Noise in valves

he

Interesting idea. To cool the electrons you'd need some scattering mechanis m, or possibly some means of selecting a small chunk of velocity space--oth erwise a grid can only change the average velocity, not its distribution.

Particle accelerators use tricks like that, but they waste a lot in the pro cess.

A fun fact is that free electrons are stable in pure N2, so you might be ab le to use differential pumping to make a quiet electron source. Or just use a liquid gallium field emitter.

Cheers

Phil Hobbs

If cathode limited I believe you are correct. However the space charge around the cathode means noise is predominantly shot noise.

I don't see how adding more grids, which are mainly to decouple the grid gain from the anode voltage, and to reduce secondary emission. My understanding is that screen grid current will actually contribute to this noise.

BICBW

You need to read the paper Adrian linked. ;)

Yup.

Cheers

Phil Hobbs

Yes, thanks, now reading!!

So got one part right :-)

If you are adding the grids for that purpose (i.e. to create a conventional pentode), that is what will happen. I wasn't suggesting that.

I was suggesting using a valve with more than one grid (such as an off-the-shelf pentode) to produce a triode gain section which used G2 or G3 (or both) as the control grid , fed with electrons by a virtual cathode generated by G1. This would demonstrate whether it was possible to create a lower temperature virtual cathode which would reduce the Johnson noise generated by the internal resistance of the valve, which normally behaves as though it is at the temperature of the real cathode.

The partition noise of a pentode, where the electron stream is divided between G2 and anode, would not occur because none of the grids would be passing grid current; they would all be acting as control grids.

Can any tubes get close to the noise level of a good jfet? As in below

Tube noise specs tend to be obscure.

At very low frequencies?

Most noise specs tend to be obscure.

Curves of equivalent input noise voltage and current, nV and pA per root Hz, are pretty good. I've never seen that for a tube.

I think tubes are less repeatable than fets, and need to be selected for low noise. The quality of the vacuum apparently matters.

So, would a cold-cathode tube be quieter?

So, you can make a filament for incandescent lights that's only (or largely) emissive in the visible, and is thus more efficient.

Can you pattern a cathode material such that it only matches the wavelength of one energy of electron, thereby making the electrons coming from the cathode 'cooler'?

It would have less Johnson noise, for sure. The physics of small field emit ters isn't something I know a lot about, so maybe there are other drawbacks besides a tendency to run out of gallium.

Focused ion beam (FIB) cutting tools run the gallium tip at positive bias, so that the ions get launched rather than the electrons. I imagine that eve n when running the field emitter at negative bias you'd still lose a fair a mount of gallium due to electromigration-type effects, and of course a sing le atomically-sharp emitter would be working pretty hard to reach even the microamps.

Cheers

Phil Hobbs

There is some work on lowering the temperature of cathodes (BaO-Sc2O3 works a hundred degrees cooler than W-ThO2).

But the linear accelerator folk also use photocathodes, and heated material with a laser excitation gets a good high current density.

More exotic, there's a secondary-emission scheme with diamond plates that lets you chill the secondary emitter

And there's a favorite old trick, using a foil with a negative work function; you just fire the electron beam through the cold material, and it chills the electrons during the short transit time. Only problem: it only works with positively charged electrons (positrons). No joke; SLAC gets long positron-storage-ring lifetimes.

the

ism, or possibly some means of selecting a small chunk of velocity space--o therwise a grid can only change the average velocity, not its distribution.

Slits and a magnetic field perhaps.

Reminds me of an experiment I want to try again. Heat up a piece of metal and measure the change in noise.

I tried this with a little tungsten light bulb, but the resistance was a bit small, and complicated by the fact that the resistance of tungsten changes a lot with T. (well and all the vibrations!)

I'm thinking some nichrome wire maybe.... I want to heat it by passing current through it... I guess I'm going to be stuck with a fairly low resistance, to get enough V^2/R heating w/o going to a HV power supply.

George H.

rocess.

able to use differential pumping to make a quiet electron source. Or just u se a liquid gallium field emitter.

I'm no expert, but at low frequency you hit the 1/f corner... (The noisier Jfet opamps I've used have a corner near a few kHz.)

George h.

m the

In theory I think that would work.

George H.