Carrier frequency measurement after resampling

[ThomasKeyser]

Hey,

I'm not an rf expert. Is it possible to measure the carrier frequency of an ASK modulated Signal after resampling it down?

I would like to cover two frequency bands (433.2MHz & 434.64MHz). I fixed the Center frequency of the RFSA to e.g. 433.92MHz with an IQ rate of 2MS/s. Is it possible to get the carrier frequency of the signals after resampling it down to e.g. 9600 x 16Samples per Symbol?

Thanks in advance.

Regards Thomas

[EzerB]

Hello Mr. Keyser,

First of all, let me ensure that I am understanding correctly what you are doing. According to your description, you are using the RFSA API, so you are acquiring the RF signal with a Vector Signal Analyzer (VSA). If you are just interested in measuring the carrier frequencies of your ASK modulated signals, the easiest way would be using the RFSA Soft Front Panel. You just have to enter the center frequency and the span, and then you will see two peaks corresponding to the two carrier frequencies of your ASK signals, if these frequencies fall within the bandwidth of the acquired signal. This is the case for the values you gave us, since you would be acquiring a signal centred at 433.92 MHz with a bandwidth of ca. 0,8*2MS/s = 1,6 MHz. Please find under http://www.ni.com/tutorial/7786/en/ a short tutorial on how to use the RFSA Soft Front Panel. If you need to measure the frequencies programmatically, you should apply a VI which is able to calculate the spectrum of the acquired signal (e.g. the FFT Power Spectrum and PSD VI), so that you can see the peaks on the corresponding carrier frequencies.

I hope I was able to answer your question. Please let me know If you still have any questions regarding this topic.

Best regards,

Ezer Bennour

RF System Engineer

National Instruments

[ThomasKeyser]

Thanks for your reply. I will more explain what I want to do.

At the Moment I acquire IQ data with the RFSA API to demodulate the Signal and to get the data Bytes (payload). I also want to get the spectrum of this Signal to check the carrier frequency Peak. All in one acquirement. Right now I'm using the spectral measurement toolkit to get the spectrum, but directly from the fetched IQ data (2MS/s). I would like to do this after resampling down the iq data (before the demodulation), to save Memory usage for further handling. But this is not working. The frequency values are not correct anymore. Is there a way to do this ? Maybe with some calculation?

Regards Thomas

[EzerB]

Thank you for your explanation. Let me recall some key figures related to your setup:

- You are using a sampling frequency Fs = 2 MS/s. By doing so, the maximal bandwidth of the acquired signal would be Bw= ca. 0.8*Fs= 1.6 MHz, i.e. the highest frequency Fmax of your signal after downconversion should be below Bw/2= 800 kHz.

- The center frequencies of your payload signal are very close to this edge: 433.92 MHz - 433.2MHz = 720 kHz, but since you are using Fs= 2 MS/s, the Nyquist Theorem is fulfilled (Fs> 2*Fmax), so applying an FFT to your fetched IQ data gives you the right spectrum.

- If you downsample your acquired data, the Nyquist Theorem is no longer fulfilled. In your example you are talking about Fs= 9,6 kS/s after downsampling, which is way smaller than 2*FMax. That is why the values of the carrier frequencies after computing the FFT are no longer correct. The spectrum is namely greately distorted since Fs << 2*Fmax

- Since your main goal is to reduce resource usage, I would recommend you to apply the so called zoom FFT to compute the spectrum portion which is relevant for you. The basic principle of the zoom FFT is explained in the following: After downconversion, the signal of interest in the acquired data would be centered at +/- 720 kHz. By applying the zoom FFT, this function shifts the signal of interest to 0, downsamples the signal and then computes the FFT of the signal. The reason why downsampling the signal does not affect the spectrum in this case is, that your signal of interest has a very small bandwidth ( some tens of kHz which is the bandwidth of your modulated ASK signal). Therefore, after shifting your signal of interest to 0, you will just need to fulfill the Nyquist Theorem for the small portion in order to get a valid spectrum, so your new sampling frequency will be in the range of some tens or hundreds of kHz.

- There is a built-in Zoom FFT function in LabVIEW, which is available with the Sound and Vibration Measurement Suite (cf. http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/zoom_fft_analysis/). But it is also possible to implement it on your own if the basic principle is clear. 

Best regards,

Ezer Bennour

RF System Engineer

National Instruments