View HSP43124_1984708.PDF datasheet online --- IC-ON-LINE

Datasheet File OCR Text:

1 tm fn3555.6 HSP43124 serial i/o filter the serial i/o filter is a high performance filter engine that is ideal for off loading the burden of filter processing from a dsp microprocessor. it supports a variety of multistage filter configurations based on a user programmable filter and fixed coefficient halfband filters. these configurations include a programmable fir filter of up to 256 taps, a cascade of from one to five halfband filters, or a cascade of halfband filters followed by a programmable fir. the half band filters each decimate by a factor of two, and the fir filter decimates from one to eight. when all six filters are selected, a maximum decimation of 256 is provided. for digital tuning applications, a separate multiplier is provided which allows the incoming data stream to be multiplied, or mixed, by a user supplied mix factor. a two pin interface is provided for serially loading the mix factor from an external source or selecting the mix factor from an on- board rom. the on-board rom contains samples of a sinusoid capable of spectrally shifting the input data by one quarter of the sample rate, f s /4. this allows the chip to function as a digital down converter when the filter stages are configured as a low-pass filter. the serial interface for 3- input and output data is compatible with the serial ports of common dsp microprocessors. coefficients and configuration data are loaded over a bidirectional eight bit interface. features ? 45mhz clock rate ? 256 tap programmable fir filter ? 24-bit data, 32-bit coefficients ? cascade of up to 5 half band filters ? decimation from 1 to 256 ? two pin interface for down conversion by f s /4 ? multiplier for mixing or scaling input with an external source ? serial i/o compatible with most dsp microprocessors applications ? low cost fir filter ? filter co-processor ? digital tuner block diagram ordering information part number temp. range ( o c) package pkg. no. HSP43124pc-45 0 to 70 28 ld pdip e28.6 HSP43124pc-33 0 to 70 28 ld pdip e28.6 HSP43124sc-45 0 to 70 28 ld soic m28.3 HSP43124sc-33 0 to 70 28 ld soic m28.3 HSP43124si-40 -40 to 85 28 ld soic m28.3 programmable fir filter input formatter din sclk syncin mxin syncmx output formatter dout syncout clkout a0-2 c0-7 wr control interface rd fsync fclk half band filter #5 half band filter #2 half band filter #1 data sheet may 1999 caution: these devices are sensitive to electrostatic discharge; follow proper ic handling procedures. 1-888-intersil or 321-724-7143 | intersil (and design) is a trademark of intersil americas inc. copyright ? intersil americas inc. 2002. all rights reserved
2 pinout 28 lead pdip, soic top view sclk syncin gnd fsync v cc fclk wr rd a0 a1 a2 v cc din syncout clkout v cc c7 c5 gnd c3 c2 c1 c0 dout c6 c4 28 27 26 25 24 23 22 21 20 19 18 17 16 15 2 3 4 5 6 7 8 9 10 11 12 13 14 syncmx mxin 1 HSP43124
3 pin description name type description v cc -+5v power supply gnd - ground din i serial data input. the bit value present on this input is sampled on the rising edge of sclk. a ?high? on this input represents a ?1?, and a low on this input represents ?0?. the word format and operation of serial interface are con- tained in the data input section. syncin i data sync. the HSP43124 is synchronized to the beginning of a new data word on din when sclk samples syn- cin ?high? one sclk before the first bit of the new word. note: syncin should not maintain a ?high? state for longer than one sclk cycle. sclk i serial input clk. the rising edge of sclk clocks data on din and mxin into the part. the following signals are synchronous to this clock: din, syncin, mxin, syncmx. mxin i mix factor input. mxin is the serial input for the mix factor. it is sampled on the rising edge of sclk. a ?high? on this input represents a ?1?, and a low on this input represents ?0?. also used to specify the weaver modulator rom output as a part of the two pin f s/ 4 down conversion interface. details on word format and operation are contained in the mix factor section. syncmx i mix factor sync. the HSP43124 is synchronized to the beginning of a serially input mix factor when sclk samples syncmx ?high? one sclk before the first bit of the new mix factor. note: syncmx should only pulse ?high? for one sclk cycle. also used to specify weaver modulator rom output as a part of the two pin f s /4 down con- version interface. fclk i filter clock. the filter clock determines the processing speed of the filter compute engine. clock rate require- ments on fclk for particular filter configurations is discussed in the filter compute engine section. this clock may be asynchronous to the serial input clock (sclk). fsync is synchronous to this clock. fsync i filter sync. this input, when sampled low by the rising edge of fclk, resets the filter compute engine so that the data sample following the next syncin cycle is the first data sample into the filter structure. if a data stream is currently being input, the current sum of products and the input data are ?canceled? and the din pin is ignored until the next syncin cycle occurs. wr i write. the falling edge of wr loads data present on c0-7 into the configuration or coefficient register specified by the address on a0-2. the wr signal is asynchronous to all other clocks. note: wr should not be low when rd is low. rd i read. the falling edge of rd accesses the control registers or coefficient ram addressed by a0-2 and places the contents of that memory location on c0-7. when rd returns ?high? the c0-7 bus functions as an input bus. the rd pin is asynchronous to all other clocks. note: rd should not be low when wr is low. a0-2 i address bus. the a0-2 inputs are decoded on the falling edge of both rd and wr . table 1 shows the address map for the control registers. c0-7 i/o control and coefficient bus. this bidirectional bus is used to access the control registers and coefficient ram. clkout o output clock. programmable bit clock for serial output. note: assertion of fsync initializes clkout to a high state. syncout o output data sync. synyout is asserted high for one clkout cycle before the first bit of a new output sample is available on dout. dout o serial data output. the bit stream is synchronous to the rising edge of clkout. (see the serial output formatter section for additional details.) HSP43124
4 functional descriptions the HSP43124 is a high performance digital filter designed to process a serial input data stream. a second serial interface is provided for mix factor inputs, which are multiplied by the input samples as shown in figure 1. the multiplier result is passed to the filter compute engine for processing. the filter compute engine centers around a single multiply/accumulator (mac). the mac performs the sum-of- products required by a particular filter configuration. the processing rate of the mac is determined by the filter clock, fclk. increasing fclk relative to the input sample rate increases the length of filter that can be realized. the filtered results are passed to the output formatter where they are rounded or truncated to a user defined bit width. the output formatter then generates the timing and synchronization signals required to serially transmit the data to an external device. filter configuration the HSP43124 is configured for operation by loading a set of eight control registers. these registers are written through a bidirectional interface which is also used for reading the control registers. the interface consists of an 8-bit data bus, c0-7, a 3-bit address bus, a0-2, and read/write lines, rd and wr. the address map for the control registers is shown in table 1. data is written to the configuration control registers on the falling edge of the wr input. this requires that the address, a0-2, and data, c0-7, be stable and valid on the falling edge of the wr , as shown in figure 2. note: wr should not be active low when rd is active low. data is read from the configuration control registers on the falling edge of the rd input. the contents of a particular register are accessed by setting up an address, a0-2, to the falling edge of rd as shown in figure 2. the data is output on c0-7. the data on c0-7 remains valid until rd returns high, at which point the c0-7 bus is three-stated and functions as an input. for proper operation, the address on a0-2 must be held until rd returns ?high? as shown in figure 2. note: rd should not be active low when wr is active low. m u x mux holding reg mix factor holding serial multiplier round/ saturate coefficient ram r e g variable length shift register (8 to 24 bits) + round/ saturate weaver modulator rom mux control mxin syncin 25 24 32 48 57 halfband coefficient rom reg register file + variable length shift register (8-24-bits) din syncin syncmx mxin a0-2 c0-7 wr rd fsync fclk sclk multiply/ input formatter filter compute engine syncmx output formatter accumulator dout syncout clkout figure 1. serial filter block diagram input # bits ? msb f/2 ? format ? mix sel ? # bits ? format ? filt en ? # hbs ? ? ? decimation rate ? fir sym ? rd en ? filter length ? ram access ? round ? format ? gain cor ? # bits ? fclk ? msb f/l ? clkout ? indicates configuration control word data parameter. control parameters wr a0-2 write timing c0-7 rd a0-2 read timing c0-7 figure 2. read/write timing HSP43124
5 writing coefficients the HSP43124 provides a register bank to store filter coefficients for configurations which use the programmable filter. the register bank consists of 128 thirty-two-bit registers. each register is loaded by 4 one byte writes to the bidirectional interface used for loading the configuration registers. the coefficients are loaded in order from least significant byte (lsb) to most significant byte (msb). the coefficient registers are loaded by first setting the coefficient read enable bit to ?0? (bit 4 of the filter configuration register). next, coefficients are loaded by setting the a2-0 address to 010 (binary) and writing one byte at a time as shown in figure 3. the down loaded bytes are stored in a holding register until the 4th write cycle. on completion of the fourth write cycle, the contents of the holding register are loaded into the coefficient ram, and the write pointer is incremented to the next register. if the user attempts to write more than 128 coefficients, the pointer table 1. configuration control register functional description address register description bit positions bit function 000 filter configuration 2-0 specifies the number of halfbands to use. number ranges from 0 to 5. other values are invalid. 3 filter enable bit. 1 = enable. 0 = minimum filter bypass (either the fir or hbf must be enabled to get an output). 4 coefficient read enable. when set to 1, enables reading and disables writing of coefficient ram. note: this bit must be set to 0 prior to writing the coefficient ram. 7-5 fir decimation rate. range is 1-8 (8 = 000). 001 programmable fir filter length 7-0 number of taps in the programmable fir filter. for even or odd symmetric filters, values range from 4- 256, 1 to 3 are invalid, and 0000000 = 256. for asymmetric filters, the value loaded in this register must be two times the ac- tual number of coefficients. 010 coefficient ram access 7-0 coefficient ram is loaded by multiple writes to this address. (see writing coefficients section for additional details.) 011 input format 4-0 number of bits in input data word, from 8 (01000) to 24 (11000). values out- side the range of 8 - 24 are invalid. 5 number system. 0 = two?s complement, 1 = offset binary. 6 serial format. 1 = msb first, 0 = lsb first. 7unused 100 output timing 4-0 number of fclks per clkout. range 1 to 32. (00000 = 32 fclks) 5 1 = msb first, 0 = lsb first. 6-7 unused 101 output format 4-0 number of bits in output data word, from 8 to 32. a value of 32 is represented by 00000, and values from 1 to 7 are invalid. 5 round select. 1 = round to selected number of bits, 0 = truncate. 6 number system. 0 = two?s complement, 1 = offset binary. 7 gain correction. 1 = apply scale factor of 2 to data. 0 = no scaling. 110 filter symmetry 1-0 00 = even symmetric fir coefficients 01 = non-symmetric coefficients 10 = odd symmetric fir 7-2 reserved: must be 0. 111 mix factor format 4-0 number of bits in mix factor, from 8 (01000) to 24 (11000). values outside the range of 8 - 24 are invalid. 5 serial format. 1 = msb first, 0 = lsb first. 6 mix factor select. 1 = serial input, 0 = weaver modulator look-up-table. 7 unused HSP43124
6 halts at the 128th register location, and writing is disabled. the coefficient address pointer is reset when any other configuration register is written or read. note: a new coefficient set may be loaded during a filter calculation at the risk of corrupting output data until the load is complete. the number of coefficients that must be loaded is dependent on whether the coefficient set exhibits even symmetry, odd symmetry, or asymmetry (see figure 4). for filters that exhibit either even or odd symmetry, only the unique half of the coefficient set must be loaded. the coefficients are loaded in order starting with the first filter tap and ending with the center tap. the coefficient associated with the first tap is the first to be multiplied by an incoming data sample as shown in figure 5. for even/odd symmetric filters of length n, n/2 coefficients must be loaded if the filter length is even, and (n+1)/2 coefficients must be loaded if the filter length is odd. for example, a 17 tap symmetric filter would require the loading of 9 coefficients. enough storage is provided for a 256 tap symmetric filter. for asymmetric filters the entire coefficient set must be loaded. the coefficients are loaded in order starting with the first tap and ending with the final filter tap (see figure 5 for tap/coefficient association). enough storage is provided for a 128 tap asymmetric filter. for asymmetric filters the value loaded into the programmable filter length register addressed must be twice the actual number of coefficients. reading coefficients the coefficients are read from the storage registers one byte at a time via c0-7 as shown in figure 6. to read the coefficients, the user first sets the coefficient read enable bit to 1 (bit 4 of filter configuration control register). setting this bit resets the ram read pointer and disables the ram from being written. next, with a2-0 = 010, multiple ?high? to ?low? transitions of rd , output the coefficients on c0-7, one byte at a time, in the order they were written. note: rd should not be ?low? when wr is ?low? . data input data is serially input to the HSP43124 through the din input. on the rising edge of sclk, the bit value present at din is clocked into the variable length shift register. the beginning of a serial data word is designated by asserting syncin ?high? one sclk prior to the first data bit as shown even symmetric note: filters with even symmetric coefficients exhibit symme- try about the center of the coefficient set. most fir filters have coefficients which are symmetric in nature. odd symmetric note: odd symmetric coefficients have a coefficient envelope which has the characteristics of an odd function (i.e. coeffi- cients which are equidistant from the center of the coefficient set are equal in magnitude but opposite in sign). coefficients designed to function as a differentiator or hilbert transform ex- hibit these characteristics. asymmetric note: asymmetric coefficient sets exhibit no symmetry. figure 4. coefficient characteristics a0-2 c0-7 a0-2 = 010 (binary) lsb msb lsb msb first coefficient second coefficient wr figure 3. coefficient loading odd length point of symmetry even length 0.5 -0.5 0.25 -0.25 0.1 -0.1 center of coefficient set z -1 z -1 + c0 c1 c2 x(n) input y(n) output first filter tap last filter tap figure 5. three tap transversal filter architecture x 2 x 1 x 0 y(n) = c 0 x 2 + c 1 x 1 + c 2 x 0 a0-2 c0-7 a0-2 = 010 (binary) lsb msb lsb msb first coefficient second coefficient rd figure 6. coefficient reading HSP43124
7 in figure 7. on the following sclk, the first data bit is clocked into the variable length shift register. data bits are clocked into the shift register until the data word, of user programmable length (8 to 24 bits), is complete. at this point, the shifting of data into the register is disabled and its contents are held until syncin is asserted on the rising edge of sclk. when this occurs, the contents of the variable length shift register are transferred to the input holding register, and the shift register is enabled to accept serial data on the following sclk. the serial data word may be two?s complement or offset binary and may be input most significant bit (msb) first or least significant bit (lsb) first as defined in the input format register (see table 1). if a data word is specified to be less than 24 bits, the least significant bits of the input holding register are zeroed. note: syncin should not be ?high? for longer than one sclk cycle . mix factor the HSP43124 provides a second serial interface for loading values which are multiplied by the input samples in the serial multiplier. these values, or mix factors, are input using the mxin and syncmx pins. aside from being used as a serial input, this interface can also be used to select mix factors from the weaver modulator rom. the mix factor source is specified in the mix factor format register (see table 1). note: data is passed unmodified through the serial multiplier by selecting the weaver modulation rom as the mix factor source and tying both syncmx and mxin ?high?. the procedure for loading serial mix factors is similar to that for the loading of data via the din input. the bit value present on mxin is clocked into the variable length shift register by the rising edge of sclk. the beginning of the serial word is designated by the assertion of syncmx one sclk prior to the first bit of the serial word as shown in figure 7. after the serial word has been clocked into the shift register, the shifting of bits into the register is disabled and its contents are held until the next assertion of syncmx. when syncmx is asserted on the rising edge of sclk, the contents of the variable length shift register are transferred into the mix factor holding register. the parallel output of the mix factor holding register feeds directly into the serial multiplier. the mix factor data word is programmable in length from 8 to 24 bits and may be input msb or lsb first as specified in the mix factor format register. if a data word is specified to be less than 24 bits, the least significant bits of the mix factor holding register are zeroed. the mxin and syncmx inputs can function as two pin interfaces to weaver modulator rom addresses. used in proper sequence, down conversion by f s /4 can be achieved. these inputs are latched on the rising edge of sclk when syncin is high as shown in figure 9. the mapping of syncin and mxin to rom outputs is given in table 2. when syncin is high on the rising edge of sclk, the output of the rom is transferred to the mix factor holding register, and the syncmx and mxin inputs are decoded to produce a new rom output. as a result, there is a latency of one syncin cycle between when the syncmx and mxin inputs are decoded and when the rom output is loaded into the mix factor holding register. serial multiplier the serial multiplier multiplies the mix factor holding register by the contents of the input holding register. the multiplication cycle is initiated when syncin is sampled high by the rising edge of sclk. this transfers the contents of the variable length shift register to the input holding register, and loads the output of the mix factor holding register into the serial multiplier. on subsequent sclks, the contents of the input holding register are shifted into the serial multiplier for processing. when the last data bit is shifted into the multiplier, the multiplication cycle is complete and the result is written to the register file on the next rising edge of fclk. the synchronization between a data sample and the mix factor it is to be multiplied by is dependent on which mix factor source is specified. for mix factors which are input serially, the mix factor is loaded concurrently with the data sample to be multiplied (see figure 8). din/ lsb lsb sclk syncin/ sycnmx mxin note: assumes data is being loaded lsb first. figure 7. serial input timing for either din or mxin inputs sync leads data table 2. weaver modulator rom decoding syncmx mxin mix factor 000 01-1 100 111 HSP43124
8 note: figure 8 shows the loading of a data sample, x0, such that it will be multiplied by a mix factor designated by m0. for mix factor bit widths which are less than the input bit width, syncmx may be asserted before syncin if desired. if the mix factor is generated by the weaver modulator rom, the mix factor must be specified on mxin and syncmx one syncin before that which precedes the target data word (see figure 9). filter compute engine the filter compute engine centers around a multiply accumulator which is used to perform the sum-of-products required for a variety of filtering configurations. these configurations include a cascade of up to 5 halfband filters, a single symmetric filter of up to 256 taps, a single asymmetric filter of up to 128 taps, or a cascade of halfband filters followed by a programmable filter. the filter configuration is specified by programming the filter configuration register (see table 1). the cascade of up to five halfband filters is an efficient decimating filter structure. each fixed coefficient filter in the chain introduces a decimation of two, and the aggregate decimation rate of the entire halfband filtering stage is given by: thus, a cascade of 3 halfband filters would decimate the input sample stream by a factor of 8. figure 10a is a block diagram of the halfband filter section. the normalized frequencies for each halfband stage is labeled. figure 10b is an illustration of a cascaded filter composed of five halfband filters. the final stage filter output is clocked at fclk/32. since the output of each filter is at half the rate of the input, the five halfband filter passband characteristics can be viewed on a single plot whose x axis is normalized to the filter output clock rate. notice that all halfband filters, by design, have 120db passbands that are less than the output rate divided by 2. since the alias profile is well below -120db in the filter passband, alias concerns are eliminated. the frequency responses of the five filters are presented graphically in figure 10c and in tabular form in table 3. notice that the 6db passband bandwidth (f = 0.25) is identical for all five halfband filters. the width of the transition band, however, is different for each filter. the transition band for the fifth halfband filter, hb5, is the narrowest while that for the first halfband filter, hb1, is the widest. the cascade of the halfband filters always terminates with hb5 and is preceded by filters in order of increasing transition bandwidth. for example, if the HSP43124 is configured to operate with three halfbands, the chain of filters would consist of hb3 followed by hb4 and terminated with hb5. if only one halfband is selected, hb5 is used. din sclk syncin syncmx mxin lsb lsb x0 m0 msb lsb lsb msb figure 8. data/mix factor synchronization for serially input mix factors sync leads data din sclk syncin x0 sycnmx/ mxin m0 msb lsb lsb figure 9. data/mix factor synchronization weaver modulator mix factors sync leads data dec hb = 2 (number of halfband filters selected) (eq. 1) HSP43124
9 figure 10a. block diagram of fixed coefficient halfband filters bits2-0 100 0 1 11 1 101 1 1 11 1 010 0 0 01 1 011 0 0 11 1 000 0 0 00 0 001 0 0 00 1 mux1 = (bit2 and bit0) and bit1 mux2 = bit2 mux3 = (bit1 and bit0) or bit2 mux4 = bit1 or bit2 mux5 = bit0 or bit1 or bit2 invalid = bit2 and bit1 multiplexers? decoder table and equations mux5 mux4 mux3 mux2 mux1 hb5 hb4 hb3 hb2 hb1 input to f normalized = f s f s f hb1 = f s or f s /2 f s /2 f hb1 /2 f hb2 = f hb1 or f hb1 /2 f normalized = f hb1 f hb3 = f hb2 or f hb2 /2 f hb2 /2 f normalized = f hb2 f hb3 /2 f normalized = f hb3 f hb4 = f hb3 or f hb3 /2 f hb4 /2 f normalized = f hb4 output of halfband section halfband section 01 mux5 mux4 mux3 mux2 mux1 01 01 01 01 35 tap 23 tap 19 tap 11 tap 7 tap figure 10b. spectral composition of five cascaded halfbands figure 10c. composite response of fixed coeffi- cient halfband filters with respect to the normalized frequency shown in fig- ure 10a fclk 4 fclk 2 3fclk 4 fclk fclk 2 fclk 2 f s 2 f s 2 fclk 4 fclk 4 fclk 4 fclk 4 fclk 8 fclk 8 fclk 8 fclk 8 fclk 16 fclk fclk fclk fclk fclk fclk 32 fclk 32 fclk 64 fclk 64 stage 1 stage 2 stage 3 stage 4 stage 5 posite filter fclk 16 com- -120 -140 -160 -180 -200 0 0.125 0.25 0.375 0.5 magnitude (db) -60 -40 -20 -80 -100 6db bandwidth hb1 transition bw normalized frequency hb2 hb3 hb4 hb5 (normalized to output frequency) HSP43124
10 the coefficient set for each of the halfband filters is given in table 4. these values are the 32-bit, two?s complement, integer representation of the filter coefficients. scaling these values by 2 -31 yields the fractional two?s complement coefficients used to achieve unity gain in the filter processor. if a specific frequency response is desired, a programmable fir filter may be activated. the filter compute engine takes advantage of symmetry in fir coefficients is by summing data samples sharing a common coefficient prior to multiplication. in this manner, two filter taps are calculated per multiply accumulate cycle. if an asymmetric filter is specified, only one tap per multiply accumulate cycle is calculated. the processing rate of the filter compute engine is proportional to fclk. as a result, the frequency of fclk must exceed a minimum value to insure that a filter calculation is complete before the result is required for output. in configurations which do not use decimation, one input sample period is available for filter calculation before an output is required. for configurations which employ decimation, up to 256 input sample periods may be available for filter calculation. the following equation specifies the minimum fclk rate required for configurations which use the programmable filter as an fir filter. in this equation f s is the input sample rate (sclk/# bits in ser word), taps is the number of taps in the fir filter (0 to 256), dec fir is the decimation rate of the programmable fir (1 to 8), hb clks is a compute clock factor based on the number of halfband filters in the configuration (see table 5), and dec hb is the aggregate decimation rate for the cascade of halfband filters (see table 5). for example, if the input sample rate is 800khz, a 128 tap fir filter with no decimation is selected, and a cascade of 2 halfband filters is used, calculate the minimum fclk rate as follows: thus, the min fclk is 19.6mhz. note: for configurations in which the halfband filters are used, the fclk rate must exceed 14f s . the longest length fir filter realizable for a particular configuration is determined by solving the above equation for taps. the resulting expression is given below. the maximum throughput sample rate may be specified by solving the above equation for f s . the resulting equation is note: for configurations using filters with asymmetric coeffi- cients, the term taps in the above equations should be multi- plied by two in order to determine the correct fclk. the filter compute engine is synchronized with an incoming data stream by asserting the fsync input. when this input is sampled low by the rising edge of fclk, the compute engine is reset, and the data word following the next assertion of syncin is recognized as the first data sample input to the filter structure. min fclk f s dec hb ------------------- - taps/(2*dec fir ) hb clks 1 ++ () or at least14f s when halfbands are used ? ? ? ? ? = (eq. 2) (eq. 3) min fclk 800khz 4 --------------------- ?? ?? 128 2 1 --------- - ?? ?? 33 1 ++ ?? ?? 200khz () 64 33 1 ++ [] 19.6mhz = or at least 14 (800khz) 11.2mh z = ? ? ? ? ? ? ? = max taps = 2dec fir ((fclk/f s )dec hb - hb clks - 1) (eq. 4) max f s = fclk*dec hb /(taps/(2*dec fir ) + hb clks + 1). (eq. 5) table 3. frequency response of halfband filters normalized frequency halfband #1 halfband #2 halfband #3 halfband #4 halfband #5 0.000000 -0.000000 0.000000 0.000000 -0.000000 -0.000000 0.007812 0.000000 -0.000000 -0.000000 -0.000000 -0.000000 0.015625 -0.000113 -0.000000 -0.000000 -0.000000 -0.000000 0.023438 -0.000677 -0.000006 -0.000000 -0.000000 -0.000000 0.031250 -0.002243 -0.000052 -0.000000 -0.000000 -0.000000 0.039062 -0.005569 -0.000227 -0.000000 -0.000000 0.000000 0.046875 -0.011596 -0.000719 -0.000001 0.000000 -0.000000 0.054688 -0.021433 -0.001859 -0.000009 -0.000000 -0.000000 0.062500 -0.036333 -0.004165 -0.000041 -0.000000 -0.000000 0.070312 -0.057670 -0.008391 -0.000149 -0.000001 -0.000000 0.078125 -0.086916 -0.015557 -0.000448 -0.000012 -0.000000 0.085938 -0.125619 -0.026983 -0.001175 -0.000066 -0.000000 0.093750 -0.175382 -0.044301 -0.002767 -0.000258 -0.000000 HSP43124
11 0.101562 -0.237843 -0.069457 -0.005963 -0.000815 -0.000000 0.109375 -0.314663 -0.104701 -0.011924 -0.002208 -0.000000 0.117188 -0.407509 -0.152566 -0.022368 -0.005313 -0.000000 0.125000 -0.518045 -0.215834 -0.039695 -0.011613 -0.000000 0.132812 -0.647925 -0.297499 -0.067100 -0.023435 -0.000031 0.140625 -0.798791 -0.400727 -0.108640 -0.044186 -0.000287 0.148438 -0.972266 -0.528809 -0.169262 -0.078552 -0.001468 0.156250 -1.169959 -0.685131 -0.254777 -0.132639 -0.005427 0.164062 -1.393465 -0.873129 -0.371785 -0.214009 -0.016180 0.171875 -1.644372 -1.096269 -0.527552 -0.331613 -0.041152 0.179688 -1.924262 -1.358019 -0.729872 -0.495620 -0.092409 0.187500 -2.234728 -1.661842 -0.986908 -0.717181 -0.187497 0.195312 -2.577375 -2.011181 -1.307047 -1.008144 -0.349593 0.203125 -2.953834 -2.409468 -1.698769 -1.380771 -0.606862 0.210938 -3.365774 -2.860128 -2.170548 -1.847495 -0.991193 0.218750 -3.814917 -3.366593 -2.730783 -2.420719 -1.536664 0.226562 -4.303048 -3.932319 -3.387764 -3.112694 -2.278126 0.234375 -4.832037 -4.560817 -4.149669 -3.935463 -3.250174 0.242188 -5.403856 -5.255675 -5.024594 -4.900864 -4.486639 0.250000 -6.020599 -6.020600 -6.020600 -6.020600 -6.020600 0.257812 -6.684504 -6.859450 -7.145791 -7.306352 -7.884833 0.265625 -7.397981 -7.776287 -8.408404 -8.769932 -10.112627 0.273438 -8.163642 -8.775419 -9.816921 -10.423476 -12.738912 0.281250 -8.984339 -9.861469 -11.380193 -12.279667 -15.801714 0.289062 -9.863195 -11.039433 -13.107586 -14.352002 -19.344007 0.296875 -10.803663 -12.314765 -15.009147 -16.655094 -23.416153 0.304688 -11.809574 -13.693460 -17.095793 -19.205034 -28.079247 0.312500 -12.885208 -15.182171 -19.379534 -22.019831 -33.409992 0.320312 -14.035372 -16.788332 -21.873730 -25.119940 -39.508194 0.328125 -15.265501 -18.520315 -24.593418 -28.528942 -46.509052 0.335938 -16.581776 -20.387625 -27.555685 -32.274414 -54.604954 0.343750 -17.991278 -22.401131 -30.780161 -36.389088 -64.087959 0.351562 -19.502172 -24.573368 -34.289623 -40.912403 -75.444221 0.359375 -21.123947 -26.918915 -38.110786 -45.892738 -89.610390 0.367188 -22.867725 -29.454887 -42.275345 -51.390583 -108.973686 0.375000 -24.746664 -32.201569 -46.821358 -57.483341 -152.503693 0.382812 -26.776485 -35.183285 -51.795181 -64.272881 -153.443375 0.390625 -28.976198 -38.429543 -57.254162 -71.898048 -158.914017 0.398438 -31.369083 -41.976673 -63.270584 -80.556969 -156.960175 0.406250 -33.984089 -45.870125 -69.937607 -90.550629 -153.317627 0.414062 -36.857830 -50.167850 -77.378593 -102.379677 -161.115540 0.421875 -40.037594 -54.945438 -85.762718 -117.007339 -153.504684 table 3. frequency response of halfband filters (continued) normalized frequency halfband #1 halfband #2 halfband #3 halfband #4 halfband #5 HSP43124
12 0.429688 -43.585945 -60.304272 -95.332924 -136.890198 -158.650345 0.437500 -47.588165 -66.385063 -106.462181 -185.130432 -154.637756 0.445312 -52.164894 -73.392075 -119.793030 -187.297241 -153.870453 0.453125 -57.495132 -81.640152 -136.802948 -182.300125 -161.882385 0.460938 -63.861992 -91.658478 -175.030167 -203.460876 -152.278915 0.468750 -71.755898 -104.468010 -158.939362 -174.691895 -164.329758 0.476562 -82.156616 -122.641861 -157.095886 -174.737076 -153.535690 0.484375 -97.627930 -166.537369 -155.613434 -175.108841 -153.507477 0.492188 -139.751450 -165.699081 -154.708450 -169.966568 -167.665482 table 4. halfband filter coefficients (32 bits, un-normalized) coefficient halfband #1 halfband #2 halfband #3 halfband #4 halfband #5 c0 -67230275 12724188 624169 -197705 23964 c1 00000 c2 604101076 -105279784 -6983862 2303514 -242570 c3 1073741823 0 0 0 0 c4 604101076 629426509 38140187 -13225905 1306852 c5 0 1073741827 0 0 0 c6 -67230275 629426509 -145867861 51077176 -4942818 c7 0000 c8 -105279784 650958284 -161054660 14717750 c9 0 1073741793 0 0 c10 12724188 650958284 657968488 -37027884 c11 0 1073741825 0 c12 -145867861 657968488 84032070 c13 000 c14 38140187 -161054660 -191585682 c15 000 c16 -6983862 51077176 670589251 c17 0 0 1073741824 c18 624169 -13225905 670589251 c19 00 c20 2303514 -191585682 c21 00 c22 -197705 84032070 c23 0 c24 -37027884 c25 0 c26 14717750 c27 0 c28 -4942818 c29 0 table 3. frequency response of halfband filters (continued) normalized frequency halfband #1 halfband #2 halfband #3 halfband #4 halfband #5 HSP43124
13 serial output formatter the output formatter serializes the parallel output of the filter compute engine and generates the timing and synchronization signals required to support a serial interface. the formatter produces serial data words with programmable lengths from 8 to 32 bits. the data words may be organized with either most or least significant bit first. also, the data word may be rounded or truncated to the desired length and the format of the output data may be specified as either two's complement or offset binary. to simplify applications where the serial i/o filter is used as a down converter, the output formatter can be configured to scale the output by a factor of 2. the above options are programmed via the output format and output timing registers detailed in table 1. the HSP43124 outputs a bit stream through dout which is synchronous to a programmable clock signal output on clkout. the output clock, clkout, is derived from fclk and has a programmable rate from 1 to 1 / 32 times fclk. the duty cycle of clkout is 50% for rates that have an even number of fclks per clkout. for rates that have and odd number of fclks per clkout the high portion of the clkout waveform spans (n+1)/2 fclks and the low portion spans (n-1)/2 fclks where n is the number of fclks. external devices synchronize to the beginning of an output data word by monitoring syncout. this output is asserted ?high? one clkout prior to the first bit of the next data word as shown in figure 11. input and output data formats the data formats for the input, output and coefficients are fractional two?s complement. the bit weightings in the data words are given in figure 12. input or output data words programmed to have less than 24 bits, map to the most significant bit positions of the 24-bit word. for example, an input word defined to be 8 bits wide would map to the bit positions with weightings from -2 0 to 2 -7 . c30 1306852 c31 0 c32 -242570 c33 0 c34 23964 table 4. halfband filter coefficients (32 bits, un-normalized) (continued) coefficient halfband #1 halfband #2 halfband #3 halfband #4 halfband #5 table 5. performance envelope parameters number of halfbands hb clks dec hb 001 1132 2334 3698 4 125 16 5 221 32 lsb msb lsb msb clkout dout syncout note: assumes data is being output lsb first. figure 11. serial output timing sync leads data fractional two?s complement format for 24-bit input and output 242322212019181716151413121110987654321 -2 0 . 2 -1 2 -2 2 -3 2 -4 2 -5 2 -6 2 -7 2 -8 2 -9 2 -10 2 -11 2 -12 2 -13 2 -14 2 -15 2 -16 2 -17 2 -18 2 -19 2 -20 2 -21 2 -22 2 -23 fractional two?s complement format for 32-bit coefficients 3231302928272625242322212019181716151413121110987654321 -2 0 . 2 -1 2 -2 2 -3 2 -4 2 -5 2 -6 2 -7 2 -8 2 -9 2 -10 2 -11 2 -12 2 -13 2 -14 2 -15 2 -16 2 -17 2 -18 2 -19 2 -20 2 -21 2 -22 2 -23 2 -24 2 -25 2 -26 2 -27 2 -28 2 -29 2 -30 2 -31 note: the negative sign on the msb implies 2?s complement formt. figure 12. data formats HSP43124
14 fclk/sclk uncertainty region figure 13 shows a clocking relationship for the HSP43124 serial i/o filter that could result in an uncertainty at the output. for simplicity, the frequency of fclk and sclk are assumed to be equal to each other, and clkout is assumed to be equal to flck. when the rising edge of fclk lags behind the rising edge of sclk by a small amount of time (t scfc ), then the fclk edge on which samples are read into the filter compute engine is determined by a race condition. in order to insure proper function for the HSP43124, t scfc must be greater than 3.8ns. if exact timing (a particular clock edge for a specific data bit) then make sclk and fclk synchronous. if fclk and sclk are asynchronous, there will be jitter (a specific data bit will be output as 1 of 2 possible clock edges depending on the fclk to sclk phasing). for multiple part applications, use synchronous clocks or use separate syncs on what receives each data, as the outputs may vary by a clock cycle. asynchronous fclk and sclk if fclk and sclk are asynchronous clocks, then the output sample rate (tracked by syncout) of the HSP43124 might jitter in a real time system. this jitter will be demonstrated using an sclk with a period that is 3/2 times the period of fclk (i.e., f fclk /f sclk = 3/2), as shown in figure 14a and figure 14b. if the lsb occurs when there are two fclk edges in one sclk period (see figure 14a), then a null data bit will occur in the dout data stream. if the lsb occurs when there is one fclk edge in one sclk period for the lsb (see figure 14b), then no null data bit will occur. given the 3/2 period relationship between fclk and sclk, the user can see that the syncout jitters by one clock. for example, if the output data is represent by 16 bits, then the number of clkout rising edges between syncout pulses should jitter between 15 and 16. the syncout jitter demonstrated by the 3/2 frequency example can be generalized to other asynchronous f fclk /f sclk ratios. setting the frequencies for fclk and sclk at integer multiples of one another eliminates timing jitter in the output sample rate. sclk syncin din fclk syncout dout t scfc msb msb lsb lsb msb msb figure 13. fclk/sclk uncertainty region figure 14a. number of clkout = number of bits + 1 for the time period between syncouts where f fclk /f sclk = 3/2 figure 14b. number of clkout = number of bits for the time period between syncouts where f fclk /f sclk = 3/2 sclk syncin din clkout syncout dout msb msb lsb null msb lsb msb 1 2 1 2 sclk syncin din clkout syncout dout msb msb msb msb lsb lsb 1 2 1 2 HSP43124
15 absolute maximum ratings thermal information supply voltage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . +7.0v input, output voltage . . . . . . . . . . . . . . . . .gnd -0.5v to v cc +0.5v esd rating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . class 1 operating conditions voltage range (commercial). . . . . . . . . . . . . . . . . . . 4.75v to 5.25v voltage range (industrial) . . . . . . . . . . . . . . . . . . . . . 4.75v to 5.25v temperature range (commercial) . . . . . . . . . . . . . . . . 0 o c to 70 o c temperature range (industrial). . . . . . . . . . . . . . . . . -40 o c to 85 o c thermal resistance (typical, note 1) ja ( o c/w) soic package. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 pdip package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 maximum junction temperature . . . . . . . . . . . . . . . . . . . . . . 150 o c maximum storage temperature . . . . . . . . . . . . . . . . -65 o c to 150 o c maximum lead temperature (soldering 10s) . . . . . . . . . . . . 300 o c (soic - lead tips only) die characteristics gate count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40,304 caution: stresses above those listed in ?absolute maximum ratings? may cause permanent damage to the device. this is a stress o nly rating and operation of the device at these or any other conditions above those indicated in the operational sections of this specification is not implied. note: 1. ja is measured with the component mounted on an evaluation pc board in free air. dc electrical specifications v cc = 5.0v + 5%, t a = 0 o to 70 o c commercial, t a = -40 o to 85 o c industrial parameter symbol test conditions min max units power supply current i ccop v cc = max, fclk = sclk = 45mhz notes 2, 3 -203ma standby power supply current i ccsb v cc = max, outputs not loaded - 500 a input leakage current i i v cc = max, input = 0v or v cc -10 10 a output leakage current i o v cc = max, input = 0v or v cc -10 10 a clock input high v ihc v cc = max, fclk and sclk 3.0 - v clock input low v ilc v cc = min, fclk and sclk - 0.8 v logical one input voltage v ih v cc = max 2.0 - v logical zero input voltage v il v cc = min - 0.8 v logical one output voltage v oh i oh = -5ma, v cc = min 2.6 - v logical zero output voltage v ol i ol = 5ma, v cc = min - 0.4 v input capacitance c in fclk = sclk = 1mhz all measurements referenced to gnd. t a = 25 o c, note 4 -10pf output capacitance c out -10pf notes: 2. power supply current is proportional to frequency. typical rating is 4.5ma/mhz. 3. output load per test circuit and c l = 40pf. 4. not tested, but characterized at initial design and at major process/design changes. HSP43124
16 ac test load circuit ac electrical specifications (note 5) v cc = +4.75v to +5.25v, t a = 0 o c to 70 o c (commercial) v cc = +4.75v to +5.25v, t a = -40 o c to 85 o c (industrial) parameter symbol notes 45mhz 40mhz 33mhz units minmaxminmaxminmax fclk, sclk period t cp 22 - 25 - 30 - ns fclk, sclk high t ch 8-10-12-ns fclk, sclk low t cl 8-10-12-ns setup time din, mxin, syncin, syncmx to sclk t ds 8-8-9-ns hold time din, mxin, syncin, syncmx from sclk t dh 0-0-0-ns setup time fsync to fclk t ss 8-8-8-ns hold time fsync from fclk t sh 0-0-0-ns setup time c0-7, a0-2 to falling edge of wr t ws 10 - 10 - 10 - ns hold time c0-7, a0-2 from falling edge of wr t wh 3-3-3-ns setup time a0-2 to falling edge of rd t rs 10 - 10 - 10 - ns hold time a0-2 from rising edge of rd t rh 0-0-0-ns wr high t wrh 10 - 10 - 12 - ns wr low t wrl 10 - 10 - 12 - ns rd high t rdh 10 - 10 - 10 - ns rd low to data valid t rdo -25-25-25ns rd high to output disable t od -6-6-6ns fclk to clkout t foc -12-13-14ns clkout to syncout, dout t do -8-9-10ns output rise, fall time t rf note 6 -3-3-3ns notes: 5. ac tests performed with c l = 40pf, i ol = 5ma, and i oh = -5ma. input reference level for fclk and sclk is 2.0v, all other inputs 1.5v. test v ih = 3.0v, v ihc = 4.0v, v il = 0v. 6. controlled via design or process parameters and not directly tested. characterized upon initial design and after major proces s and/or changes. equivalent circuit c l (note) i oh 1.5v i ol dut switch s1 open for i ccsb and i ccop s 1 note: test head capacitance. HSP43124
17 all intersil u.s. products are manufactured, assembled and tested utilizing iso9000 quality systems. intersil corporation?s quality certifications can be viewed at www.intersil.com/design/quality intersil products are sold by description only. intersil corporation reserves the right to make changes in circuit design, soft ware and/or specifications at any time without notice. accordingly, the reader is cautioned to verify that data sheets are current before placing orders. information furnishe d by intersil is believed to be accurate and reliable. however, no responsibility is assumed by intersil or its subsidiaries for its use; nor for any infringements of paten ts or other rights of third parties which may result from its use. no license is granted by implication or otherwise under any patent or patent rights of intersil or its subsidiari es. for information regarding intersil corporation and its products, see www.intersil.com waveforms figure 15. timing relative to wr figure 16. output rise and fall times figure 17. input data timing figure 18. timing relative to read figure 19. timing relative to flck and clkout wr c0-7, t ws t wh t wrh a0-2 t wrl t rf t rf 0.8v 2.0v sclk din, mxin, t ds t dh syncin, syncmx t cl t ch t cp rd a0-2 t rs t rdo c0-7 t od t rh t rdh clkout syncout dout t do fclk t ss t sh fsync t cp t cl t foc t ch HSP43124

▲Up To Search▲

Price & Availability of HSP43124

	To Download HSP43124 Datasheet File
If you can't view the Datasheet, Please click here to try to view without PDF Reader .