w4kek/WSJT-X
1
0
Fork 0
mirror of https://github.com/saitohirga/WSJT-X.git synced 2025-10-31 13:10:19 -04:00
Code Issues Packages Projects Releases Wiki Activity

Fix the nagging 'KA1R' problem with decoding after change in TRperiod.

Browse Source
This commit is contained in:
Joe Taylor 2020-07-11 15:24:21 -04:00
parent f77b6bf71a
commit 3e61688229
1 changed files with 74 additions and 72 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

146
lib/fst240_decode.f90
View File

@ -46,8 +46,6 @@ contains
complex, allocatable :: c2(:)
complex, allocatable :: cframe(:)
complex, allocatable :: c_bigfft(:) !Complex waveform
real, allocatable, target :: r_data(:)
complex, pointer, dimension(:) :: c_data_ptr
real llr(240),llra(240),llrb(240),llrc(240),llrd(240)
real candidates(100,4)
real bitmetrics(320,4)
@ -121,7 +119,7 @@ contains
hiscall0=''
first=.false.
endif
l1=index(mycall,char(0))
if(l1.ne.0) mycall(l1:)=" "
l1=index(hiscall,char(0))
@ -213,14 +211,10 @@ contains
nfft1=nfft2*ndown
nh1=nfft1/2
allocate( r_data(1:nfft1+2) )
call c_f_pointer (c_loc (r_data), c_data_ptr, [(nfft1+2)/2]) ! c_data_ptr shares memory with r_data
allocate( c_bigfft(0:nfft1/2) )
allocate( c_bigfft(0:nfft1/2+1) )
allocate( c2(0:nfft2-1) )
allocate( cframe(0:160*nss-1) )
if(ndepth.eq.3) then
nblock=4
if(hmod.eq.1) nblock=4 ! number of block sizes to try
@ -239,13 +233,11 @@ contains
! The big fft is done once and is used for calculating the smoothed spectrum
! and also for downconverting/downsampling each candidate.
r_data(1:nfft1)=iwave(1:nfft1)
r_data(nfft1+1:nfft1+2)=0.0
call four2a(c_data_ptr,nfft1,1,-1,0)
c_bigfft=cmplx(r_data(1:nfft1+2:2),r_data(2:nfft1+2:2))
! write(*,3001) iwspr,nfa,nfb,nfsplit,ndepth
!3001 format('a',5i5)
! iwspr=1 !### For hardwired tests ###
do i=1,nfft1/2
c_bigfft(i)=cmplx(float(iwave(2*i-1)),float(iwave(2*i)))
enddo
c_bigfft(nfft1/2+1)=0.
call four2a(c_bigfft,nfft1,1,-1,0)
if(iwspr.eq.0) then
itype1=1
itype2=1
@ -295,10 +287,9 @@ contains
minsync=1.5
endif
! Get first approximation of candidate frequencies
call get_candidates_fst240(c_bigfft,nfft1,nsps,hmod,fs,fa,fb, &
minsync,ncand,candidates,base)
minsync,ncand,candidates,base)
ndecodes=0
decodes=' '
@ -317,41 +308,49 @@ contains
call fst240_downsample(c_bigfft,nfft1,ndown,fc0,sigbw,c2)
call timer('sync240 ',0)
do isync=0,1
if(isync.eq.0) then
fc1=0.0
if(emedelay.lt.0.1) then ! search offsets from 0 s to 2 s
is0=1.5*nspsec
ishw=1.5*nspsec
else ! search plus or minus 1.5 s centered on emedelay
is0=nint(emedelay*nspsec)
ishw=1.5*nspsec
endif
isst=4*hmod
ifhw=12
df=.1*baud
else if(isync.eq.1) then
fc1=fc2
is0=isbest
ishw=4*hmod
isst=1*hmod
ifhw=7
df=.02*baud
endif
smax=0.0
do if=-ifhw,ifhw
fc=fc1+df*if
do istart=max(1,is0-ishw),is0+ishw,isst
call sync_fst240(c2,istart,fc,hmod,nsyncoh,nfft2,nss,fs2,sync)
if(sync.gt.smax) then
fc2=fc
isbest=istart
smax=sync
endif
enddo
fc1=0.0
if(emedelay.lt.0.1) then ! search offsets from 0 s to 2 s
is0=1.5*nspsec
ishw=1.5*nspsec
else ! search plus or minus 1.5 s centered on emedelay
is0=nint(emedelay*nspsec)
ishw=1.5*nspsec
endif
smax=-1.e30
do if=-12,12
fc=fc1 + 0.1*baud*if
do istart=max(1,is0-ishw),is0+ishw,4*hmod
call sync_fst240(c2,istart,fc,hmod,nsyncoh,nfft2,nss, &
ntrperiod,fs2,sync)
if(sync.gt.smax) then
fc2=fc
isbest=istart
smax=sync
endif
enddo
enddo
fc1=fc2
is0=isbest
ishw=4*hmod
isst=1*hmod
smax=0.0
do if=-7,7
fc=fc1 + 0.02*baud*if
do istart=max(1,is0-ishw),is0+ishw,isst
call sync_fst240(c2,istart,fc,hmod,nsyncoh,nfft2,nss, &
ntrperiod,fs2,sync)
if(sync.gt.smax) then
fc2=fc
isbest=istart
smax=sync
endif
enddo
enddo
call timer('sync240 ',1)
fc_synced = fc0 + fc2
@ -579,29 +578,29 @@ contains
return
end subroutine decode
subroutine sync_fst240(cd0,i0,f0,hmod,ncoh,np,nss,fs,sync)
subroutine sync_fst240(cd0,i0,f0,hmod,ncoh,np,nss,ntr,fs,sync)
! Compute sync power for a complex, downsampled FST240 signal.
use timer_module, only: timer
include 'fst240/fst240_params.f90'
complex cd0(0:np-1)
complex, allocatable, save :: csync1(:),csync2(:)
complex, allocatable, save :: csynct1(:),csynct2(:)
complex ctwk(8*nss)
complex csync1,csync2,csynct1,csynct2
complex ctwk(3200)
complex z1,z2,z3,z4,z5
logical first
integer hmod,isyncword1(0:7),isyncword2(0:7)
real f0save
common/sync240com/csync1(3200),csync2(3200),csynct1(3200),csynct2(3200)
data isyncword1/0,1,3,2,1,0,2,3/
data isyncword2/2,3,1,0,3,2,0,1/
data first/.true./,f0save/-99.9/,nss0/-1/
save first,twopi,dt,fac,f0save,nss0
data f0save/-99.9/,nss0/-1/,ntr0/-1/
save twopi,dt,fac,f0save,nss0,ntr0
p(z1)=(real(z1*fac)**2 + aimag(z1*fac)**2)**0.5 !Compute power
if(nss.ne.nss0 .and. allocated(csync1)) deallocate(csync1,csync2,csynct1,csynct2)
if(first .or. nss.ne.nss0) then
allocate( csync1(8*nss), csync2(8*nss) )
allocate( csynct1(8*nss), csynct2(8*nss) )
nz=8*nss
call timer('sync240a',0)
if(nss.ne.nss0 .or. ntr.ne.ntr0) then
twopi=8.0*atan(1.0)
dt=1/fs
k=1
@ -618,22 +617,25 @@ contains
k=k+1
enddo
enddo
first=.false.
nss0=nss
fac=1.0/(8.0*nss)
nss0=nss
ntr0=ntr
f0save=-1.e30
endif
if(f0.ne.f0save) then
dphi=twopi*f0*dt
phi=0.0
do i=1,8*nss
do i=1,nz
ctwk(i)=cmplx(cos(phi),sin(phi))
phi=mod(phi+dphi,twopi)
enddo
csynct1=ctwk*csync1
csynct2=ctwk*csync2
csynct1(1:nz)=ctwk(1:nz)*csync1(1:nz)
csynct2(1:nz)=ctwk(1:nz)*csync2(1:nz)
f0save=f0
nss0=nss
endif
call timer('sync240a',1)
i1=i0 !Costas arrays
i2=i0+38*nss
@ -662,11 +664,11 @@ contains
if(i5+is+ncoh*nss-1.le.np) then
z5=sum(cd0(i5+is:i5+is+ncoh*nss-1)*conjg(csynct1(is+1:is+ncoh*nss)))
endif
s1=s1+abs(z1)/(8*nss)
s2=s2+abs(z2)/(8*nss)
s3=s3+abs(z3)/(8*nss)
s4=s4+abs(z4)/(8*nss)
s5=s5+abs(z5)/(8*nss)
s1=s1+abs(z1)/nz
s2=s2+abs(z2)/nz
s3=s3+abs(z3)/nz
s4=s4+abs(z4)/nz
s5=s5+abs(z5)/nz
enddo
else
nsub=-ncoh
@ -718,7 +720,7 @@ contains
subroutine get_candidates_fst240(c_bigfft,nfft1,nsps,hmod,fs,fa,fb, &
minsync,ncand,candidates,base)
complex c_bigfft(0:nfft1/2) !Full length FFT of raw data
complex c_bigfft(0:nfft1/2+1) !Full length FFT of raw data
integer hmod !Modulation index (submode)
integer im(1) !For maxloc
real candidates(100,4) !Candidate list