CPFSK demodulation and synchronization

Darol wrote:

> On Dec 2, 1:54 pm, Vladimir Vassilevsky <nos...@nowhere.com> wrote:
> 
>>>One non-coherent approach is to use a differential detector followed
>>>by a decision feedback equalizer (DFE). The output from the
>>>differential detector could be sampled at twice the baud rate before
>>>being fed to the DFE, and the phase of of the sampling doesn't have to
>>>be controlled because the DFE will adapt.
>>
>>He uses RECTANGULAR pulses. What is the point of DFE ?
>  
> Even with rectangular modulating pulses, isn't a DFE beneficial
> because of the memory required to implement CPFSK?

That memory is already taken into account by the differential detector.

>>>This method may not be generic enough for you because it works best
>>>with signals that contain a preamble.
>>
>>Any method works with or without preamble. Preamble is a small matter of
>>convenience, no more then that.
> 
> With a preamble, cannot the DFE train faster and with greater accuracy
> than using a blind technique (e.g. constant modulus)? This would lower
> BER, especially for bursty signals.

By preamble, I meant a regular sequence for carrier and sync 
acquisition. Unless for some special cases, equalizer can't train on the 
regular sequence.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com