QRZ! Ham Radio 1

home *** CD-ROM | disk | FTP | other *** search

/ QRZ! Ham Radio 1 / QRZ Ham Radio Callsign Database - December 1993.iso / ucsd / dsp / wefax.arc / WEFDEMOD.ASM < prev next >

Wrap

Assembly Source File | 1988-02-17 | 14.5 KB | 655 lines

; AMSAT/TAPR DSP WEFAX demodulator ; Sample rate should be set to 10000 samples/sec. ; Code by Bob McGwier N4HY. Copyright 1988, ALL RIGHTS RESERVED! ; ; Needs: post frequency synthesis filtering and adaptive equalization ; to achieve "perfection" :-) org 0 b go org 0FF0H dw 3932 ; 2400Hz at 10000 samples/sec ; ; Okay Jose' set up equates for data memory etc. ; org 10H ; code should begin at memory address 16 cause David ; does something strange with locations 0-15. Does ; he in fact document this? Is it a TMS32010 feature? sine: equ 0 ; will store sine values from tone one: equ 1 ; guess what goes here duhhhhh! freq: equ 2 ; PLL frequency stored here phase: equ 3 ; PLL phase stored here maskl: equ 4 ; mask for fine or low order bits of phase mask: equ 5 ; mask for doing modulo 16384 arithmetic with phase sinx: equ 6 ; used to store coarse sine value (high order bits) cosx: equ 7 ; as above for cosine mone: equ 8 ; minus one stored here wkph: equ 9 ; different phases (PLL, remodulator tone etc) are stored hr ; for frequency synthesis masko: equ 10 ; mask for converting on board PCM to DAC format mps: equ 11 ; multiplier for quadrant determination for sine mpc: equ 12 ; multiplier for quadrant determination for cosine siny: equ 13 ; fine correction value for use in SIN(X+Y) = sinx*cosy+ cosy: equ 14 ; cosx*siny cosine: equ 15 ; cosine is also needed for Q arm in Costas loop coph: equ 16 ; working number holder modem: equ 17 ; Used to choose between complex tone for arms and real tone ; in the modulator as explained below contr: equ 18 freqo: equ 20 ; freqo is assigned one of the values above for remod phaseo: equ 71 ; phaseo is the phase of the remodulator output. phase is ; as continuous as DSP will allow on this board xn0: equ 21 ; place for storing all the values for the "I" or data arm xn1: equ 22 xn2: equ 23 xn3: equ 24 xn4: equ 25 xn5: equ 26 xn6: equ 27 xn7: equ 28 xn8: equ 29 xn9: equ 30 xn10: equ 31 xn11: equ 32 xn12: equ 33 xn13: equ 34 xn14: equ 35 xn15: equ 36 xn16: equ 37 xn17: equ 38 xn18: equ 39 xn19: equ 40 xn20: equ 41 xn21: equ 42 xn22: equ 43 xn23: equ 44 yn0: equ 45 ; storage for values in "Q" or phase error arm. yn1: equ 46 yn2: equ 47 yn3: equ 48 yn4: equ 49 yn5: equ 50 yn6: equ 51 yn7: equ 52 yn8: equ 53 yn9: equ 54 yn10: equ 55 yn11: equ 56 yn12: equ 57 yn13: equ 58 yn14: equ 59 yn15: equ 60 yn16: equ 61 yn17: equ 62 yn18: equ 63 yn19: equ 64 yn20: equ 65 yn21: equ 66 yn22: equ 67 yn23: equ 68 c0: equ 69 ; Phase error maskf: equ 70 ; frequency mask so that frequency doesn't try and leave ; proper range tester: equ 72 ; remodulator output -300 to 300 millivolts on DAC's data: equ 73 h10: equ 74 h11: equ 75 armfilt: dw 6861 dw 9601 sintbl: dw 0 ; coarse sine table in steps of PI/64 radians to PI/2 dw 804 dw 1607 dw 2410 dw 3211 dw 4011 dw 4807 dw 5601 dw 6392 dw 7179 dw 7961 dw 8739 dw 9511 dw 10278 dw 11038 dw 11792 dw 12539 dw 13278 dw 14009 dw 14732 dw 15446 dw 16150 dw 16845 dw 17530 dw 18204 dw 18867 dw 19519 dw 20159 dw 20787 dw 21402 dw 22004 dw 22594 dw 23169 dw 23731 dw 24278 dw 24811 dw 25329 dw 25831 dw 26318 dw 26789 dw 27244 dw 27683 dw 28105 dw 28510 dw 28897 dw 29268 dw 29621 dw 29955 dw 30272 dw 30571 dw 30851 dw 31113 dw 31356 dw 31580 dw 31785 dw 31970 dw 32137 dw 32284 dw 32412 dw 32520 dw 32609 dw 32678 dw 32727 dw 32757 dw 32767 ; PI/2 fines: dw 0 ; fine sine table in steps of PI/(64*64) radians to PI/64 dw 12 dw 25 dw 37 dw 50 dw 62 dw 75 dw 87 dw 100 dw 113 dw 125 dw 138 dw 150 dw 163 dw 175 dw 188 dw 201 dw 213 dw 226 dw 238 dw 251 dw 263 dw 276 dw 289 dw 301 dw 314 dw 326 dw 339 dw 351 dw 364 dw 376 dw 389 dw 402 dw 414 dw 427 dw 439 dw 452 dw 464 dw 477 dw 490 dw 502 dw 515 dw 527 dw 540 dw 552 dw 565 dw 578 dw 590 dw 603 dw 615 dw 628 dw 640 dw 653 dw 665 dw 678 dw 691 dw 703 dw 716 dw 728 dw 741 dw 753 dw 766 dw 779 dw 791 finec: dw 32767 ;ditto to above for fine sine dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32766 dw 32765 dw 32765 dw 32765 dw 32765 dw 32765 dw 32765 dw 32765 dw 32765 dw 32764 dw 32764 dw 32764 dw 32764 dw 32764 dw 32764 dw 32764 dw 32763 dw 32763 dw 32763 dw 32763 dw 32763 dw 32762 dw 32762 dw 32762 dw 32762 dw 32762 dw 32761 dw 32761 dw 32761 dw 32761 dw 32760 dw 32760 dw 32760 dw 32760 dw 32759 dw 32759 dw 32759 dw 32759 dw 32758 dw 32758 dw 32758 dw 32758 dw 32757 dw 32757 go: ldpk 0 ; make sure that we are pointing to 0 data page lack one ; make and store one sacl one lac one,11 ; make and store DAC converter mask sacl masko lac one,13 ; make and store frequency mask sub one sacl maskf zac sub one sacl mone ; make and store minus one lac one,14 sub one sacl mask ; make and store mask for modulo 16384 phase arithmetic lac one,6 sub one sacl maskl ; make and store fine part of address mask; zac sacl phase lac one,8 ; make and store frequency address in memory sub one sacl freq lac freq,4 tblr freq ; read PLL initial frequency larp 0 lark ar0,7 lack armfilt tblr h10 add one tblr h11 ; Okay the BS is over lets get to work ! wait: bioz fire ; is it time for a new sample b wait ; nope go wait in the corner fire: in xn0,pa3 ; get a new sample here ; as the following routines have a variable length lac xn0,4 sub one,15 ; Change ADC format to two's complement sacl xn0 banz ntyt out data,pa7 lac contr add one and one sacl contr dont: out contr,pa5 lark ar0,7 ntyt: lac tester,6 addh masko sach yn0 out yn0,pa4 ; send wefax tone to the speaker ; end output ; Now begin demodulator lack one sacl modem ; store control that makes complex tone lack one ; sacl mps ; Restore sine and cosine quadrant multipliers sacl mpc ; lac phase call tones ; go make complex tone lac sine sacl tester lt xn0 ; load the current sample mpy cosine; multiply by cosine pac sach yn0,1 ; store it Q channel current sample mpy sine ; multiply sample by sine pac sach xn0,1 ; now store it in current sample place for I channel ; Low pass for each sample cutoff is at 400 Hz. >1000Hz is 30+ db down ; Roll off is smooth between 400 and 1000. ; First lowpass I arm product zac ; lt xn23 mpyk -717 ; ltd xn22 mpyk -214 ; ltd xn21 mpyk -337 ; ltd xn20 mpyk -903 ; ltd xn19 mpyk -865 ; ltd xn18 mpyk -57 ; ltd xn17 mpyk -1356 ; ltd xn16 mpyk -1855 ; ltd xn15 mpyk -479 ; ltd xn14 mpyk 2845 ; ltd xn13 mpy h10 ; ltd xn12 mpy h11 ; ltd xn11 mpy h11 ; ltd xn10 mpy h10 ; ltd xn9 mpyk 2845 ; ltd xn8 mpyk -479 ; ltd xn7 mpyk -1855 ; ltd xn6 mpyk -1356 ; ltd xn5 mpyk -57 ; ltd xn4 mpyk 865 ; ltd xn3 mpyk 903 ; ltd xn2 mpyk 337 ; ltd xn1 mpyk -214 ; ltd xn0 mpyk -717 ; apac ; sach xn0,1 ; Now do the Q arm ; zac ; lt yn23 mpyk -717 ; ltd yn22 mpyk -214 ; ltd yn21 mpyk -337 ; ltd yn20 mpyk -903 ; ltd yn19 mpyk -865 ; ltd yn18 mpyk -57 ; ltd yn17 mpyk -1356 ; ltd yn16 mpyk -1855 ; ltd yn15 mpyk -479 ; ltd yn14 mpyk 2845 ; ltd yn13 mpy h10 ; ltd yn12 mpy h11 ; ltd yn11 mpy h11 ; ltd yn10 mpy h10 ; ltd yn9 mpyk 2845 ; ltd yn8 mpyk -479 ; ltd yn7 mpyk -1855 ; ltd yn6 mpyk -1356 ; ltd yn5 mpyk -57 ; ltd yn4 mpyk 865 ; ltd yn3 mpyk 903 ; ltd yn2 mpyk 337 ; ltd yn1 mpyk -214 ; ltd yn0 mpyk -717 ; apac ; sach yn0,1 ; Okay we now have filtered products with data bauding pseudo base banded ; Let's close the loop and lock on lac xn0; Hard limit data line for phase error detector bgez pdps ; is data baud a plus one ? lt yn0 ; get ready for a multiply mpyk -1 ; multiply phase error arm by -1 pac b ny ; go close the loop pdps: lac yn0 ny: sacl c0 ; store sign corrected phase error. ; error updates are computed compute new phase and freq lac c0,13 ; make phase gain 1/8 of phase error sach cosy ; lac cosy ; bgez delph ; is phase error nonnegative? add one,15 ; No so bring it back from -angle to between 0 & 2*PI delph: add phase ; add the old phase add freq ; add the phase increment (frequency) and mask ; modulo 2*pi sacl phase ; store it for next sample complex tone generation step lac xn0 bgz posd lt mone mpy xn0 pac posd: sacl data b wait ; go output and then do it again ; complex tone generator if modem>0 if modem<0 sine wave generator tones: sacl wkph ; store a working copy lac wkph,4 subh one blz getem ; is it in a quadrant bigger than first? subh one bgez thfr; is it in a quadrant greater than two? lac 1,13 ; nope so load pi sub wkph ; subtract phase so that it maps back into 1st quad sacl wkph ; store lac mone ; load -1 sacl mpc ; store it in the cosine multiplier b getem ; go read tables thfr: lac mone ; multiplier for bottom half sacl mps ; store lac wkph ; sub one,13 ; map angle back to upper half and go do it again b tones getem: lac wkph,10 ; take 1st quadrant phase equiv for sine and pick ; off coarse part of phase address sach coph ; store it lack sintbl ; load sine table offset into accumulator add coph ; add coarse address off set tblr sinx ; read the coarse sine value lac one,6 ; load pi/2 sub coph ; subtract the coarse phase to get cosines offset sacl coph lack sintbl add coph ; do same as above for coarse cosine tblr cosx lac wkph ; load working phase and maskl ; mask off fine addres sacl coph ; store it lack fines ; locate fine table offset add coph ; add for offset into fine table tblr siny ; read table lack finec add coph tblr cosy zac lt sinx ; load sinx mpy cosy ; multiply by cosy lta siny ; load t reg with fine sin and accumulate previous prod. mpy cosx ; multiply by coarse cosx to use sin(X+Y) apac ; add the result to coarse sine sach sine ; store it. lac modem blz mult lac 1,12 ; load full address pi/2 sub wkph ; subtract the working phase to get cosine sacl wkph lac wkph,10 ; from here to the later MARK it is identical to above sach coph lack sintbl add coph tblr sinx lac one,6 sub coph sacl coph lack sintbl add coph tblr cosx lac wkph and maskl sacl coph lack fines add coph tblr siny lack finec add coph tblr cosy zac lt cosy mpy sinx lta siny mpy cosx apac ; MARK sach cosine ; store it in the cosine mult: lt mps; now we need to do a few multiplies by sign changes due to mpy sine ; to quadrant part of phase address pac; multiply sine by sine sign (:-) and sacl sine ; store the result lac modem bgz cosm ret cosm: mpy cosine; now multiply cosine by the same pac sacl cosine ; store it lt mpc; load cosine differentiator from sine sign (:-) mpy cosine; multiply pac sacl cosine ; store ret end ; Ebbly Ebbly Ebbly thats all folks