Electronics

1.     Feedback in electronic engineering

In the good old days of electronics, amplifiers were fairly basic, and had a number of problems. The main one being that different frequencies were amplified by different amounts. Typically, high and low tones were amplified less than middle tones, so recorded music for instance would sound rather ‘tinny’. There were other problems too – sounds could be distorted if the volume was turned up.

In 1927, Harold Black invented a method of getting round these problems using negative feedback, with part of the signal coming out of the amplifier being fed back to the input signal to reduce the amplification. This natty technique solved several problems at once, and so became standard practice in electronic sound amplifiers.

Not only did it tend to amplify all frequencies the same – giving what is called a ‘flat frequency response’, but it also reduced distortion and production variations. Mass produced electronic components are not identical, they vary in their electronic properties, and this can cause problems when mass producing electronic equipment; negative feedback helps reduce this problem. Not only that, it also helps reduce the distortion, so it was a winner all round. There is a small price to pay however, because the negative feedback reduces the gain (amount of amplification). This is not much of a problem, and the benefits greatly outweigh the reduction in gain.

How does the negative feedback produce all these good results? In essence, it tends to even things out. If one frequency is amplified more than others, then the negative feedback will reduce the amplification of that frequency more than the others. Likewise if quiet sounds are amplified more than loud ones, then the reduction in amplification will be more for those sounds, helping to reduce distortion. It’s a bit like taxation; tax reduces everybody’s income, but it reduces higher earners by a bigger percentage, so it tends to even out post tax income.

But it gets even better. Negative feedback also helps to improve stability. Positive feedback can produce instability, like it does in public address systems. If there is any tendency for positive feedback to creep into an electronic amplifier, then using negative feedback reduces this problem, and so makes for a more stable amplifier.

All this does rely however on the feedback being fast enough – which it generally is in electronic systems. But if the feedback is slower than the rate at which the input is changing, it can turn into positive feedback instead of negative. This is definitely not good news, and the output can oscillate as a result. Oscillation basically means something wiggling up and down – like a ball bouncing, or a swing going back and forth. Electronic engineers have some lovely names for these kind of problems – this oscillation is called hunting, and there is a similar but slightly more complicated problem which causes fluctuation oscillations called motorboating – because it can lead to a noise simliar to a motor boat bouncing along.

Why does the feedback have to be fast enough? Well, basically it has to get back to the input in time to affect the signal which caused it – otherwise it is affecting some new signal coming in. It then gets its timing wrong, like somebody pushing on a swing at the wrong moment. Getting back to the taxation analogy – if it takes more than a year for the tax to bite, then it can exaggerate any reduction in income by imposing tax from a more affluent period. Feedback that gets its timing wrong like this is called out of phase, because the timing or phase of the signal is wrong.

Oscillators

I have already pointed out that positive feedback can cause oscillation. This is generally a problem, but it can also be used to good effect. Oscillators are needed in electronics for a variety of uses. A signal generator produces signals of a given frequency – maybe as an audible sound of a single pitch for tuning a musical instrument, or as a much higher freqency signal for use in testing electronic equipment.

Radios use tuneable oscillators  that can be made to oscillate at different frequencies, so that the radio can tune into different stations. They do this by using positive feedback and a tuned circuit with electronic components called a capacitor and an inductor. The tuned circuit makes the positive feedback work at the desired frequency, which can be changed by having a variable capacitor. When you twiddle a tuning dial on an old fashioned manual tuned radio, you are changing the variable capacitor. The radio frequencies used here are much higher than audible frequencies, but the principle is the same. Similar oscillators are used in radio and TV transmitters.

Oscillators that work at one accurate frequency are also used, for example in electronic watches. The watch counts the signals from the oscillator to work out the time. These oscillators use a quartz crystal to set their frequency.


Quartz produces a small electrical voltage when it is squeezed. In a quartz watch, a small piece of quartz is used in an oscillator, with positive feedback used to keep it ringing at an accurate frequency. Luckily, the quartz oscillation is not very sensitive to temperature changes, so quartz watches can be very accurate.

Other examples of audio frequency oscillators are tone dialling phones, alarm clocks, computers, keyboards, personal and burglar alarms – in fact, almost anything that makes a sound these days probably has an audio oscillator in there somewhere. And it all depends on positive feedback.

Another clever use of feedback in Hi Fi systems that is being used quite often these days is Motional Feedback loud speakers, originally developed in the early 1970s by Philips. Here the sound from the speakers is fed into a microphone, and fed back to the amplification using negative feedback to smooth the performance of the speakers. Speakers always have the problem that their performance is much affected by the acoustics of the room that they are used in. Motional feedback gets around this by feeding back the sound as it is produced, and achieves good low frequency sound even from quite small speakers.

Flip Flops

Another example of the kind of names engineers like to give things, Flip Flops are an example of another kind of oscillator. This time, instead of using analog components, where values like voltage and current vary continuously, we use digital components, which switch between two different values.  A basic component of digital systems (like computers) is the gate.

I don’t want to get too technical here, so suffice to say a gate is a component that switches the ouput according to the state of the inputs. If you take two inverting gates, and connect them together so that the output of each feeds the input of the other, something interesting happens. The resulting circuit can ‘Flip Flop’ between two states, where the output of one gate is high (equivalent to 1), and the other is low (equals 0). Depending on the details, the circuit can either be stable, where it sticks in one state or the other, monostable (returns to one state), or a free running astable multivibrator, where it switches rapidly between one and the other. This is very similar to the oscillator described previously, but it creates a digital pulse output instead of a smooth wave.

                                                     

Computers usually have a clock which is some kind of pulse like this which is distributed around the parts of the computer to keep them working together, or synchronously.

As you can see, flip flops are a case of feedback, because the output of one gate is fed back into the input of the other, and hence eventually back into itself.

Before we leave electronics, I would like to describe one more feedback example. It’s a bit complicated, but its interesting, and widely used, so I will try and keep it simple. Its called a Phase Locked Loop.

Its really just another kind of oscillator, but what it does is to keep the frequency matched to that of another signal (called the reference signal). It all hinges on using a special kind of oscillator whose freqency depends on another signal. This is called a voltage controlled oscillator. What happens is that the output from the oscillator is compared with the reference signal, and the difference is used to control the oscillator frequency. So if the oscillator drifts away from the reference signal, it is automatically corrected by the change in the feedback.

Here is a picture that might help:

Think about tuning a guitar – you can tell if a string is wrong by lstening for a beat note with a tuning fork or another string. The beat note disappears when the string is correct. So you are using feedback of the beat note to adjust the string. This is similar to the phase lock loop adjusting the oscillator using the difference in the output with the reference signal.

Phase locked loops are used in all kinds of electronics: digitally tuned radios and TVs, FM radios, computer disk drives, modems, and remote controls. Lots of clever things are done with them, such as clock recovery – reconstructing a necessary timing signal from a data signal, and its all done with feedback. In fact, two lots of feedback – one in the oscillator itself, and another in the signal comparison loop.

Lasers

When lasers were first invented, people knew that they were onto something a bit special, but weren’t quite sure what. It was a classic case of a solution looking for a problem. Media journalists of course latched onto the idea of using them as death rays, but this turned out to be rather impractical. Laser beams get easily absorbed or scattered by atmospherics, so their use as death rays is not very promising, though the U.S. military still play with them.

Also, powerful lasers are rather large cumbersome things. Scientists at Lawrence Livermore labs tried to use them to kick start nuclear fusion in small pellets of fuel by firing several lasers into it from all directions, this required an awful lot of long laser tubes, and never really got anywhere. Though I believe they are still trying – but then all attempts at nuclear fusion as a source of power have proved intractable. Anyway, it turned out that the largest use of lasers would be with very small ones – as in CD and DVD players. These are semiconductor lasers, and are used to read (and write) the small pits that make up the digital signal for CDs and DVDs. Lasers are used all over the place now: in the clothing industry for cutting cloth, as measuring devices in surveying, in entertainment for laser light shows, in medicine and dentistry for micro surgery, and simply as pointing devices.

This is all fascinating, but what does it have to do with feedback? Well, it’s all to do with the way a laser works. Laser stands for ‘Light Amplification by the Stimulated Emission of Radiation’ – a bit of a mouthful, which is why we call them lasers. Actuallly, lasers came about after something similar called Masers (Microwave Amplification ….). These were somewhat esoteric devices used to amplify small microwave signals from telescopes which were rather ‘noisy’ – they had small signals and a lot of rubbish mixed in with them. Masers were good at this, but had rather limited applications, so Lasers were developed from them, and eventually turned out to be very useful.

How a Laser Works

A laser is basically a tube with very precisely flat ends. The ends are treated so that they let some light through, but not all of it – like a half mirrored piece of glass. Inside the tube is a pure substance (can be gas, liquid or solid) that has atoms in it that emit light of a desired colour when they get excited. When atoms  are excited they jump around a bit, and if they get really excited, they fling of a photon or two (a little bit of light). That’s how all light gets made. Lightbulbs have bits of wire that get very hot, and then emit light. Other lights use gas (neon tubes) or vaporised metals (mercury, sodium), but they all basically get something hot and excited.

In a laser, light still gets made something like this, but in a more controlled fashion. What happens is that the laser is ‘pumped’ by injecting some energy – this can be done in a number of ways (e.g. like a camera flash), but the end result is that the atoms in the laser get excited. Some atom fires of a photon. The photon whistles down the tube until it gets to the end, or it hits another atom on the way. If it gets to the end, it may make it through to the outside world, or it may get reflected back. If it hits another atom, that’s already pretty excited, this atom then fires of a another photon, which is in phase, or in step with the incoming photon. This photon can in turn trigger off another atom. The trick is that, because the laser has very precisely flat ends, only photons that are travelling exactly along the laser get to travel up and down several times, kicking off a lot of other photons on the way.

So all these atoms firing off photons kick off other atoms to do the same – and this all happens very fast, with lots of photons whistling up and down the tube until all the pump energy is used up, and the atoms calm down. The photons that escape come out as a pulse of light with rather peculiar properties. This is how a ‘pulse laser’ works. Continuous lasers are similar except they are continuously pumped with energy, and give out a continuous beam of laser light.

Laser light is different because it has two properties, it is pure, or monochromatic and coherent. It is also unidirectional, because most of it comes out along the axis of the laser. Most light is a bit chaotic, it consists of lots of different wavelengths all mixed up, and the photons of light are not in step with each other, they all get started at different times. When sunlight is split up, like in a rainbow, you can see all the different wavelengths from red (longest) to blue (shortest). Laser light isnt like that – it is all one pure wavelength, because it comes from just one type of atom. It is also coherent, all the photons are in step with each other, because they all got started by the one process. Think of a bunch of marching soldiers. Ordinary light is like an indisciplined rabble, they all have different stride lengths, and they are not marching in step. Laser light is like a smart bunch of well trained squaddies, all in step and with the same stride. Most light beams spread out or disperse quite quickly, but because laser beams are coherent, they can travel a long way without spreading out. This is why we can bounce a laser beam off the moon, and use them to paint pictures in the sky at rock concerts.

I suppose a purist might say that lasers are not an example of feedback, but of a cascade or avalanche effect; one thing triggers another until lots of it happens together. But if we look at the process that is going on, we see that it is another example of positive feedback.

Photons trigger excited atoms. Atoms emit more photons.

Or in a picture:

In fact, lasers are a bit like oscillators in the sense that they also produce a (fairly) pure signal by feeding energy into a tuned circuit.

New developments keep taking place in laser technology. At the time of writing, there is work going on at Cambridge University on Liquid Crystal lasers, making really miniature lasers less than the thickness of a human hair. It is possible that these could be used in future flat screen displays, with better colour than exisitng plasma and LCD displays.