Re: [Discuss-gnuradio] MAKING A NTSC TV RECEIVER

[Martin McCormick]

        I must think a little better next time.  I looked at my
posting to the list and realized I was being rather dim-witted in
that I was imagining the signal path for the video as needing the
sound card as it does when one is decoding the sound carrier or
an FM radio station.  In theory, one could do that if you could
make the circuitry in the sound card run a lot faster than it
normally does but that isn't necessary at all.

        The I and Q signals from the rtl chip first go to a
processing block to add their vectors in a floating point method
just like all other analog processing but much more frequently.
You still need a low-pass filter to prevent aliasing just as with
audio and low-speed data but the frequency is much higher since
we are taking lots more samples per second.  That block must run
a lot faster all right.  

        Each computed level represents one pixel and here's the
first spot where there could be trouble.  If the processor can't
do the math fast enough, the next sample arrives and the first
one hasn't been set yet.  We are already behind and there is no
flow-control so something is going to give and it is most likely
going to be no or badly-formed pixels which will mean no picture
at all.

        These calculations as well as the placement of each pixel
will be being done at the usb sampling rate so I bet the newest
and best hardware will probably decode the monochrome image and
systems like the 600-MHZ Pentium I do unix command-line, audio
and email on would positively smoke if I tried to do this
there.

        I believe the DSP chips used in dedicated DSP devices use
every trick in the book to process rapidly and general-purpose
computers have much more difficulty keeping up.

        Many people on this list know far more about this than I
do so feel free to call me out but in theory, if you sampled at
12 MHZ, one could generate a digital stream that if run through a
D/A converter would be a full color NTSC picture and the full
audio carrier which would need a second decoder if you wanted the
sound but it is possible.  If you ran the first D/A through an
oscilloscope, you would see the full video envelope plus the
sound carrier at 4.5 MHZ.  It would be a NTSC base band signal.

        The same is true for PAL and SECAM but one would need 14
or 16 MHZ sampling rates to capture those formats.

Martin WB5AGZ