From: firstname.lastname@example.org (Glenn Gundlach)
Subject: Re: Audio noise in diff amps
Date: 19 Dec 2002 21:32:50 -0800
NNTP-Posting-Date: 20 Dec 2002 05:32:50 GMT
"Ban" wrote in message news:...
> Glenn Gundlach wrote:
> > I work in video duplication and we're in the process of moving the
> > equipment to a new building. One of the router inputs is 'black' and
> > no audio. The router input is a differential opamp with 22K resistors
> > to the inputs. In the past, the 'black' input was simply left floating
> > but one of the guys said to short them together. FYI, the noise level
> > floating is -74 dbm (Fluke 8060) with operating level at +4. SO, the
> > question is: is it theoretically quieter shorted or floating? Is the
> > Johnson noise of the input resistors significant? When the inputs
> > float, those resistors are essentially not there so they can't
> > contribute to the noise, can they?
> when you look at the schematic of a differential amp, there are besides the
> input resistors also 1 feedback resistor from output to inv. input and 1
> resistor from non-inv. input to gnd.
> These resistors form now the input impedance and they might be bigger than
> the 22k, in case there is a gain. Lets assume they have 100k for 13dB gain.
> When you leave the inputs open, the effective input resistance is now 100k
> for each input.
> When you connect the 22k to gnd, the effective input resistance is only 18k.
> The noise is thus 7.5dB lower, quite a difference. Even with 0dB gain the
> input resistances are halfed and this results in a 3dB improvement.
> When you connect them together, the effect is not so well defined, as there
> might be no more cancellation and also an offset induced, but basically it
> would be the same as to shorten them to gnd.
Well, I can point out that this particular input is a Pesa Cougar
switcher and while I haven't checked the prints lately, I'm pretty
certain the input is unity gain or less. It has to have 24 dB of
headroom above the +4 operating level.
This was mainly a curiosity issue that I thought would be of interest.
I was sort of right but I thought Mr. Thompson, Mr Hill, & Mr. Aylward
might have comments (rather some of the name calling).
I'm a little unclear on the resistor noise. While it requires a
complete circuit, is the noise thermally induced or does it alter
current flow from an external source? Since there are 2 resistors in
the input network, does that noise level increase by 3 db? 2 discrete
resistors making random noise doesn't double the noise voltage, right?
SO, the original question was: is it more or less noisy to tie the
inputs together. As a practical matter, it is not a problem and has
not been a problem for the 25+ years I have been in the broadcast