Thermal noise study...

[dkemppai]

Yeah, I know, I never remember which part of the tube is which. I had to look it up. Anyway, my point is things are hot!

(The followin deifnition is taken from the book:"Electronic Communications systems Fundimentals through Advanced" Second Ed, by Wayne Tomasi)
(Start Snippit)
Thermal nose: Thermal noise is associated with the Brownian Movement of electrons within a conductor (taken loosely, a conductor is resistor/transistor/tube/wire -Dan). In accordance with the kinetic theory of matter, electrons within a conductor are in thermal equilibrium with the molecules and in constant random motion. Electrons within a conductor carry a unit negative charge, and the mean-square velocity of an electron is proportional to the absolute themperature. Consequently, each flight of an electron between collisions with molecules constitutes a short pulse of current. Because electron movement is totally random and in all directions, the average voltage produced in the substance by their movement is 0VDC. However, such random movement gives rise to an ac component. This AC component has several names, which include "Thermal Noise", "Brownian Noise", "Johnson Noise", "Randon noise", "Resistance Noise", and "white noise". Hence, thermal noise is the random motion of free electrons within a conductor caused by thermal agitation.
(end of snippit)

Also, a quick note for anyone worried about shot noise. It happens in every amplifier. Shot noise is a completely different noise, it has to do with the random arrival of carriers.

Anyway, back to White/Thermal noise...
So, what we can glean from this is that Heat causes a "white" current flow (because a moving electron is current). That current flow multiplied by the resistance of the circuit generates a "white" voltage noise. Therefore, the hotter the electronics get and the more resistance it has, the more internal Thermal noise it will generate.

So, My question can be modified slightly. In a tube, is the flow of electrons purely ripple free DC, or is there some noise evident because of the higher levels of heat in a Tube than in a transistor?

Also, that being said, cryogenic amplifiers are used for low noise high frequency RF amplifiers. So...

Is anyone using cryogenic amplfiers for auido? We already know that a cooler amp would be less noisy.
And, Finally for all you tube guys, I'd really like to see how you come cryogenic tube amp!

-Dan

[fdegrove]

Hi,

So, My question can be modified slightly. In a tube, is the flow of electrons purely ripple free DC, or is there some noise evident because of the higher levels of heat n a Tube than in a transistor?

If the anode voltage is free of ripple the noise will be free of it as well.
The noise present (or generated) in a tube will depend on a number of factors.
Some tubes have an equivalent noise factor of a resistor of say 180 Ohm.
The greater the electron movement the more noise is going to be generated,be that in a tube or semi-conductor or a piece of wire.

You can combat this to some extend by cooling the tube and by insuring your PSU does not induce any noise of its own.
Cryo treatment my help as well by providing a more homogenous emission.

I'm not convinced however that hotter tubes would per definition be more noisy than cooler ones.
If this were the case all Class A tube amps would be noisier than Class B ones.
I never noticed an increased noise level in Class A tube designs.

Shot noise however can't be lowered by lowering the operating temperature of the devices.

Noise is best characterized by the Fourier transform of the time-varying fluctuations in electric current, which is called the noise spectral density, S. For thermal noise, the spectral density is given by 4kT/R, where k is Boltzmann's constant and T is the temperature. Thermal noise is thus white noise - the spectral density is independent of frequency.

Noise is IMO more likely to be a problem in a semi-conductor since it is caused by the motion of the electrons and occurs in any conductor that has a resistance, R.
Since electrons travel in a vacuum in a tube,the only noise generated is by the movement of the electrons themselves.

Ciao,

[dkemppai]

Originally posted by fdegrove
Hi,

Some tubes have an equivalent noise factor of a resistor of say 180 Ohm.
The greater the electron movement the more noise is going to be generated,be that in a tube or semi-conductor or a piece of wire.

You can combat this to some extend by cooling the tube and by insuring your PSU does not induce any noise of its own.
Cryo treatment my help as well by providing a more homogenous emission.

I'm not convinced however that hotter tubes would per definition be more noisy than cooler ones.

Ciao,

I find is interesting that you stated that cooling the tube will reduce some of the noise. Yet you are unconvinced that this is true??? HUH? This is exactly my point.

Again, were looking only at thermal noise, not PSU noise or other induce noises (Internal, not external noise).

Ok, if we look at the calculating the noise for a system, assuming the noise source's resistance, and the load resistance are matched, we can say that the noise on the load will be Vnoise = SquareRoot(4RKTB) where R = resistance, K = Boltzmans constant, T = Temp in Absolute, B = bandwidth of the system. For a given bandwith system, we can see that the noise voltage is proportional to the resistance, and temperature.

Tubes are generally high impedance devices. To match the tube, the load impedance needs to be high also. (Is this not why tubes generally operate at higher voltages than transistors for a given output power?) Higher temperatures combined with the higher resistances should generate more thermal noise. Likewise, in a transistor with the lower impedance and lower temperature the thermal noise should be reduced. (And because transistors don't need heat to operate, can be cooled to reduce noise even more)

I agree that the greater electron movement will generate more noise. That movement however is referring to the internal AC movement of electrons in the device (wire/tube/transistor), and not the imposed migration of electronis due to an external source. Noise imposed upon a circuit will be less when there is less resistance in the circuit.

Another way to think about this is to think of two wires each connected to a resistor, one that is made of individual copper atoms stacked up (Think of a wire with a diameter of one copper atom wide). The other wire is extermely large. In the large wire free electrons bouncing around because of thermal agitation will have many paths. They may or may not cause a current flow in the resistor. They can go in any direction.
The Free electrons in the small wire however, will only have two paths, and therfore will be forced to create a current moving through the path (Wire and Resistor). That moving current will cause a voltage across the resistor. Statistically speaking, the larger wire will move less electrons through the resistor than the smaller wire. Calculating the resistance of each wire should be easy, and we will find that the smaller wire has a larger resistance.

Now onto the reality of the whole thing. The noise generated may or may not be detectable by ear. In both semiconductor and tube amps, the gains are probably small enough that the noise will not be noticed during normal use. Noise will always be there, it's just a matter of how big it gets to be. A better amplifier will be built from a low noise point of view.

Thermal noise is usually problematic in RF communications systems, where extermely high gains are needed (Thus the cryogenic amplifiers). Voltage gains of 1,000 to 10,000 may be typical in RF systems, however an audio amp may run more like gains of 50 to 100. With that low of a gain, you probably won't hear the noise.

Let me also state that I'm not trying to prove that tube amps are better or worse that transistor amps. All amplifiers distort, tube and transistor and fet. Some poeple like the sound of the distortion in a tube amp, others like the distortion in a transistor amp, yet others like fets. It's all a matter of taste and the individual.


My thoughs are that if solder makes a difference, then thermal noise should play an even bigger role in amplifier design!


Enough rambling for tonight!
-Dan

(All in all life is good, I have friendly forum in which to debate some enlightened passionate people!)

(Ignore all my Typos!!!)

[Nelson Pass]

Originally posted by dkemppai
Some poeple like the sound of the distortion in a tube amp, others like the distortion in a transistor amp, yet others like fets.

1) Of course, Fets are transistors

2) I don't know anybody who likes the distortions of a
bipolar transistor amp; its advocates seem to think there aren't
any (comparatively, of course).

[halojoy]

Originally posted by dkemppai
Ok,
I've heard discussions on silver in audio, and solder in audio, but what about resistors in audio... ....or wires, or semiconductors, or tubes....

Seems to me that thermal noise would be of more concern than Seebeck, or Peltier, or that stuff. Since thermail noise is porportional to temperature...
-Dan

There has been too few studies, if any,
of what kind of noise people prefer to listen to.

Why not make a study?

Let different bipolars, different tubes and different fets
generate say 0.5% along with some program material
And see what a great number of people prefer.

Just for the comparision, we should add one case
without noise.
Not that people would like it, but we could add it just for fun...



We could also make the study in a second part,
where people get exact information of what they listen to.
Not that I think that would make any difference,
but who knows?

[halojoy]

I think the best place to add noise
is in the preamplifier
or already in the CD-player.
This to avoid too high S/N ratio.

SACD suffers from way too high S/N ratio, in my opinion.
140 dB !!!!!
What joy can that bring to da music

[fdegrove]

Hi,

The noise generated by tubes (shot noise,NOT thermal noise) is that it is not bundle in a range of bands as in a semi-conductor but spread over the entire audible range.

IMO noise in tubes is less obtrusive to the human ear since it tends to be spread over the entire audible spectrum.

Distortion in tubes is (depending on the type of application/circuit) often predominantly second order distortion rather than first order,this second order harmonic didtortion is less annoying,causing less listening fatigue then first order distortion.

When you solder carelessly and apply too much solder to make a good joint the endresult will inevitably be a noisy circuit all else being equal.

Ciao,

[SY]

What is first order distortion? This is a term I'm unfamiliar with.

Second order distortion is characterisitic of any single-ended circuit, independent of the type of device used. Some will have more than others, but they'll all have it.

[fdegrove]

Hi,

First order harmonic distortion AKA odd or uneven order distortion is the first you would see when doing a distortion analysis.

Unpleasant to human hearing would be 1st,3rd,5th order distortion. (1st being the distorted fundamental,all consecutive odd numbers are overtones of this.)

Same goes for even order distortion,2nd,4th,6th order which are less obtrusive to the ear since the are actually closely related to the harmonic degradation of tones as produces by an acoustic instrument.

Hope this helps,


P.S. SY,third harmonic distorted wine would be pretty close to vinegar.

[dkemppai]

Originally posted by Nelson Pass


1) Of course, Fets are transistors

Yes, Of course they are, but I was referring to the difference between transfer characteristics for each device. (I guess, you could argue that every device is different). In reality the difference may be little, however, we all agree that small differences are detectable...