[amsat-bb] going digital

Phil Karn karn at ka9q.net
Tue Jul 22 05:00:27 UTC 2014

On 07/21/2014 02:00 PM, Gus wrote:

> Actually, DirecTV is not available for purchase in my neck of the
> woods.  I can get a system installed, but it wouldn't be my system to
> fiddle with.

Sorry about that, I was primarily thinking of AMSAT-NA members when I
said that. Forget the DirecTV receiver itself and consider their
offset-fed Ku-band dishes and 0.7 dB NF LNBs that are readily available
on the surplus market. Since DirecTV isn't available in Barbados, paint
the dish with the logo of whatever DBS provider you like. Nobody will
know the difference.

> Phil, I'm not saying it /should/ be, but that if it is, you won't get
> many users on the system.  So, part of the system design for any such
> bird had better include design of low cost, easy to find, simple to use
> hardware that will allow me -- or any other doofus like me -- to get QRV
> without too much hardship.

And here I am with you 100%. So many hams seem to automatically assume,
for no good reason, that getting on any new digital mode must
necessarily require exotic, expensive and hard to find components and a
PhD to use. But the required hardware is now very common, and nearly all
of the "complexity" that defines any particular digital communication
system these days is implemented almost entirely in software that costs
nothing to copy and distribute once written.

The hardware to run said software already exists in virtually every ham
shack. Even a low-end general purpose PC is a very capable DSP engine,
and this has been true for almost 20 years since Intel introduced MMX,
their first set of DSP instructions for the x86 CPU. Since then there
has been SSE, SSE2, SSE3, SSSE3, SSE4 and now AVX and AVX2, Advanced
Vector Extensions 2.

Here's an example of how far consumer DSP has come just in the time I've
been involved with AMSAT. I wrote the demodulation and decoding software
now used by the ISEE-3 Reboot Project. It takes the A/D output of an
Ettus USRP software defined radio, tracks the RF carrier, demodulates
the phase modulation, tracks and demodulates the binary symbols, and
performs error correction to produce raw telemetry frames.

ISEE-3 uses a rate 1/2 constraint length k=24 convolutional error
correction code. When it was launched in 1978, decoding it took highly
specialized, dedicated hardware. Someone here who worked for Linkabit,
the manufacturer of that decoding hardware, can probably jump in with
the exact cost in millions of dollars and the number of 6' equipment
racks it filled.

I am doing all these functions on my laptop using not only the Fano
algorithm used by that original Linkabit hardware, but also the Viterbi
algorithm, which is better at correcting errors in noise.

The Viterbi algorithm is traditionally used with much shorter codes
(k=7, first used on the Voyagers, is still common) because its
complexity doubles each time you increase it by one. Had I suggested, in
1978, using it for a k=24 code (taking 131,072 times as much effort as
k=7) NASA would have laughed me out of the room.

The same reaction would still have been justified in 1985 when they
renamed it ICE and sent it out of earth orbit to the first-ever comet
rendezvous, requiring Herculean efforts to overcome the extra path loss.

Now I'm doing it all on my 2-year-old Sony laptop.

That's how much computing power we hams now have in front of us. All we
need is a little imagination and vision as to what it makes possible.
Instead there seems to be a contest to see who can produce the most
objections, valid or otherwise. Is this what the amateur service (or
AMSAT) has become?


More information about the AMSAT-BB mailing list