 |
| Author |
Message |
|
|
 Posted: May 31, 2012 - 12:11 PM |
|
Joined: Sep 20, 2003
Posts: 4069
Location: Surrey, England
|
|
| Probably a stupid question, but do the input values (the waveform samples) for the Goertzel algorithm need to be centred around zero, or can I use the raw values, complete with DC offset? |
|
|
| |
|
|
|
|
|
|
Posted: May 31, 2012 - 12:18 PM |
|
Joined: Dec 11, 2007
Posts: 6848
Location: Cleveland, OH
|
|
Well, we'll see if anybody has a different idea...
Goertzel is basically a modified DFT, (FFT), that only looks for a few, specific frequencies, instead of the entire spectrum.
A DC offset in the signal would normally show up as the 0 bin component in an FFT.
One would therefore not think that it would impact ones ability to look for the other frequencies of interest.
JC |
|
|
| |
|
|
|
|
|
|
Posted: May 31, 2012 - 12:38 PM |
|
Joined: Jan 08, 2009
Posts: 1152
Location: Lund, Sweden
|
|
| I did some testing with an example in Matlab. It made no difference if I added a (large) DC offset. |
|
|
| |
|
|
|
|
|
|
Posted: May 31, 2012 - 01:17 PM |
|
Joined: Sep 20, 2003
Posts: 4069
Location: Surrey, England
|
|
Many thanks. It avoids having to perform some sort of running average or min/max detection and offset removal.
John |
|
|
| |
|
|
|
|
|
|
|
FAQ
Search
Register
Log in to check your private messages
Log in
Maximize
RSS
