Newbie question!

Hello!

I am interested in FFT and sampling and have just started a project at
home, which involves programming and dsp/signal theory/sampling.

My setup is a shortwave radio and I want to decode some simple 50 baud
FSK signals in an application that I am currently building. So far I
have successfully sampled an audio signal, so I have raw sample data
from my sound card, in chunks of for example 1024 samples. I also have
made use of a forceful mathematics library and now my application
sports a realtime FFT spectrogram of the audio input, sampled in 8000,
11025 or 22050 Hz. I can also change number of FFT pins in order to
raise or lower the spectrum resolution.

But now to my question!

I wonder if anyone could describe to me how I go from a sampled audio
buffer to for example showing an eye diagram., or how to show the
instantaneous frequency plotted continously, clearly showing the
levels. In general terms.

When I've done this I aim at do the same for PM signals and showing a
QAM plot in real time. It's great fun but right now I'm a bit stumped
on how to go from sampled audio to FSK levels and then to eye diagram
and then bits =)

Regards,
Jonas Hallgren
Sweden

Jonas Hallgren wrote:

> It's great fun but right now I'm a bit stumped
> on how to go from sampled audio to FSK levels and then to eye diagram
> and then bits =)

I'm a newbie too, but I've accomplished what you describe. :0)

For classical 2-tone BFSK-signals ( RTTY,Sitor-B, ARQ-E, ARQ-E3, etc..) 
the basic steps are as follows:

1. The raw sound samples are fed through 2 filters ( IIR, FIR ), one 
for the mark frequency, one for the space frequency. I am currently 
using 125 Hz wide FIR filters.

2. The filter outputs are then fed through  envelope detectors to get 
rid of the audio frequency component. I used a simple low pass filter.

3. The detector outputs then go to a decision (comparator) circuit, the 
output of which will be 1/0 value according to the received signal.

4. For synchronous signals, the comparator output is sent to a 
PLL-circuit which samples the signal at the bit center ( the PLL 
obviously needs to know the baud rate) . I've used a simple 
edge-detector + counter + a heavy low pass filter to minimize the 
effect of noise spikes.
   For asynchronous RTTY signals, the comparator output is sent to a 
classical start/stop detector: wait for a 1>0 transition, check for 
startbit=0 1/2 bittime later, sample N bits, check stop bit, etc...

5. For synchronous signals, the next step is to get into frame sync, 
and to decode the signal depending on the code/alphabet used.

Example:
<http://users.pandora.be/dirk.claessens2/DSP/scrshot.JPG>


BTW: I could not have accomplished all the above without the *lot* of 
help & tips I got in this group...

Thank you very much!

I will try to implement a filter first then :)

Regards,
Jonas

On 9 Apr, 12:09, DSP-Newbie <N...@way.invalid> wrote:
> Jonas Hallgren wrote:
> > It's great fun but right now I'm a bit stumped
> > on how to go from sampled audio to FSK levels and then to eye diagram
> > and then bits =)
>
> I'm a newbie too, but I've accomplished what you describe. :0)
>
> For classical 2-tone BFSK-signals ( RTTY,Sitor-B, ARQ-E, ARQ-E3, etc..)
> the basic steps are as follows:
>
> 1. The raw sound samples are fed through 2 filters ( IIR, FIR ), one
> for the mark frequency, one for the space frequency. I am currently
> using 125 Hz wide FIR filters.
>
> 2. The filter outputs are then fed through  envelope detectors to get
> rid of the audio frequency component. I used a simple low pass filter.
>
> 3. The detector outputs then go to a decision (comparator) circuit, the
> output of which will be 1/0 value according to the received signal.
>
> 4. For synchronous signals, the comparator output is sent to a
> PLL-circuit which samples the signal at the bit center ( the PLL
> obviously needs to know the baud rate) . I've used a simple
> edge-detector + counter + a heavy low pass filter to minimize the
> effect of noise spikes.
>    For asynchronous RTTY signals, the comparator output is sent to a
> classical start/stop detector: wait for a 1>0 transition, check for
> startbit=0 1/2 bittime later, sample N bits, check stop bit, etc...
>
> 5. For synchronous signals, the next step is to get into frame sync,
> and to decode the signal depending on the code/alphabet used.
>
> Example:
> <http://users.pandora.be/dirk.claessens2/DSP/scrshot.JPG>
>
> BTW: I could not have accomplished all the above without the *lot* of
> help & tips I got in this group...