I would be very interested in your previously unknown data.
I would also like to see a demodulated version that shows the time history phase shifted data from I and Q as my simple sample showed.
Hi,
Regardless of the signal, in general to calculate its spectrum you should follow the procedure of the example here below, do you agree?
Greetings
FM
Hello MF,
Thank you for all the help and for taking your time.
I see you have done two analysis for IQ, please correct me if I am wrong: FFT and discrete FFT.
However, the results in terms of shape and bandwidth are different between the two analysis.
The DFT replica is somewhat resembles the shape of the Matlab and measured results but BW is not correct. It should be 20MHz.
The FFT result is also not showing the correct BW. I used your IQ spectra file and entered sampling time as 2.222E-08 i.e. Sampling frequency is about 45MHz. That gave me a BW of 10MHz instead of the expected 20MHz. I made an attempt to use the log for y-axis, but I had no success. Please see plot below.
For the FFT plot shown above, I left the signal frequency as it was in your file i.e. 1/T2sp. But my understanding is that 1/T2sp should yield the sampling frequency and not signal frequency. Do you think that is contributing the error in the FFT plot?
NB: I measured the LTE 20MHz wide signal in the signal analyser (in IQ domain). The IF BW was set to 25MHz to capture the complete signal.
Hello,
Thank you. I got the markers operation working after activation. I am still struggling of importing Excel data into the table. I resolve this issue by copying and pasting from the previous file.
Regarding my last question, here is my explanation.
IQ Data
In the last part, inverse FFT is used to recover the original signal. The net results are shown below. As you can see there are several observations:
- What is f2k and FEk?
- The results do not correspond to the original IQ after applying inverse FFT. Am I Missing a point?
-
I see that re-center command is used to Shift zero-frequency component to center of spectrum. How does the command know to which samples to re-centre?
- What is mg10, mg11, mg12 etc?
- What is auto-correlation graph showing us? What can we get out of it for analysis point of view?
- Having captured the I/Q data, is there any way to plot on constellation. The reason being is for calculating the EVM from reference I/Q.
- Is there a way to calculate EVM from a given capture I/Q data? The issue I can see is preamble synchronisation between the signal generator and the analyser in terms of where does the constellation starts to determine EVM?
Hi,
The last part is only indicative of what else can be done, since I used cfft, to reconstruct the signal I have to use icfft and not ICFFT. Unfortunately in the reconstruction of the signal with "makeri" the phases are altered, so you get a different signal from the original as you can see in the relative mathcad help. mg10, mg11, mg12, are the Fourier transforms of the autocorrelation functions of the signal I, of the signal Q and of the complex signal IQ, considered as random processes. Furthermore, to represent the IQ signal on the constellation, you first need to quantize each sample, read with a click, and convert from analog to digital ( four bits for example). Then you have to represent it on the corresponding constellation and so on.
