Spurious signals in downconverter

Hi,

I have written a matlab code for downconverter.
I have tried like this.

fftsize = 1024;
fi      = 83;             % input frequency
fs      = 200;            % sampling frequency
flo     = 50;             % local oscillator frequency
order   = 15;
df      = 5;              % filter cutoff frequency
dfactor = fs / (2 * df);  % decimation factor
siglen  = dfactor * fftsize;

sequence = 0 : (siglen - 1);
X        = sin(2 * pi * sequence * fi / fs);
cosine   = cos(2 * pi * sequence * flo / fs);
sine     = sin(2 * pi * sequence * flo / fs);
Xc       = X .* cosine;
Xs       = X .* sine;

b = fir1(order - 1, (2 * df / fs),'low', hann(order));
Xi = filter(b, 1, Xc);
Xq = filter(b, 1, Xs);

I = downsample(Xi, dfactor);
Q = downsample(Xq, dfactor);

Z = complex(I, (-1 * Q));
Y = fft(Z, fftsize) / fftsize;

f = 2 * df * linspace(0, 1, fftsize);
plot(f, 2 * abs(Y(1: fftsize))) 
title('Amplitude Spectrum of y(t)')
xlabel('Frequency (Hz)')
ylabel('|Y(f)|')

I observe some spurious signal in the output for any local oscillator
frequecny and LPF frequecy. What am I doing wrong?

Thanks...

	 

_____________________________		
Posted through www.DSPRelated.com

>Hi,
>
>I have written a matlab code for downconverter.
>I have tried like this.
>
>fftsize = 1024;
>fi      = 83;             % input frequency
>fs      = 200;            % sampling frequency
>flo     = 50;             % local oscillator frequency
>order   = 15;
>df      = 5;              % filter cutoff frequency
>dfactor = fs / (2 * df);  % decimation factor
>siglen  = dfactor * fftsize;
>
>sequence = 0 : (siglen - 1);
>X        = sin(2 * pi * sequence * fi / fs);
>cosine   = cos(2 * pi * sequence * flo / fs);
>sine     = sin(2 * pi * sequence * flo / fs);
>Xc       = X .* cosine;
>Xs       = X .* sine;
>
>b = fir1(order - 1, (2 * df / fs),'low', hann(order));
>Xi = filter(b, 1, Xc);
>Xq = filter(b, 1, Xs);
>
>I = downsample(Xi, dfactor);
>Q = downsample(Xq, dfactor);
>
>Z = complex(I, (-1 * Q));
>Y = fft(Z, fftsize) / fftsize;
>
>f = 2 * df * linspace(0, 1, fftsize);
>plot(f, 2 * abs(Y(1: fftsize))) 
>title('Amplitude Spectrum of y(t)')
>xlabel('Frequency (Hz)')
>ylabel('|Y(f)|')
>
>I observe some spurious signal in the output for any local oscillator
>frequecny and LPF frequecy. What am I doing wrong?
>
>Thanks...
>
>	 
>
>_____________________________		
>Posted through www.DSPRelated.com

your signal @ +/- 83 single tone
your osc @ +50 sin/cos
so you get tone @ 33 and 67(200 - sum)
so your filter is just cutting them off

Kaz

---------------------------------------
Posted through http://www.DSPRelated.com

Kanchan,

After the multiplication by sine and cosine you have _two_ replicas of 
the input signal, and so you may want to suppress one of those with a 
low-pass filter.

To do that, it's essential that you
(1) understand what much of a passband do you need and what level of 
attenuation in the stopband do you want,
(2) design a low-pass filter with the desired parameters (note that for 
greater stopband attenuation you may need to increase the filter order),
(3) display the output of your simulations in dB, rather than linear 
units, so that you can compare what you've got with what you've 
originally planned for.

In your example, both replicas of the input signal are in the 
stopband/transition band of the low-pass filter, so you see one replica 
killed by your filter and the other replica not entirely killed.

Hope that helps.

Regards,
Evgeny.


On 02.04.2015 15:42, Kanchan Salvekar wrote:
> Hi,
>
> I have written a matlab code for downconverter.
> I have tried like this.
>
> fftsize = 1024;
> fi      = 83;             % input frequency
> fs      = 200;            % sampling frequency
> flo     = 50;             % local oscillator frequency
> order   = 15;
> df      = 5;              % filter cutoff frequency
> dfactor = fs / (2 * df);  % decimation factor
> siglen  = dfactor * fftsize;
>
> sequence = 0 : (siglen - 1);
> X        = sin(2 * pi * sequence * fi / fs);
> cosine   = cos(2 * pi * sequence * flo / fs);
> sine     = sin(2 * pi * sequence * flo / fs);
> Xc       = X .* cosine;
> Xs       = X .* sine;
>
> b = fir1(order - 1, (2 * df / fs),'low', hann(order));
> Xi = filter(b, 1, Xc);
> Xq = filter(b, 1, Xs);
>
> I = downsample(Xi, dfactor);
> Q = downsample(Xq, dfactor);
>
> Z = complex(I, (-1 * Q));
> Y = fft(Z, fftsize) / fftsize;
>
> f = 2 * df * linspace(0, 1, fftsize);
> plot(f, 2 * abs(Y(1: fftsize)))
> title('Amplitude Spectrum of y(t)')
> xlabel('Frequency (Hz)')
> ylabel('|Y(f)|')
>
> I observe some spurious signal in the output for any local oscillator
> frequecny and LPF frequecy. What am I doing wrong?
>
> Thanks...
>
> 	
>
> _____________________________		
> Posted through www.DSPRelated.com
>

and you are decimating down to 10 (from 200), why??

kaz
---------------------------------------
Posted through http://www.DSPRelated.com