Why not infinite noise?

[Chris Carlen]

Hi:

Off-topic electronics question:

1/f noise density is asymptotically approaching infinity towards 0Hz.
Then why isn't measured noise infinite in reality? It seems this would
imply that a stable DC amplifier is impossible. That one should measure
random output fluctuations limited only by the rails. Yet this is not
seen. Why? If not, then there effectively must be some horizontal
asymptote to the 1/f noise distribution.

I want to compute the RMS noise for an instrument amp with DC-20kHz BW
for instance. What noise density do I use? The 1/f corner frequency
value? For an LT1167 example, with G=1 this is about 100nV/(Hz^0.5), so
I'd come up with about 14uV.

But I am unsure if this is the right noise density value to use.

Comments appreciated.


Good day!


--
_______________________________________________________________________
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA
[email protected] -- NOTE: Remove "BOGUS" from email address to reply.

 

[Phil Hobbs]

Hi:

Off-topic electronics question:

1/f noise density is asymptotically approaching infinity towards 0Hz.
Then why isn't measured noise infinite in reality? It seems this
would imply that a stable DC amplifier is impossible. That one should
measure random output fluctuations limited only by the rails. Yet
this is not seen. Why? If not, then there effectively must be some
horizontal asymptote to the 1/f noise distribution.

I want to compute the RMS noise for an instrument amp with DC-20kHz BW
for instance. What noise density do I use? The 1/f corner frequency
value? For an LT1167 example, with G=1 this is about 100nV/(Hz^0.5),
so I'd come up with about 14uV.

But I am unsure if this is the right noise density value to use.

Comments appreciated.


Good day!

It depends on what you're trying to do. Flicker (1/f) noise does not go
off to infinity at f=0, and neither does drift, even though its
transform theoretically goes as 1/f**2. Applying infinite time
integrals to these simple functional forms implies unphysical
assumptions about the system, especially the assumption that it has
existed from time t=-infinity, and will never be turned off. Usually a
bit of clear thinking will solve the problem: e.g., if you're trying to
specify how much noise you'll see if you glance at the scope, integrate
the noise spectrum over the duration of the glance. That will give you
a nice finite dc value.

Cheers,

Phil Hobbs

 

[Roy McCammon]

Hi:

Off-topic electronics question:

1/f noise density is asymptotically approaching infinity towards 0Hz.

You may want to look up old postings about this. The upshot
is if the unit only operates continuously for 1 second,
then the use 1 Hz for the lower frequency limit. If it operates
continuously for a year then 1/year is the lower cutoff
frequency. If you recalibrate once every six months then use
2/year as the lower frequency. You will find the numbers
are reasonable.

The rational is that at the moment you turned on the
unit (or zeroed it), all the fourier components in the
noise had to add up to zero and it takes a darn long
while for them to develop enough phase difference to
add up to something egregiously large.

 

[Nicholas O. Lindan]

1/f noise density is asymptotically approaching infinity towards 0Hz.
Then why isn't measured noise infinite in reality?

But it is. Big bang was one hell of a noisy event, n'est ce pas?

It seems this would imply that a stable DC amplifier is impossible.
That one should measure random output fluctuations limited only by the rails.
Yet this is not seen.

You aren't waiting long enough.

It's like flipping coins. You have to flip a whole lot of times to
get a string of 1000000 heads. Though in the build up you will encounter
2 strings of 999,999 heads, 4 strings of 999,998 heads ....

And, the fluctuations are not limited by the rails.

It's cosmic, man. Cosmic.

 

[Rich Grise]

You may want to look up old postings about this. The upshot
is if the unit only operates continuously for 1 second,
then the use 1 Hz for the lower frequency limit. If it operates
continuously for a year then 1/year is the lower cutoff
frequency. If you recalibrate once every six months then use
2/year as the lower frequency. You will find the numbers
are reasonable.

Of course! It's obvious when you put it that way.

The rational is that at the moment you turned on the
unit (or zeroed it), all the fourier components in the
noise had to add up to zero and it takes a darn long
while for them to develop enough phase difference to
add up to something egregiously large.

So, then, I guess, 1/eternity = The Big Bang, right?

Cheers!
Rich

 

[ChrisGibboGibson]

How can you have infinite noise at DC ?

Surely by definition DC is noiseless? Otherwise it wouldn't be DC

Am I drunk?

Gibbo

 

[Rich Grise]

How can you have infinite noise at DC ?

Surely by definition DC is noiseless? Otherwise it wouldn't be DC

Am I drunk?

Gibbo

The Universe couldn't contain a pulse of infinite amplitude, so it
broke up into billions and billions of pieces in a huge explosion.
These pieces have been trying to reconstitute themselves ever since.
;-)

Cheers!
Rich