[Eeglablist] Very large amplitude values - STUDY design

[Arnaud Delorme]

Dear Marco,

Your calculation is correct. EEGLAB uses the pwelch Matlab for computing power in STUDY (and for single datasets as well) by default. Power in pwelch is in dB/rad/sample so this might be at the source of the scaling problem you are encountering. You might want to look at the Matlab discussion lists on this subject.
Best regards,

Arno

On Jul 11, 2012, at 9:24 AM, Marco Montalto wrote:

> Any feedback re. problem below?
> 
> Thanks!
> 
> Marco
> On 09 May 2012, at 20:27, Marco Montalto wrote:
> 
>> Dear Arnaud,
>> 
>> By any chance have you any feedback regarding the problem I reported in an earlier email (pasted below)? Thanks a lot for all the advise you may give me.
>> 
>> Regards,
>> Marco
>> 
>> Begin forwarded message:
>> 
>>> From: Marco Montalto <montaltomarco at onvol.net>
>>> Subject: Fwd: Very large amplitude values - STUDY design
>>> Date: 18 March 2012 18:51:50 GMT+01:00
>>> To: Arnaud Delorme <arno at ucsd.edu>
>>> 
>>> In reference to my previous email, I am hereby adding an example from Channel Fp1, Subject 1, Condition 1, at 8 Hz:
>>> 
>>> Power value, log (dB) scaled: 
>>> 54.27199936
>>> 
>>> Power value, now transformed to absolute power value (10^(Value/10):
>>> 267423.7262
>>> 
>>> Square root of the above absolute power value in order to get amplitude:
>>> 
>>> 517.1302798
>>> 
>>> 
>>> One question: should I be moving the decimal point by one maybe? Because then the amplitude values make sense.
>>> 
>>> Thanks very much and looking forward to your reply.
>>> 
>>> Regards,
>>> Marco
>>>  
>>> Begin forwarded message:
>>> 
>>>> From: Marco Montalto <montaltomarco at onvol.net>
>>>> Subject: Very large amplitude values - STUDY design
>>>> Date: 17 March 2012 12:13:27 GMT+01:00
>>>> To: Arnaud Delorme <arno at ucsd.edu>
>>>> 
>>>> Dear Arnaud,
>>>> 
>>>> I have just one more question: in STUDY design, after extracting the log(dB) scaled power values from 8-13 Hz (with a bin resolution of 0.5 Hz) and then transforming the values into absolute power values (10^(Power Value/10), if I then take the square root of the resulting values to check the amplitude, the values are very high, beyond 150 microvolts, even though I am 100% sure that none of the epochs exceed ±100 microvolts. Would you know why this is so? Am I missing out something?
>>>> 
>>>> Thanks once again for any information you may supply me with.
>>>> 
>>>> Regards,
>>>> Marco
>>> 
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20120712/a31b98ef/attachment.html>