Re: Trying to understand CIC interpolation

On Jul 4, 10:40 pm, Krishna <krishna.pil...@xxxxxxxxx> wrote:
Dirk,

I was trying to comment on the statement:
2) The CIC interpolation filter is really intended to be used on> > something that is already oversampled, your test signal is not.

I interpreted this as - "The input signal is not oversampled - hence
the code is wrong".

Krishna,

I have no idea why you would make that interpretation. The comment had
nothing to do with the code, just the test signal.

Remember the CIC filter is really a pretty poor filter as a general
lowpass filter. When interpolating, it greatly attenuates images over
a very small region/regions, so the original lowpass bandwidth whose
images (resulting from the CIC filter) you want filtered out must be
very narrow bandwidth to start with. Normally the input signal would
be extremely lowpass bandlimited (compared to its sample rate), either
because it was originally bandlimited, or it was interpolated up to
be, by interpolation methods more appropriate for a signal that has
not been severely bandlimited relative to its original sample rate. A
CIC interpolation filter is often used as the final filter(s) in a
chain of interpolation filters that might be [FIR filter,
interpolating halfband filters, CIC filters], such as part of a chip
that does 'Digital Up Conversion(DUC)'.

The test signal is not in a range that you would normally use a CIC
filter on for interpolation. You would probably only use it because
you know what the output should be for that particular input, as one
of your test signals for the code. A further shortcoming of the test
signal is explained below.

Maybe I made the wrong interpretation :)
In the previous post, I tried to point out that the code has an
upsampling stage inbetween differentiator and integrator stage
(Refer the architecture shown in Figure9 ofhttp://www.us.design-reuse.com/articles/article10028.html).
Hence your statement on 'upsampling not being done' was not accurate.


I never made the statement that "upsampling not being done". What
post are you refering to?

Additionally, I do not understand the comment:

If you
interpolate alternating signed ones (magnitude 1 sine wave sampled attem
peaks, ... 1,-1,1,-1,1,...) and get pairs of alternating signed ones
(...1,1,-1,-1,1,1,-1,-1,,...) that does not correspond to samples from
an amplitude 1 sign wave.


First, excuse my proofreading, that should be "sine wave" not "sign
wave". Also, assuming the test signal is intending to represent a
sinusoid, my implication that it had to come from an amplitude 1
sinusoid is incorrect. In fact, the sinusoidal signal amplitude of
the signal that the test signal could have come from cannot be
determined from the samples, because the signal frequency is half the
sample rate. With the appropriate phase it could be samples of a
sinusoid with signal frequency of half the input sample rate and ANY
amplitude >= 1.0. So you cannot use the test signal to determine a
non-zero system gain at the test signal frequency. On the other hand,
the output from the test signal (assuming a sinusoid) does have an
unambiguous amplitude of sqrt(2).

Repeating the same sample twice is a form of filtering with almost
sinc() shaped transfer
function. Agree?

Repeating the same signal twice is adequate for interpolation only if
your input is already sufficiently oversampled. If you are looking for
a flat requency response, a sinc() shaped transfer function is pretty
poor.

Regards,

Dirk


Regards,
Krishnahttp://dsplog.blogspot.com

.

Relevant Pages

  • Re: Minimum useful sinc length
    ... I think it's important to focus on the interpolation factor and the ... sample rate then sincs can be used to "reconstruct" which means infinite ... Convolution in time with a sinc. ... Perfect lowpass filter of the samples in frequency. ...
    (comp.dsp)
  • Re: Questions regarding polyphase filter
    ... structure and polyphase filter. ...  for example for interpolation. ... What you have here is a polynomial interpolation technique. ...
    (comp.dsp)
  • Re: Questions regarding polyphase filter
    ... as a polyphase filter. ... =A0for example for interpolation. ... What you have here is a polynomial interpolation technique. ...
    (comp.dsp)
  • Re: Questions regarding polyphase filter
    ... as a polyphase filter. ... =A0for example for interpolation. ... polynomial technique, with the order of the polynomial restricted to ...
    (comp.dsp)
  • Re: Interpolation
    ... regardless of how much bandwidth reduction the filter has done, ... If the samples are preserved the filter has ... no clear definition of interpolation or any of the other terminology. ... will still preserve the input samples in the output. ...
    (comp.dsp)
Loading