Reduced Heater voltage to reduce noise

[DualTriode]

Good Morning,
I opened a 1987 issue of The Audio Amateur and read a letter to the editor by L. Rusch of Glenham NY. Mr. Glenham discussed a 10 watt amplifier he had built following a circuit in a RCA tube manual. The 6.3 volt tube heaters were supplied with 5.4 volts and demonstrated lower noise as a result.
I have heard of this effect before but not experienced or explored it myself.
Does anyone in the community have experience or knowledge of the effect of adjusting heater voltage vs. tube noise?
DT
All just for Fun!

[Arnulf]

Johnson?Nyquist noise - Wikipedia, the free encyclopedia ?

[Rod Coleman]

With low-power valves, reducing the heater temperature (by lowering voltage) will usually reduce noise.

The mechanism: at full working cathode temperature, with a tube in new condition, the emitted electrons will show a distribution of energy levels. Some will be energetic enough to leave the cathode and reach the grid [if grid-cathode voltage is low] - and if this makes up a continuous flow it will be measurable as grid current.

Reducing the cathode temperature will lower the energy (hence velocity) of the electron emission, and so reduce grid current.

Grid Current is the focus when dealing with Noise in valves. We should distinguish between two types of noise - Noise from the thermal agitation of electrons (thermal noise/Johnson noise) and noise from the discreteness of the electrical charge (shot noise).

The grid current described is small and haphazard because electrons are emitted by the cathode randomly and independently - their arrival at the grid is like rain on a tin roof. This is Shot Noise, rather than Johnson [thermal] noise, and is the dominant noise effect in a triode.

Operating the valve is such a way that grid current is lowered will minimise noise.

- lower cathode temperature
- low anode voltage
- lowered anode current

These lowered parameters will generally reduce gm, which may increase thermal noise. But with a tube, Shot Noise effects overwhelm thermal noise by a big margin.

With transistors, increased collector current gives lowered noise, and many a sand-head gets into trouble with tube design on this count.

Also, one must choose a sample set of tubes with low gas, since ionisation grid current is also plagued by shot noise.

Old tubes may also get noisier due to accumulation of cathode particles on the grid, leading to grid-emission current - also a shot-noise prone mechanism. Warming the cathode before applying B+ is a possible precaution to minimise this particular risk.

[DualTriode]

Hello Rod Coleman and All,
Interesting insists into noise in circuits.
I suppose that when the ‘’valve” was designed heater voltage, noise, gm, mu and the complete list of variables was taken into consideration and the important variables were optimized. Now 70 plus years later we (I am) are rediscovering what they learned then. Perhaps the valve used where low noise was more important grid current was minimized. I am also thinking that not every circuit required this consideration.
Now we are digging through what remains of NOS valves and using them for other than their original application. Perhaps this is the other end of the scale of trying to maximize the audio power output of a television sweep tube.
It may be worthwhile to breadboard a tube like a 5670 vary the heater voltage and observe the noise floor on one of the new fangled FFT PC setups. I say 5670 because I have some NOS JAN GE’s. They were manufactured to tight standards. The noise variable was specified and controlled. It may be possible to lower the noise floor of a valve RIAA preamplifier.
DT
All just for Fun!

[DF96]

Grid noise should not be an issue at audio frequencies. It becomes noticeable from about 20Mhz and up. If you have sufficient grid current for the noise to be an issue then you probably have enough grid current to cause distortion too.

The main noise source in a triode is called shot noise, and comes from the discrete nature of electrons as they hit the anode. However, the space charge around the cathode smooths things out considerably so shot noise is much less than it could be. It ends up with the shot noise being affected by cathode temperature as described by Rod Coleman. This makes me suspect that so-called 'shot' noise is really another form of Johnson/thermal noise, but complicated by the fact that there is not thermal equilibrium in the valve. Shot noise is equivalent to the thermal noise current produced by a resistance equal to 1/gm, at an absolute temperature of about half the cathode temperature. Lowering the cathode temperature will reduce noise at first, provided that there is still enough emission to provide smoothing. Reduce it too much and the noise gets worse, as the smoothing goes.

Pentodes have an extra noise source - partition noise. This is similar to shot noise.

[DualTriode]

Hello All,
More Interesting insights into noise in circuits.
So far we have Shot noise, Thermal noise and Partition noise with Shot noise identified as the most prominent.
Shot noise is identified as being related to the absolute temperature and resulting velocity of the electrons (charged particles) leaving the cathode. I am not sure is the noise is the result of grid current or electrons hitting the anode perhaps both. Not to discount other possible explanations.
If I was one the research group at MIT in 1941 studying noise in electron tubes I would have been more suspecting of high velocity electrons hitting the grid rather than the anode. In Class A there is a cloud of electrons striking the anode, a couple more or less would get lost in the cloud. A small number of electrons with velocity high enough to overcome the repulsion of the negatively charged grid seems to me a likely culprit. I do not know, just thinking noise.
More Thoughts?
Anyone ever adjust the heater voltage to reduce Valve Noise?
DT
All just for Fun!

[Wavebourn]

I used EF86 wit 5.5V regulated filament voltage, 60V on G2, grounded cathode, 150K in anode, 4M7 in 1'st grid, grounded cathode. I've found it was the optimum for a mic pre.

[DF96]

An electron captured by the grid instead of the anode would cause exactly the same noise current (antiphase) at both electrodes. This would be an example of partition noise. It is not true that among so many at the anode you would not notice a few missing. The difference would be down to different circuit impedances (so the same noise current causes different noise voltages) and gain from the grid. Grid partition noise is not likely to be much of a problem in a healthy triode when properly biased. The grid leak bias recommended by Wavebourn guarantees a very low grid current, so no noise from this.

The official line is that triodes do not have thermal noise, but shot noise related to cathode temperature as this affects the statistics. As I said, I find this suspicious. True shot noise is not temperature-related, but comes just from the discrete charge of the electron. This is how a noise diode works (used for measuring the noise factor of a radio receiver).

There is another noise source - the cathode. It is believed that cathode impurities are responsible for flicker noise (1/f noise). Unlike shot noise and partition noise, flicker noise depends on the detail of valve manufacture (e.g. cleanliness) and material quality so some manufacturers will be better than others, and valves designed for audio are likely to be better than those intended for RF or switching.

[DualTriode]

Tube (Valve) noise is a complex topic, I have found.
Shot noise Partition noise, Thermal noise all add up from many interacting causes. We as users have limited control. We can do a few things. Previously I did not given it much thought, noise happens.
Today I spent some time reading about Valve noise in VACUUM TUBE AMPLIFIERS Valley and Wallman Editors and ELECTRONIC DESIGNERS HANDBOOK Landee, Davis & Albrecht . The bulk of the information was about RF receivers, not limited to audio frequencies.
As stated by Rod Coleman and DF96 the primary noise is shot noise. From what I read it is not so much individual electrons striking the surface of electrodes as rain on a tin roof but random variations of the intensity of the electron cloud in terms of energy and number of electrons. The random uneven flow of electrons being the cause of the shot noise. Individual high velocity electrons striking a surface can dislodge multiple electrons; this is secondary emission, another kettle of fish.
Besides selecting tubes for their noise there are two things I took away from the books we can control regarding “Valve” noise; absolute cathode temperature and plate voltage. Cathode temperature affects the number of electrons and the velocity of the electrons in the cloud. Reducing anode voltage has a limiting affect on electron or current flow through the Valve. With reduced flow of electrons there is lower average random variation of the electron cloud and reduced Shot noise as a result.
Just a couple of things to think about, perhaps play with.
DT
All just for fun!

[DualTriode]

Hello All,
I have been reading through the old school stuff. Even bought a hard back copy of MIT Vacuum Tube Amplifiers edited by Valley and Wallman from Abe Books. I also have ordered a copy of High Fidelity Amplifier Design by Chrowherst and Cooper.
Noise in valve amplifiers is not a simple topic. The old school for low noise was detection of weak radio signals. The MIT book, a result of the WWII effort has 220 plus pages of valve noise discussion and low noise circuits.
Some tubes were designed for low noise some like the 6922 have lower noise because of “modern” mass production techniques. High gm is typical.
There is a lot to sort through.
Does anyone have any noise as it relates to audio references to toss out?
DT
All just for fun!