Re: [Discuss-gnuradio] Tutorial on BPSK bursts

[Andy Walls]

On Tue, 2016-03-08 at 21:04 +0000, Jesse Reich wrote:
> Thank you much Andy. I'll take a look and give it a try tonight when I
> get home. 
> 
> 
> Is the BPSK that is used unusual because of the 1.1 rad phase shifts

Yes.  One would expect BPSK constellation points to be pi radians apart
(aka half way around the unit circle), like at +pi/2 and -pi/2 or at 0
and pi.  Being at +0.35pi and -0.35pi is just weird, but probably done
for robustness of reception (safety of life at sea; someone needs to
hear it!).


>  and is the related to what I've heard that the signal still has power
> in the pure carrier throughout the message?

The FFT of the tx burst shows that.

When phase locked to the carrier, the In-Phase component never vanishes;
it remains at cos(1.1 rad) while the Quadrature part moves between +/-
sin(1.1 rad).  When phase locked to the carrier during the 160 ms of
unmodulated carrier, the In-Phase component is at 1.0.


> 
> Also, do you know if the T.001 beacons use differential BPSK or not? 

This document doesn't make any mention of differential encoding:
http://www.cospas-sarsat.int/images/stories/SystemDocs/Current/cs_t.001_oct_2013.pdf


> 
> 
> I was able to capture a file sink of a PLB transmission last night so
> I'll have to give it a try.  But the frequency I saw seemed unusual;
> around 406.008 MHz. I believe it should be at 406.037 based on the
> beacon I was using. 

Well, if your SDR receiver is not locked to a 10 MHz reference or a 10
MHz reference from a GPSDO, then you could be off freq.

Beacon Test mode is also supposed to use a special freq.

FWIW, the freq plan is here:

http://www.cospas-sarsat.int/images/stories/SystemDocs/Current/cs_t012_oct_2013.pdf

I honestly find these C/S specification to be written in a very
confusing manner.  Just figuring out the allowed bandwidth of a
transmitter was confusing.

Have Fun.

-Andy
 
> 
> 
> Thanks again,
> Jesse 
> 
> On Tue, Mar 8, 2016 at 3:47 PM Andy Walls
> <address@hidden> wrote:
> 
>         On Sat, 2016-03-05 at 12:00 -0500,
>         address@hidden
>         wrote:
>         > Date: Sat, 5 Mar 2016 00:46:12 -0500
>         > From: Tim K <address@hidden>
>         
>         >
>         > It's not terrifically designed, but you might find my PSK31
>         repo
>         > helpful.
>         > According to Balint though, the PLL is supposed to go after
>         clock
>         > recovery,
>         > not before.
>         >
>         > https://github.com/tkuester/gr-psk31
>         >
>         > - Tim
>         >
>         > On Fri, Mar 4, 2016 at 8:56 PM, Jesse Reich
>         <address@hidden> wrote:
>         >
>         > > So I've been playing with GNU Radio for about 2 months and
>         trying to
>         > learn
>         > > about signals and signal processing in general. Mostly out
>         of
>         > curiosity
>         > > that has sprung up from my profession. I am an Aerospace
>         engineer by
>         > > training and I work as a satellite ground systems engineer
>         for the
>         > > Cospas-Sarsat program.
>         > >
>         > > I've started with what I thought should be an easy task
>         with
>         > aspirations
>         > > of more complex tasks. The first task, decode a BPSK
>         signal
>         > modulated on a
>         > > 406 MHz carrier. The signals consist of a 160 ms carrier,
>         24 sync
>         > bits and
>         > > then 120 data bits at 400 bits/sec. I'm just trying to
>         decode and
>         > write the
>         > > bits. Sounded to me to be an easy problem to tackle.
>         Unfortunately I
>         > can't
>         > > seem to even get off of the ground. Does anyone know of
>         any
>         > tutorials that
>         > > would be good.
>         > >
>         > > I've tried the gnuradio tutorials and honestly they don't
>         really
>         > seem to
>         > > help. Any help would be greatly appreciated.
>         > >
>         > > Thanks,
>         > > Jesse
>         
>         
>         Hi Jesse:
>         
>         See the attached, disasterous flowgraph:
>         a) the top half is a simulated EPIRB transmitter and noisy
>         channel
>         b) the bottom half is an EPIRB PSK receiver I winged together
>         
>         The receiver probably needs tweaking on the M&M clock recovery
>         parameters.  Also the receiver implementation ends at the
>         recovered,
>         manchester encoded bits.  You have to decode the bits and
>         frame them
>         yourself from that point.  Tim K probably has good stuff in
>         his github
>         repos to do that.
>         
>         The C/S T.001 stuff isn't your normal BPSK, so the all-in-one
>         PSK demod
>         blocks in GNURadio won't work well.  (Heck, the costas loop
>         block gets
>         confused on the 160 ms of unmodulated carrier.)
>         
>         The C/S T.001 BSPK modulation has an unusual constellation:
>         
>         (1.0          , 0.0          ): unmodulated carrier
>         (cos( 1.1 rad), sin( 1.1 rad)): '1' symbol
>         (cos(-1.1 rad), sin(-1.1 rad)): '0' symbol
>         
>         This makes it easy to lock on to with a PLL with a very
>         sluggish
>         response.  That is what I did in the attached GRC.
>         
>         Regards,
>         Andy