How to plot S-Curve for TED ?

hi  all ,
       I simulated OQpsk modulated data which has 4 samples per symbol (base band). I also simulated the squaring and differentiating based Timing error detector .Now i want to know how well is it performing for that i want to simulate and draw the  s -curve.

                  can anyone suggest me a best method to plot s-curve! 
thanks in advance... 
          

On 25.06.2015 12:57, Eshwar varma wrote:
> hi  all ,
>         I simulated OQpsk modulated data which has 4 samples per symbol (base band). I also simulated the squaring and differentiating based Timing error detector .Now i want to know how well is it performing for that i want to simulate and draw the  s -curve.
>
>                    can anyone suggest me a best method to plot s-curve!
> thanks in advance...
>
>

Eshwar, it's one of the things you "just" do. The only thing you need is 
understanding what an S-curve is. I suggest you to use Matlab (or any 
analogue) for simulations / plotting, since it's really efficient for 
such tasks.

Regards,
Evgeny.

>hi  all ,
>       I simulated OQpsk modulated data which has 4 samples per symbol
>(base band). I also simulated the squaring and differentiating based
Timing
>error detector .Now i want to know how well is it performing for that i
want
>to simulate and draw the  s -curve.
>
>                  can anyone suggest me a best method to plot s-curve! 
>thanks in advance...

Lets say, you have 10 samples per symbol and lets say you have 10 symbols.
This makes in total of 100 samples in a packet.

Now lets say you receive these 100 samples noise free at receiver and
assuming that you are using rectangular pulse both at Tx and Rx, what will
be the output waveform when you convolve the received 100 samples with the
rectangular pulse?

To save you sometime, you will get a triangular waveform (you should
simulate it), where the peaks will indicate ideal sampling instant. 

Now since you have received noise free samples and if you sample at
correct instant, and get the error for the 10 symbols (not samples), the
mean error should be zero.

But if you either advance or delay the sampling instant then your mean
error will not be zero. (Once again, you should simulate it)

Then plot these mean error values you get and it should follow a pretty
good S-curve.

Hope this helps.

Sia
---------------------------------------
Posted through http://www.DSPRelated.com