Re: [Discuss-gnuradio] GMSK at 768kb/sec and other good stuff

[Robert McGwier]

Eric Blossom wrote:

> Matt, Bob McGwier (N4HY) and I spent last week working on GNU Radio
> face to face.  We got an incredible amount of stuff done, and not much
> sleep!
>
> Below are some of the highlights:
>
>  
---------- snip ----------------------

You left out why it was possible to work this hard for a week and get a 
lot done.  The Blossom family hospitality is second to none.   I stayed 
in one of the newest and most lavishly decked out casino's in Reno 
(cheap, $50 a night!) and did not put one red cent into a single slot 
machine.  The only time I managed to even enter the casino,  it was to 
walk through to the diner!  It was a busy week, but an extremely 
valuable learning experience for me and I am very happy Matt and Eric 
persevered in my training until I could see how to do things.  For those 
that do not understand how things work behind the scenes,  it is a 
remarkable piece of work.  With Stephane's recent contributions of the 
SSE/3DNOW based routines for filtering and the FFTW SSE based FFT 
routines, this code is remarkably efficient and is truly beginning to 
meet all of its great promise.

My apologies on the Kaiser window.  I got trapped like everyone else 
that does a quick search.  They get the Numerical Recipes code (usually 
without a single line of attribution) as I did.  Go to Julius O. Smith's 
CCRMA resampling treatise and you will find the entire rationale for 
what was done here for the rational (integer up, integer down) 
resampler.  Smith's fractional resampler is also very neat but we did 
not need it for the week since the Eric/Matt had already implemented a 
"compromise" but very fast fractional resampler.  From my observations, 
and where we are using it,  it is not much of a compromise and as I 
said, it is very fast.

Matt and I started the Costas loop and we all detoured to the PLL 
blocks. We can and should finish the Costas loop to enable a fancy DSB 
detector,  QPSK demodulator, etc.

In gnuradio-examples/python/usrp  you will find usrp_wfm_rcv_pll.py.  
The difference between it and usrp_wfm_rcv.py is the latter uses the 
arctan demodulator and the new one uses the PLL demodulator.  The PLL 
design parameters probably need tweaking.  Give it a try if you have the 
tuner.  To my ear, it sounds brighter, but your ear might hear 
differently.  Let us know.  I thought it might be more expensive 
computationally but if it is,  it is not by much.   Matt ran it on his 
abacus,  err uhhh,  1.8 GHz (constantly throttled back for thermal to 
600 MHz) laptop.

We looked at the FlexRF board for 70 cm and as Eric mentioned,  we used 
it for the GMSK.  Matt's abacus could not send faster than 756 kbps 
without glitching so I am sure it will run at least 1 mbps on a 
desktop.  In looking at the FlexRF,  we managed to do the necessary 
adjustments to suppress the LO to an extraordinary (and entirely 
unnecessary) and to suppress the image to > 30 dB.  This will be fine 
for the 70cm, 900, 1296, and 2304 bands.  I have two transverters for 
1296 and 2304 that do not have this level of suppression.  Matt used a 
quadrature mixer in a neat package and it works.

The other thing we discussed as Eric mentioned and still need to make 
progress on is the modification to the bit file for the USRP and then 
implement the code to enable the phased array code.  The need to modify 
the bit file and the discussion of the applications of this hardware and 
software to the C to C band transponder for AMSAT=NA's Eagle spacecraft 
mission kept us from  attempting an implementation of the phased array code.


My apologies to Matt.  We had great fun at his computer's expense though 
it worked just fine. I brought my great laptop only to discover that it 
has USB 1.1 and had to spend 3 days getting SUSE 10 set up to compile 
and run GnuRadio to even find that out!

What else did we forget?

Bob


> I'm sure I've left out something important!
>
> This wouldn't have been possible without the enthusiastic efforts of
> Bob and Matt!  Thanks guys!
>
> Eric
>
>
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>
>  


--
Laziness is the number one inspiration for ingenuity.  Guilty as charged!