for finite acquisition ,why is it desirable that the sample rate be 10 times the samples to be read

[RexQ]

i m usinng a 6035E daq card and using DAQmx read function in a for loop+i m using a waveform chart to display the acquired waveform , while using the finite sample acquisition mode is it favourable that  the samples per channel be one tenth the sample rate...................is there any hard n fast rule ragarding this ........while using DAQ assistant also we come across such a ratio b/w samples to read and sample rate.

[Wayne.C]

RexQ,

A sampling rate of 10X greater than the frequency being sampled is pretty standard.  It depends on how you intend to analyze the acquired data.  The Nyquist theorem states that a signal must be sampled at 2X to accurately measure its frequency without aliasing.  That becomes the bare minimum sampling rate.  If you intend to measure say the RMS value of a sine wave, then you need more points of data to increase your accuracy.  The same would be true if you were going to use a FFT function.

[GovBob]

I don't mean to hijack this thread but I think I have the same question concerning number of samples.

Let's say I have a signal who's highest frequency componant is 1000Hz. According to Nyquist I need to sample this signal at 2000Hz to determine the frequency. If I want a better representation of the signal I need to sample at 10000 Hz.

Now the question is how many samples? I too have seen a rule of thumb that I would want 1000 samples (1/10) at the 10000Hz rate. The  question is what is the reason for the 1/10 rule?

Message Edited by GovBob on 08-19-2008 04:30 PM

[rpursley8]

A higher sampling rate would be important for time-domain processing because it gives better time resolution (RMS, peak measurements, etc).

A longer sampling time would be important for frequency-domain processing because it gives better frequency resoution (bandwidth, peak frequency measurements, etc).