1. Home
  2. General Electronics
  3. Noise in parallel

Noise in parallel

Hi,

I'm having a discussion about placing two noise sources in parallel such as Johnson noise. I would appreciate it if a EE would please tell us who is correct? Here is the conversation -->

I wrote:

++++++ Any EE knows that parallel noise sources decreases.

Here is a Spice noise analysis on one single 1N914 diode, full bandwidth -->

32.324 uV rms

Here is a Spice noise analysis on two 1N914 diodes in parallel, full bandwidth -->

22.856 uV rms

22.856 uV / 32.324 uV =3D 0.70709

And here's the noise source of two 1N914 diodes in *series* -->

45.713 =B5V rms

45.713 =B5V / 32.324 uV =3D 1.4142

As stated, two noise source in *series* is relative to sqrt(n).

The net noise from noise sources in parallel decreases. The net noise from noise sources in series increases.

++++++

The other pers------ if you base your existence on simulation software rather than reality no wonder you are chasing a stupid idea. Noise does not negate because of increased samples you moron.

I spent 6 years as a senior geophysicist for Exxon and I know my stuff when it comes to noise. Seismic data is all about signal and noise and it is universal in information theory.

Prove it to yourself fool by simply generating random numbers in a basic program and see that it is impossible to get a noise peak of 100 or -100 if the random number is between 0 and 1 with a -0.5 bias, that is just so obvious, then run it accumulating a million loops and look at the highest responses, there will be some greater than 1, some greater than 10 etc etc.

If there is a coherent signal bias then the signal adds directly to the n samples but the noise adds to the sqrt of n samples x rms noise.

So PL when you start attacking peoples knowledge you want to do so with a bit more justification than your 2 sample spice data.

You are the most arrogant and stupid pseudo scientist I have ever read.

There are a lot of people here that are real scientists that have given you a hearing and you have proved to be the proverbial Whackadoo.

Grow up Paul, get a life and do something you can achieve, you will never qualify as a scientist!!!!

BTW read my original post on noise and see that it mentions the issue of increased load, that is the real issue!

A spice model if run properly will reflect that the load increases with resistors in parallel, so too did I. So in my post I said that diode arrays would not work and so do you. Also in my post I pointed out your stupidity for if not in parallel then what use is pA currents for the real World. MORON.

PL F off!!

------

I wrote:

++++++ Resistance decreases when paralleling. I'm recalling when you tried to show some math on calculating power where you even forgot to write the voltage, LOL.

I already explained your error, as you forget to include phase angle. When you include phase angle in addition to amplitude you will clearly see your error. That is why the Johnson noise equation clearly shows I am correct -->

Vn =3D sqrt(4 k T R BW)

R is resistance. When you place two resistors in parallel you get half the resistance. Do the math, will you.

++++++

The other pers------ You really are showing yourself to be so stupid.

The issue of noise is an issue of measured noise across a load.

Noise itself does not decrease but measured noise decreases because the load increases.

If you are going to quote a formula do us the service of understanding it yourself first.

You argue for me not against me.

read again moron!

I said load increases with resistors in parallel!!.

You laugh when you cannot read or understand apparently.

Paul I know you have never held a science post in your life and that seems to have come from an incomplete education experience, perhaps you flunked science. It is fair to say that you have been ridiculed by many in your life which is why you stupidly attack people.

I can only suggest that you start listening a bit more and go back to college if you can afford to do so.

------

Reply to
Paul
Loading thread data ...

Were the diodes run at some DC bias current? If so, what was the current of each individual diode in the above situations?

Diodes don't necessarily behave like resistive noise generators because diode impedance and noise depend on bias currents.

John

Reply to
John Larkin

Hi,

It was done in LTspice at zero bias, at RT, which I believe is 27C in LTspice. So the noise is Johnson noise at Rz if we ignore Ct. In this case, the noise is easily calculated from kTC noise, Vn =3D sqrt(k T / C).

So who's correct? He's saying the noise from two noise sources in parallel increases. I'm saying it decreases.

Regards, Paul

Reply to
Paul

Given two identical noise sources having identical source impedances, if the noise voltage of one unloaded source is E, the noise voltage of the paralleled sources is 0.707 * E.

That can be arrived at several different ways.

But be careful: the default behavior of the diodes that come with LT Spice is very wrong. But that won't affect the generalization about paralleling.

John

Reply to
John Larkin

l

Thanks, that's what I was telling the guy, but he kept calling me a stupid moron among other names.

What type of behaviors? I haven't seen anything to far off in LTspice given a good detailed model. The problem is a lot of diode models are to generalized. Also, the same part # made by different manufactures are often different. For example the Ge diodes such as the 1N314A vary so much from manufacture to manufacture that a generalized model is almost worthless.

One limitation is LTspice is the diode parameters are static. The parameters in a good microwave Si diode don't change much, but the

1N34A does. The difference in emission coefficient between zero bias and 10 mA could be significant.

That's enough of my rant. :)

Paul

Reply to
Paul

"Paul"

I'm having a discussion about placing two noise sources in parallel such as Johnson noise. I would appreciate it if a EE would please tell us who is correct?

** Johnson noise sources implies resistances.

Two resistances in parallel follow a simple rule for computing the combination.

The noise voltage (Vn) in any given bandwidth is related to that combined value by the usual formula.

Vn = sq.rt. 4kTBR

Vn is proportinal to the sq.rt. of the combined resistance value - if all other factors remain the same.

..... Phil

Reply to
Phil Allison

Set up LT Spice to curve-trace its 1N914 model.

John

Reply to
John Larkin

l

tell

ll

/

The accuracy is limited to the model. LTspice only models fundamental components. As you know, a real component is modeled using numerous ideal components. A real diode is complex, and sometimes requires using multiple diodes just to accurately model one single diode. I've seen manufactures using two diodes to model ultra low ZBD's that have unusually high reverse leakage.

I have not matched LTspice with the full I-V curve of a particular

1N914, but the 1N914 model that comes with LTspice is remarkably close to a Radio Shack 1N914 diode at zero bias. I'd imagine it's close at higher current levels as well.

Paul

Reply to
Paul

Basically, you are correct. All things being equal of course.

Funny he mentions that, I've spent more than 10 years in the marine Seismic industry designing seimsic data aquisition equiment. Parallel components in seismic analog front ends are a common technique to lower the noise figure. Not uncommon in other industries either, but I wouldn't say "every EE knows", I've found it is one of the lesser known techniques.

Seems he has a bee in his bonet, good luck with that!

Dave.

--
---------------------------------------------
Check out my Electronics Engineering Video Blog & Podcast:
http://www.alternatezone.com/eevblog/
Reply to
David L. Jones

l

Won't the current noise increase in lock step. Parallel impedances reduce the thermal voltage noise and increase the thermal current noise. The thermal noise power is a function of kT.

George H.

George H.

Reply to
George Herold

l

Won't the current noise increase in lock step. Parallel impedances reduce the thermal voltage noise and increase the thermal current noise. The thermal noise power is a function of kT.

George H.

George H.

Reply to
George Herold

You only have current noise if there's a finite-impedance load to have current into.

If a noise source dumps I amps RMS into a short, adding an identical noise source in parallel will result in 1.414 * I into the short.

John

Reply to
John Larkin

l

tell

ll

/

Photodiode into TIA op-amp. Until the frequency gets high the op-amp looks like a short. But I can measure lots of current noise.

Yup, that seems right. Resistors have sqrt(4kTR) of volatge noise and sqrt(4kT/R) of current noise.

George H.

Reply to
George Herold

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.