Low Pass filter and DPSK params

[david]

Hi,

I have questions about the parameters in the DBPSK and Low Pass Filter
blocks in GRC.

Low Pass Filter block:
1) If I set the lowpass filter block to interpolate by a factor of 10
should the sampling rate parameter by set to the input sampling rate or
the output sampling rate since they will obviously differ by a factor of
10?
2) If I set the lowpass filter block to interpolate by a factor of 10 is
the filtering operation performed before interpolation or after
interpolation? (Can the filter get rid of the spectral copies centered
at multiples of the input sample rate that are created during
interpolation or do I need another lowpass filter afterwards to remove
these spectral copies?)

DBPSK block:

1. What exactly is the Samples / Symbol parameter mean? I am trying
   to figure out what the relationship is between the input sample rate
   (bytes per second) to the output sample rate in (complex samples per
   second).
2. Since the input is bytes is each bit of the byte converted to a
   separate symbol meaning get 8 symbols for each imput sample(each byte)?

Thanks,
Dave

[david]

On 4/12/2010 3:25 PM, David B. wrote:

interpolation? (Can the filter get rid of the spectral copies centered
at multiples of the input sample rate that are created during
interpolation or do I need another lowpass filter afterwards to remove
these spectral copies?)

One of my favorite things about GRC is that it allows you to quickly
create sample programs to test stuff just like this. You can lay down a
signal source, interpolating filter, and FFT and scope sinks to see what
happens when you adjust the parameters.

Specifically, 1) yes, the sample rate should be at the output rate of
the filter (it’s generally true that you want to build your filter at
the highest sampling rate), and 2) yes, an interpolating filter will
interpolate before filtering to remove the spectral images (in actual
fact, the implementation details are more complicated, but the effect is
the same that the spectral images are removed).

DBPSK block:

1. What exactly is the Samples / Symbol parameter mean? I am trying
   to figure out what the relationship is between the input sample rate
   (bytes per second) to the output sample rate in (complex samples per
   second).
2. Since the input is bytes is each bit of the byte converted to a
   separate symbol meaning get 8 symbols for each imput sample(each byte)?
   Thanks,
   Dave

With BPSK, yes, you get 8 bits out and each of those bits is converted
to 1 symbol. We then interpolate this by the samples_per_symbol >= 2 in
order to satisfy Nyquist for the pulse shape filtering. So each sample
out of the shaping filter is represented by samples_per_symbol number of
samples. Play around with this value in GRC and see the effects of using
2, 4 or 8 samples per symbol in both the frequency and time domain.

Tom