frequency range of "power spectrum " function

[bon2b]

I'm using signalexpress 2009. To analyze acquired time domain signal from previous step in the same project, I applied power spectrum function.

The DAQmx Aquire step is setup as following:

Acquisition Mode: continuous samples

samples to read: 1024

Rate(Hz): 1024

With this setup, I expect a frequency range from power spectrum analysis of 0-512Hz. Instead, the power spectrum function gave me frequency components of 0-820 Hz. 

How well can I trust the result from 512Hz on?

I attached the screenshots of:

1)my application project configuration

2) frequency domain result

[dsb@NI]

Without more info, I will have to speculate:

The actual sample rate was probably coerced to the minimum sample rate of the cDAQ module (which is higher than 1024 Hz).

As to confidence in the high-frequency half of the spectrum. You can only have confidence in the alias-free portion of the spectrum. If the cDAQ module has anti-aliasing filters, the rule of thumb is that you have confidence in the spectrum up to Fs/2.56. On National Instruments dynamic signal acquisition (DSA) modules, the anti-aliasing filters are typically sharper (allowing for more usable bandwidth). If the module does not have anti-aliasing, your confidence in any part of the frequency spectrum can only be based on your confidence that the acquired signals were inherently band-limited.

FYI - The Power Spectrum Step does not return the Nyquist bin. The Power Spectrum does not truncate the spectrum to the alias-free bandwidth.

[bon2b]

The cDAQ module I used for data acquisition is NI9234 without current excitation. the spec of NI9234 says this:

data rate range(fs) using internal master timebase:

Minimum.....1.652kS/s

Maximun.....51.2kS/s

So you mean when the data acquisition rate I set in SignalExpress(1024Hz) is lower than NI9234's minimus data rate (1625Hz), the actual data acqusition rate was 1625Hz, and that's why I observed a frequency range of 0-826Hz when using power spectrum function on my acquired data. Am I correct?

[Efrain_G]

You might find this Knowledge Base helpful:

http://digital.ni.com/public.nsf/allkb/593CC07F76B1405A862570DE005F6836?OpenDocument

[dsb@NI]

yes, I think your understanding is correct, and you can verify the actual sample rate by looking at the dt (dt=1/Fs) of the acquired waveforms or looking at the block duration (T=N/Fs).

Also, I encourage you to post your projects and VIs when you post a question.

[bon2b · Accepted solution]

Very useful information, thank you both for helping!