[hpsdr] Is there a DSP or DttSP guru in the house?

[Robert McGwier]

Bill Tracey wrote:
> ***** High Performance Software Defined Radio Discussion List *****
>
> Do we have any DSP or DttSP folks on the list?
>   

Was this a rhetorical question with tongue in cheek?  ;-)
> Looking to clean up our support on the Janus prototype for mixed sampling 
> rates.  Currently our system runs all of the channels (IQ in, mic in, 
> received audio out,  IQ out) at the same sampling rate.   We want to change 
> this such that we can run IQ in at 48, 96 or 192 khz and hold the other 
> three channels at 48 khz sampling.   My understanding of the current DttSP 
> and the PowerSDR code is that there is one and only one sampling rate for 
> all channels, so I'm looking at doing this by upsampling/interpolating the 
> mic in channel to the IQ in rate and then downsampling the IQ out and 
> received out streams when IQ in is running at 96 or 192 khz.
>   
> I believe the output side downsampling is easy  - just throw away the 
> appropriate number of samples.   The input side is not so clear to me.  If 
> I zero stuff to upsample I believe I will get images of the spectrum at 
> multiples of the original sampling rate.    This would suggest i need to 
> follow my upsampling with some sort of low pass filter to remove the extra 
> image.
>   
This is not correct because you have forgotten our support of Virtual 
Audio Cable.  You will see in the DttSP source resample.C and resampleF.c

These routines optimally do polyphase resampling based on rational rate 
changes.  They work by having two pointers rotating at the various rates.

If you have input at rate N and output at rate M,   let us suppose N and 
M have at least one factor they do not share in common so this is 
interesting ;-).

Let H be the least common multiple of N and M.  For this illustration, 
assume N < M (the arguments are symmetric for these purposes).  You must 
construct the proper low pass filter running at the rate H that cuts off 
below the Nyquist for N (N/2).   Then the two pointers step through the 
filter applying it to only those taps in the filter that would have non 
zero samples underneath them and the other steps through these results 
at the output rate.

Bob

> Does anyone know if this is actually needed with the current DttSP/PowerSDR 
> code?  I'm sort of hoping there's already an LPF in the mic input chain 
> that will take care of the images I'm introducing by upsampling.  If not, 
> anyone have any suggestions/code pointers to some code to use to do the 
> needed filtering.
>
> Regards,
>
> Bill (kd5tfd)
>
>
>
> _______________________________________________
> HPSDR Discussion List
> To post msg: hpsdr at hpsdr.org
> Subscription help: http://lists.hpsdr.org/listinfo.cgi/hpsdr-hpsdr.org
> HPSDR web page: http://hpsdr.org
> Archives: http://lists.hpsdr.org/pipermail/hpsdr-hpsdr.org/
>
>   


-- 
AMSAT VP Engineering. Member: ARRL, AMSAT-DL, TAPR, Packrats,
NJQRP/AMQRP, QRP ARCI, QCWA, FRC. ARRL SDR Wrk Grp Chairman
Laziness is the number one inspiration for ingenuity.  Guilty as charged!


 1147623593.0