Re: Hilbert transform

[Ben Abbott]

On Jul 7, 2012, at 10:25 AM, Sergei Steshenko wrote:

> ----- Original Message -----
>> From: Ben Abbott <address@hidden>
>> To: Sergei Steshenko <address@hidden>
>> Cc: Juan Pablo Carbajal <address@hidden>; "address@hidden" <address@hidden>
>> Sent: Saturday, July 7, 2012 7:11 AM
>> Subject: Re: Hilbert transform
>> 
> [snip]
>> 
>> The hilbert function relies upon the fft(). Which implies the signal being 
>> transformed should be periodic.
>> 
>> You'll need to add the negative frequencies to the frequency_response.
>> 
>>     frequency_response = [frequency_response, 
>> fliplr(frequency_response(2:end))]
>> 
>> And then use the Hilbert transform to recover the phase information.
>> 
>> Ben
>> 
> 
> Thanks a lot, that _really_ helps with the Butterworth filter example.
> 
> However, it makes things worse with my experimental data. The data may be 
> problematic - it intentionally has measured data up to _half_ Nyquist 
> frequency (sample_rate/4), so data above it is noise. Maybe this noise 
> baffles the transform.

Can you attach the data?

Ben