>I don't have much experience specing out op-amps. What does nV/rtHz
>mean? nanovolts per ?hertz? Guess the lower the number the better?
Henry:
It means what it says. If this is being used in the front end of a DC
receiver, 4 nV/SqrtHz really only means anything given a bandwidth. If
my receiver is going to have an audio filter 500 Hz wide, then the noise
of this op-amp referenced to the input side of the op-amp is:
4 * Sqrt(500) = 4 * 22.4 = 89.4 nV of noise
Now you need to include the series input resistors on the op-amp also.
If these were 1K and there were one one the + side, and one on the -
side,
you have 2K total (at least that is what the app note said). The noise
contribution of those resistors are 0.13*Sqrt(R)*Sqrt(Bandwidth) with a
units
of nV.
0.13 * Sqrt(2000) * Sqrt(500) = 130 nV of noise.
Notice that the 1K resistors have more noise than the op-amp! Since the
noise
is uncorrelated, the total input noise is the RMS sum or:
Sqrt(89.4*89.4 + 130*130) = 157 nV of RMS noise.
Now if the receiver has 100 db of gain (100,000 x) the "white noise" of
the
reciver at full gain will be 100,000 * 157e-9 = 15.7 mV RMS of noise
which will
indeed be audable.
On the other hand nV/SqrtHz needs to be closely evaluated. The data
sheet for
the op-amp will show a plot for noise vs frequency. Sometimes, a
non-audio
op-amp will show the rated noise level only at a higher frequency such
as 1 KHz.
The noise level is typically worse at lower frequencies, improving at
higher
frequencies. For cw type audio, I would like to see low noise
performance starting
at 200 Hz or so.
If the op-amp is used on the speaker side of the volume control (after
the normal
audio filter shaping), simple audio filtering such as a cap in parrallel
with the
feedback resistance will help kill the gain at high frequencies so that
you don't
hear very broad band, high frequency hiss.
- Dan Tayloe, N7VE; Phoenix, Az; Az ScQRPions
|