Maker Pro
Maker Pro

noise quuestion - µV vs nV sqrt(hz) ?

L

leye

Jan 1, 1970
0
How does one compare noise figures expressed in µV v.s. more famillar
nV SQRT(Hz) ?

More precisely, if MAX412 has 2.2 nV SQRT(Hz) and NJM2068 0.56µV -
which one is noiser under same conditions (frequency, impedance,
temerature, etc).


Thanks,


l.
 
R

Robert Baer

Jan 1, 1970
0
leye said:
How does one compare noise figures expressed in µV v.s. more famillar
nV SQRT(Hz) ?

More precisely, if MAX412 has 2.2 nV SQRT(Hz) and NJM2068 0.56µV -
which one is noiser under same conditions (frequency, impedance,
temerature, etc).

Thanks,

l.

If the noise of a device is speced in uV and another specd in
nV/rootHz, thenone must beconverted to the other by applying the same
bandwidth criteria.
For example, the uV spec should have a bandwidth stated or speced, so
nV/rootHz could be calculated.
Likewise, one might be able to correctly apply that bandwidth to the
nV/rootHz spec and calculate the total noise.
That takes care of the frequency part you asked about.
Impedance (of the source) is a totally different aspect and may or may
not be defined.
And nobody has ever specified noise at a temperature other than room
temperature, so that aspect cannot be relevant.
 
L

leye

Jan 1, 1970
0
If the noise of a device is speced in uV and another specd in
nV/rootHz, thenone must beconverted to the other by applying the same
bandwidth criteria.
For example, the uV spec should have a bandwidth stated or speced, so
nV/rootHz could be calculated.
Likewise, one might be able to correctly apply that bandwidth to the
nV/rootHz spec and calculate the total noise.
That takes care of the frequency part you asked about.
Impedance (of the source) is a totally different aspect and may or may
not be defined.
And nobody has ever specified noise at a temperature other than room
temperature, so that aspect cannot be relevant.

Well, depends who defines it, ISO says 18deg C, most manufacturers 20.
Then, impedance 50, 75 or 110 ohm. When expressed in µV some bandwidth
curves tend to be quite exotic - just pointing out how hard is to
compare anything when manufacturers chose methods that flatter the
most.
Guess it would be easier just to put ICs in the socket and A/B them.

l.
 
Top