1

I have a file CSV which contains DSBSC modulated signal data.The first column represents time and second column represents amplitudes. I have to demodulate this DSBSC signal to obtain message signal.
The DSBSC signal has a DC offset. By plotting the data, it looks like a positive offset is added to positive amplitude values and negative offset is added to negative offset values.

The plot is as shown in the link below.

DSBSC signal with DC offset plot

As both positive and negative offset is present, finding DC offset by calculating mean was not helpful. How do I calculate the offset for this signal?

karthik
  • 29
  • 5
  • You may be confusing terms. DC is not changing, therefore there is no DC offset that changes depends on the data. Also, demodulation means that there is an enveloping sing wave of lower frequency, but that means its not DC. Also, I can't see the DC offsets you mention in the figure, care to elaborate? – Ander Biguri Aug 19 '20 at 14:04
  • @AnderBiguri The low frequency envelop is a cosine signal which can be seen in that attached image. Looking at the plot, I thought a constant value is added for the positive amplitudes and subtracted for negative amplitudes(not sure if this can be called as DC). Now I want to demodulate the DSBSC signal to obtain the cosine signal. Because of offset when I multiply the DSBSC signal with the carrier signal(which is the first step for demodulation), I am not getting the envelop of cosine signal. – karthik Aug 19 '20 at 14:44
  • So, in my analysis, first I have to remove the offset from the DSBSC signal and then multiply it with the carrier and then use a low pass filter to obtain the cosine signal. For this process, I need to find out the offset value to subtract from the DSBSC signal. – karthik Aug 19 '20 at 14:48
  • The DC offset and the mean are the same thing. Therefore, if subtracting the mean does not solve the problem, then the problem isn't a DC offset. Can you attach your data? – Harry Aug 19 '20 at 15:42
  • @Harry : Please find the data file DSBSC_DATA.csv in the below link. https://github.com/Karthiknimit/publicfile/blob/master/DSBSC_DATA.csv – karthik Aug 19 '20 at 16:33
  • The mean is approximately -0.005073945674. When I subtract this, it looks pretty well centred on 0. I think this should be comfortably good enough to demodulate. What specifically is your problem? – Harry Aug 19 '20 at 21:06
  • @Harry: For me,After the mean is subtracted the data in not well centered on 0, which means the offset is not removed completely. Considering after mean subtraction, the offset is removed, I tried to demodulate the signal with carrier sin signal of frequency 9892.896[sin(2*numpy.pi*9892.896*time)]. The demodulated data is not as expected. – karthik Aug 19 '20 at 21:53
  • From the plot, I can make out the offset is been added for positive amplitudes and subtracted for negative amplitudes which you have mentioned. In this case taking mean will not help as positive offset and negative offset adds up to 0. – nithin Aug 19 '20 at 21:58
  • @Harry I speculate the smallest DC offset i.e, -0.00507 is because of the noise present in the signal. – karthik Aug 19 '20 at 22:16
  • I don't see any particular difference for positive and negative amplitudes. The signal looks fine to me. – Harry Aug 20 '20 at 13:25
  • @karthik it looks like your [DSBSC_DATA.csv](https://github.com/Karthiknimit/publicfile/blob/master/DSBSC_DATA.csv) file has a very small mean (~ -0.0057 or ~1/60th of the peak-to-peak amplitude), unlike the [graph in your post](https://i.stack.imgur.com/6W3YJ.png) which seems to have a mean of ~ -0.05 (or ~1/6th or the peak-to-peak amplitude). Also, with the csv file provided the demodulation looks more like a sinusoidal if using a frequency of 9891.42 rather than 9892.896. – SleuthEye Aug 25 '20 at 01:44

0 Answers0