digitized waveform in frequency domain

[Alann]

Hey,

Is there any way to see the spectrum of the digitized waveform. Actually I want to disable DDC (in NI-5142) and observed the phenomenon of aliasing. 

[Steve_B]

Hi Marry,

You can disable the Anti-Aliasing and Noise Filters by specifying the maximum frequency using the niScope Configure Chan Characteristics property. You can find additional information about this in the NI High Speed Digitizers Help. You can open this on a Windows OS by navigating to Start>>Programs>>National Instruments>>NI-Scope>>Documentation>>NI High Speed Digitizers Help. When it opens you can expand Devices>>NI 5142 Overview>>Signal Conditioning>>Filters. You will see information about the filters and a note about how you can change them using the property I described. You can set set maximum frequency to a value above 35MHz and below the maximum bandwidth of the card, which with the filters off is 100MHz.

[Alann]

Thanks Steve,

I set the parameter to the options given in the guide but I cant see aliasing. The VI is attached. The span of the spectrum is also a big confusion. When I disable DDC, it sets span according to some processsing in SMT VIs which i do not understand. Can you please comment on the working of this VI.

My only interest is to disable the anti-alias filter and observe aliasing.

[Steve_B]

Hi Marry,

I took a look at your code and you are setting some parameters for the DDC though you are disabling it. You should be able to see aliasing as you lower your sampling rate. Adjust your sample rate lower toward your offset frequency and aliasing should occur. You are setting up the Spectral Measurement Toolkit to configure the sample size to work for your measurements. The filters are disabled by setting your max frequency. What settings are you using to see the aliasing and what does your power spectrum look like?

[Alann]

Hey, thanks for looking the code.

In the fig 1, Sample rate is 17MHz (which is rounded to the next supported rate that is 20MHz). Signal frequency in 10MHz. So there should be no aliasing. But there are peaks at 9.8MHz and 10.2 MHz. Why is that?

 

In fig 2, The sample rate is changed to 14 MHz. Peaks occur after every 0.05 MHz.

 

Where as in figure 3. If I set the sample rate below bandwidth. (17MHz is rounded to 20MHz). Power spectrum is different.

[Steve_B]

Hi Marry,

I looked at the images you posted and the spikes in figure 1 appear to be higher in power than the noise, so it appears that you are measuring a signal at roughly 200kHz. In figure 2 it looks like you are measuring a signal that has a frequency of roughly 50kHz and you are seeing the harmonics at multiples of that frequency. Figure 3 looks to be the exact same as figure 1 but the size of the window has changed and you have zoomed out to see a larger frequency range. The window size was changed by changing the span setting.

There are a few things that may make working with your program easier:

     The window size for your FFT is proportional to your sample rate and can be limited below that using the span parameter. Setting your span above your sample rate will not change the window size.

     The frequency shift parameter is the amount that will offset the frequencies in software. This is like adding a value 'X' to every value in the frequency axis of your measurement. So you will actually see aliasing based on the physical signal and its relationship to your sample rate, not necessarily the representation of the signal on your power spectrum graph.

     The reason you see a weak signal at the frequency shift value is because there is a slight DC offset in the physical signal which is being shifted up from DC to the frequency you set, but this shift is only occurring in software.

You should be able to set up a test with a signal generator connected to your digitizer, run this program with the appropriate settings to see the power spectrum, and adjust the frequency of your signal higher until the aliasing begins to occur. You should set your span to a value above your sample rate and have your sample rate low enough to be able to generate a signal at roughly half of that rate. You will see the frequency appear to decrease in your power spectrum even though you are still increasing the frequency once the frequency becomes high enough in relationship to your sample rate. I would also recommend that you turn off the averaging while doing this in order to see the actual frequency that is being generated as it is shifting.