mirror of
https://github.com/pavel-demin/ft8d.git
synced 2024-11-14 16:11:48 -05:00
43 lines
972 B
Fortran
43 lines
972 B
Fortran
subroutine ft8_downsample(dd,newdat,f0,c1)
|
|
|
|
! Downconvert to complex data sampled at 200 Hz ==> 32 samples/symbol
|
|
|
|
parameter (NMAX=15*12000,NSPS=1920)
|
|
parameter (NFFT1=192000,NFFT2=3200) !192000/60 = 3200
|
|
|
|
logical newdat
|
|
complex c1(0:NFFT2-1)
|
|
complex cx(0:NFFT1/2)
|
|
real dd(NMAX),x(NFFT1)
|
|
equivalence (x,cx)
|
|
save cx
|
|
|
|
if(newdat) then
|
|
! Data in dd have changed, recompute the long FFT
|
|
x(1:NMAX)=dd
|
|
x(NMAX+1:NFFT1)=0. !Zero-pad the x array
|
|
call four2a(cx,NFFT1,1,-1,0) !r2c FFT to freq domain
|
|
newdat=.false.
|
|
endif
|
|
|
|
df=12000.0/NFFT1
|
|
baud=12000.0/NSPS
|
|
i0=nint(f0/df)
|
|
ft=f0+8.0*baud
|
|
it=min(nint(ft/df),NFFT1/2)
|
|
fb=f0-1.0*baud
|
|
ib=max(1,nint(fb/df))
|
|
k=0
|
|
c1=0.
|
|
do i=ib,it
|
|
c1(k)=cx(i)
|
|
k=k+1
|
|
enddo
|
|
c1=cshift(c1,i0-ib)
|
|
call four2a(c1,NFFT2,1,1,1) !c2c FFT back to time domain
|
|
fac=1.0/sqrt(float(NFFT1)*NFFT2)
|
|
c1=fac*c1
|
|
|
|
return
|
|
end subroutine ft8_downsample
|