Re: [Discuss-gnuradio] AR5000 and downconversion

[John Gilmore]

> I'm looking to receive and demodulate an Australian HDTV signal. We use a 
> COFDM (Coded Orthogonal Frequency Division Multiplexing) modulation scheme 
> instead of the ATSC scheme used in the US but the bandwidth requirements 
> are approximately the same, around 6-7MHz. This, I believe, rules out the 
> possibilities of trying to use undersampling techniques to sample the 
> signals, as I don't think I can fit the entire signal into a Nyquist region.

For US HDTV (ATSC), the symbol rate is 21+ Msymbols/sec, and our
hardware would only receive at 20 Msamples/sec.  We experimented with
overclocking our hardware, and actually got it working with external
clock generation, but ultimately realized that we were going to have
to resample the signal on symbol boundaries in software anyway.  The
receiver's clock will never be in sync with the transmitter's clock --
at least not for long.  Once we had written and debugged some
resampling code, we realized that we could just sample the original
signal at the standard 20 Msamples/sec, and once our software figured
out where the symbol boundaries are (in between the samples, usually),
it was still capable of resampling the symbols without trouble.

You may be able to do some similar processing.

By the way, check the GNU Radio Wiki at http://comsec.com/wiki .  The
HowtoHdTv page describes the tuner hardware that we used.  There's a
Microtune TV tuner available, which can tune from 50-860 MHz in 6 MHz
chunks.  A version with an I.F. of 5.75 or 7.125 MHz is available.
The Wiki has a link to detailed tech specs (and to a simple circuit to
interface it to your computer).  The idiots who market the tuner
require that you get a temporary email address and register with their
idiotic web site to get a password, so they can track you to a
temporary email address, before they'll let you download the data
sheet to see if you like their part.  It's a wonder that some
companies remain in business.  If you find a more responsive company
with a similar part, *please* let us know.

It would be great to have a COFDM demodulator for GNU Radio.  I think
COFDM would be a lot easier to code up than the highly obscure ATSC
decoding, and would probably run in a lot closer to realtime.  ATSC
decoding runs about 40x slower than realtime on a fast Athlon.

        John