power in a frequency range: bandpass or PSD?

I need to estimate the power in a signal in the frequency range of 30
to 40 Hz.

1) should I simply create a butterworth bandpass filter to estimate
power in that range? Or, do I need to estimate activity in this range
using power spectral density estimation (using pwelch or similar
approach)


2) what is the critical difference between bandpass filtering a signal
and getting estimate of activity in a frequency range using PSD?


Thanks,

IT

On Mar 10, 6:05�pm, inverse.tang...@gmail.com wrote:
> I need to estimate the power in a signal in the frequency range of 30
> to 40 Hz.
>
> 1) should I simply create a butterworth bandpass filter to estimate
> power in that range? Or, do I need to estimate activity in this range
> using power spectral density estimation (using pwelch or similar
> approach)

You can do either.

> 2) what is the critical difference between bandpass filtering a signal
> and getting estimate of activity in a frequency range using PSD?

The difference between the two is mainly
a matter of on-line or off-line procesing.

With both methods you might need to calibrate
the system, if you want to process measured data.

Rune

On Mar 10, 1:05 pm, inverse.tang...@gmail.com wrote:
> I need to estimate the power in a signal in the frequency range of 30
> to 40 Hz.
>
> 1) should I simply create a butterworth bandpass filter to estimate
> power in that range? Or, do I need to estimate activity in this range
> using power spectral density estimation (using pwelch or similar
> approach)
>
> 2) what is the critical difference between bandpass filtering a signal
> and getting estimate of activity in a frequency range using PSD?
>
> Thanks,
>
> IT

Without further details, I'd say the filter is the most
straightforward of the two approach, each of which can work. With a
filter, you get an output for every input sample. With a PSD, you only
get an estimate after a block of samples have been consumed. There is
also the matter of lining up the PSD bins with the desired band.
Depending on your sample rate and platform, that might dictate an
unfriendly or overly large FFT size.

John