SoylentNews is people
SoylentNews
from the I-like-making-"choo-choo"-sounds dept.
The hissing sound you hear in the background when you turn up the volume of your music player is called "noise". Most of this hiss is due to the thermal motion of electrons in the music-player circuitry. Just like molecules in a hot gas, electrons in the circuitry are constantly jiggling about in a random fashion, and this motion this gives rise to an unwanted noise signal.
But there is another type of noise that only comes into play when we have an electrical current flowing. This noise is known as shot noise. Obstacles that generate shot noise in this way are found in many electronic components, such as diodes and some transistors, and electronic engineers take great efforts to try to get rid of the effects of all sources of noise, including shot noise, in their designs.
Now a new study has shown that shot noise can be eliminated at its microscopic origin. And to do so, they have borrowed an idea from an unlikely source – the early days of the steam engine.
[...] In the 18th century, James Watt was struggling to get his steam engine to run at a constant speed. To solve this problem, he came up with the "centrifugal governor" in 1788, a contraption that consisted of two metal balls rotating on a vertical spindle driven by the steam engine. If the engine ran too fast, the balls would move upwards under the centrifugal force (a force acting on a body moving in a circular path is directed away from the centre around which the body is moving).
[...] The new experiment focuses on an ultra-small electronics device known as the single-electron transistor, which... [is] somewhat like ordinary transistors, which switch electronic signals, but taken to the extreme limit of miniaturisation such that electrons move through them one at a time. This happens via quantum tunneling, which means the current through a single-electron transistor suffers from the randomness of shot noise.
Using sensitive charge measurements, the researchers were able to detect exactly when an electron had tunnelled through the transistor. Based on this electron counting, they then adjusted the voltages of the transistor, following Watt's recipe for the centrifugal governor: if more electrons than normal had tunnelled, they changed the voltages to reduce the flow; if fewer had tunnelled, the voltages were changed to increase the flow.
(Score: 0) by Anonymous Coward on Friday November 11 2016, @08:12PM
Not sure about in practice. Seems like this requires an independent feedback loop for every last transistor. That's a lot of complexity to drag around when the goal is miniaturization...
(Score: 5, Interesting) by KilroySmith on Friday November 11 2016, @08:33PM
Depends. In many analog amplifier circuits, there are a sequence of amplifiers, each with a finite small gain, that combine to give the overall larger gain of the circuit. For example, in the product we build, there are roughly six stages of amplifiers with an average gain of 10 to go from the microvolt signals that we receive to the volt-level signals that we actually digitize. Roughly 95% of the noise added by the entire chain of amplifiers comes from noise added by the very first amplifier - a microvolt of noise added to the 10 microvolts of signal gets amplified by all the other amplifiers to still be 10% of the signal at the last stage. A microvolt of noise added by the last stage where the signal is now 1,000,000 microvolts is supremely unimportant.
If this can be used in a low-signal, low-noise amplifier, it could be quite important in uses ranging from guitar amplifiers to radio telescopes to cell phones to brain research, when the signal you're interested in looking at is so small that current amplifiers add objectionable quantities of noise. The size isn't so important - using it in only the first stage of amplification might give you 95% of the benefit of using it in all stages; using it in the first two stages might give you 99.5% of the benefit of using it in all stages.
(Score: 2) by dak664 on Friday November 11 2016, @11:21PM
In principle you can do everything with a digital filter if you sample fast enough. This feedback mechanism may be cheaper for equivalent bandwidth, but i doubt it.
(Score: 2) by mcgrew on Saturday November 12 2016, @03:07PM
Sampling rate has nothing to do with noise and everything to do with alias distortion.
mcgrewbooks.com mcgrew.info nooze.org