Archived from groups: rec.audio.pro (
More info?)
On Tue, 16 Aug 2005 17:04:38 +0200, Robert Angst
<robert.angst@tu-berlin.de> wrote:
>
>>
>> ** No - all you need is two op-amps.
>
>Ok, another thougt:
>if I connect the symetrical output of the two OPs into a single ended
>input (e.g. RNC) with + hot, - and shield grounded, I will loose the
>entire benefit of the differential construction between mic and output.
>If I'd leave the shield unconnected it should work. Is this correct?
>
>I sure want to be fully balanced but want to keep compatibility with
>single ended gear as well. Do I in that case have to go the classic
>route: first differential amp and then splitting it up again in two
>differential pairs?
I was thinking there was something 'bad' about your original idea
of keeping the positive and negative signals separate, and now you've
mentioned it.
Another reason(s) to combine them is to cancel out any common-mode
signal from the input. Passing on such a common-mode signal reduces
the maximum signal (if the signal path can handle 15V peaks and your
common-mode noise is 10V peak, your biggest signal is 5V before
clipping. It also relies on the next device having good CMRR to cancel
common-mode stuff, which if the next device has an unbalanced input,
it obviously won't have.
You absolutely need the differential-conversion stage (with
highly=precise, matched resistors to get good CMRR) for a mic preamp,
as the mic signal could be a few millivolts and the common-mode signal
can be several volts. To turn the situation around so that the mic
signal is much larger instead, you need to cancel the common signal
with a very high precision.
The instrumentation amplifier configuration helps by giving (up to
about) 1,000 times gain for differential signals, and only a gain of 1
for common-mode signals, but you still need the differential stage to
cancel out the common-mode signals.
>In that case I could include an insert jack for my
>RNC fairly easy..
>
>As an OP I would use an LT1469/8. I plugged them in my SX202 (as
>suggested in this NG) and realy like the results.
-----
http/www.mindspring.com/~benbradley