Spectrum analysis, strange behaviour

[mforbiz]

Hi!.

I've created this VI that perform a spectral analysis of a simple sinusoidal signal of wich I can change the frequency in a range from 100 to 1000 Hz.

I've noted that if I choose a frequency's value like 340,350,..., 460... and so on, I see in the spectral graph exactly the expected signal's amplitude value, in this example: 4

Instead, If I set a value like 341, 342, 343,... and so on until the next "round" value I don't see the same exact amplitude value (4) but a lowest value and the graph is more scattered around the max value:

Any idea?

Thanks in advance

[mcduff · Accepted solution]

You take 1000 samples which is 100 ms of data; your frequency resolution is 10 Hz. When your frequency doesn’t fall exactly on a 10 Hz bin, it is spread out over multiple bins depending on the window.

mcduff

[BertMcMahan]

McDuff is right, add another 0 to your sample rate and you'll get perfect 1-Hz matching again.

[altenbach]

You only get a single line if the sample contain an integer number of cycles.

If you don't have an integer number of cycles in the signal, you get spectral leakage.

[mforbiz]

Hi mcduff.

Sorry for delay.

You're right.

I'll try with windowing

Thanks

[mforbiz]

Hi Bert.

I think you intend to add another 0 to number of sample.