oren的博客

Analogue Sign Processing

The actual environment is analogue! So why is every single machine or instrument about us while in the home or at work increasingly described as 'Digital'? To answer this question, we had better establish what we mean by the word: 'Analogue'.

All measurable factors in life vary continuously in amplitude (size) with time: the skin temperature, the velocity of the car or even daylight. We can change a different temperature to the different electrical voltage making use of a sensor. We now have an electrical analogue of the original effect. As the temperature varies so does the voltage. Now we have this analogue sign, we can method it utilizing other electronic elements and display a temperature reading on the simple pointer instrument. The thing to remember is the fact natural parameters vary continuously, not in discrete steps; even devices that appear to operate in the discrete or electronic fashion may be deceptive.

Measuring the Environment

If we need to measure a length, we could use a ruler or tape measure. But what if we want a machine to choose the measurement? Perhaps we would like the measurement displayed in different units according to your switch position, or even contain the machine use the information itself and perform some appropriate action. We need an electrical or electronic method that converts whatever we are trying to measure to an electrical signal, maybe performing some sign processing before displaying or outputting the end result.

Simple Analogue Units

A Single-Unit Measure-Display Procedure calls for no processing. Examples include the mercury thermometer, mercury barometer plus the moving-coil ammeter. Notice that they are really all ‘Analogue’ Techniques, converting a parameter directly to the visible display.

The Two-Part Measure-Display Procedure usually involves little or no processing and before cars became ‘digitalised’, examples of analogue measure-display could be uncovered using the fuel gauge and water temperature instrumentation. In both cases, we have a separate sensor converting the parameter to an electrical signal, in these cases current, in addition to a display instrument. The latter utilised a hot-wire method to transform the current value to some pointer position.

Measure-Process-Display Systems tend to be more complex electronically than the examples above because some sort of signal conditioning circuits are included between the sensor as well as the display/output system. Traditionally these will be 'analogue' circuits composed of transistors, resistors, capacitors and far more recently integrated circuits or 'chips'. Note that not all chips are electronic. A typical piece of 'signal processing' might be to get rid of high-frequency electrical noise from a nearby electrical motor. The circuit would probably be a Low-Pass Filter in this case.

What's wrong with Analogue Processing?

The programs discussed above a described as 'traditional' because they represent an era of measurement and electrical/electronic techniques that go back centuries. They had the advantage of remaining reliable and cheap to make (in large quantities). Analogue sign processing was kept to the minimum because electronic parts were being expensive, unreliable and required skilled style and design engineers to make it work. Let's appear at this in a lot more detail.

Component tolerances are a major headache for the analogue hardware designer. Very specific values of resistors or capacitors might be necessary to realize a particular specification, but only specific preferred values are manufactured. This might mean resorting to variable components at greatly increased cost plus the need for setting up adjustments after production.

Component ageing is less of a problem nowadays thanks to new materials, but it can still be significant. Such as, a resistor might have had a certain resistance value when it left the factory, but years later it may have changed enough to get the circuit outdoors its original specification or even to cause comprehensive failure.

Electrical noise or Interference induced from the analogue circuit can sometimes be removed by added circuitry if it might be distinguished from your wanted signal. Far more often than not, the electronics cannot tell the variance between noise and signal. Consider the old vinyl record player (now back in fashion for reasons that escape me): it’s impossible to get rid of the needle 'scratch', turntable bearing 'rumble', clicks, pops and hiss without removing chunks of your music as well. Your brain can sort it all out, but even the most sophisticated analogue processing technique cannot. The most beneficial you can hope for is to reduce overall noise to an acceptable level.

Complicated components style and design is needed for even simple processing tasks. Even if all you want to do is implement a low-pass filter, that is, take away all frequency elements above a specified value through the sign, you will find it no straightforward process. Provided a precise performance specification, there are a large variety of possible techniques, every of which has an even larger number of possible circuit implementations. The tolerance problems come in to play, and if that wasn't enough, the layout and structure from the printed circuit board (PCB) it can be all built on may add 'stray' capacitance effects major to instability inside of a high-frequency layout. Design compromises are inevitable.

Difficulties in debugging, modifying, or updating an analogue components design and style cause the product to generally be expensive at the outset, with considerably wastage of effort later. Mistakes inside the circuit design and style direct to the physical replacement of factors and remaking of PCBs. Updates later on will often involve similar physical changes, so a great deal to ensure usually it is not really worth bothering and also the whole process is designed again from scratch.

Electronic on the rescue…

By now you could be forgiven for thinking that designing and making any new electronic program is fraught with these types of difficulty, that it truly is a miracle any 'high-tech' products are manufactured at all. Fortunately, salvation is at hand while using the invention from the computer and Discrete-Time or Electronic Sign Processing. During the 1920's work by a telegraph engineer known as Harry Nyquist formed the basis of what we now call digital signal processing, although even he based mostly his ideas on a great deal earlier work by others. In order to realize the benefits of DSP, we must move within the Continuous-Time processing that we have been utilizing approximately now, to Discrete-Time processing.

What's meant by 'Discrete-Time'? Nyquist and other people ended up able to show mathematically that you could work on samples of the signal taken at regular intervals and still get a satisfactory output. It seems bizarre, but it truly is true: you can sample a continuous waveform or sign, then reconstruct the original continuous sign accurately from those samples. It gets better. The rule that governs this sampling, often known as the Nyquist-Shannon Sampling Theorem, is very simple but without it there could well be no electronic sign processing. It’s not necessary to know the complicated mathematical proof behind this very simple equation in get to use it:

fs > 2B wherever fs is the sampling rate and B would be the bandwidth in the signal being sampled.

So, one example is, if you have got an audio sign which has a maximum frequency limit of 15kHz, then you will need a sampling rate of far more than 30000 samples/second. Naturally, there are one or two ‘catches’ which I’ll talk about later, but basically if you sample the signal at this rate it is possible to recover the original analogue waveform just.

Related links：

The Value of Electronic Sign Processing

The Fundamentals of Digital Signal Processing

Basic Elements of Electronic Sign Processing Method

Distinctions between Analog Sign and Digital Sign

Advantage of Electronic About Analog Sign Processing