effect of over sampling rate on the meaurement

[chandhuu]

Hello,

please let me know the following things.

1)Effect of over sampling rate  on the measurement.

2)what is the criterion for selection of sampling rate for acquisition of transient signals? for example rise time of the signal or highest frequency of signal etc...

awaiting for ur precisious suggestions

regards

chanduu

[CCornish]

The normal way to determine sampling rate is the Nyquist Theorem.

Determine the highest frequency signal you are trying to capture/view, multiply by 2 and you will have a good starting point for sampling rate.

For example... highest frequency I am trying to view/capture is 500Hz, I would want to sample greater than or equal to 1000Hz.

I have never found oversampling to be a problem, unless it allows me to see that a signal is doing something I was not expecting.

[JÞB]

The answer depends entirely on what measurements you want  to take and on what type of signal.  If you are measuring frequency of a sine wave increased sampling rates improve resolution.  if you are measuring an RMS voltage sample rate it will have no significant impact on accuracy (or trivial ones that are hidden in the rest of the measurment system for the purests).  For DC of a square wave raising samle rates first adds speed of the measurement until you pass 2x the signal frequency then you increase error!

What are you trying to measure?

[MattBradley]

From my experience, the minimum that you want to sample at is 2.56 x the maximum frequency of the signal.  If you measure at the Nyquist frequency, you will preserve frequency information but not amplitude.  If possible, I like to be at least 10 x the maximum frequency of the signal.  Another consideration that must be addressed is the length of the sample.  This will impact the resolution of any frequency analysis, so we would need to know more about what you are looking for there.  Again, I like to extend the sampling to give a resolution at least twice as good as the desired resolution.

Nobody ever said they wished they took less data.  OK, maybe somebody said it, but a lot more said they wished they took more.

[chandhuu]

Hello,

let me put my question in different words.. Suppose for an experiment, test duration is 1-2 milli sec( which includes tansient and steady state), max frequency of the signal is around 10Khz, during transident rise time of the signal is 7-8 micro sec. I want to capture the transient and steady state portion of the signal. what is the sampling rate i have to put for this perticular problem.

[RavensFan]

You would probably want to measure every  microsecond which would mean a 1 MHz sampling frequency.

10 kHz means a samle every 100 microseconds (1/10th millisecond).  You would completely miss the transient rise and would only have about 20 data points for a 2 millisecond test.

[Coq_Rouge]

You are only doing measurements for a limited period of time. So I would say use as high sample rate as your daq board permits. Also remember that the max sample rate in the manual almost always is aggregate.