Re: [Discuss-gnuradio] Improve FSK4 decodes in OP25

[Andy Walls]

On Sun, Jan 22, 2017 at 7:53 PM, Andy Walls <address@hidden> wrote:
> On Sun, Jan 22, 2017 at 7:19 PM, Luke Berndt <address@hidden> wrote:
>
>> Well, there was definitely a lot of noise! and I am pretty sure the source
>> was my computer. I got a couple of USB extension cables so I could get away
>> from the computer and the USB hub and that helped a lot. I was able to noise
>> floor by about 20db.
>>
>> Almost every transmission is coming in clear now.
>
> Yay!
>
>
>> Oddly enough though when I
>> do a debug recording, it still doesn’t give me the 4 distinct levels. Any
>> thoughts on what could be up?
>
> You might be looking at a section of the IQ where there is no
> transmission ongoing?
>
> Or you've tweaked something (e.g. baseband level with a multiply
> const) such that the
> joint tracking loop in the 4FSK demod block loses track of the symbol clock?
>
> IDK.

FWIW, the gr-op25 block I'm using came from:

git://op25.osmocom.org/op25.git

the head of the master branch.  At the time I checked it out,
the head commit was:

commit d34fbe52edcfcb79dac349f221e5181adcf2eb8f
Author: Max <address@hidden>
Date:   Wed Oct 7 16:08:29 2015 -0400

    reduce skip time


My GNURadio is from the head of the master branch from a few weeks ago:
v3.7.10.1-191-g0971afba


- Andy

>> Here is an example of the new and improved reception:
>> https://www.dropbox.com/s/7is60h94ctf6d0t/34800-1485122982_4.90888e%2B08.raw
>
> With this file, and the flowgraph I provided to you, I get the four
> levels  (see one of the
> two attached pictures).  The levels are a little noisy, but the symbol
> slicer should sort
> them out with only a few errors, and error correcting code should fix
> the stray symbol errors.
>
> There are parts in the IQ data where no transmission is being sent,
> which will look like a
> mess.  See the second attached picture.
>
>
>
>>  - Luke
>
> Regards,
> Andy