Re: Resampling radio data

[Marcus Müller]

You clearly want the rational resampler, as your resampling ratio is rational.

A 25/23-rate resampler is computationally expensive, no matter how you 
implement it,
because to keep the original bandwidth intact, you'd need a low-pass filter 
with a
transition width below 2/25th of the bandwidth. Not cool.

However,
you don't care about keeping the full original bandwidth intact, you only need 
20/23 of
the original bandwidth, meaning that you get up to 5/25=1/5 in transition 
width. Which is
not that hard.

So, design a low-pass filter with a normalized cutoff frequency of 4/10, a 
transition
width of 1/5, and live with the aliasing of "do not care" signal into "do not 
care"
bandwidths.

I don't foresee this to become a problem, considering your beefy laptop, so try 
with the
rational resampler and a custom low pass filter tap vector, and you should be 
fine.

Best regards,
Marcus
On 16.02.21 23:21, Mark Koenig wrote:
> The BWs are 20mhz, 40mhz and 80Mhz respectively.  The host can range from a 
> server with 48
> cores(2.6Ghz), 256GB ram to a laptop with 8 cores(3.0Ghz), 64GB ram.
> 
> Get Outlook for iOS <https://aka.ms/o0ukef>
> ------------------------------------------------------------------------------------------
> *From:* Brian Padalino <bpadalino@gmail.com>
> *Sent:* Tuesday, February 16, 2021 5:17:32 PM
> *To:* Mark Koenig <mark.koenig@iubelttechnologies.com>
> *Cc:* discuss-gnuradio@gnu.org <discuss-gnuradio@gnu.org>
> *Subject:* Re: Resampling radio data
>  
> On Tue, Feb 16, 2021 at 4:14 PM Mark Koenig 
> <mark.koenig@iubelttechnologies.com
> <mailto:mark.koenig@iubelttechnologies.com>> wrote:
> 
>     Hello,____
> 
>       ____
> 
>     I am receiving data from a radio at 23Msps, and I would like to resample 
> to a rate of
>     25Msps in which my software program can ingest it.  I am aware of the 
> Fractional
>     Resampler and the Rational Resampler.  Is one better than other, take 
> less CPU cycles,
>     etc.?  Is there a different block which would suit my needs more 
> efficiently?  Should
>     I instead look to implement the resampling within C++ as opposed to 
> adding another
>     block to my flowgraph?____
> 
>     __ __
> 
>     I am going to do the above with the following rates:  46Msps -> 50Msps, 
> 92Msps -> 100Msps.
> 
> 
> How much of your bandwidth are you actually looking to keep intact?  How fast 
> is your host
> CPU?
> 
> Brian
>

smime.p7s
Description: S/MIME Cryptographic Signature