Re: [Discuss-gnuradio] Decimation Rate with a fraction part

[Martin Braun]

On Fri, Mar 04, 2011 at 08:51:28AM -0500, Nick Othieno wrote:
> Thanks guys. The ideas looks theoretically sound, but for some reason the grc
> setup I have created does not seem to like the idea of very large 
> interpolation
> and decimation values. It does not print out any errors, it just hangs.

Hi Nick,

don't use decimations/interpolations that large, the appropriate filters
are impossible to get working.

Rather, have a look at gr_pfb_arb_resampler_ccf. You need to rationally
resample to an approximate sampling rate and let the block do the rest
by clever algorithms.

Even if you can reach a sampling rate rationally, I usually try and
cascade it to get where I want (and I think that's how it's usually
done).
From 16e6 to 16.3676e6 is, as Ed already mentioned, 40919/40000.
GR standard tools won't give you any good anti-aliasing filters for
this, but you could go in several steps, such as
29/20 × 17/20 × 83/100.

However, working that out already takes longer than setting up
gr_pfb_arb_resampler_ccf, or, even easier pfb_arb_resampler_ccf.py, so
I'd recommend just using that.

MB


> 
> I have attached a copy of my grc setup.
> 
> 
> On Thu, Mar 3, 2011 at 3:39 PM, Ed Criscuolo <address@hidden>
> wrote:
> 
> 
>     On 3/3/11 3:04 PM, Nick Othieno wrote:
> 
>         Hi all,
> 
>         I would like to get an output of 16.3676 Ms/s. Is it then possible to
>         set a decimation values of 6.109631223? I am wondering whether
>         decimation values have to be whole numbers, or whether decimation
>         values
>         with fraction parts are allowed.
> 
> 
>     Decimation numbers must be integers.
> 
>     What you need to do is set the USRP to acquire at some value close to
>     what you need, say 16 Ms/s.
> 
>     Then use a rational resampler with the proper numbers for interpolation
>     and decimation to get the desired rate.  For instance, at 16 Ms/sec,
>     set interpolation to 163676 and decimation to 160000.  In this case, both
>     are exactly divisible by 4, so it could even be 40919 & 40000.
> 
>     @(^.^)@  Ed
> 
> 


> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/discuss-gnuradio


-- 
Karlsruhe Institute of Technology (KIT)
Communications Engineering Lab (CEL)

Dipl.-Ing. Martin Braun
Research Associate

Kaiserstraße 12
Building 05.01
76131 Karlsruhe

Phone: +49 721 608-43790
Fax: +49 721 608-46071
www.cel.kit.edu

KIT -- University of the State of Baden-Württemberg and
National Laboratory of the Helmholtz Association

pgpfXmJBCApjW.pgp
Description: PGP signature