Shot noise

[Fred Chen]

There are tiny fets, "SETs", that can resolve gate charge to a small
fraction of e. I think it may be possible to do a neat demonstration
using an eprom... possible student project maybe.

John

Yes, these are Single-Electron-Transistors (or also
Single-HOle-Transistors (SHOTs)). It may be an expensive project. You
either need a 1 nm quantum dot at room temperature, or more commonly
it can be 10s of nm's but near absolute zero. The reason is you want
the separation between the quantum states to exceed thermal excitation
(kT).

Fred

 

[John Larkin]

Yes, these are Single-Electron-Transistors (or also
Single-HOle-Transistors (SHOTs)). It may be an expensive project. You
either need a 1 nm quantum dot at room temperature, or more commonly
it can be 10s of nm's but near absolute zero. The reason is you want
the separation between the quantum states to exceed thermal excitation
(kT).

Fred

Wouldn't signal averaging help? The charge is still quantized even if
it is below the rt noise floor.

John

 

[Winfield Hill]

Mike Engelhardt wrote...

I thought that it was about being able to smoothly turn a
MOSFET on or off due to the discreteness of electron charge.
You went after the number of electrons it took to drive the
gate voltage anywhere. The gate wasn't floating, so I just
noticed that again you seemed to be applying charge quantization
in a situation that it didn't apply, since you can basically
charge a gate with any fraction of an electron you want. I
just noticed it because it matched the erroneous statement
in AoE. I don't know if you still have trouble with the
concept or not.

http://groups.google.com/groups?hl=en&[email protected]
The gate _was_ floating and had no observable resistive connection
to the outside world, yet the MOSFET's conductance changed a bit
with time. After accounting for temperature, etc., I calculated
that the observed changes would involved more than 15 electrons
per second into the gate. Given the insulated gate, there was no
resistive conductor with long-range electron interaction, etc, to
consider. [Note, in my experiment the drain voltage was *below*
the gate voltage (695mV vs 2.04V), yet the gate voltage apparently
increased (by 45mV), which could not have been from internal FET
leakage paths. Since no Faraday shield was used, a proper rerun
of the experiment is in order, but this hasn't been done yet.]

 

[Mikko Kiviranta]

Wouldn't signal averaging help? The charge is still quantized even if
it is below the rt noise floor.
John

Averaging is no good if your integration time gets longer than 1/f
noise corner. SETs tend to have rather large 1/f corner frequencies,
assumedly because of the background charge fluctuation.

Regards,
Mikko

 

[Ken Smith]

http://groups.google.com/groups?hl=en&[email protected]
The gate _was_ floating and had no observable resistive connection
to the outside world, yet the MOSFET's conductance changed a bit
with time. After accounting for temperature, etc., I calculated
that the observed changes would involved more than 15 electrons

Where you using a block of Radium as the base plate?

 

[Mike Engelhardt]

Win,

http://groups.google.com/groups?hl=en&[email protected]
The gate _was_ floating ...

Sure, if the gate is isolated, then you basically look directly
at charge quantization. But that's not the thread I was talking
about. I was talking about
http://groups.google.com/[email protected]&rnum=1.
In that thread you seemed to be still unaware that you don't
directly see the effects of charge quantization in a conductor
where the electrons' waveforms overlap. I don't know if you
get it now or not.

BTW, I've thought about this defense you posted that that your
mistake in the 1st Ed. of AoE was the mainstream thinking. That
is nonsense. Yes, it was a common error, but I don't think it
was the mainstream thinking at all. Even SPICE from the 70's was
programmed not to make that mistake. SPICE doesn't put shot noise
in resistors, but it does in, e.g., a diode where the electrons
cross a potential barrier.

* C:\xp\noshot.asc
R1 N001 OUT 1K
R2 OUT 0 1K
V1 N001 0 {V}
V2 N002 0 AC 1
..step param V list 1 2 3
..noise V(out) V1 oct 4 1 1Meg
..end

V1 is stepped, which varies the current in the resistors, but
not the output noise. The schematic is below for those using
LTspice.

Don't get me wrong, I liked AoE's discussion of stochastics.
Things like considering the 1/f noise of resistors was a nice
touch. Also, I really liked the discussion of equivalent input
noise current and voltage of opamps. Though as I recall, you
failed to mention that many opamps don't physically follow
that model. For example, many JFET input opamps have noisy
tail currents. If the input is perfectly balanced, that current
does not get into the signal flow path. But there's an impedance
imbalance on the two inputs, it does. That's a case where the
En/In noise density picture falls down.

--Mike

--- noshot.asc ---
Version 4
SHEET 1 880 680
WIRE 288 336 288 320
WIRE 288 128 288 112
WIRE 288 240 288 224
WIRE 288 112 144 112
WIRE 144 112 144 176
WIRE 144 256 144 336
WIRE -96 112 -96 176
WIRE -96 256 -96 320
WIRE 288 224 400 224
WIRE 288 224 288 208
FLAG 144 336 0
FLAG 288 336 0
FLAG -96 320 0
FLAG 400 224 OUT
SYMBOL res 272 112 R0
SYMATTR InstName R1
SYMATTR Value 1K
SYMBOL res 272 224 R0
SYMATTR InstName R2
SYMATTR Value 1K
SYMBOL voltage 144 160 R0
SYMATTR InstName V1
SYMATTR Value {V}
SYMBOL voltage -96 160 R0
SYMATTR InstName V2
SYMATTR Value AC 1
TEXT 80 408 Left 0 !.step param V list 1 2 3
TEXT 56 440 Left 0 !.noise V(out) V1 oct 4 1 1Meg

 

[John Larkin]

BTW, I've thought about this defense you posted that that your
mistake in the 1st Ed. of AoE was the mainstream thinking. That
is nonsense.

Mike,

How many error-free, thousand-page books have you published?

John

 

[Mike Engelhardt]

John,

How many error-free, thousand-page books have you published?

None. Like I said, I liked AoE's discussion of stochastics.
Heck, that's why I'm familiar with it. But I don't see the
sense in being defense about the error or trying to defend
it today when there was even a public domain program so
widely used as SPICE2 that didn't get it wrong way before
AoE.

The fallacy of AoE's statement jumped up at me when I
designed some commercial nuclear instrumentation preamps so
I mention it. I honestly mean no disrespect for Win at all.

--Mike