Academic Company Events Community Support Solutions Products & Services Contact NI MyNI
This Document is not yet Rated

How Do I Improve The Frequency Resolution of My AutoPowerSpectrum Function?



Primary Software:
Primary Software Version: 6.0
Primary Software Fixed Version: N/A
Secondary Software: Measurement Studio>>Professional Edition (Base Package), Measurement Studio>>Standard Edition (Starter Kit)

Problem:
I am trying to measure the power spectrum of my signal at frequencies less than 1Hz but my frequency range goes to 50kHz and the resolution of frequencies is not accurate enough for my application.

Solution:
There are two key parts of this discussion: the logistics of using the AutoPowerSpectrum function and also general usage tips to obtain specific data from the function.

First of all, AutoPowerSpectrum computes the single-sided scaled auto power spectrum of a time domain signal. The AutoPowerSpectrum function follows the formula { FFT(x) FFT*(x) } / n^2. There is also a wealth of function-specific information in the Measurement Studio Reference Help located at Start»Programs»National Instruments»Measurement Studio»Help»Measurement Studio Reference.

In general, there are three main points to observe when using this function. To adequately handle these three points we need to know the frequency range of interest and the resolution of interest.
  1. Implement a Low Pass Filter in either hardware or software to filter out frequencies outside of your desired range. This provides several advantages, including eliminating aliasing of higher frequencies when sampling at a lower frequency and reducing the overall power of the signal which could help reduce noise.
  2. Sample at a frequency only 2-5 times higher than the highest frequency of interest in your range. The bandwidth or high limit frequency of the AutoPowerSpectrum function will be 1/2 of your sampling frequency. If you are not using low pass filters then you will need to sample at 2 times the highest frequency component in your signal to prevent aliasing. This effectively decreases your resolution for the same number of samples since the AutoPowerSpectrum is now dealing with a larger range.
  3. Increase the number of samples acquired from your signal to increase the resolution within your range (df = fs/n).

Therefore, decreasing your sampling rate or increasing the number of samples will increase your resolution.

Related Links:
Developer's Zone Article: Resolution Bandwith(RBW)


Attachments:





Report Date: 01/24/2003
Last Updated: 12/28/2011
Document ID: 2TNAC86I

Your Feedback! poor Poor  |  Excellent excellent   Yes No
 Document Quality? 
 Answered Your Question? 
  1 2 3 4 5
Please Contact NI for all product and support inquiries.submit