(Root) raised cosine matching filters

[Joel Kolstad]

I've done some reading lately (in the likes of Proakis, Lathai, etc.), and
they all discuss the use of raised cosine, root raised cosine, or similar
"pulse" signals that occupy approximately constant bandwidths, are designed to
largely eliminate ISI, etc. That's great. However, this is usually all in
the context of digital (or at least discrete time) systems. Is anyone
familiar with how you might go about building a matched filter for, say, a
root raised cosine (RRC) pulse using traditional (continuous time) filtering?
(The digital case is rather trivial! ) I can see that, if one were to
just build a relatively low order bandpass filter (say, just a Butterworth
filter), the result probably wouldn't be _that_ far from the mark, but I
imagine there's a somewhat more systematic way to do this?

Thanks,
---Joel Kolstad

 

Digital implementation of a Finite Impulse Response filter is - as you
say - relatively trivial.

The analogue continuous time equivalent is the tapped delay line
filter, in which you hook up an op amp to each tap of a delay line and
combine the weighed outputs of the taps with a summing amplifier. Since
most tapped delay lines are actually sequences of linear-phase low pass
filters, this is usually an inefficient way of realising your filter,
but since you can buy tapped delay lines off the shelf, it can be the
way to go for one off and small volume projects.

------------
Bill Sloman, Nijmegen

 

[Ian]

"Joel Kolstad" <JKolstad71HatesSpam@yahoo.com> wrote in message
news:11dr1kgo9lsa0a3@corp.supernews.com...

I've done some reading lately (in the likes of Proakis, Lathai, etc.), and
they all discuss the use of raised cosine, root raised cosine, or similar
"pulse" signals that occupy approximately constant bandwidths, are
designed to
largely eliminate ISI, etc. That's great. However, this is usually all
in
the context of digital (or at least discrete time) systems. Is anyone
familiar with how you might go about building a matched filter for, say, a
root raised cosine (RRC) pulse using traditional (continuous time)
filtering?
(The digital case is rather trivial! ) I can see that, if one were to
just build a relatively low order bandpass filter (say, just a Butterworth
filter), the result probably wouldn't be _that_ far from the mark, but I
imagine there's a somewhat more systematic way to do this?

Thanks,

There were a bunch of papers on how to do this back in
the '60s and '70s, for example in the Bell System TJ.
The widespread use of DSP has led to this being done
digitally these days.

However, if you really want to do this in analog, Nuhertz Technologies
Filter Solutions program will synthesize RC and RRC filters for
various alpha factors.

http://www.nuhertz.com/filter/

Regards
Ian

 

[Joel Kolstad]

Thanks for the information, Ian and Bill...

"Ian" <Ian_Buckner_not_@agilent.com> wrote in message
news:1121852072.264141@newsreg.cos.agilent.com...

However, if you really want to do this in analog

Actually I don't, but there's a perception that getting a matched filter and a
detector done in analog is going to be simpler (faster) than digitizing and
doing it in an FPGA. We'll see...

---Joel