[hpsdr] Decimation Question

[Georg Prinz]

Graham,
one word more, because you were asking about the decimation especially.
As you know Decimation is a reduction of the sample rate by lowpass
filtering with following compression. This, as I noticed, is done after
the AD to reduce the data stream caused by oversampling.  This will not
effect the SFDR.
SFDR can be improved e.g. by dithering. But there are other techniques
existing, too.
In the meantime I found the referenced article. If you are interested I
can send you it.

Vy73, Georg, dl2kp


On Mo, 2009-03-23 at 13:40 -0500, Graham / KE9H wrote:
> Georg:
> 
> Thank you for the discussion of decimation.  Everyone looks at it slightly
> different, and every time I see or hear another explanation, I understand it
> a little better.
> 
> Phil and I agree with your guess that SFDR is not
> improved through decimation.  I am hoping that one of the university
> professor experts will comment and show us an explanation.
> 
> As to the tutorial...
> I was referring to Phil's  4 hour  tutorial at DCC last September.
> 
> TAPR is selling his presentation on DVD at
> 
> http://www.arvideonews.com/dcc2008/
> 
> You want DVD #6 ...
> Software Defined Radio"Through the Looking Glass"
> by Phil Harman VK6APH.
> 
> I found it an excellent tutorial on the design of Mercury,
> step by step.
> 
> Thank you, again,
> --- Graham
> 
> =
> 
> Georg Prinz wrote:
> > Hello Graham,
> >
> > can you tell me where I can find the tutorial?
> >
> > The SNR can be improved by oversampling of the imput signal. The imput
> > signal will be considered as Lowpass signal with a maximum frequency
> > fmax which will be sampled with a rate fs>>2fmax. The OverSampling Rate
> > OSR = fs/(2fmax) shows how far the minimum sample rate (Nyquistrate)
> > will be exceeded. Now you have the situation that the power density
> > spectrum of the quantization noise will be distributed over a much
> > broader bandwidth at the same average power. The noise energy, which
> > superimpose the imput signal, will be smaller as when used with the
> > Nyquistrate for sampling.
> > If you have a ideal lowpass filtering after the AD, you will get a 
> > better SNR by +10logOSR. That means if you double the sampling rate will
> > gain 3dB. Further more it will lower the demand on the anti-aliasing
> > filter in front of the AD. But you will get a higher data rate from the
> > AD. This is the reason why usually a reduction of the sampling rate is
> > done after the first filtering.
> >
> > The dynamic range states for which input amplitudes the AD produces
> > positive SNR [dB] and is defined as the ratio of the power of a sinus
> > signal with the best SNR (= max. amplitude), devided by the minimum
> > signal power of a sinus signal where SNR is equal 0 dB. 
> > When oversampling used, the dynamic range will increase by +10log(OSR),
> > too.
> > You have to reallize that in real ADs there are additional errors
> > generated by thermic noise, jitter, intermodulation, comparator
> > blurrings, etc.
> >
> > When using oversampling the SFDR defines how far strong error signals
> > can be detected in the input frequency band without overloading the AD.
> >
> > At the moment I can guess only that the SFDR will not improve.
> >
> > Unfortunately I didn't find the exact definition of the SFDR as a
> > formula but found a literature reference : "Wepman, J.A.:
> > Analog-to-Digital Converters and Their Applications in Radio Receivers.
> > IEEE Communications Magazine, 33(5):39-45, May 1995".
> > Maybee someone has access to the IEEE magazines. I would be interested
> > in a copy.
> >
> > So, I hope that my english is good enough to make me understandable, HI.
> >
> > Vy73,
> > Georg, dl2kp
> >
> > On Mo, 2009-03-23 at 10:40 -0500, Graham / KE9H wrote:
> >   
> >> ***** High Performance Software Defined Radio Discussion List *****
> >>
> >> All:
> >>
> >> In Phil's tutorial on SDR design, he showed that SNR increases 
> >> by 6 dB, or effective bits of resolution increases by 1 bit of 
> >> resolution for every decimation by a factor of 4.
> >>
> >> My question is... 
> >> What happens to spur free dynamic range during decimation?  
> >> Does it also improve? or does it stay constant?
> >>
> >> --- Graham
> >>
> >>
> >> _______________________________________________
> >> 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/
> >>     
> >
> >
> >   
> 


 1237885432.0