I have done the IsSpurInBand() more resonable, although I don't know if
it matches exactly the flowchart (unsure about the Fifbw, which I can't
find in the code, but it might correspond to the spectrum-parameters).
Patch against 0.7.101:
368,369c368,369
< int n1=1,n2,f;
<
---
> int n1=1,nmax=5,n2,f;
>
385c385
< if( (f>spectrum_from) && (f<spectrum_to))
---
> if( (f>spectrum_from) && (f<spectrum_to)) {
387c387,389
< } while ( (f>(f2-spectrum_to)) || (n2>-5));
---
> return 1; // true
> }
> } while ( (f>(f2-spectrum_to)) || (n2>-nmax));
389c391
< } while (n1<5);
---
> } while (n1<nmax);
391c393
< return 1;
---
> return 0; // false
Whole function:
// IsSpurInBand()?
static int mt2032_spurcheck(int f1, int f2, int spectrum_from,int
spectrum_to)
{
int n1=1,nmax=5,n2,f;
f1=f1/1000; //scale to kHz to avoid 32bit overflows
f2=f2/1000;
spectrum_from/=1000;
spectrum_to/=1000;
dprintk("spurcheck f1=%d f2=%d from=%d
to=%d\n",f1,f2,spectrum_from,spectrum_to);
do {
n2=-n1;
f=n1*(f1-f2);
do {
n2--;
f=f-f2;
dprintk(" spurtest n1=%d n2=%d ftest=%d\n",n1,n2,f);
if( (f>spectrum_from) && (f<spectrum_to)) {
printk("mt2032 spurcheck triggered: %d\n",n1);
return 1; // true
}
} while ( (f>(f2-spectrum_to)) || (n2>-nmax));
n1++;
} while (n1<nmax);
return 0; // false
}
Is this correct now? I might take a look at the function that calls this
too.
On Tue, Jan 07, 2003 at 12:18:24PM +0100, Marius Kotsbak wrote:
> On Tue, Jan 07, 2003 at 10:01:10AM +0100, Peter Bienstman wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > Hi Marius,
> >
> > I have the same card and I too have been looking at that particular piece of
> > code. I also think that's the problem. I even downloaded that data sheet and
> > played with the idea of implementing this.
>
> For radio, it seems like the spurcheck isn't nessecary, but it is
> probably sometimes needed for TV, so this might be the reason why you
> see better picture with pinnacles driver. BTW: I don't think it is so
> much work to implement it, as the algoritm is clearly specified in the
> datasheet.
>
> >
> > BTW, I tried radio under XP, and it doesn't work well either.
>
> It works with my card under w2k. You could try with some 3-party drivers
> / applications.
>
> >
> > Incidentally, did you try watching TV under Windows with this card?
> > Unfortunately, the image quality is a lot better under XP,
>
> I havent't seen much difference, but that could be caused by different
> frequencies in norwegian channels, and of your channels.
>
> > even after trying
> > all sorts of interlacing and filtering options under Linux (mplayer,
> > timefortv,...). Do you have the same experience? Which drivers settings and
> > TV software do you use?
>
> I just user tuner=33, and use zapping most (overlay mode).
>
> >
> > Cheers,
> >
> > Peter
> >
> > On Tuesday 07 January 2003 01:46, Marius Kotsbak wrote:
> > > I have started reviewing the code in tuner.c to find why the radio on
> > > PCTV PRO with MT2030-tuner isn't working.
> > >
> > > First I need to know if it has worked well before?
> > >
> > > I have found one possible error source, and that is the spurcheck, which
> > > is as I can see missing. The mt2032_spurcheck() just print some
> > > debug-information, but never touch other than local variables, don't
> > > talk to any hardware, and always return 1. Where it is called, it is
> > > commented "should recalcuulate lo1...".
> > >
> > > What I am not sure of is if this is necessary for radio tuning. In the
> > > datasheet it is noted that the spur removal isn't required for north
> > > Ameracan FM, which seems to have the same range as the norwegian (which
> > > doesn't work well).
> > >
> > > Does anyone (particulary the driver authors) know any other part of the
> > > code I can search for errors? The symtoms as I have said is that the
> > > tuning of radio is bad, som I can only hear sound some seconds before it
> > > turns the sound off. Sound might come back later automatically and
> > > last for some seconds.
> > >
> > > Marius K
> >
> > - --
> > - ------------------------------------------------
> > Peter Bienstman
> > Ghent University, Dep. of Information Technology
> > Sint-Pietersnieuwstraat 41, B-9000 Gent, Belgium
> > tel: +32 9 264 34 45, fax: +32 9 264 35 93
> > email: Peter.Bienstman@xxxxxxxxx
> > - ------------------------------------------------
> > -----BEGIN PGP SIGNATURE-----
> > Version: GnuPG v1.2.1 (GNU/Linux)
> >
> > iD8DBQE+GpdZ4dgPAIjyquoRAqPpAKDevDaEVSeiu0fSp+p27yMiO+zHcACfVBq0
> > yGIl4Vq96CbWTnkFsnNXCEw=
> > =FUQ2
> > -----END PGP SIGNATURE-----
>
>
>
> --
> video4linux-list mailing list
> Unsubscribe mailto:video4linux-list-request@xxxxxxxxxx?subject=unsubscribe
> https://listman.redhat.com/mailman/listinfo/video4linux-list
