured as outputs for h, v, and f output data. p1 12 carries h data, p2 12 carries v data and p3 12 carries f data (refer to figure 4c). when output/input is set high and when hvf_out is set low, p1 12..0 , p2 12..0 , p3 12..0 , p4 10..0 are configured as output video data ports and p5 12..0 , p6 12..0 , p7 12..0 , p8 10..0 are configured as input video data ports (refer to figure 4b). when output/input is set high and when hvf_out is set high, p1 11..0 , p2 11..0 , p3 11..0 , p4 10..0 are configured as output video data ports and p5 12..0 , p6 12..0 , p7 12..0 , p8 10..0 are configured as input video data ports. in this mode, p1 12 , p2 12 , p3 12 are configured as outputs for hvf output data. p1 12 carries h data, p2 12 carries v data and p3 12 carries f data (refer to figure 4d). note: no bi-directional i/os should be driven until after the output/input and the hvf_out control bits have been set (unless dp_en is set high to tri-state the outputs). this will ensure that any potential conflicts between input and output data buses are avoided. output/input and hvf control bit output/input hvf_out description 00p1 12..0 , p2 12..0 , p3 12..0 , p4 10..0 are configured as input video data ports. p5 12..0 , p6 12..0 , p7 12..0 , p8 10..0 are configured as output video data ports. refer to figure 4a. 01p1 11..0 , p2 11..0 , p3 11..0 , p4 10..0 are configured as input video data ports. p1 12 , p2 12 , p3 12 are configured as h, v and f outputs, respectively. p5 12..0 , p6 12..0 , p7 12..0 , p8 10..0 are configured as output video data ports. refer to figure 4c. 10p1 12..0 , p2 12..0 , p3 12..0 , p4 10..0 are configured as output video data ports. p5 12..0 , p6 12..0 , p7 12..0 , p8 10..0 are configured as input video data ports. refer to figure 4b. 11p1 11..0 , p2 11..0 , p3 11..0 , p4 10..0 are configured as output video data ports. p1 12 , p2 12 , p3 12 are configured as h, v and f outputs, respectively. p5 12..0 , p6 12..0 , p7 12..0 , p8 10..0 are configured as input video data ports. refer to figure 4d. for h, v, f output timing refer to the timing reference signal section of this data sheet.
521 - 88 - 03 9 data port enable dp_en is used for synchronously enabling and disabling the bi-directional data ports of the GF9105A. when dp_en is set high, the data ports are disabled and set to a high impedance state. when dp_en is set low, all data ports are enabled. input clock (clk) for standard video signals, the clock input (clk) of the GF9105A runs at one of three rates: 13.5/18mhz, 27/36mhz or 54mhz. the 18 mhz and 36 mhz variations on main clock frequencies are used in 16 x 9 video applications where luminance is sampled at 18 mhz. the use of a 27/36mhz clock with the GF9105A is the most common application. these clocks can be used with any format of input or output data with the exception of single link mode. figures 7a and 7c show multiplexed and non-multiplexed input data with a 27/36mhz clock. when the GF9105A is used with either smpte rp174 compliant single link input or output data, the input clock must run at 54 mhz (see figure 7b). a 13.5/18 mhz input clock speed can only be used when both the input and output data are in a non multiplexed format (see figure 7d). this clock rate was added to the GF9105A for use when the device is operating with non-multiplexed input and output data, since in this case a 27.0mhz clock may not be available. to use the 13.5 mhz input clock rate, the lowf control bit must be set high. when input clock rates of 27.0 mhz or 54.0 mhz are used, the lowf control bit must be set low. please note, when using the GF9105A with non-multiplexed 4:2:2:4 or 4:4:4:4 input data and an input clock rate of 27/36mhz, two rising edges of the 27/36mhz input clock are required to latch in a 13.5/18mhz input data rate (see figure 7c). basic operation of the GF9105A the basic operating mode for the GF9105A is selected via the int/dec control bit ( see host programming section ). the effective block diagram of the GF9105A processing core depends on the state of int/dec . when int/dec is set high, the internal fir filters are set for interpolation and are placed in front of the programmable 3x3 color space converter. refer to figures 5a and 5c for a functional block diagram of the GF9105A processing core when int/dec is set high. when int/dec is set low, the internal fir filters are set for decimation and are placed after the programmable 3x3 color space converter. refer to figures 5b and 5d for a functional block diagram of the GF9105A with int/dec set low. in these figures, static control bits (signals loaded via the asynchronous parallel interface) are shown at the top of the diagram and control signals with dedicated input pins are shown at the bottom of the diagram. dp_en control pin dp_en description 0 output data ports enabled. 1 output data ports disabled (high impedance state). input clock summary input clock rate (mhz) modes 13.5/18 mhz non-multiplexed input data and non-multiplexed output data (lowf=1) 27/36 mhz all input / output data formats except single link 54 mhz smpte rp174 single link input or output data int/dec control bit int/dec description 0 fir filters set for decimation. fir filters placed after the 3x3 multiplier as in figure 5b and 5d. 1 fir filters set for interpolation. fir filters placed before the 3x3 multiplier as in figure 5a and 5c.
521 - 88 - 03 10 there are seven basic blocks that make up the GF9105A. these are: ? input de-multiplexer horizontal blanking and input offset adjustment fir filters 3x3 color space converter and key scaler output offset adjustment output clipping output multiplexer since the GF9105A processing core functionality depends on the state of int/dec , device operation will be described first for the case where int/dec is set high and then for the case where int/dec is set low. GF9105A operation in interpolation mode (int/dec = 1) refer to figures 5a and 5c for a functional block diagram of GF9105A operation with int/dec = 1 bit weighting although the input data ports are physically 13 bits or 11 bits wide, the GF9105A processing core is limited to processing 10 or 8-bit unsigned input data while int/dec is set high. it should be noted that while int/dec is set low, the GF9105A processing core will accept up to 13 bit input data. refer to later sections for a description of processing core functionality while int/dec is set low. as mentioned above, the GF9105A is limited to processing 10 or 8-bit unsigned input data while int/dec is set high. this input data must be properly embedded within the input data ports. the following table illustrates how to properly embed 10 or 8-bit data within the 13 bit data ports. note that when output/input =0 and hvf_out=1, p1 12 , p2 12 and p3 12 (which corresponds to b 12 ) are outputs rather than inputs. these 3 outputs are used for presenting output h, v and f output signals. the user should be careful to ensure that p1 12 , p2 12 and p3 12 ,are not driven by upstream logic when output/input =0 and hvf_out=1. other unused inputs should be set low by the user. input de-multiplexer the muxed_in and sl/dl _in control bits ( see host programming section ) determine the input data format. the muxed_in control bit is used to identify whether the incoming data is in a multiplexed or non-multiplexed format. the sl/dl _in control bit is used to identify whether the incoming data is in a single link or dual link format. dual link (sl/dl _in = 0) while muxed_in is set low, input data is assumed to be two 10 bit streams in 4:2:2:4 or 4:4:4:4 data format as shown in figure 7a. the input de-multiplexer separates the 4:2:2:4 or 4:4:4:4 input signals into four channels of y/g, c b /b, c r /r and key data. these four data streams are then passed to the next processing section. when operating with multiplexed 4:2:2:4 or 4:4:4:4 input data, the 4:2:2 data stream enters the GF9105A processing core from processing core input port c1. while output/input =0 processing core port c1 corresponds to device data port p1 output/input = 0, hvf_out = 0 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p1 12..0 to p3 12..0 embedded 10 bit signal 000b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p4 10..0 embedded 10 bit signal na na 0 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p1 12..0 to p3 12..0 embedded 8 bit signal 000b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 input port: p4 10..0 embedded 8 bit signal na na 0 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00
521 - 88 - 03 11 (refer to figures 4a and 4c). while output/input =1 processing core port c1 corresponds to device data port p5 (refer to figures 4b and 4d). the key:2:2 or key:xx:xx data enters the GF9105A processing core from processing core input port c4. while output/ input =0, processing core port c4 corresponds to device data port p4 (refer to figures 4a and 4c). while output/ input =1, processing core port c4 corresponds to device data port p8 (refer to figures 4b and 4d). when muxed_in is set high, input data is assumed to be 4:2:2:4 or 4:4:4:4 data in a non-multiplexed format as shown in figure 7c. since the incoming data is already non-multiplexed, the input data is passed on to the next processing section unmodified. in this mode of operation, input data is presented to all four processing core input ports. while output/ input =0, processing core ports c1-c4 correspond to device data ports p1-p4 (refer to figures 4a and 4c). while output/ input =1 processing core ports c1-c4 correspond to device data ports p5-p8 (refer to figure 4b and 4d). output/input = 0, hvf_out = 1 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p1 12..0 to p3 12..0 embedded 10 bit signal h, v, or f output 00b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p4 10..0 embedded 10 bit signal na na 0 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p1 12..0 to p3 12..0 embedded 8 bit signal h, v, or f output 00b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 input port: p4 10..0 embedded 8 bit signal na na 0 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output/input = 1, hvf_out = 0 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p5 12..0 to p7 12..0 embedded 10 bit signal 000b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p8 10..0 embedded 10 bit signal na na 0 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p5 12..0 to p7 12..0 embedded 8 bit signal 000b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 input port: p8 10..0 embedded 8 bit signal na na 0 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output/input = 1, hvf_out = 1 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p5 12..0 to p7 12..0 embedded 10 bit signal 000b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p8 10..0 embedded 10 bit signal na na 0 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p5 12..0 to p7 12..0 embedded 8 bit signal 000b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 input port: p8 10..0 embedded 8 bit signal na na 0 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00
521 - 88 - 03 12 single link (sl/dl _in = 1) when operating with single link input data, the 4:4:4:4 data stream (smpte rp174 compliant) enters the GF9105A processing core from processing core input c1. while output/input = 0 processing core port c1 corresponds to device data port p1 (refer to figures 4a and 4c). while output/input = 1 processing core port c1 corresponds to device data port p5 (refer to figures 4b and 4d). in this mode, the input clock (clk) is operating at 54 mhz. also, note that the muxed_in control bit must be set low (muxed_in = 0). synchronization in order to properly synchronize the input de-multiplexer, the GF9105A requires a sync_cb control signal input. for multiplexed input data, sync_cb should change from high to low at the start of an even numbered cb sample. after synchronizing the device with the incoming data stream, sync_cb can remain low until re-synchronization is desired. refer to figure 7a for timing of sync_cb with a dual link multiplexed input data stream. refer to figure 7b for timing of sync_cb with a single link multiplexed input data signal. the timing shown may be referred to as ?standard sync_cb timing?. in order to simplify overall system design, the hsync output from the gs9001 edh coprocessor may be used as a sync_cb signal when operated with a 4:2:2 or dual link 4:4:4:4 input signal. in this mode of operation, the 10 bit multiplexed data entering the GF9105A must be fed from the output of the gs9001 and the GF9105A?s sync_cb input must be fed from the gs9001?s hsync output (refer to figure 8a). to use this mode of operation the GF9105A?s gs9001 control bit ( refer to host programming section ) must be set high. when operated with a 4:2:2 or a dual link 4:4:4:4 input signal and when the gs9001 control bit is set high, the gs9001?s hsync, vsync, and field output signals may also be used to drive the gs9105a?s output multiplexer. refer to the timing reference signal section for information regarding this. when dealing with single link 4:4:4:4 input or output signals ?standard? sync_cb timing above must be used. when using standard sync_cb and hvf timing, the gs9001 control must be set low. the gs9020 may be used to provide such standard sync_cb timing and hvf. when operated in this manner, the 10 bit multiplexed data entering the GF9105A must be fed from the output of the gs9020 and the GF9105A?s sync_cb and hvf inputs must be fed from the gs9020?s h, v, f outputs. the same gs9020/GF9105A configuration may also be used when interfacing the GF9105A to a standard 4:2:2 or dual link 4:4:4:4 link input signal. in this case, the gs9001 control bit must still be set low. with non-multiplexed input data, sync_cb must change from high to low at the start of an even-numbered cb sample. it is important to note that sync_cb changes from high to low on an even-numbered cb sample and not an odd-numbered sample. after synchronizing the device with the incoming data stream, the sync_cb signal can remain low until re- synchronization is desired. refer to figure 7c for timing of sync_cb with non-multiplexed input data. following the input de- multiplexer, data is passed to the horizontal blanking section of the device. horizontal blanking when h_blank is high, all four channels of input are forced to a user selectable set of levels. when h_blank is low data is passed through the horizontal blanking section of the device unmodified. refer to figures 10a and 10b for typical timing of h_blank with multiplexed input data and figure 10c for typical timing with non-multiplexed input data. in these figures, a muxed_in and sl/dl _in control bits muxed_in sl/dl _in description 0 0 input is in a dual link multiplexed format. 0 1 input is in a single link multiplexed format. 1 xx input is in a non-multiplexed format. gs9001 control bit gs9001 description 0 standard sync_cb and h,v,f timing. simple interface to gs9020. 1 modified sync_cb and h, v, f timing. simple interface to gs9001. note: standard sync_cb and h, v, f timing must be used when receiving or generating single link 4:4:4:4 signals.
521 - 88 - 03 13 prime () indicates to which samples the h blanking will be applied. the hb 1 and hb 0 control bits (see host programming section) determine which of the four sets of blanking levels are selected. input offset adjustment following the horizontal blanking function, a fixed set of offsets may be added to the input data. the ioa 1 and ioa 0 control bits (see host programming section) specify which of the four possible input offset adjustments will be applied to the data. as an example, the interpolation/decimation filters operate on two?s complement data, so for y/c b /c r input, ioa 1 and ioa 0 should both be set low to remove the inherent offset from the incoming data. fir filters following the input offset adjustment, data is passed to the fir filtering section of the device. these filters, when enabled , will up-sample c b and c r data by a factor of two so that 4:2:2:4 data is sample-rate converted to 4:4:4:4 data. subsequent processing of co-sited y, c b and c r samples may take place on such 4:4:4:4 data. the frequency response of these ccir- 601 compliant fir filters is shown in figures 6a and 6b and the characteristics are listed in figure 6c. in order to maintain proper synchronization between all four channels of input data, the y/g and key channels are passed through a digital delay line that matches the fir filter latency. output resolution from the fir filters depends on the state of the rnd8/10 and fil_rnd control bits ( see host programming section ). rnd8/10 should always be set to match the data format being output by the device (high for 8 bit data or low for 10 bit data). fil_rnd should be set low unless the GF9105A is being used in a mode where the 3x3 matrix is set for unity gain bypass mode. ( see 3x3 color space converter and key scaler section) . in this case, fil_rnd should be set high. the fir filter only takes 10-bit input in interpolation mode and proper input offset ha s to be used. hb 1 and hb 0 control bits hb 1 hb 0 description 0 0 blanking levels of 64, 512, 512 and 64 applied to y/g, c b /b, c r /r and key channels respectively. 0 1 blanking levels of 64, 64, 64 and 64 applied to y/g, c b /b, c r /r and key channels respectively. 1 0 blanking levels of 0, 0, 0 and 0 applied to y/g, c b /b, c r /r and key channels respectively. 1 1 blanking levels of 0, 512, 512 and 0 applied to y/g, c b /b, c r /r and key channels respectively. ioa 1 and ioa 0 control bits ioa 1 ioa 0 description 0 0 offsets of -64, -512, -512 and -64 added to the y/g, c b /b, c r /r and key channels respectively. 0 1 offsets of -64, -64, -64 and -64 added to the y/g, c b /b, c r /r and key channels respectively. 1 0 offsets of 0, 0, 0 and 0 added to the y/g, c b /b, c r /r and key channels respectively. 1 1 offsets of 0, -512, -512 and 0 added to the y/g, c b /b, c r /r and key channels respectively. rnd8/10 and fil_rnd control bits rnd8/10 fil_rnd description 0 0 output has minimum rounding for high accuracy for a non-identity matrix, using 10-bit input data. 0 1 more rounding is performed to increase overall accuracy when matrix is being bypassed, using 10-bit input data. 1 0 output has minimum rounding for high accuracy for a non-identity matrix, using 8-bit input data. 1 1 more rounding is performed to increase overall accuracy when matrix is being bypassed, using 8-bit input data.
521 - 88 - 03 14 the bypass_f control bit (see host programming section) can be used to bypass the interpolation filters. when this bit is set low, the filters are enabled and normal operation occurs. when this bit is set high, the filters are bypassed and the data is passed through the filter section unmodified. fil_rnd should be set low when bypass_f is set high. total latency through the filter is independent of the bypass_f control signal. note that after changing the state of bypass_f, an initialization period corresponding to the device?s latency is required before valid data is available at the output of the device. 3x3 color space converter and key scaler in this section, a 3x3 matrix multiplication (color space conversion) may be performed on the y/g, c b /b and c r /r data. the 3x3 matrix multiplier has 13-bit two?s complement coefficients and maintains full precision throughout the 3x3 calculation. the nine 13-bit coefficients ( see host programming section ) used in this 3x3 calculation determine the color space conversion that the GF9105A will perform. these coefficients are referred to as cm ij , where i refers to the row and j refers to the column in which cm ij is found. the matrix multiplication can be shown as: the nine matrix coefficients have 13-bit two?s complement resolution and cover a range from -4 to +3.9990234375. bit weighting for the coefficients is as follows: matrix bypassing can be accomplished by setting fil_rnd high and loading an identity matrix, by setting cm 11 , cm 22 and cm 33 to unity and setting the remaining six coefficients to zero. in this mode, gain through the matrix stage is 1.000. typical examples of matrix coefficients that will provide full range rgb to yc b c r , and yc b c r to full range rgb conversions are: key signals may also be scaled by a programmable scaling factor. the key scaling coefficient (see host programming section) has the same resolution and bit weighting as the nine 3x3 matrix multiplier coefficients. typical examples of key scaler values that could be used are: key scaler = 0.8563 for full range rgb to yc b c r conversions. key scaler = 1.1677 for yc b c r to full range rgb conversions. bypass_f control bit bypass_f description 0 filters are enabled. data is sample-rate converted from 4:2:2:4 to 4:4:4:4 data. 1 filters are disabled. data is passed through the filter section unmodified. coefficient bit b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 weighting -2 2 2 1 2 0 2 -1 2 -2 2 -3 2 -4 2 -5 2 -6 2 -7 2 -8 2 -9 2 -10 y/g out c b /b out c r /r out cm 11 cm 12 cm 13 cm 21 cm 22 cm 23 cm 31 cm 32 cm 33 y/g in c b /b in c r /r in = y c b c r 0.5027 0.0976 0.2561 0.2899 ? 0.4376 0.1477 ? 0.3633 ? 0.0711 ? 0.4374 g b r g b r 1.1677 0.3931 ? 0.8164 ? 1.1677 2.0248 0 1.1677 0 1.6025 y c b c r = =
521 - 88 - 03 15 matrix output resolution full precision is maintained within the 3x3 matrix multiplier until the output is rounded to a 13-bit or 11-bit word, depending on the state of the rnd8/10 control bit. output from the key scaler is also rounded and clipped based on the state of the rnd8/10 control bit. output offset adjustment output offset adjustment is provided to allow a specified set of offsets to be added to the data streams. the control bits ooa 1 and ooa 0 (see host programming section) determine which set of offsets is applied to the data. output clipping in the output clipping block, the data is clipped to a specific number of bits. the clp_d 1 and clp_d 0 control bits (see host programming section) determine the clipping mode that will occur. output multiplexer the muxed_out , 4:4:4:4/4:2:2:4 _out, sl/dl _out and hvf_out control bits (see host programming section) determine the output data format. dual link (sl/dl _out = 0) when muxed_out and 4:4:4:4/4:2:2:4 _out are both set low, the device will multiplex the three channels of y/g, c b /b and c r /r data into a single channel of 4:2:2 data as prescribed by smpte 125m. key information will be presented in a key:2:2 format where the c b /c r samples in the key channel are set to color blanking levels as outlined in itu-r-601. this mode can only be used when the output data has been rounded to 10-bit or 8-bit unsigned data. the 4:2:2 data stream is presented on processing core output data port c5 and the key:2:2 data is presented on processing core output data port c8. when muxed_out is set low and 4:4:4:4/4:2:2:4 _out is set high, the device will multiplex the four channels of y/g, c b /b, c r /r and key information into two streams of 4:2:2 and key:2:2 data as prescribed by smpte 125m. this mode can only be used when the output data has been clipped to 10-bit or 8-bit unsigned data. the 4:2:2 data stream is presented on processing core output data port c5 and the key:2:2 data is output on processing core output data port c8. timing reference signals (trs) may be inserted into the output data streams with such trs signals conforming to the eav/sav 3x3 matrix mulitplier output resolution for y/g, c b /b and c r /r channels rnd8/10 description 0 matrix output channels rounded to 13 bits. 1 matrix output channels rounded to 11 bits. key scaler output rnd8/10 description 0 output of key scaler rounded to 11 bits. 1 output of key scaler rounded to 9 bits. ooa 1 and ooa 0 control bits ooa 1 ooa 0 description 0 0 offsets of 64, 512, 512 and 64 are added to the y/g, c b /b, c r /r and key channels respectively. 0 1 offsets of 64, 64, 64 and 64 are added to the y/g, c b /b, c r /r and key channels respectively. 1 0 offsets of 0, 0, 0 and 0 are added to the y/g, c b /b, c r /r and key channels respectively. 1 1 offsets of 0, 512, 512 and 0 are added to the y/g, c b /b, c r /r and key channels respectively.
521 - 88 - 03 16 formats as outlined in smpte 125m. (see trs insertion section). when output/input is set high, processing core output port c5 corresponds to device data port p1 and processing core output port c8 corresponds to device data port p4. while output/input is set low, processing core output port c5 corresponds to device data port p5 and processing core output port c8 corresponds to device data port p8. single link (sl/dl _out = 1) when generating single link output data, the 4:4:4:4 data stream (smpte rp174 compliant) exits the GF9105A processing core from processing core output c5. while output/input = 0 processing core port c5 corresponds to device data port p5(refer to figure 4a). while output/input = 1 processing core port c5 corresponds to device data port p1 (refer to figure 4b). in this mode, the input clock (clk) is operating at 54 mhz. also, note that the muxed_out control bit must be set low (muxed_out = 0) and the 4:4:4:4/4:2:2:4 _out control bit must be set high (4:4:4:4/4:2:2:4 _out = 1) . when the device is configured for outputting non-multiplexed data and rnd8/10 is set low, 13-bit two?s complement,12-bit two?s complement, or 10-bit unsigned data may be output on processing core output data ports c5-c7 and 11-bit two?s complement or 10-bit unsigned data output on the processing core output data port c8. the output data will be embedded within the physical 13-bit output ports as shown in the following tables. note that when hvf_out=1 and output/input =1 b 12 of the GF9105As 13-bit two?s complement output is not available. in this case, the 13-bit output data is clipped to a 12-bit two?s complement number. bit 12 (b 12 ) of the 13 bit physical interface is used to output the h, v and f output signals. in this case, 13 bit output data is clipped 13 bit to a 12 bit two?s complement number. clp_d 1 and clp_d 0 control bit operation rnd8/10 clp_d 1 clp_d 0 description 000y/g, c b /b, c r /r channels: clipped to a 13-bit two?s complement number (values -4096 to 4095) key channel: clipped to an 11-bit two?s complement number (values -1024 to +1023) 001y/g, c b /b, c r /r channels: clipped to a 12-bit two?s complement number (values -2048 to 2047) key channel: clipped to an 11-bit two?s complement number (values -1024 to +1023) 010y/g, c b /b, c r /r, key channels: clipped to a 10-bit unsigned number (values 0 to +1023) 011y/g, c b /b, c r /r, key channels: clipped to a 10-bit unsigned number (values +4 to +1019) 100y/g, c b /b, c r /r channels: clipped to a 11-bit two?s complement number (values -1024 to 1023) key channel: clipped to an 9-bit two?s complement number (values -256 to +255) 101y/g, c b /b, c r /r channels: clipped to a 10-bit two?s complement number (values -512 to +511) key channel: clipped to an 9-bit two?s complement number (values -256 to +255) 110y/g, c b /b, c r /r, key channels: clipped to a 8-bit unsigned number (values 0 to +255) 111y/g, c b /b, c r /r, key channels: clipped to a 8-bit unsigned number (values +1 to +254) muxed_out , 4:4:4:4/4:2:2:4 _out and sl/dl _out control bits muxed_out 4:4:4:4/4:2:2:4 _out sl/dl _out description 0 0 0 output data in a 4:2:2:4 dual link multiplexed format. 0 1 0 output data in a 4:4:4:4 dual link multiplexed format. 0 1 1 output data in a 4:4:4:4 single link multiplexed format. 1 0 xx output 4:2:2:4 data in a non-multiplexed format. 1 1 xx output 4:4:4:4 data in a non-multiplexed format.
521 - 88 - 03 17 when the device is configured for outputting non-multiplexed data and rnd8/10 is set high, the output bit weighting is slightly modified. for non-multiplexed output formats, the device may output 11-bit two?s complement,10-bit two?s complement or 8-bit unsigned data on processing core output data ports c5-c7, and 9-bit two?s complement or 8-bit unsigned data may be output on processing core output data port c8. note that when hvf_out=1 and output/input =1 the msb of the 11-bit two?s complement output is not available. in this case, the output data is clipped to a 10-bit two?s complement number (-512 to + 511). bit 12 (b 12 ) of the physical interface is used to output the h, v and f output signals. the output data will be embedded within the physical 13-bit output ports as shown below. output/input = 0, hvf_out = 0 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p5 12..0 to p7 12..0 13-bit two?s complement output b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p5 12..0 to p7 12..0 12 bit two?s complement output (b 11 extended) b 11 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p5 12..0 to p7 12..0 10 bit unsigned output 000b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p8 10..0 11-bit two?s complement output na na b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p8 10..0 10 bit unsigned output na na 0 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output/input = 0, hvf_out = 1 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12 , p2 12 or p3 12 h, v or f output ------------ output port: p5 12..0 to p7 12..0 13-bit two?s complement output b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p5 12..0 to p7 12..0 12-bit two?s complement output (b 11 extended) b 11 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p5 12..0 to p7 12..0 10 bit unsigned output 000b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p8 10..0 11 bit two?s complement output na na b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p8 10..0 10 bit unsigned output na na 0 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0
521 - 88 - 03 18 when the device is configured for outputting multiplexed data, 8-bit or 10-bit unsigned data is transferred to the output data ports. consult the tables of the bit weighting section for embedding 8 or 10 bits within 13 bit data ports. note that when hvf_out=1 and output/input =1, the msb of the GF9105As 13 bit of the physical interfaces are used to output the h, v and f output signals. output/input = 1, hvf_out = 0 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12..0 to p3 12..0 13 bit two?s complement output b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12..0 to p3 12..0 12-bit two?s complement output (b 11 extended) b 11 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12..0 to p3 12..0 10 bit unsigned output 000b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p4 10..0 11 bit two?s complement output na na b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p4 10..0 10 bit unsigned output na na 0 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output/input = 1, hvf_out = 1 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12..0 to p3 12..0 12 bit two?s complement output h, v or f output b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12..0 to p3 12..0 10 bit unsigned output h, v or f output 00b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p4 10..0 11 bit two?s complement output na na b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p4 10..0 10 bit unsigned output na na 0 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output/input = 0, hvf_out = 0 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p5 12..0 to p7 12..0 11 bit two?s complement output b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p5 12..0 to p7 12..0 10 bit two?s complement output (b 9 extended) b 9 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p5 12..0 to p7 12..0 8 bit unsigned output 000b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p8 10..0 9 bit two?s complement output na na b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p8 10..0 8 bit unsigned output na na 0 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00
521 - 88 - 03 19 output/input = 0, hvf_out = 1 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12 , p2 12 or p3 12 h, v or f output ------------ output port: p5 12..0 to p7 12..0 11 bit two?s complement output b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p5 12..0 to p7 12..0 10 bit two?s complement output (b 9 extended) b 9 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p5 12..0 to p7 12..0 8 bit unsigned output 000b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p8 10..0 9 bit two?s complement output na na b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p8 10..0 8 bit unsigned output na na 0 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output/input = 1, hvf_out = 0 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12..0 to p3 12..0 11 bit two?s complement output b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p1 12..0 to p3 12..0 10 bit two?s complement output (b 11 extended) b 9 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p1 12..0 to p3 12..0 8 bit unsigned output 000b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p4 10..0 9 bit two?s complement output na na b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p4 10..0 8 bit unsigned output na na 0 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output/input = 1, hvf_out = 1 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12..0 to p3 12..0 10 bit two?s complement output h, v or f output b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p1 12..0 to p3 12..0 8 bit unsigned output h, v or f output 00b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p4 10..0 9 bit two?s complement output na na b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output port: p4 10..0 8 bit unsigned output na na 0 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00
521 - 88 - 03 20 GF9105A operation in decimation mode (int/dec =0) refer to figure 5b for a functional block diagram of the GF9105A operation with int/dec =0 bit weighting when using multiplexed input data, the processing core is limited to processing either 8-bit or 10-bit unsigned input data. the input data should be embedded within the 13-bit data port as shown in the tables of the bit weighting subsection of the interpolation mode section. note that when hvf_out=1, p1 12 , p2 12 and p3 12 (which corresponds to b 12 ) are outputs rather than inputs. these 3 outputs are used for presenting h, v and f output signals. the user should be careful to ensure that p1 12 , p2 12 and p3 12 are not driven by upstream logic when hvf_out=1. other unused inputs should be set low. when using non-multiplexed input data, the GF9105A processing core can accept up to 13-bit two?s complement data from processing core input ports c1-c3 and up to 11-bit two?s complement data from processing core input port c4. note that signed or unsigned numbers that fit within the relevant 13-bit or 11-bit dynamic range may also be presented to the device inputs. this type of input data must still be formatted as a 13-bit or 11-bit two?s complement number, with appropriate sign extensions. input bit weighting is as shown below. output/input = 0, hvf_out = 0 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p1 12..0 to p3 12..0 13 bit two?s complement input (10 bit based data) b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p1 12..0 to p3 12..0 11 bit two?s complement input (8 bit based data) b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 input port: p4 10..0 11 bit two?s complement input (10 bit based data) na na b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p4 10..0 9 bit unsigned input (8 bit based data) na na b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output/input = 0, hvf_out = 1 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 output port: p1 12 p2 12 or p3 12 h, v or f output ------------ input port: p1 11..0 to p3 11..0 12 bit two?s complement input (10 bit based data) -b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p1 11..0 to p3 11..0 10 bit two?s complement input (8 bit based data) -b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 input port: p4 10..0 11 bit two?s complement input (10 bit based data) na na b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p4 10..0 9 bit unsigned input (8 bit based data) na na b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00
521 - 88 - 03 21 input de-multiplexer refer to the input de-multiplexer discussion in the interpolation mode section. horizontal blanking refer to the horizontal blanking discussion in the interpolation mode section. input offset adjustment refer to the input offset adjustment discussion in the interpolation mode section. 3x3 color space converter and key scaler refer to the 3x3 color space converter and key scaler discussion in the interpolation mode section. matrix output resolution full precision is maintained throughout the 3x3 matrix multiplication. to ensure that maximum precision is maintained by the GF9105A, rounding of the y/g, c b /b and c r /r channels depends on the state of the rnd8/10 and bypass_f control bits (see host programming section) . output/input = 1, hvf_out = 0 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p5 12..0 to p7 12..0 13 bit two?s complement input (10 bit based data) b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p5 12..0 to p7 12..0 11 bit two?s complement input (8 bit based data) b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 input port: p8 10..0 11 bit two?s complement input (10 bit based data) na na b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p8 10..0 9 bit unsigned input (8 bit based data) na na b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 output/input = 1, hvf_out = 1 13 bit physical interface data port reference b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p5 12..0 to p7 12..0 13 bit two?s complement input (10 bit based data) b 12 b 11 b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p5 12..0 to p7 12..0 11 bit two?s complement input (8 bit based data) b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00 input port: p8 10..0 11 bit two?s complement input (10 bit based data) na na b 10 b 9 b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 input port: p8 10..0 9 bit unsigned input (8 bit based data) na na b 8 b 7 b 6 b 5 b 4 b 3 b 2 b 1 b 0 00
521 - 88 - 03 22 fir filters with int/dec set low, the internal fir filters will be set for decimation of up to 12 input bits, and the c b /b and c r /r channels will be decimated by a factor of two. as a result, 4:4:4:4 data will be sample rate converted to 4:2:2:4 data. the frequency response of the decimation filters are shown in figure 6a and figure 6b and the characteristics are listed in figure 6c. resolution out of the fir filters is 13 bits when the device is operated with 10-bit input data and 11 bits when the device is operated with 8-bit data. the fil_rnd control bit (see host programming section) should always be set low while int/dec is set low. the bypass_f control bit ( see host programming section) can be used to bypass the decimation filters. when this bit is set low, the filters are enabled and normal operation occurs. when this bit is set high, the filters are bypassed and the data is passed through the filter section unmodified. total latency through the filter is independent of the bypass_f control signal. note that after changing the state of bypass_f, an initialization period corresponding to the latency of the chip is required before valid data is available at the output of the device. output offset adjustment refer to the output offset adjustment discussion in the interpolation mode section. output clipping refer to the output clipping discussion in the interpolation mode section. output multiplexer refer to the output multiplexer discussion in the interpolation mode section. timing reference signal (trs) timing reference signals (trs) may be inserted into the output data stream of the GF9105A. in order for the trs signals to be inserted, the GF9105A?s h, v and f inputs must be driven with external horizontal (h), vertical (v) and field (f) signals. such signals should be synchronized with the incoming data stream. a low to high transition of h triggers the insertion of an eav sequence and a high to low transition triggers the insertion of an sav sequence. figures 9a, 9b and 9c show the standard timing (gs9001 control bit is set low) relationships between input data and the h, v and f inputs for multiplexed data, non-multiplexed data, and single link (4:4:4:4) data, respectively. 3x3 matrix multiplier resolution for y/g, c b /b and c r /r channels rnd8/10 bypass_f description 0 x lsbs are rounded off leaving 10-bit core data and 2 msb extension bits. 1 0 minimal rounding is performed, leaving the 8-bit core data and msb/lsb extensions for high accuracy when filter is not bypassed. 1 1 more rounding is performed, leaving 8-bit core data and msb extension bits to increase overall accuracy when filter is being bypassed. filter output rounding for c b /b and c r /r channels rnd8/10 description 0c b /b and c r /r channels rounded to 13-bit output resolution. 1c b /b and c r /r channels rounded to 11-bit output resolution. bypass_f control bit operation bypass_f description 0 filters are enabled. data is sample-rate converted from 4:4:4:4 to 4:2:2:4 data. 1 filters are disabled. data is passed through the filter section unmodified.
521 - 88 - 03 23 when the gs9001 control bit is set high and when operating with a multiplexed 4:2:2 data stream or a dual link (4:4:4:4) data stream, hvf input signals required for trs insertion may be supplied by the gs9001. in this case, the multiplexed data being fed to the GF9105A comes from the gs9001output data bus and the h input of the GF9105A is fed from the hsync output of the gs9001. in addition, the v and f inputs of the GF9105A are fed from the vblank and the field outputs of the gs9001. the relative timing of the input data and the h, v and f input signals in this mode of operation is shown in figure 9d. the h, v and f output signals of the GF9105A are derived from the h, v and f input signals. figures 9e, 9f and 9g show the timing relationship between the data output and the h, v and f output signals for multiplexed, non-multiplexed and single link output data. these timing relationships will be valid provided the timing relationships between the input data and the input h , v and f input signals are maintained as shown in figures 9a through to 9e. n o t e t h e g s 9 00 1 b i t d o e s n o t a f f e c t t h e ou t p u t h , v an d f t i m i n g . device latency when the device is working with dual link input and output data, latency through the device is 68 clock cycles and is constant regardless of which mode the device is in. when the device is working with single link input or output data, latency through the device is 136 clock cycles and is constant regardless of which mode the device is in. the latency is counted by starting at the clock cycle that latches in the input data, and counting the number of clock cycles that occur until the corresponding output data is clocked out of the device, as illustrated in figure 13. host programming the GF9105A has a host interface that allows programming of the 9 matrix coefficients, the key scaler coefficient, and several static control bits that are used to set the operating mode of the GF9105A. this data is loaded into 23 memory locations. the host interface consists of a 5-bit address bus (addr[4:0]), an 8-bit bi-directional coefficient port (coeff_port[7:0]), a read/ write pin (r/w ), and a chip select pin (cs ). to write to a specific memory location, the r/w pin must be set low (putting the coefficient port in input mode). in addition, the address and coefficient buses must be set. following this, the cs pin should be changed from high to low. data will then be clocked into the specified address. the settings of a specific memory location can be observed by performing a read operation. this is carried out by setting the r/w pin high (thus putting the coefficient bus in output mode) and setting the address bus before changing the cs pin from high to low to clock-in the address. this causes the data stored in the corresponding address to be output on the coefficient bus. the standard timing for host writing and reading is shown in figures 11a and 11b. this is the simplest method of using the host interface because r/w and the address bus (and coefficient bus when writing) all change at the same time, a minimum of 20 ns before and after the falling edge of cs . the maximum frequency for cs using this mode of operation is 25 mhz. faster (more advanced) reading and writing can be achieved by meeting certain timing requirements, as shown in figures 12a and 12b. the r/w signal setup time must be met before the first falling edge of cs . in addition, normal setup and hold times must be provided on the address bus (and the data bus when writing) with respect to the falling edge of cs . the maximum frequency using this timing is 40 mhz. note that the coefficient (coeff_port[7:0]) i/o are tri-stated when cs =1 or when r/w =0. since the memory is random access, it is not necessary to write to or read from memory locations sequentially. the memory can be considered as separate from the GF9105A processing core and can be programmed independently of the system clock. since the nine matrix coefficients and the key scaler coefficient are 13 bits wide and the memory locations are only 8 bits wide, each coefficient requires 2 memory locations. for each coefficient, the 5 lsbs are loaded into the msbs of the first memory location and the 8 msbs are loaded into the second memory location. for example, when loading a 13-bit coefficient into address n and address n+1, bit placement shall be as follows: the memory is organized such that the 9 matrix coefficients and the key scaler coefficient occupy addresses 0 through 19 (addr[4:0]=00000 through addr[4:0]=10011). addresses 20 (addr[4:0]=10100), 21 (addr[4:0]=10101), and 22 (addr[4:0]=10110) contain the static control bits that control the operation of the GF9105A. note that even if only one control bit is to be altered, the entire word must be reprogrammed.
521 - 88 - 03 24 addressing of split-up coefficients address coefficient port assignment addr[4:0] = n coeff_port[7] = b4 coeff_port[6] = b3 coeff_port[5] = b2 coeff_port[4] = b1 coeff_port[3] = b0 coeff_port[2] = x (don ? t care) coeff_port[1] = x (don ? t care) coeff_port[0] = x (don ? t care) addr[4:0] = n+1 coeff_port[7] = b12 coeff_port[6] = b11 coeff_port[5] = b10 coeff_port[4] = b9 coeff_port[3] = b8 coeff_port[2] = b7 coeff_port[1] = b6 coeff_port[0] = b5 memory location assignments for programming the host interface address port coefficient port assignment addr[4:0] = 00000 coeff_port[7:3] = cm 11 [4:0] addr[4:0] = 00001 coeff_port[7:0] = cm 11 [12:5] addr[4:0] = 00010 coeff_port[7:3] = cm 12 [4:0] addr[4:0] = 00011 coeff_port[7:0] = cm 12 [12:5] addr[4:0] = 00100 coeff_port[7:3] = cm 13 [4:0] addr[4:0] = 00101 coeff_port[7:0] = cm 13 [12:5] addr[4:0] = 00110 coeff_port[7:3] = cm 21 [4:0] addr[4:0] = 00111 coeff_port[7:0] = cm 21 [12:5] addr[4:0] = 01000 coeff_port[7:3] = cm 22 [4:0] addr[4:0] = 01001 coeff_port[7:0] = cm 22 [12:5] addr[4:0] = 01010 coeff_port[7:3] = cm 23 [4:0] addr[4:0] = 01011 coeff_port[7:0] = cm 23 [12:5] addr[4:0] = 01100 coeff_port[7:3] = cm 31 [4:0] addr[4:0] = 01101 coeff_port[7:0] = cm 31 [12:5] addr[4:0] = 01110 coeff_port[7:3] = cm 32 [4:0] addr[4:0] = 01111 coeff_port[7:0] = cm 32 [12:5] addr[4:0] = 10000 coeff_port[7:3] = cm 33 [4:0] addr[4:0] = 10001 coeff_port[7:0] = cm 33 [12:5] addr[4:0] = 10010 coeff_port[7:3] = key[4:0] addr[4:0] = 10011 coeff_port[7:0] = key[12:5]
521 - 88 - 03 25 addr[4:0] = 10100 coeff_port[7] = output/input coeff_port[6] = int/dec coeff_port[5] = bypass_f coeff_port[4] = muxed_in coeff_port[3] = muxed_out coeff_port[2] = rnd8/10 coeff_port[1] = fil_rnd coeff_port[0] = 4:4:4:4/4:2:2:4 _out addr[4:0] = 10101 coeff_port[7] = hb 1 coeff_port[6] = hb 0 coeff_port[5] = ioa 1 coeff_port[4] = ioa 0 coeff_port[3] = ooa 1 coeff_port[2] = ooa 0 coeff_port[1] = clp_d 1 coeff_port[0] = clp_d 0 addr[4:0] = 10110 coeff_port[7] = gs9001 coeff_port[6] = sl/dl _in coeff_port[5] = sl/dl _out coeff_port[4] = s* coeff_port[3] = 0 (reserved) coeff_port[2] = hvf_out coeff_port[1] = 0 (reserved) coeff_port[0] = lowf with single link output data, the s control bit should be set low for gbr signals (s=0) and set high for y, c b , c r signals (s=1). memory location assignments for programming the host interface address port coefficient port assignment
521 - 88 - 03 26 fig. 6a interpolation/decimation filter frequency response (sampling at 13.5 mhz) fig. 6b interpolation/decimation filter pass band (sampling at 13.5 mhz) fig. 6c interpolation/decimation filter characteristics at sampling frequency of 13.5 mhz parameter value filter order 57 pass band ripple < ?.0089 db pass band edge 2.850 mhz dc gain 0.000 db 3.375 mhz ( ? s /4) attenuation 12.058 db minimum stop band attenuation 58.615 db stop band edge 3.669 mhz 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 20.00 0.00 -20.00 -40.00 -60.00 -80.00 -100.00 -120.00 -140.00 frequency (mhz) magnitude (db) 0.00 0.50 1.00 1.50 2.00 2.50 3.00 frequency (mhz) magnitude (db) 0.030 0.020 0.010 0.00 -0.010 -0.020 -0.030
521 - 88 - 03 27 fig. 7a timing of sync_cb signal with dual link 4:2:2:4 or dual link 4:4:4:4 input data fig. 7b timing of sync_cb signal with single link multiplexed 4:4:4:4 input data fig. 7c timing of sync_cb signal with non-multiplexed 4:2:2:4 or 4:4:4:4 input data key:x:x or key:2:2 data 4:2:2 data cb 0 xxx or cb 1 sync_cb 27/36mhz clk y 0 cr 0 y 1 cb 2 y 2 cr 2 y 3 key 0 xxx or cr 1 key 1 xxx or cb 3 key 2 xxx or cr 3 key 3 4:4:4:4 data cb 0 / b 0 sync_cb 54 mhz clk y 0 / g 0 cr 0 / r 0 key 0 cb 1 / b 1 y 1 / g 1 cr 1 / r 1 key 1 c r /r c b /b y/g sync_cb 27/36mhz clk key y 0 /g 0 y 1 /g 1 y 2 /g 2 y 3 /g 3 cr 0 /r 0 xxx or cr 1 /r 1 cr 2 /r 2 xxx or cr 3 /r 3 cb 0 /b 0 xxx or cb 1 /b 1 cb 2 /b 2 xxx or cb 3 /b 3 key 0 key 1 key 2 key 3
521 - 88 - 03 28 fig. 7d timing of sync_cb signal with non-multiplexed 4:2:2:4 or 4:4:4:4 input data (low frequency mode) fig. 8a GF9105A optionally coupled to the gs9001 edh coprocessor for multiplexed input and non-multiplexed output data, gs900 1 control bit set high fig. 8b GF9105A optionally coupled to the gs9001 edh coprocessor for multiplexed input and output data, gs9001 control bit se t high cb 0 /b 0 xxx or cb 1 /b 1 cb 2 /b 2 xxx or cb 3 /b 3 y 0 /g 0 y 1 /g 1 y 2 /g 2 y 3 /g 3 c r /r c b /b y/g sync_cb 13.5/18mhz clk key cr 0 /r 0 xxx or cr 1 /r 1 cr 2 /r 2 xxx or cr 3 /r 3 key 0 key 1 key 2 key 3 c1 c4 sync_cb 4:2:2 4:2:2 data out hsync gennum gs9001 gennum GF9105A key:x:x or key:2:2 key:x:x or key:2:2 data in data out gennum gs9001 data in c5 c6 c7 c8 h v f c1 c4 sync_cb 4:2:2 4:2:2 data out hsync gennum GF9105A key:x:x or key:2:2 key:x:x or key:2:2 data in data out v blank field gennum gs9001 data in c5 c8 gennum gs9001
521 - 88 - 03 29 fig. 9a relative timing of multiplexed input data and h, v and f inputs (gs9001 = 0) fig. 9b relative timing of non-multiplexed input data and h, v and f inputs 4:2:2 data in h in v in f in 3ff 000 000 trs-id 3ff 000 000 trs-id cb/b y/g cr/r k v in h in f in las = last active sample las xx xx xx xx xx xx xx xx xx xx fas las xx xx xx xx xx xx xx xx xx xx fas las xx xx xx xx xx xx xx xx xx xx fas las xx xx xx xx xx xx xx xx xx xx fas fas = first active sample
521 - 88 - 03 30 fig. 9c relative timing of single link input data and h, v and f inputs fig. 9d relative timing of data, sync_cb, h, v and f inputs (gs9001 = 1) fig. 9e relative timing of dual link 4:2:2:4 or dual link 4:4:4:4 multiplexed output data and h, v and f output signals 4:4:4:4 data in h in v in f in 3ff 000 000 trs-id 3ff 000 000 trs-id sync_cb and h from hsync output of gs9001 4:2:2 data in from data output of gs9001 v from vertical output of gs9001 f from field output of gs9001 3ff 000 000 trs-id 3ff 000 000 trs-id 4:2:2 data out h out v out f out 3ff 000 000 trs-id 3ff 000 000 trs-id
521 - 88 - 03 31 fig. 9f relative timing of non-multiplexed output data and h, v and f output signals fig. 9g relative timing of single link 4:4:4:4 multiplexed output data and h, v and f output signals cb/b y/g cr/r k v out h out f out las = last active sample las xx xx xx xx xx xx xx xx xx xx fas las xx xx xx xx xx xx xx xx xx xx fas las xx xx xx xx xx xx xx xx xx xx fas las xx xx xx xx xx xx xx xx xx xx fas fas = first active sample 4:4:4:4 data out h out v out f out 3ff 000 000 trs-id 3ff 000 000 trs-id
521 - 88 - 03 32 fig. 10a typical timing of h_blank signal with dual link multiplexed 4:2:2:4 or 4:4:4:4 input data fig. 10b typical timing of h_blank signal with single link multiplexed 4:4:4:4 input data fig. 10c typical timing of h_blank signal with non-multiplexed 4:2:2:4 or 4:4:4:4 input data key:x:x or key:2:2 data_in 4:2:2 data hblank xxx or cb' n+1 key' n xxx or cr' n+1 key' n+1 xxx or cb n+3 key n+2 xxx or cr n+3 key n+3 cb' n y' n cr' n y' n+1 cb n+2 y n+2 cr n+2 y n+3 4:4:4:4 data hblank cb' n /b' n y' n /g' n cr' n /r' n key' n cb n+1 /b n+1 y n+1 /g n+1 cr n+1 /r n+1 key n+1 c r /r c b /b y/g h_blank key y' n /g n y n+1 /g n+1 y n+2 /g n+2 y n+3 /g n+3 cr' n /r n xxx or cr n+1 /r n+1 cr n+2 /r n+2 xxx or cr n+3 /r n+3 cb' n /b n xxx or cb n+1 /b n+1 cb n+2 /b n+2 xxx or cb n+3 /b n+3 key' n key n+1 key n+2 key n+3
521 - 88 - 03 33 fig. 11a standard host interface timing for writing to the GF9105A fig. 11b standard host interface timing for reading from the GF9105A fig. 12a advanced host interface timing for writing to the GF9105A coeff_port addr 20ns 20ns n n+1 n+2 n n+1 n+2 r/w cs 20ns 20ns coeff_port addr n n+1 n+2 n n+1 n+2 r/w cs 20ns 20ns 20ns 20ns t hen t hdis coeff_port addr n n+1 n+2 n n+1 n+2 r/w cs t rws t s t h
521 - 88 - 03 34 fig. 12b advanced host interface timing for reading from the GF9105A fig. 13 illustration of device latency and i/o timing addr n n+1 n+2 r/w cs t s t h t rws coeff_port n n+1 n+2 t hen t hdis data in data out clk 0 1 2 3 4 69 70 71 72 1 2 3 4 1 2 3 4 t od t h t oh t s t cy
521 - 88 - 03 35 absolute maximum ratings parameter value/units supply voltage (v dd ) -0.3 to +7.0 v input voltage range (any input) -0.5 v to (v dd +0.5) v operating temperature range 0 c to 70 c 0 mhz <= ? clk <= 54mhz storage temperature range -65 c <= t s <= 150 c lead temperature (soldering, 10 seconds) 260 c electrical characteristics conditions: v dd = 5 v, t a = 0 to 70 c unless otherwise shown. parameter symbol conditions min typ max units supply voltage v dd 4.75 5 5.25 v supply current quiescent i ddq v dd = max, v in = 0v - 0.5 ma supply current unloaded i ddu v dd = max, dp_en = v dd , ? = 27 mhz - 73 - ma input voltage, logic low v il --0.8v input voltage, logic high v ih 2--v output voltage, logic low v ol v dd = min, i ol = 2 ma - - 0.4 v output voltage, logic high v oh v dd = min, i oh = -2 ma 2.4 - - v input capacitance c in t a = 25 c, ? = 1 mhz - - 10 pf output capacitance c out t a = 25 c, ? = 1mhz - - 10 pf switching characteristics conditions: v dd = 5 v, t a = 0 to 70 c unless otherwise shown. parameter symbol conditions min typ max units input setup time t s 6--ns input hold time t h 2--ns output delay time t od v dd = min, c l = 25 pf - - 13 ns output hold time t oh v dd = max, c l = 25 pf 4 - - ns output enable time t en v dd = min, c l = 25 pf - - 12 ns output disable time t dis v dd = min, c l = 25 pf - - 12 ns host interface r/w setup time t rws 20 - - ns host interface output enable time t hen v dd = min, c l = 25 pf - - 12 ns host interface output disable time t hdis v dd = max, c l = 25 pf - - 12 ns
521 - 88 - 03 36 clock cycle time t cy non-multiplexed input data and non- multiplexed output data lowf=1 55 - - ns all input/output data formats except single link 27 - - ns smpte rp174 single link input or output data 18 - - ns clock pulse width low t pwl as a percentage of min. clock cycle time 40 - 60 % clock pulse width high t pwh as a percentage of min. clock cycle time 40 - 60 % device latency 1 low frequency mode (lowf = 1) 34 34 34 clks dual link input and output data 68 68 68 clks single link input or output data 136 136 136 clks note 1: latency is defined as the number of clock cycles between the time when the data is latched into the device and when th e corresponding output data is clocked out of the device. refer to figure 13. switching characteristics conditions: v dd = 5 v, t a = 0 to 70 c unless otherwise shown. parameter symbol conditions min typ max units
521 - 88 - 03 37 gennum corporation mailing address: p.o. box 489, stn. a, burlington, ontario, canada l7r 3y3 tel. +1 (905) 632-2996 fax. +1 (905) 632-5946 shipping address: 970 fraser drive, burlington, ontario, canada l7l 5p5 gennum japan corporation c-101, miyamae village, 2-10-42 miyamae, suginami-ku tokyo 168-0081, japan tel. +81 (03) 3334-7700 fax. +81 (03) 3247-8839 gennum uk limited 25 long garden walk, farnham, surrey, england gu9 7hx tel. +44 (0)1252 747 000 fax +44 (0)1252 726 523 gennum corporation assumes no responsibility for the use of any circuits described herein and makes no representations that th ey are free from patent infringement. ?copyright june 1999 gennum corporation. all rights reserved. printed in canada. fig. 14 160 pin mqfp 160 1 31.20 31.20 ?.25 28.00 ?.10 28.00 ?.10 4.10 max 3.42 ?.25 0.65 bsc 0.30 ?.08 1.6 ref 0.13/0.30 max radius 0.13 min radius 0.88 ?.15 0.40 min 10 ?? 10 ?? 0? min 7? max 8? max 0? min revision notes: updated input clock (clk) information. for latest product information, visit www.gennum.com document identification data sheet the product is in production. gennum reserves the right to make changes at any time to improve reliability, function or design, in order to provide the best product possible. caution electrostatic sensitive devices do not open packages or handle except at a static-free workstation

▲Up To Search▲

Price & Availability of GF9105A

	To Download GF9105A Datasheet File
If you can't view the Datasheet, Please click here to try to view without PDF Reader .