HELP !!!!!!!!!!!!!!!about the windowing technic

[alex123]

HI ,every 1:
a two component signal consists a primary sinusoid of large amplitude and a secondary periodic nonsinusoid of much smaller amplitude.say use a Chebyshev window, how to estimate the MAG of the fundamental and harmonic Freq of the Second signal,reffered to the Mag of the Primary signal Freq. how to estimate the ratio of the fundamental Freq of the secondary signal to the Freq of the primary signal. what can be said about the waveform of the secondary signal?
and what is the inherent limitation of windowing?

the spectrum is attached
Thanks !!!!!!

 

[brmadhukar]

Hi,
1) You cannot say anything about the time domain nature of the second waveform just looking at the magnitude spectrum. However with phase spectum you may be able to.
2) The windowing has a limitation that the Bandwidth of the individual frequency bins increases. This is the reason why the reactangular window is more popular in signal detection applications.

B R
M

 

[masai_mara]

first convert the db value to linear scale and then divide the resultant value by the gain of the window you used to get the actual magnitude.
The fundamental freq of the primary signal is just what is there as its a pure sinusoid but for the second signal take the lowest freq number(apprx 25 khz as I see)
About the wave form of secondary signal we can see that it has got upto 4 harmonics within your sampling range and each have decreasing magnitude starting from the fundamental.
The inherent limitation of windowing is simply that its a window! and can only include a finite samples of input for estimation.

 

[alex123]

thanks , all. i still cann't understand.
1. how to measure the magnitude from that gragh
2. how to measure the freq of the primary and secondary signal?
from the x- axis of picture, i got 6 freq approx
30 khz,50-70 khz, 170 ,280 ,380 ,500
are these freq = primary' fundamental, secondary's Fundamental, secondary's 4 harmonics?

 

[brmadhukar]

Hi,
1) You just read the value from the graph. But I do not really know why you should worry about the second peak. It is so low. (Q: How are you capturing the signal?. You seem to have avery high value)
2) Frequency offcet. You have to measure the number of bins and multiply by the resolution.

If you are worried about the difference in the graphs it is expected as the noise power you measure reduces with resolution bandwidth.
brmadhukar