[LAU] DC offset

[jsuomine]

On 01/30/2012 09:33 AM, Fons Adriaensen wrote:
> On Mon, Jan 30, 2012 at 08:41:28AM +0530, Rustom Mody wrote:
>
>> Immediate context:
>> Im trying to capture from firefox using the .asoundrc/jack settings shown
>> http://jackaudio.org/routing_alsa
>> [Track is http://www.raaga.com/play/?id=183659
>> I would just buy it if it were for sale; its not. You may have to click
>> past an ad]
>
> I just recorded the first minute using the alsa-jack plugin, this
> works perfectly and their is no offset in the signal. So it's not
> the source or firefox or the plugin.
>
>> However Ive noticed earlier that when I use my (cheap headset) mike to
>> record directly into audacity, this similar DC offset occurs. Since I am no
>> audio pro I did not know what to google for before seeing the tabs in the
>> normalize dialog of audacity.
>
> Could it be that while recording from firefox you had the soundcard
> (auto)connected to the same track as well ? That would add the offset
> if the cause is your soundcard (which I suspect), even if the headset
> is not plugged in.
>
>> So if you ask me exactly how much DC offset I dont know how to find out in
>> figures.  The audacity track shows it 1/4th to 1/3rd way up. Center is 0.0
>> top is 1.0 inbetween is 0.5.
>> This is little more than half way between 0 and 0.5.
>
> Which is way too much, and indicates broken hardware.

My old EEE PC 4G has offset in its input connection. This is because the 
microphone in has phantom power connected to it. I think this is 
documented feature, but what is not, is that in computer hardware the AD 
conversion is done before getting rid of the offset the phantom power is 
causing. I think I even checked the schematics to confirm this. So 
everything recorded is  way of and will easily be limited. To put it 
more accurately, if normally offset in scale -MAXINT - MAXINT should be 
0, it would now be something like MAXINT/2. Could you be facing similar 
problem here?

.j