Correct FFT analysis

[Valentin87]

I analyze the Harmonic Distortion Analyzer VI. And I have a question for the correctness of the measuring of harmonics. To the letter I attached VI, in which the calculation of the harmonic amplitudes. The values of the harmonics on FFT graph, in my opinion, too high. For example, the measured 1st harmonic = ~ 83 dB, and in my opinion it should be ~ 77 dB (the difference between the calculated amplitude of the second harmonic (from the graph about 83) and the amplitude of the fundamental tone(from the graph about 6)). Please explain how to evaluate the results of measurements?

[mikeporter]

You could check the help description of the routine for an explanation of how it works. Often with analysis there can be subtle differences in algorithms and setting that can lead to apparent differences. But given the amount of effort that NI puts into verifying and validating its code, if LV says its 83, its 83.

Mike..

[johnsold]

One thing to consider is that the amplitude of a spectral component may NOT be equal to the highest magnitude element in the spectral array due to spectral leakage.  Here is an expanded view of the second harmonic peak.  The data point at 20 MHz is at -83.0546 dB.  The second harmonic component is -79.7999 dB. The estimated frequency is 20.0000626 MHz. See below for how I got these numbers.  Note the asymmetry of the data points near the peak. Both the asymmetry and the non-zero width of the peak are indications of spectral leakage.

I did not dig into the workings of the Harmonic Distortion Analyzer.vi but it may take those things into account. 

When I take the components level output, convert it to dB (re 1 V), and subtract the fundamental value from the others I get the second harmonic -76.50 dB relative to the fundamental. 

Interpreting the spectral analysis of a sampled data signal is not always a straightforward process. If you can tell us what you want to do with the data after you anaylze it, someone may be able to recommend a process which will produce valid results for you.

In the attached VI I got the best results for finding thespectral peaks with a threshold fo -94 dB.  I did not set that as a default value before saving.

Lynn

[Valentin87]

In VI on spectrum the amplitude of fundamental tone = ~6.31 dB. But when calculate the components level dB the amplitude of fundamental tone = ~3.3 dB. Why that values are different and how they linked to amplitude of input signal?