On Wed, 7 Mar 2007 20:48:47 +0100
"Michel Jullian" <[EMAIL PROTECTED]> wrote:
To avoid the voltage drop associated with a diode, which
is huge compared to the noise signal, one could use smart
auto-controled switches (fets) instead, which would only
connect the noise source to the capacitor when the source
is at a higher potential than the capacitor. This kind of
diodelesss rectification scheme is used in low voltage
switchmode power supplies, and is called "synchronous
rectification".
A kind of "sample and hold" which would resample every
time the noise signal gets higher than the storage
voltage.
Now would the whole system be able to power itself plus
some excess in isothermal conditions, where the switches
themselves exhibit thermal noise? That's the question.
Michel
Charlie's comment 1:
I think that gates and drains on one buss / sources on the
other buss FETs could substitute for diodes in an array.
Johnson noise in the channels will be rectified as it
interacts with the gate.
This would be harder to fabricate even if the gate is a
metal mesh. Sampling and holding is not needed.
> If rectifying noise is to work, I should think you'd
want to use
> something (like a Schottky diode) with a very low
forward drop.
>
> Second, nearly all your noise is very close to zero
volts, and close
> to zero volts, diodes are close to linear. They
conduct as something
> like I = I_s * (exp(x*v) - 1) where "x" is a constant
I don't feel
> like writing out. This is from a book but the general
form is easy
> enough to verify in a lab (though tedious). Very
close to zero volts,
> this formula is very close to
>
> I = I_s * x * v
>
> or, in other words, the diode is (nearly) linear at
zero volts. That
> suggests that it might leak really badly in an
application where
> the signal strength is totally minute.
>
> These may just be practical concerns but it's not
clear how to get
> past them.
Charlie's comment 2:
I_s is the saturation current carried by thermal
electron / hole pairs under reverse bias short of
avalanche breakdown. It is directly proportional to
junction area. A smaller junction diode will be more
nonlinear. This is one way that small diodes operate
reasonably at small power levels of Johnson noise. The
other ways that small diodes operate reasonably are (2)
low capacitance to handle the full bandwidth and (3)
Reasonable power density for 1 / 2 kTB per junction. The
smallness of the diodes allows a greater number to be
fabricated in a given area or volume for more aggregation
of 1 / 2 kTB power.
> I don't know enough about LEDs to answer this
additional question: Can
> an LED operate "backwards", as a solar cell? This may
be a concern at
> very low emission rates.
As you pointed out, LED's emit appreciable energy above
forward voltage.
Although according to real experiments and Spice sims
the LED just does not
suddenly go from emitting to not emitting photons. In
fact, there's no magical
level where an LED suddenly changes.
Charlie's comment 3:
Nanometer scale LEDs would also have sharper nonlinearity.
They may have enough internal resistance to simplify the
fabrication by not needing individual resistors.
> >
> > Lets consider photovoltaic cells. Even at room
temperature in
> > complete darkness (no solar) there are visible
light photons
> > striking the cell. I calculate a 10 cm x 10 cm
common solar cell
> > would generate roughly 1E-30 volts. Not much
voltage, lol, but
> > still something nonetheless. The amount of
radiated blackbody
> > energy is small in the visible region. Although
the FIR region is
> > another story. Both sides of a thin sheet of 1m x
1m material
> > radiates roughly 920 watts continuously in complete
darkness at room
> > temperature. Technology is improving, thereby
allowing photovoltaic
> > cells to capture lower and lower frequencies. A
Canadian university
> > succeeded in creating a 1355 nm photovoltaic cell!
That's only
> > 1/11th the wavelength away from the peak 15000 nm
920 watts/m^2
> > blackbody 300 K radiation. BTW, blackbody radiation
at 1355 nm is
> > 2E+18 times greater than visible region of 600 nm.
To calculate this
> > I compared the radiation from 16667 to 16677 cm^-1,
which is
> > 3.907E-29 watts to 7380 to 7390 cm^-1, which is
7.499e-11 watts.
> >
> > University of Toronto in Canada achieves 1355 nm
photovoltaic cell:
> > http://nanotechweb.org/articles/news/4/1/7/1
> >
> > Eventually technology will reach the peak 15000 nm
region where a
> > thin double sided 1m x 1m sheet receives ~920
watts. It's difficult
> > for a person to believe they are surrounded by a
source "free
> > energy" because we don't see such energy with our
eyes.
>
> This one still really bugs me. I don't understand
solar cells well
> enough to know if this could work, but it just seems
/wrong/ to me
> that a cell at the same temperature as a blackbody
could generate
> useful electricity from the blackbody's radiation!
(Even a solar
> cell made in Canada!) :-)
Charlie's comment 4:
The Canadians should observe ambient IR output from their
photocell which could then be called a thermovoltaic cell.
I will visit their website immediately after posting this.
Later comments here indicate that the Canadians are
persuing ambient IR power. I believe that this is another
way to recover energy from Carnot death. I would fabricate
many thin and wide junctions and ohmic contact layers in a
multilayer structure that would put many cells in series.
I accept the concept because there is an asymmetry of
materials which will impose an asymmitry of behavior on
elecritity producing processes.
The problem with typical visible light photovoltaic
cells is there's hardly any
black body radiation in the visible light spectrum. I
calculated/guesstimated
that such a cell would generate less than 1E-30 DC
volts. On the other hand the
university in Canada made a significant breakthrough in
photovoltaic cells by
allowing such cells efficiently absorb up to 1355 nm.
Now that 1355 nm doesn't
sound like much as compared to 600 nm visible light, but
blackbody radiation at
1355 nm is 2E+18 times greater than at 600 nm visible
light.
To give an idea just how much energy is available from
blackbody radiation alone
--> A two sided thin sheet of material at room
temperature (300K) radiates ~920
watts, which is peak at around 15000 nm. It will be very
interesting when
technology increases from 1355 nm 15000 nm! :-) Last
week I came across a forum
posting about the Canadian breakthrough, but cannot find
it for the life of me.
There were physicists (or at least they sounded like
physicists) saying things
that absolutely shocked me. Statements to the effect,
"we're almost to the point
of collecting continuous ambient temperature radiation."
Nature's not that cruel, right? I'm mean, surely we
could design a device to
collect energy from a room full of bouncing basketballs.
What about large
molecules? What about typical air molecules? What
about room temperature atoms
vibrating at ~20 THz? What about vibrating electrons
traveling ~1/200 c? Is
there a sudden magical size where nature says, "No!
These particles are off
limits. You cannot have their energy!" I don't thinks
so because of the simple
fact that a capacitor is able to charge to a certain
point due to thermal
electrical noise.
IMHO the task is in trying to find a design that is able
to capture enough
ambient temperature energy without an appreciable loss.
Some realistic
possibilities would include an LED array consisting of a
trillion LED & R units.
Or perhaps improving photovoltaic cells to reach 15000
nm's. Those methods
sound complex. Truly if this was my task alone then I
would first begin by
learning how to create a simple diode. I already know
how to use a computer to
operate a circuit through the parallel port or USB.
Therefore, I would design a
device that allowed the computer to create micro scale
diodes and resistors.
Initially this may sound to complex and expensive, but I
would beg to disagree.
I like the old saying, "Where there's a will there's a
way."
Truthfully I see a clear method of extracting such
ambient temperature by means
of magnetic avalanches, but time will tell.
Regards,
Paul Lowrance
