kopia lustrzana https://github.com/roncarr880/QRP_LABS_WSPR
Freq calc changes
Trying a different way to generate the 1.46 hz frequency shifts. Previous way worked but perhaps shifts were not 1.46 hz.master
roncarr880
GitHub
rodzic
43d06e0735
commit
9b6761f17f
|
|
@ -24,6 +24,14 @@
|
|||
// R dividers in the SI5351. Dividers 1 - rx and 4 - tx will cover 1mhz to 30mhz
|
||||
// Dividers 16 - rx and 64 - tx will cover 40 khz to 2 mhz
|
||||
|
||||
// When using even output divider, PLL c parameter...
|
||||
// For 1 hz resolution, c = CLOCK / divider.
|
||||
// For 1.46 hz resolution, c = CLOCK / ( 1.46 * divider ).
|
||||
// Can then send WSPR by just adding to the b parameter 0,1,2,or3.
|
||||
// Smallest divider for 1 hz resolution, about 26.
|
||||
// Max value of c is 1048575
|
||||
// Use of the R dividers will complicate this.
|
||||
|
||||
#include <FreqCount.h> // one UNO I have does not work correctly with this library, another one does
|
||||
#include <EEPROM.h>
|
||||
#include <OLED1306_Basic.h>
|
||||
|
|
@ -53,7 +61,7 @@
|
|||
#define CAT_MODE 0 // computer control of TX
|
||||
#define FRAME_MODE 1 // self timed frame (stand alone mode)
|
||||
#define MUTE A1 // receiver module T/R switch pin
|
||||
#define START_CLOCK_FREQ 2700449800LL // *100 for setting fractional frequency ( 446600 )
|
||||
#define START_CLOCK_FREQ 27004498L // ( 446600 )
|
||||
//#define START_CLOCK_FREQ 2700466600 // test too high
|
||||
//#define START_CLOCK_FREQ 2700426600 // test too low
|
||||
|
||||
|
|
@ -80,8 +88,8 @@ char val_print = ' ';
|
|||
// we are using the post dividers and can receive down to 40khz
|
||||
// vco 600 to 900
|
||||
|
||||
uint64_t clock_freq = START_CLOCK_FREQ;
|
||||
uint64_t drift; // calibrate freq result used as a measure of temperature and mapped to correct
|
||||
uint32_t clock_freq = START_CLOCK_FREQ;
|
||||
uint32_t drift; // calibrate freq result used as a measure of temperature and mapped to correct
|
||||
// 27 master clock drift
|
||||
uint32_t freq = FREQ; // ssb vfo freq
|
||||
const uint32_t cal_freq = 3000000; // calibrate frequency
|
||||
|
|
@ -620,12 +628,12 @@ void clock_correction( int8_t val ){ // time keeping only, change the fudge f
|
|||
}
|
||||
|
||||
void temp_correction( ){ // short term drift correction with a linear map
|
||||
uint64_t local_drift; // corrects for drift due to day/night temperature changes
|
||||
uint32_t local_drift; // corrects for drift due to day/night temperature changes
|
||||
|
||||
if( wspr_tx_enable ) return; // ignore this when transmitting
|
||||
|
||||
local_drift = map( cal_result, 2999900, 3000000, 2000, 0 );
|
||||
|
||||
local_drift = map( cal_result, 2999900, 3000000, 20, 0 );
|
||||
|
||||
if( local_drift != drift ){
|
||||
drift = local_drift;
|
||||
si_pll_x(PLLB,cal_freq,cal_divider,0);
|
||||
|
|
@ -635,9 +643,11 @@ uint64_t local_drift; // corrects for drift due to day/night temperature c
|
|||
|
||||
|
||||
// correct errors on bits that do not change often
|
||||
// this version moves the counters up to a hard limit and moves down on different bit decoded
|
||||
// decodes from trends only for 1 missed bit
|
||||
// this version moves the counters up to a hard limit and moves down on different bit decoded
|
||||
// slips in time on weak signal as is called from frame_sync
|
||||
// this version does not use bit_errors but instead requires trends to be at the limit in order to decode
|
||||
// less decodes, less false decodes
|
||||
// bump the trends type for each minute to be what is expected
|
||||
char wwvb_trends( char val, uint8_t dat ){
|
||||
static int i = 60;
|
||||
static int unslip;
|
||||
|
|
@ -704,17 +714,17 @@ static int s_count; // counts from last sync - detect when spaced 9 apart. M
|
|||
// if( decodes == 0 && ( count >= LIMIT/2 || trend_t == SYNC || i == 8 ) ) part_decode(i);
|
||||
|
||||
// return history on errors with only 1 bit incorrect
|
||||
if( new_t == ERR && count == LIMIT ){
|
||||
if( trend_t == ZEROS ){
|
||||
if( bit_errors(dat,0xfc,i) < 2 ) val = 'o';
|
||||
}
|
||||
if( trend_t == ONES ){
|
||||
if( bit_errors(dat,0xf0,i) < 2 ) val = 'i';
|
||||
}
|
||||
if( trend_t == SYNC ){
|
||||
if( bit_errors(dat,0xc0,i) < 2 ) val = 's';
|
||||
}
|
||||
}
|
||||
// if( new_t == ERR && count == LIMIT ){
|
||||
// if( trend_t == ZEROS ){
|
||||
// if( bit_errors(dat,0xfc,i) < 2 ) val = 'o';
|
||||
// }
|
||||
// if( trend_t == ONES ){
|
||||
// if( bit_errors(dat,0xf0,i) < 2 ) val = 'i';
|
||||
// }
|
||||
// if( trend_t == SYNC ){
|
||||
// if( bit_errors(dat,0xc0,i) < 2 ) val = 's';
|
||||
// }
|
||||
// }
|
||||
|
||||
|
||||
if( i == 0 && val == '.' ) val = 'Z'; // view index on no decode no history
|
||||
|
|
@ -735,6 +745,7 @@ static int s_count; // counts from last sync - detect when spaced 9 apart. M
|
|||
}
|
||||
|
||||
|
||||
/*
|
||||
// return the number of bits different between mask and value
|
||||
int bit_errors( uint8_t val, uint8_t mask, int i ){
|
||||
int count;
|
||||
|
|
@ -763,7 +774,7 @@ uint8_t b;
|
|||
return count;
|
||||
|
||||
}
|
||||
|
||||
*/
|
||||
|
||||
/*
|
||||
// correct errors on bits that do not change often
|
||||
|
|
@ -1013,7 +1024,7 @@ static unsigned int one_second;
|
|||
}
|
||||
|
||||
// set the frequency
|
||||
si_pll_x(PLLA,Rdiv*4*(freq+audio_freq),divider,Rdiv*4*146*wspr_msg[i]);
|
||||
si_pll_x(PLLA,Rdiv*4*(freq+audio_freq),divider,Rdiv*4*wspr_msg[i]);
|
||||
if( i == 0 ) tx_on();
|
||||
++i;
|
||||
}
|
||||
|
|
@ -1051,31 +1062,39 @@ void i2cd( unsigned char addr, unsigned char reg, unsigned char dat ){
|
|||
/****** SI5351 functions ******/
|
||||
|
||||
void si_pll_x(unsigned char pll, uint32_t freq, uint32_t out_divider, uint32_t fraction ){
|
||||
uint64_t a,b,c;
|
||||
uint64_t bc128; // floor 128 * b/c term of equations
|
||||
uint64_t pll_freq;
|
||||
uint64_t cl_freq;
|
||||
uint32_t a,b,c;
|
||||
uint32_t bc128; // floor 128 * b/c term of equations
|
||||
uint32_t pll_freq;
|
||||
uint32_t cl_freq;
|
||||
|
||||
uint32_t P1; // PLL config register P1
|
||||
uint32_t P2; // PLL config register P2
|
||||
uint32_t P3; // PLL config register P3
|
||||
uint64_t r;
|
||||
|
||||
uint32_t r;
|
||||
uint32_t f; // fudge freq and fraction if remainder is near max ( c )
|
||||
|
||||
cl_freq = clock_freq;
|
||||
|
||||
//if( pll == PLLA ) cl_freq += drift; // drift not applied to 3 mhz calibrate freq
|
||||
cl_freq += drift; // drift applied to 3 mhz. Pick one of these.
|
||||
|
||||
c = 1000000; // max 1048575
|
||||
pll_freq = 100ULL * (uint64_t)freq + fraction; // allow fractional frequency for wspr
|
||||
pll_freq = pll_freq * out_divider;
|
||||
if( pll == PLLB ) c = 1000000; // max 1048575, cal freq pll
|
||||
else{
|
||||
c = (float)(cl_freq) / ( 1.46 * (float)(out_divider));
|
||||
while( c > 1048575 ) c /= 2, fraction /= 2; // !!! fraction /2 or *2 ?
|
||||
}
|
||||
// pll_freq = 100ULL * (uint32_t)freq + fraction;
|
||||
pll_freq = freq * out_divider;
|
||||
a = pll_freq / cl_freq ;
|
||||
r = pll_freq - a * cl_freq ;
|
||||
b = ( c * r ) / cl_freq;
|
||||
b = ( (uint64_t)c * (uint64_t)r ) / (uint64_t)cl_freq;
|
||||
f = Rdiv*4*3 + 1;
|
||||
if( b + f >= c ) b -= f; // fudge b if go over max b value on wspr steps, tx off desired freq by 5 hz
|
||||
bc128 = (128 * r)/ cl_freq;
|
||||
b += fraction; // wspr offset
|
||||
P1 = 128 * a + bc128 - 512;
|
||||
P2 = 128 * b - c * bc128;
|
||||
if( P2 > c ) P2 = 0; // avoid negative numbers
|
||||
//if( P2 > c ) P2 = 0; // avoid negative numbers
|
||||
P3 = c;
|
||||
|
||||
i2cd(SI5351, pll + 0, (P3 & 0x0000FF00) >> 8);
|
||||
|
|
@ -1678,7 +1697,7 @@ static uint8_t err, vp, earl, lt;
|
|||
case 4:
|
||||
Serial.print(" FF "); Serial.print(FF);
|
||||
Serial.write(' '); Serial.print(ff);
|
||||
Serial.print(" Drift "); Serial.print((int)drift/100);
|
||||
Serial.print(" Drift "); Serial.print((int)drift);
|
||||
++dbug_print_state;
|
||||
break;
|
||||
case 5:
|
||||
|
|
|
|||
Ładowanie…
Reference in New Issue