Tweak decoder for better decodes under poor conditions
This commit is contained in:
@@ -5,15 +5,23 @@ subroutine ft8_downsample(dd,newdat,f0,c1)
|
||||
parameter (NMAX=15*12000,NSPS=1920)
|
||||
parameter (NFFT1=192000,NFFT2=3200) !192000/60 = 3200
|
||||
|
||||
logical newdat
|
||||
logical newdat,first
|
||||
complex c1(0:NFFT2-1)
|
||||
complex cx(0:NFFT1/2)
|
||||
real dd(NMAX),x(NFFT1)
|
||||
real dd(NMAX),x(NFFT1),taper(0:100)
|
||||
equivalence (x,cx)
|
||||
save cx
|
||||
data first/.true./
|
||||
save cx,first,taper
|
||||
|
||||
if(first) then
|
||||
pi=4.0*atan(1.0)
|
||||
do i=0,100
|
||||
taper(i)=0.5*(1.0+cos(i*pi/100))
|
||||
enddo
|
||||
first=.false.
|
||||
endif
|
||||
if(newdat) then
|
||||
! Data in dd have changed, recompute the long FFT
|
||||
! 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
|
||||
@@ -23,9 +31,9 @@ subroutine ft8_downsample(dd,newdat,f0,c1)
|
||||
df=12000.0/NFFT1
|
||||
baud=12000.0/NSPS
|
||||
i0=nint(f0/df)
|
||||
ft=f0+8.0*baud
|
||||
ft=f0+8.5*baud
|
||||
it=min(nint(ft/df),NFFT1/2)
|
||||
fb=f0-1.0*baud
|
||||
fb=f0-1.5*baud
|
||||
ib=max(1,nint(fb/df))
|
||||
k=0
|
||||
c1=0.
|
||||
@@ -33,6 +41,8 @@ subroutine ft8_downsample(dd,newdat,f0,c1)
|
||||
c1(k)=cx(i)
|
||||
k=k+1
|
||||
enddo
|
||||
c1(0:100)=c1(0:100)*taper(100:0:-1)
|
||||
c1(k-1-100:k-1)=c1(k-1-100:k-1)*taper
|
||||
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)
|
||||
|
||||
Reference in New Issue
Block a user