John Curl's Blowtorch preamplifier part II

[jneutron]

Originally Posted by DF96

Your point of confusion is what I suspected it to be. If you apply your reasoning to a resistor, then you would prove that a resistor cannot generate a thermal voltage as there is no net movement of electrons there either.

I knew you'd think in that fashion.

You are confusing the direct measurement of electron movement with the coupling of a magnetic field caused by the movement. A hot wire does not produce external magnetic field more than 5 to 10 nanometers from the surface.

Originally Posted by DF96

I have no need to Google mean free path. Scattering of electrons might produce some noise, but that would not be shot noise. In a good conductor the electrons spend most of their time not being scattered - that is why it is a good conductor.

Interestingly enough, I pulled shot noise from a text. Granted, first edition was '59, second in '70.. Maybe Schwartz was confused...

Your definition of "good conductor" and mine are worlds apart. You live in the world of ampere per square mm, I live in the world of kiloampere per square mm and no electron-lattice interaction.

Read my previous post on "excess noise". Perhaps that will help you understand what I'm speaking of..

Cheers, John

[DF96]

Sorry, when you said shot noise I assumed you were talking about shot noise. Excess noise may also be present, of course, and scattering may well be a component of this. And thermal noise from all dissipation mechanisms.

As often on this site (and this thread), I wonder what all the fuss is about. We seem to have finally established that thermal noise is a significant (dominant, only?) contributor to transformer noise, and that this can be calculated by measuring the resistance at a particular frequency and estimating the temperature. This will be ambient for an input transformer - otherwise we would have to do separate calculations for windings, core and any other source of loss as they may have different temperatures.

If thinner laminations give lower Q and higher noise, then that is exactly what we would expect if eddies are a factor.

[DF96]

So you have a one-way coupling? Windings couple to core, but core does not couple to windings? You can have dissipation without fluctuation? What sort of physics is that?

[john curl]

Interesting point, but SMPTE IM is a subset of real IM, and I would think not relevant in this case.

[janneman]

Originally Posted by DF96

Your point of confusion is what I suspected it to be. If you apply your reasoning to a resistor, then you would prove that a resistor cannot generate a thermal voltage as there is no net movement of electrons there either.

[snip].

Dave,

If a resistor, by itself, generates noise, presumably we can harvest that as energy. Where's the energy coming from? Would we cool down the universe when we would harvest it?

jan didden

[jneutron]

Originally Posted by DF96

Sorry, when you said shot noise I assumed you were talking about shot noise.

No problem..I figured this is what was the sticking point..

It is what Schwartz called it in "Information Transmission, Modulation, and Noise..McGraw Hill, 1970. Page 512 on..

Originally Posted by DF96

Excess noise may also be present, of course, and scattering may well be a component of this. And thermal noise from all dissipation mechanisms.

It is the dissipation mechanism of eddy currents that we've been discussing.

Originally Posted by DF96

As often on this site (and this thread), I wonder what all the fuss is about.

Well that's an entirely different question.

The fuss is because every person is an individual entity, living, breathing, and in need of being convinced that their opinion is not the correct one, and tht everybody MUST conform to (my) opinion...

That was an easy question...

Originally Posted by DF96

We seem to have finally established that thermal noise is a significant (dominant, only?) contributor to transformer noise, and that this can be calculated by measuring the resistance at a particular frequency and estimating the temperature.

At these power levels, actual temp is unimportant. An effective temp could be calculated, but I believe the effective resistance provides a better handle for building and designing..

What was asserted was that hf noise is somehow dependent on lamination thickness, and I attribute that to eddy current loss.

Originally Posted by DF96

If thinner laminations give lower Q and higher noise, then that is exactly what we would expect if eddies are a factor.

I believe you typo'd and meant if thicker lams..

Originally Posted by DF96

So you have a one-way coupling? Windings couple to core, but core does not couple to windings? You can have dissipation without fluctuation? What sort of physics is that?

Um, standard physics.

Do not confuse the magnetic coupling between the coil and core with thermal motion which does not produce a macroscopic magentic field.

The random thermal movement of electrons within the core do not create a magnetic field which the coils can see. As I stated, 5 to 10 nanometers from the surface, local magnetic fields are gone, integrated out.. That is below the surface oxides.

Cheers, John

[john curl]

WELL SAID, DF96. However, thicker lams give more noise, skinny lams less noise, but are hard to make and work with.

[DF96]

If you could harvest the energy, then yes you would cool the resistor. This energy comes from whatever heated the resistor in the first place. Doing this would not cool the universe because the energy you extract will all end up as heat again eventually. Remember what thermodynamics says: you can't win, you can't even break even for long.

Almost any method you think of for harvesting the energy would end up being either inefficient, or putting just as much energy back. One possibility is to take another resistor of the same value, which just happens to be near absolute zero in temperature. (We will gloss over how much energy it took to cool it). Connect the two resistors together with wires which are good electrical conductors, but poor heat conductors (if you can find them). Then the hot resistor will cool, and the cold resistor will warm up. They will reach equilibrium at their original mean temperature. Don't hold your breath, as this may be a slow process unless they are very tiny in size. kT is not a lot of energy, unless you are trying to make a radio receiver or a phono pickup.

[DF96]

Sorry, yes "thicker lams".

Temperature does matter, unless we simply assume that all relevant items are at the same temperature. May be OK for a phono preamp, as they will be near ambient temp. However, we are used to calculating noise in voltage or power terms and for that you do need temperature.

I am not convinced that there is one-way coupling. You seem to have created a magnetic black hole! As I said, dissipation without fluctuation. A loss mechanism which generates no noise. I don't believe that.

[jneutron]

Originally Posted by DF96

One possibility is to take another resistor of the same value, which just happens to be near absolute zero in temperature.
(We will gloss over how much energy it took to cool it).

Resistance Temco causes a problem here..stainless isn't too bad, it has almost the same value at 4.5 Kelvin as at room.

And 4.5K requires 1 kilowatt of refridgeration power for every watt dissipated. 1.88 K requires double that..

If anybody can find a good electrical but bad thermal conductor, I know a few facilities that will buy TONS of it.

Cheers, John