On Jun 18, 2007, at 3:27 PM, Charles M. Brown wrote:
REference: Mon, 18 Jun 2007 23:54:22 +0100
"Nick Palmer" <[EMAIL PROTECTED]> wrote <in part>:
I once tried to point out the error in his thinking that so-
called thermal noise could be rectified to produce useful power
I believe that Johnson noise can be rectified then aggregated. I
will listen intently to your comments on this.
There were some good posts by John Winterflood and others in 2006
about this. Here are some of John's:
On Nov 14, 2006, at 8:54 AM, John Winterflood wrote:
R Stiffler wrote:
...
Carbon resistors generate more thermal voltage
noise than Metal film resistors....
This is not really true. We may divide the noise sources in Carbon
composition resistors into two types:
1) True "Thermal noise" (also called "Johnson" or "Nyquist" noise)
which "is the noise generated by the thermal agitation of the
charge carriers (the electrons) inside an electrical conductor in
equilibrium, which happens regardless of any applied
voltage." (From http://en.wikipedia.org/wiki/Thermal_noise). This
noise source is absolutely fundamental and is completely unvarying
regardless of type of resistor, and its power sourcing capability
is completely unvarying regardless of value of resistance, number
in parallel/series, size, etc. It is simply 4kT watts per Hz of
bandwidth. (The bandwidth presumably goes up to some very high
limit determined by the mean free path of the electrons being
scattered in the resistive conductor).
2) Excess noise (see http://en.wikipedia.org/wiki/Flicker_noise) -
which is generated by current passing through the resistor and may
well be due to thermal (or thermally induced microphonic) effects,
but is not rightly referred to as thermal noise (at least amongst
physicists). It is readily overcome with better technology. The
excess noise present in a carbon composition resistor is produced
by random effects driven by the power fed in and will only be a
very small fraction of this applied power - ie very far from
overunity!
With regard to Johnson noise, if you short or open the resistor,
then the entire 4kT watts generated is simply dissipated back into
the sourcing resistor as heat and there is no net power flow. If
you load it with a matched resistance then you can draw off half of
this power, but if the resistor you load it with is at the same
temperature, then it also generates this same power back in the
first resistor and again there is no net power flow.
Coupling to it via a transformer is no different to using a
different value of resistor as the source - the voltage to current
ratio changes but the power available remains constant. Similarly
connecting many such resistors in series or parallel simply changes
the impedance (or voltage to current ratio) without changing the
available power.
A diode is not of course a very good switch and has a gently
changing V/I slope (ie impedance) near zero bias. Thus it must
also generate Johnson noise by the same mechanism (whenever there
is a path for electrical power to be dissipated as heat, then there
is the reverse path in which the heat bath can generate electrical
power - this is called the "fluctuation dissipation theorem" in
physics). Presumably this noise power source/sink will vary
slightly in impedance with the voltage/current fluctuations - but I
am sure nature will have organised it such that no configuration
you can dream up will allow net power to be generated from thermal
energy!
If a cold resistor and a hot resistor are connected through
electrically conducting wires which are perfectly thermally
isolating (if such things existed), then thermal energy will flow
electrically from the hot resistor to the cold resistor until they
become the same temperature. However this is no more exciting (and
much slower) than simply providing a thermal conduction path.
What is more interesting is that you can synthesize a "cold"
resistor from a low-noise op-amp and room temperature resistors and
actually "chill" a remote warm resistor (or more usefully a
mechanical system coupled through a transducer) electrically. This
is called "cold damping". Of course the power to refrigerate or
pump heat from the warm system to the synthesised cold one is
coming from the op-amp power supply. With modern op-amps you can
synthesise a resistor with a temperature of less than 1 Kelvin! (if
I remember rightly).
Preface: Radiation resistance generates no thermal
noise.
I would guess that the best you could do with any antenna pointing
into deep space would be to pick up the 2.7 K microwave background
- which would probably be indistinguishable from 2.7K thermal noise
being generated in the radiation resistance seen via the antenna.
On Nov 14, 2006, at 7:32 PM, John Winterflood wrote:
R Stiffler wrote:
I guess his mail is getting messed up, the comments you
make reference to were by Paul Lowrance and not myself.
My mistake - sorry about that. Your formatting (without
caretted indenting) together with my sloppy editing caused
that.
Paul wrote:
... Radiation resistance
generates no thermal noise.
That may be true but the argument would be entirely semantic.
The exact same process of "fluctuation-dissipation" occurs
and once some thermal power has been dissipated by an
antenna, then noise with thermal characteristics comes back
in via your antenna and looks identical to a warm resistor
generating Johnson noise.
One might likewise argue that a resistor itself generates no
thermal noise (since in a zero degree K thermal bath it
certainly doesn't) and blame the effect on something going
on in the resistor with another name - eg brownian motion of
the electrons or something, but again that would just be
semantics.
The fact remains that whatever mechanism is available to
_dissipate_ electrical power into the radiation resistance
of the aether, must also act in reverse to produce electrical
_fluctuations_ from the energy previously or similarly
dissipated (hence the "dissipation-fluctuation theorem").
Once equilibrium with the surroundings is reached, the power
flowing from a warm resistor to a warm environment via an
antenna will be exactly equal to the power flowing back from
the warm environment to the warm resistor.
In order to have no thermal noise being sourced into a
circuit from an antenna, one would have to locate it so
that its entire visible environment was at absolute zero
(which is very similar to what is required for a resistor
to generate no thermal noise).
... the antenna connected to a carbon
resistor does indeed radiate more power than an
antenna connected to a metal film resistor.
Only if you provide power in the form of a current through
the carbon resistor to get the flicker mechanism oscillating
(see the wikipedia flicker noise reference I offered
previously). But then a powered integrated RF oscillator
connected to an antenna will radiate even more power than
a powered carbon resistor!
On Nov 14, 2006, at 7:43 PM, John Winterflood wrote:
R Stiffler wrote:
I guess his mail is getting messed up, the comments you
make reference to were by Paul Lowrance and not myself.
My mistake - sorry about that. Your formatting (without
caretted indenting) together with my sloppy editing was the
cause.
Paul wrote:
... Radiation resistance
generates no thermal noise.
That may be true but the argument would be entirely semantic.
The exact same process of "fluctuation-dissipation" occurs
and once some thermal power has been dissipated by an
antenna, then noise with thermal characteristics comes back
in via your antenna and looks identical to a warm resistor
generating Johnson noise.
One might likewise argue that a resistor itself generates no
thermal noise (since in a zero degree K thermal bath it
certainly doesn't) and blame the effect on something going
on in the resistor with another name - eg brownian motion of
the electrons or something - but again that would just be
semantics.
The fact remains that whatever mechanism is available to
_dissipate_ electrical power into the radiation resistance
of the aether, must also act in reverse to produce electrical
_fluctuations_ from the energy previously or similarly
dissipated (hence the "_dissipation-fluctuation_ theorem").
Once equilibrium with the surroundings is reached, the power
flowing from a warm resistor to a warm environment via an
antenna will be exactly equal to the power flowing back from
the warm environment to the warm resistor.
In order to have no thermal noise being sourced into a
circuit from an antenna, one would have to locate it so
that its entire visible environment was at absolute zero
(which is very similar to what is required for a resistor
to generate no thermal noise).
... the antenna connected to a carbon
resistor does indeed radiate more power than an
antenna connected to a metal film resistor.
Only if you provide power in the form of a current through
the carbon resistor to get the flicker mechanism oscillating
(see the wikipedia flicker noise reference I offered
previously). But then a powered integrated RF oscillator
connected to an antenna will radiate even more power than
a powered carbon resistor!
On Nov 15, 2006, at 12:12 AM, John Winterflood wrote:
Robin van Spaandonk wrote:
A diode is not of course a very good switch and has a gently
changing V/I slope (ie impedance) near zero bias.
Which is precisely why you put the transformer in between. That
shifts the voltage up the curve of the diode away from the zero
bias point.
Bear in mind that we are talking of AC (noise) voltages so one
cannot expect to work with any DC bias. Also a diode has an
exponential V/I relationship (Shockley equation) and so with
appropriate scaling I believe it can be considered to work
just as well in a high impedance circuit with zero bias as it
does in a low impedance circuit when biased to the 0.65 volt
so-called "knee" (which value is entirely dependent on the
scale on which you choose to plot the exponential - scale the
axes in microamps and millivolts instead of volts and milliamps
and the "knee" moves down to zero).
However you would need an incredible transformer ratio,
and the resulting minute current on the diode side may be
"lost" in the noise of the diode. This depends somewhat
on whether or not these purported signals ...
There is no "purported"-ness about these signals. It is a
standard experiment performed by 3rd year physics students
to measure this noise voltage and from it determine absolute
temperatures (to ~4 digits with ~hours of integration), or
knowing a single temperature, to determine Boltzmann's
constant from the noise voltage.
... from the resistor can be "ganged"
together. Since they would have random phase relative to one
another, they would likely at least on occasion enhance one
another leading to a "spike" that might be transformed and
rectified.
My point was that a transformer provides nothing that
simply choosing a different valued resistor would provide.
A high value resistor gives high voltage with low current
but still only 4kT watts per Hz of bandwidth.
Similarly ganging resistors together provides nothing
different from what a single resistor would with the same
value as the ganged set. (There is a small difference -
and that is how well the resistor is heat sunk or connected
to the heat bath - but for the power flows under discussion
better connection to the heat source/sink is hardly an issue!)
Thus it must also generate Johnson noise by the same
mechanism (whenever there is a path for electrical power to
be dissipated as heat, then there is the reverse path in
which the heat bath can generate electrical power - this is
called the "fluctuation dissipation theorem" in physics).
Presumably this noise power source/sink will vary slightly
in impedance with the voltage/current fluctuations
The transformer "transforms" the impedances, so that there
is a deliberate mismatch between resistor and diode.
I think you missed my meaning - the exponential V/I
relationship (Shockley equation) of the diode means that
it will behave just like a resistor who's resistance (or
impedance) varies (only minutely with thermal level I&V)
as the voltage or current in it varies. This is after
all what provides the rectification effect - current in the
forward direction sees the diode as a much lower valued
resistor than current in the reverse direction. It is just
possible that this effect could produce some net cohering
of the statistical fluctuations. But like I said, I doubt
if nature would make it that easy to beat its second law!
- but I am sure nature will have organised it such that no
configuration you can dream up will
allow net power to be generated from thermal energy!
A solar cell already does this, it just operates at a
higher "ambient" temperature.
A steam engine also works well when you have a significant
temperature difference - such as that between the surface
of the sun and the ambient on earth. But beating the 2nd
law requires that it work without a temperature difference
- ie turn random thermal energy into ordered electrical
energy which can then be used to say heat an isolated
resistor above ambient while slightly cooling the ambient
heat bath in the process.
Its built in diode, acts like a 0 K heat sink.
More like a 300 K heat sink!
On Nov 15, 2006, at 5:12 PM, John Winterflood wrote:
Paul wrote:
I really don't see it that way. The carbon resistor is
made of atoms containing charged particles. The noise
is relative to the temperature of the charged
particles.
Neither do I. I was trying to illustrate that
assigning the noise source to the radiation
resistance itself or some other thing such as
the E-M radiation that is bouncing around in it,
is similar to trying to separate the resistance
of the conductive paths in a resistor from the
electrons that are bouncing along them.
We don't know what the aether is made of, but we
do know that it supports electromagnetic waves
and fluctuations. The spectrum of these
fluctuations can be used to assign it a black
body temperature. The temperature of deepest
darkest space determined by this spectrum comes
out around 2.7K. If the same measurement was
done in a lab it would indicate an "aether
black body temperature" of ~300K. If you attempt
to couple to this aether with an antenna, then
this radiation temperature will comes in through
your antenna and the radiation resistance seen by
the circuit looks identical to a ~300K warm
resistor.
Why are you interject flicker noise with this example?
It's thermal noise.
I have tried several times to educate you to the
fact that the extra (or excess) noise found in
carbon resistors is _not_ true thermal noise but
is produced by DC current passing through the
resistor. Why don't you read the wiki for yourself?
Here is what it says:
"Flicker noise is found in carbon composition
resistors, where it is referred to as excess noise,
since it increases the overall noise level above
the thermal noise level, which is present in all
resistors. In contrast, wire-wound resistors have
the least amount of flicker noise. Since flicker
noise is related to the level of DC, if the current
is kept low, thermal noise will be the predominant
effect in the resistor, and the type of resistor
used will not affect noise levels."
Please note the last phrase: "the type of resistor
used will not affect the noise levels". Maybe you
wish to disagree with common experimental knowledge?
If so you should provide some reference for your as
yet baseless assertion.
If you are right, then it would be true that you
could beat the 2nd law! But you wouldn't need an
antenna - simply connecting two resistors with
different thermal noise generation levels
electrically would be sufficient to create a
temperature difference between them in an otherwise
uniform ambient.
On Nov 20, 2006, at 8:51 AM, John Winterflood wrote:
Paul wrote:
...
I was just looking at a real experiment on flicker
noise graph of carbon resistor. As you know the noise
is relative to 1/f and current. It's dependent on
current, not DC current. Here's an interesting quote
from
http://www-tcad.stanford.edu/tcad/pubs/theses/goo.pdf
"Carbon composition resistors exhibit
current-dependent excess noise due to the random
formation and extinction of macro arcs among
neighboring carbon granules."
The "macro arcs" sound very interesting
Also, here's another interesting quote,
http://www.dsprelated.com/showmessage/23702/1.php
"In most resistors there are two other noises which
are far larger than the Johnson noise. There is the
generally larger so-called 'shot noise' which is
proportional to the current through the resistor and
which unlike the white [flat] Johnson noise actually
gets larger below a certain corner frequency, i.e.
it's a 1/f noise effect, plus there is also the larger
so-called current noise which is proportional to the
voltage applied and is usually rated in uVoltsrms/Volt
which also rises below a corner frequency, i.e. also a
1/f effect!"
Notice it refers to two causes of shot noise. ...
Resistors do not exhibit "shot noise" as the term is standardly
applied
in physics. Maybe this guy is referring to flicker noise as shot
noise
because it sounds like "shots" or something. Who knows! Both of
the noise sources he describes (current dependent and voltage
dependent) would standardly be referred to as "1/f noise" or
"excess noise" or even "flicker noise".
Shot noise on the other hand is generally only observed when one
puts some sort of an energy barrier in the circuit such that the
electrons can no longer "sit on the fence" anywhere around the
circuit as they can in a typical resistor, but must decide which
side of the barrier they are going to be on. When they gain enough
energy to jump the barrier, then the granular nature of the current
(ie an integer number of electrons passing the barrier) becomes
evident as shot noise.
A typical energy barrier might be a reversed biased diode for
instance and electrons can be given enough energy to jump the
barrier by various means such as thermal activation, photo-electric
activation, voltage breakdown, etc.
(There was a time when this noise business was a black art to me
and I would have appreciated answers such as I have been
providing. So maybe there are others who might be helped by these
clarifications.)
1. Proportional to current. 2. Proportional to voltage. It seems
#2 remains unchanged to the amount of
resistance, whereas #1 changes with resistance.
...
Anyhow, back to the experiment. Since there will be
current flowing in the carbon resistor due to thermal
noise, it will exhibit more voltage noise ...
Kinda hard to get "macro arcs" (see your comment above) at thermal
noise levels of induced voltage don't you think!
End quotes
Regards,
Horace Heffner
