Fundamental frequency is not the maximum peak

Hi. I have a signal with high sampling rate (500MHz). Our fundamental frequency of our transducer is 20kHz which is seen as a peak, but it is not the maximum peak. Why is that? Is it wrong? Isn't the fundamental (driving) frequency supposed to be the max(abs(fft))?

Thank you

[Deepak Gupta]

Hi Brad,

Have you tried doing this whole thing with a sinusoidal signal of 20Khz instead of transducer data? If you get only one peak in your FFT plot with sinusoidal signal, it means you are caluculating FFT correctly.

In that case, your transdusor data has other high frequency components in it other than 20Khz signal.

[Brad]

Hi Deepak. How can I check that?

Does it also depend on sampling rate?

Thank you

[Star Strider]

Assuming your Fourier transform code is correct, and the sampling intervals are constant, the system you are observing appears to have specific eigenmodes at 165, 175, and 185 kHz. It is likely acting as a bandpass filter for those frequencies. Identifying it using the System Identification Toolbox may be appropriate.

[Brad]

Thanks Star for your reply. These are the harmonics of acousic problem we are studying, I guess. So we kind of look for these harmonics.

1. My question is can they be bigger (in amplitude) than the driving frequency? or the driving one must always be the biggest?

2. Also when we use different sensors for detection, for one of them, the driving freq has the biggest amplitude and for the other not so (like the above figure), while we use the same algorithm for both. So does this mean that the second sensor is wrong or not necessrily?

[Star Strider]

As always, my pleasure!

1. The eigenmodes can be bigger than the driving frequency, if the driving frequency is sufficientlyfar from the eigenmodes of the structure you’re studying, and the amplitude of the driving signal is powerful enough to excite them. If the driving frequency was equal to one of the eigenmodes, it could (at least theoretically) destroy the system you’re studying.

2. I wouldn’t say the second result is ‘wrong’, only that you’ll need to investigate the reason the two transducers are giving such disparate results.

Answer 1

[Star Strider]

My pleasure.

I’m not certain what your experimental setup is, and whether the ‘transducer’ is a sensor or exciter, since it can be either.

The system you are studying is likely nonlinear, at least to the extent of modulating the excitation signal. If you give it a broadband stimulus (such as an impulse), its response will favor its own eigenmodes. (For example, consider striking a bell.) The resonant frequency of the sensor is apparently broadband enough to pick up the resonant frequency of the transducer as well as the eigenmodes of the system you’re studying. You likely want the eigenmodes.

Assuming that the system and the system setup remains unchanged, the different sensors likely have different dynamics. One may be acting as a bandpass filter (so it emphasizes its own dynamics) and the other (perhaps the one that produced the spectrum in the plot image) has more broadband characteristics. (Editorially, I would use the sensor less likely to emphasize its own dynamics and more likely to have a generally uniform broadband response.)

[Brad]

Yes, the transducer is an exciting one, doing sonication.

1. You said you prefer the one with uniform broadband response, you mean shown here? but how about the driving frequency not being the max? So it doesn't matter? (because that was my original inquiry).

2. How do we know if the sensor is emphsizeing its own dynamcis? That would lead to a nosiy specturm and a P-f plot full of random peaks? This would be the criteria? and important is it for it to pick up the driving frequency?

Thanks

[Star Strider]

1. Likely the best way to characterise the system is to excite it with a broadband impulse, then use broadband reccording. That will be most likely way to produce the data you need to characterise the system.

2. You probably do not, and I am not certain excatly how to test that.

I am not certain what you are doing or what your objective is. Testing the system with a specific driving frequency will produce the response at the driving frequency. Exciting the system with a broadband (impulse) signal will produce the eigenmodes of the system.