noise in LT Spice

OK: LT Spice resistors seem to not have shot noise.

John

Probably because resistors don't have shot noise!

However, resistors do have 1/f noise when a dc current is present in them. This can be very, very significant in i.c. design. This 1/f noise goes with

1/sqrt(L.W). Poly resistors can be particularly troublesome.

Kevin Aylward B.Sc.

If John Larkin says resistors have shot noise... they must ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.

Well, some of the resistors in this guy's TV probably do now:

--
SCNR, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Yup, no shot noise in resistors as Kevin says. (Or perhaps better is to say greatly reduced.) If you've got some very big resistors in the TIA then it might be possible for the RC corner to be below the 1/f noise corner. This might sorta look like shot noise.... an extra noise term that is proportional to the current (voltage/R)... but the spectrum of the noise should show it's 1/f and not shot noise. (Mind you I'm half talking out my arse, I've not tried measuring the noise in big resistors... say 100 Meg and greater)

George H.

Kevin Aylward a écrit :

Not only DC current, but also AC... 1/f noise comes from random resistor changes, which also has impact on AC current. But instead of generating 1/f noise, it generates 1/f sidebands around your AC spectrum line...

--
Thanks,
Fred.

Some do, like high-value thickfilms. Or in my current case, liquid resistors. Past about 10 megs, it gets harder to buy low-noise metal film resistors.

For a flowing liquid, I could imagine lots of low frequency noise!

Do poly resistors have shot noise? Semiconductor currents generally have full shot noise.

I could fake some noise current to simulate my resistor shot noise, and even make it adaptive to realtime current with a multiply somewhere.

John

We tested a lot of high-ohm resistors, and the thickfilms do have shot noise. But the resistor I'm dealing with now is a liquid, ionic conduction, and according to the literature, they have full shot noise.

There's something special about metals that smooths out electron flow. There's not a lot available online about shot noise in various resistor elements, and most of the high-ohm resistor manufacturers don't know much about the subject, either.

We're buying some Dale RN55 metal films now, 50 megohms, and they're great. But the leadtime is 10-12 weeks, and they are something like $3.50 each.

John

Well according to Landauer it's the scattering that 'smooths' out the electron flow in resistors. Shot noise happens in semiconductors because it's only those 'few' random electrons (holes) on the far edge of the thermal distribution that have enough energy to get over the pn barrier. This random process happens one electron at a time. At least that's my understanding.

How do you distinguish shot noise from 1/f noise in high meg thick films?

That's interesting about ionic conduction. Do you have a reference? I thought, (though I'm often wrong) that liquid ionic conduction should be similar to that in metals. The only liquid conduction stuff I've ever done is electroplating copper onto stainless forms. And there you've got minimal resistance.... Lots of copper sulfate in the water.

George H.

Well according to Landauer it's the scattering that 'smooths' out the electron flow in resistors. Shot noise happens in semiconductors because it's only those 'few' random electrons (holes) on the far edge of the thermal distribution that have enough energy to get over the pn barrier. This random process happens one electron at a time. At least that's my understanding.

I haven't found much stuff, in books or online, about shot noise in non-metallic conductors. Metallic resistors seem to develop shot noise when the dimensions are very small; there are some papers on that.

qulab.eng.yale.edu/documents/reprints/PRL_shotnoise.pdf

I suppose a thickfilm resistor, with enough nasty boundaries, might not have the long-term coherence to get the smoothing effect. I just don't know. It is sort of academic, from a circuit design standpoint, in the sense that high-value thickfilms are clearly much noisier than metal films and can really trash a TIA.

By the spectrum, I suppose. It's difficult to measure wideband noise in, say, gohm resistors, because stray capacitance rolls off the measurement system pretty hard.

I do have a gadget that might let me see the spectrum of the noise in a 50M resistor with about 10 volts across it, up to a MHz maybe. I'll see if I can try that with various resistor types.

I have some references that refer to shot noise in conductivity-based flow cells, essentially "Coulter Counter" type devices. And others that don't. Most annoying. Nanopore devices have a tiny hole in a membrane that separates a fluid chamber, so the resistive area is sort of a very tiny thin disk, so maybe the scattering coherence is minimal. Plus there are ions blundering about, not just electrons. Hey, I'm just the circuit designer.

John

You can also use its dependence on voltage. For a fixed resistance, shot noise current goes as V**0.5, whereas 1/f current goes as V.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058

email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net

Do you mean that when there is dc current, there is wideband noise in excess of the thermal noise? This may well be the case, but I still still quibble in naming in it "shot" noise.

I would have to have a look at the this literature to see if the actual noise mechanism is the same as that of semiconductor diode and vacuum tube shot noise.

Kevin Aylward B.Sc.

Unless there is a potential barrier involved, that charges cross, then I don't see it as convential "shot" noise. There may well be other noise with another name, but I don't see that it shot noise because shot noise is, essentially, defined as that noise produced when charges cross a potential barrier. Of course definitions sometimes change.

Poly resisters are continuous sheet of poly. There is no junction/potential barrier, so no shot noise. Technically there are the end contacts, but these are heavily doped to make them ohmic by construction.

1/f noise in i.c. resisters are a particular concern of mine, and to date on chips back from fab, I have not seen any evidence of wideband excess noise due to current in poly resisters. I have seen large 1/f noise though. This is particularlly troublesome, when fab vendors models don't model it.

Kevin Aylward B.Sc.

That makes sense, the shorter it is the fewer scattering events.

Yeah it's excess noise, still if it's shot noise it's fundamental and you can't do much about it. If it's 1/f noise, or some other technical thing then there's the chance that a different 'style' of resistor will have less noise. I'm testing a few of our new noise apparatus today. (Max freq is about 1MHz, but only 100kHz if you want any real accuracy.) If I have some time later I could check out the noise in some thick film 100 Megers'. Unfortunately there's several pF of input capacitance and that limit's the frequency response... a lot... gasp less than 1kHz.

Interesting. (I never heard of the Coulter Counter.) If there is no scattering as the ion goes through the small hole then this could give shot noise! Even one or two scattering events should still leave some fraction of the shot noise. You can learn something, perhaps, by lookng at the noise,

George H.

is

the

e

ind

t .

ff

Duh, 'slaps head', of course. I was thinking they would both be linear in V.

George H.

t- Hide quoted text -

th

l

You might like the article by Rolf Landauer "Solid State Shot Noise" Phys Rev. B, 47, pg. 16 427 (June '93)

I think you can have shot noise in quantum systems without any potential barrier. Say a quantum contact that has a volume only big enough to hold one electron state at a time. (But I'm no expert.) John's ionic system might be similar in some way... if the ionic resistance is determined by a small hole through which only one ion can pass at a time.

But I think I agree with you the excess noise in thick film resistors should probably not be called 'shot noise'.

George H.

em.

al

ese

on

e s

As I understand it, there are two mechanisms in metals that reduce current noise below the Poissonian statistical value. One is e-e interaction, which is short range and cuts the full-shot noise about in half. The other is electron-phonon interaction, which only works in physically large resistors but smooths the noise almost to zero.

I can't find any definitive statement that either works in non-metallic conductors, like cermets maybe. Do phonons get very far in a messy sintered metal oxide thing?

What about ionic conductors?

What about tiny resistors inside ICs?

What about zigzag resistors? Do phonons turn corners?

A lot of the good phonon stuff is in papers that cost big bucks.

Some retired person could look into this. It's interesting but time consuming.

John

We had that same problem. 1/f noise could easily mask shot noise, especially if the bandwidth is limited. And higher voltages, as Phil notes, makes the shot signal bigger but the 1/f masking worse. It's always something.

The bottom line is that we'll use whatever resistor works best, no matter how you name the noise, but it would sure help to understand this stuff, to better analyze and simulate designs.

It's easy, a high school project, to count small particles like bacteria and blood cells with the Coulter technique. There's plenty of signal from things that big!

John

This might be of interest here:

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.