PXIe 5450 Multitone Spectrum

[mbothra]

Hello,

I am trying to generate multitones using arbitrary waveform option on PXIe 5450. I referred the example Fgen Multitone Waveform.vi. The way I understand it is that you can generate a multitone waveform at the desired sample rate and some number of samples. This waveform is then sent to 5450's memory and then it generates the signal by repeating it over time.

The problem I am having is that the spectrum of the generated signal isn't as pure as I get with the standard waveform generator on the 5450. I see lot of spectral leakage that changes with number of samples and the signal frequency. (I plot the FFT of the signals coming out and it is not as good as the standard waveform generator)

What is the impact of number of samples ? I tried coherent sampling, but that didn't help either.

Thanks

Mayank

[ZannahBanana]

Hi Mayank,

I think this excerpt from the manual will help explain what is going on with your signals. Specifically, it gives information in the second paragraph regarding the energy level of multitone measurements and the respective effect on noise sensitivity.

http://zone.ni.com/reference/en-XX/help/371361M-01/lvanlsconcepts/lvac_swept_sine_versus_multitone/

The third paragraph gives a bit more description on what can be done to mitigate the noise you are seeing. The FFT frequency resolution is limited by measurement time, so I am thinking it depends which metric you are using to measure the signal - measurement time, number of samples, or frequency.

I hope this gives a good explanation of what you are seeing!

Cheers,

Hannah

Applications Engineer

National Instruments

[mbothra]

Thanks for the reply Hannah.

It is an informative article, but not very applicable to the problem I am facing. I am collecting a good number of samples and the oscilloscope suggests that the signal has settled well enough (besides there is a >1s gap between triggering the signal generation and collecting samples)

To make a fair comparison, I did a pure single tone (standard waveform generation) and a multitone waveform generation with only a single tone being generated (second tone amplitude and frequency set to 0). There is a good amount of difference.

I have attached the pictures of results from multitone waveform VIs generating 900kHz and 1MHz separately. Both of them are not quite what you would expect and they also differ markedly.

Mayank

[ZannahBanana]

Hi Mayank,

Have you been able to see how this behavior differs at lower frequencies? Given the lack of power in the multitone signal, I am wondering if the frequency will have an impact on the clarity of the signal. Perhaps try the multitone waveform generation at varying frequencies so we can compare that to the 900 kHz generation. 

Thanks,

Hannah

[mbothra]

Hi Hannah,

I tried again and with 10 million samples setting in the multitone VI, I was getting adequate response at all frequencies of interest.

It would be helpful if you could explain how the "number of samples" setting works.

Thanks

Mayank

[ZannahBanana]

Hi Mayank,

Could you upload screenshots of the different sample sizes? 

Thanks,

Hannah

[mbothra]

Hi Hannah,

Sorry for the delay in replying.

I think I figured out what was going wrong. The "number of samples" in multitone generation works as expected.

The multitone VI creates the number of samples that it is asked to do. If the number of samples is not big enough and if we try to acquire a large number of samples from the 5450, we can see discontinuities in the time domain which creates a lot of spectral leakage in the spectrum. The discontinuity is less likely to happen if we have a very large number of samples like 10 million.

Thanks

Mayank

[ZannahBanana]

Hi Mayank,

No worries for the delay! I was hoping to see if that was the case from those screenshots and I am glad it is working as expected now!

Cheers,

Hannah

National Instruments