Despite the signal being real, as is seen in the last line of the first picture:

Bandpass-signal Power-spectrum of QAM-signal

, the exemplary spectrum of the signal shows neither an even real nor an odd imaginary part: Exemplary plot of Power-spectrum of QAM-signal

But from my theorems I know that for every real x(t) the real part of the Fourier-Transform is even and the imaginary part of it is odd?


1 Answer 1


You are right in that the Fourier Transform (FT) of every real $x(t)$ has even real part and odd imaginary part. However, your figures only show the absolute value of the bandpass spectrum, so you cant see real- and imaginary part separately. Note that the above figure (showing $|S(f)|$ is the baseband signal, which is indeed non-even and hence complex valued. But, after upconversion, the signal becomes real and if you write a small script that does the calculation of the upconversion, you will see, the FT $S_{BP}(f)$ of the bandpass signal $s_{BP}(t)$ will have mentioned properties. I have once written a walkthrough article of up-down-conversion here which might also help in understanding this process.

To clarify a bit more in math. terms:

  1. $\underline{s}(t)=s_I(t)+js_Q(t)$ is a complex baseband signal, which consists of the real part $s_I(t)$ and the imaginary part $s_Q(t)$.
  2. $S(f)=\mathcal{F}\{\underline{s}(t)\}$ is the Fourier Transform of the baseband signal. This $S(f)$ is arbitrary (i.e. not even or odd) since $\underline{s}(t)$ is a complex-valued baseband signal.
  3. $s_{BP}(t)$ is the upconverted version of $\underline{s}(t)$, and is a real-valued signal.
  4. Hence, $S_{BP}(f)=\mathcal{F}\{s_{BP}(t)\}$ has even real part and odd imaginary part.
  • $\begingroup$ But how can the baseband-signal S(f) be non-even, suggesting that there's a complex signal, when the bandpass-signal clearly is stated as $\ s(t) = s_I(t)cos(2\pif_ct) - s_Q(t)sin(2\pif_ct) \$ and then Fourier-transformed as S(f)? $\endgroup$
    – Starhowl
    Feb 27, 2017 at 9:47
  • $\begingroup$ I added some text. Is it still unclear? $\endgroup$ Feb 27, 2017 at 10:03
  • $\begingroup$ Since the set of {s_I(t), s_Q(t), cos(2\pif_ct), sin(2\pif_ct)} doesn't contain any complex elements, the function s_{BP}(t) is real itself. But every real signal is complex-conjugate symmetric, which isn't the case in the depicted spectrum? $\endgroup$
    – Starhowl
    Feb 27, 2017 at 10:06
  • $\begingroup$ no, $s(t)$ is complex, according to $s(t)=s_I(t)+\mathbf{j}s_Q(t)$. This is the baseband signal, which is complex. Only after upconversion to $s_{BP}(t)$ it becomes real-only. $\endgroup$ Feb 27, 2017 at 10:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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