[Discuss-gnuradio] DBS_RX, Gnuradio and GCC 4?

[Robert McGwier]

I was testing and debugging the new DBS_RX trying to ascertain its 
suitability for GPS
and Mode LS transponder development (AMSAT -NA, http://www.amsat.org).  We
found a very small assembly error (short left on a cap on an op amp), we 
made some
gain adjustments on the OP AMPS in the I/Q channel outputs on the DBS_RX to
get better impedance matching for the MAXIM 2118 mixer (analog quadrature
downconverter).  We successfully tuned signals on 900,  1269, 2401, and 
1575.42.
After fixing the analog imbalance imposed by the shorted cap, we were 
able to
easily do image rejection calibration and DC removal.

There were two sessions with Tom Clark, Rick Hambly, and me in the lab of
CNS systems (http://www.cnssys.com).

The first time was on RH Fedora Core 3 and the second on RHFC 4 (I think.
Whatever the latest and greatest is).  We had GCC 4, and I had to back down
to SWIG 1.3.24.  It had python 2.4.1 installed and I installed and built
wxPython 2,6.1.0 and wxGTK latest was already there.


The entire build proceeded without incident except for a test failure in 
the gnuradio
core build.

I chose to just ignore the test failure since it was in a fir filter 
test and I was going to
use dbs_rx_??? and not do any audio output,\.  It was primarily going to 
be doing
FFT's and I hoped it would not be a show stopper.

The tvrx and dbs code all seemed to run fine.  I had a special version 
of dbsrx_debug
(thanks Matt).

My preliminary results show after fixing the cap that we have nice 
sensitivity and we
could see a -90 dBm signals well above the noise in the 4 Mhz wide window we
got using the decim 16 setting.  Processing gain through further cascade 
downsamplings
raised the SNR and we viewed the signals down to as small as a few dozen 
Khz.

I noticed that the DBS_RX has some spurs from various things but for 
GPS, I believe
the decorrelation of the chipping on the satellites will spread these 
and there will be
only a small rise in the noise floor due to the spurs. These can be 
mostly overrun
by system gain before them and with the spreading of them by the receiver,
little impact should be felt from reciprocal mixing with these  A little 
more troubling was
my detection of some near in spurs to the signal of interest due almost 
surely
to an oscillator in the DBS RX system.  The phase noise, on a preliminary
look would likely be acceptable for GPS work.  I do not know about these
near in spurs.

For less demanding work,  where more signal is available (local 
transmitters from
repeaters, broadcast towers, etc.) would almost surely see this unit 
being completely
acceptable. 

It is my intention, with Frank Brickle, and Matt to use this in a 1269 
up 2401 down
prototype 100 Khz ;inear transponder which we hope to demonstrate at the 
AMSAT
annual meeting in October.

I think the DBS_RX is going to be a very inexpensive and useful unit.  I 
will tell you more
about its efficacy  for GPS after I have more time to study this.  I 
will be out of town
from tonight until Monday morning and then I will return to the CNS labs 
for more
detailed work.


Bob

Eric Blossom wrote:

> On Sat, Jul 09, 2005 at 10:12:12AM -0400, John Ackermann N8UR wrote:
>  
>
> > Followup... the gcc 4 build did fail, with a segfault in one of the
> > first tests it ran.  I changed over to gcc-3.3 and g++-3.3 and it
> > completed successfully.
> >
> > John
> >    
>
> Thanks for the data point.
>
> Eric
>
>
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>
>
>