Re: [Discuss-gnuradio] CPFSK mod/demod + strange behavior

[Marcus Müller]

Hi Olivier!

On 10.06.2016 00:05, Olivier Goyette wrote:
>
> You are wrong !
>
That being the case very often, I think the error's on your side, this
time :)
So, I will need you to *explain* how you can transport 1.somethingMb/s
through a 625kHz wide channel if every time you use that channel to
transmit a symbol, only one bit can get sent. How do these numbers add
up for you? You've got a serious problem with your pulse shaping, there.
>
> 1 sample/symbol is when I run simulation (no transmission over the air).
>
Yes, and as said, that's hardly a FSK. FSK is "I encode data by changing
the frequency. Frequency is the derivate of phase, or in the discrete
case, the phase increase between samples"; in your 2-CPFSK, you either
increase the phase of the transmit signal, or you decrease it by exactly
the same amount (assuming you send -1 and 1 bytes).
Now, with sps=1, you change that phase increase every sample – that's
nothing but a bad differential BPSK that you've involuntarily built.
>
> When I want to transmit over the air ( cable between Tx and Rx ) I
> need to use 6 samples/symbol.
>
> This is what I've told earlier, 1 sps is the only way I could send a
> message and retrieve it at the end of the transmission chain by doing
> a simulation. If I use a sps higher than 1, I can't find a way to get
> my message back
>
Well, and it's absolutely no use to try and do that over the air if the
simulation doesn't work with 6 sps.
As said, the real world usually makes things harder, not easier, and
things that don't work in simulation rarely work out in the real world
through magic.

You cited a word document with links (that's seriously the worst format
you can transport hyperlinks in, by the way). Which one of those
contains the standard that says you have to do an RRC on the FSK
modulated signal? I presume, and Achilleas seems to agree on that, that
there might be a misunderstanding here. Are you sure the RRC goes
*after* the FSK? As I mentioned in my last mail, it's not a good idea to
low pass filter after an FSK modulation. That just seems counter-productive.

Best regards,
Marcus