I want to up-convert a baseband sinusoidal signal of frequency $4000 \, \rm{Hz}$ using a carrier frequency of $20\, \rm{KHz}$. I read that one can up-convert a baseband signal to a passband signal using the following.

enter image description here

But after convolution with a carrier cosine, I get a $16\, \rm{KHz}$ signal. What could be the problem?

I used following Python code to do convolution with carrier frequency: samples * cosineSignal.

  • 1
    Multiplying a 4kHz tone with a 20kHz carrier will produce 2 tones, at 16 and 24kHz. What else were you expecting? What is your sample rate? – Juancho Oct 18 '17 at 1:01
  • I want to up-convert the signal to 20 kHz passband signal to send. I'm using 48 kHz samp rate on baseband signal. – JayHeo Oct 18 '17 at 1:34
  • 2
    You should not convolve with a carrier signal. You should multiply! – Rodrigo de Azevedo Oct 18 '17 at 13:04

A 48k sample rate is too low to reliably reproduce your upper side of 20k+4k = 24kHz. Thus you end up with a LSB (lower sideband SSB) signal, not an AM signal.

You are right in that: multiplying your signals in the time domain corresponds to a convolution in the frequency domain. A pure complex exponential carrier is basically a peak (a Dirac delta) in the frequency domain. Convolving with that corresponds to a "shift" - which is how you modulate. So far so good.

However, you forgot that your signals are real-valued sinusoidals, which have a (Hermitian) symmetric spectrum. Your baseband signal in the frequency domain then consists of two peaks, one at $4\,\mathrm{kHz}$ and one at $-4\,\mathrm{kHz}$! (The same goes for your carrier actually, but this is not relevant in this case.) Both peaks are then shifted by $20\,\mathrm{kHz}$, so that you end up with $16\,\mathrm{kHz}$ as well as $24\,\mathrm{kHz}$.

Since you seem to be simulating this, and use a sample rate of $48\,\mathrm{kHz}$ across the board, you might run into sampling issues as explained in @hotpaw2 's answer, which is why you might only see the $16\,\mathrm{kHz}$ peak.

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.