WSJT-X/lib/downsam9.f90
Joe Taylor 6339dc4b7e Many changes to the procedures for selecting decodable candidate JT9 signals.
This code may be the new benchmark?


git-svn-id: svn+ssh://svn.code.sf.net/p/wsjt/wsjt/branches/wsjtx@3274 ab8295b8-cf94-4d9e-aec4-7959e3be5d79
2013-05-16 16:02:00 +00:00

59 lines
1.4 KiB
Fortran

subroutine downsam9(c0,npts8,nsps8,newdat,nspsd,fpk,c2,nz2)
!Downsample to nspsd samples per symbol, info centered at fpk
parameter (NMAX=128*31500)
complex c0(0:npts8-1)
complex c1(0:NMAX-1)
complex c2(0:4096-1)
real s(1000)
save
nfft1=128*nsps8 !Forward FFT length
nh1=nfft1/2
df1=1500.0/nfft1
if(newdat.eq.1) then
fac=1.e-4
do i=0,npts8-1,2
c1(i)=fac*conjg(c0(i))
c1(i+1)=-fac*conjg(c0(i+1))
enddo
c1(npts8:)=0. !Zero the rest of c1
call four2a(c1,nfft1,1,-1,1) !Forward FFT
nadd=1.0/df1
j=250/df1
s=0.
do i=1,1000
do n=1,nadd
j=j+1
s(i)=s(i)+real(c1(j))**2 + aimag(c1(j))**2
enddo
! write(37,3001) i+1000,s(i),db(s(i)),nadd
!3001 format(i5,2f12.3,i8)
enddo
call pctile(s,1000,40,avenoise)
endif
ndown=nsps8/16 !Downsample factor
nfft2=nfft1/ndown !Backward FFT length
nh2=nfft2/2
fshift=fpk-1500.0
i0=nh1 + fshift/df1
fac=sqrt(1.0/avenoise)
do i=0,nfft2-1
j=i0+i
if(i.gt.nh2) j=j-nfft2
c2(i)=fac*c1(j)
enddo
call four2a(c2,nfft2,1,1,1) !Backward FFT
nspsd=nsps8/ndown
nz2=npts8/ndown
return
end subroutine downsam9