[Discuss-gnuradio] Pulse shaping filter advice needed

[Ian Wraith]

Hello
     I am continuing my work on a PI/4 DQPSK demodulator module and
thought I was making some progress with the symbol timing (I am using
the Gardner algorithm now). However I have now noticed a problem with
the pulse shaping filter that I could do with some help with.

 My system works like this - I have a homebuilt quadrature demodulator
being fed with the 10.7 MHz IF of my receiver. The quadrature demodulators
I & Q channel outputs go into the stereo line input of my onboard sound
card. In software a small test program connects the GrAudioSource module
to my module.

 Now the first thing my module does is what all the books recommend which
is to put the I & Q channels through a root raised cosine pulse shaping
filter with a beta of 0.35. I designed the filter using this web page ..

http://www-users.cs.york.ac.uk/~fisher/mkfilter/racos.html

with the following parameters ..

filtertype=Raised Cosine
samplerate=44100
corner=9000
beta=0.35
impulselen=21
racos=sqrt
comp=no
bits=   
logmin=

You will see the corner is set to 9 KHz. As the web page recommends this be
set to half the baud rate. As my symbol rate is 18 KHz  I set this to 9 KHz.
The trouble is when I look at the raw data and the filtered data in
graphical form it looks to me as though the data is being pulse shaped to
to low a frequency i.e the pulses are to wide. You can see what I mean by
looking at ..

http://borg.shef.ac.uk/~ianw/filter_problem.html

Remember the symbol rate is 18 KHz and the waveform is being sampled at
44.1 KHz. On the raw data you will see each symbol is made up of 2 or so
data points which I think is correct. However if you look at the graph
below of the same data that has been filtered you will see the pulses
are much wider - far to wide I believe.

 Could anyone tell me where I am going wrong here ? Am I misunderstanding
the filter parameters or is my mistake more basic ?

Thanks

Ian

address@hidden