History MediaNotes / AnalogVsDigital

The defining quality of an analog signal is that it's easy to amplify, attenuate, and filter out a meaningful signal, even if the signal appears to look like garbage. As an example, when tuning into a radio station, it's first static, then static and the radio station, and then finally the radio station as the tuner can filter out the proper frequency from the "noisy" signal. The radio waves that also travel in the air are very low power, but the correct signal is filtered out before being amplified and making its way to the speakers. Another advantage of this is that when the signals degrade, it's a gradual degradation.

Another issue is that due to the [[RandomNumberGod random nature of the universe]], it's very hard, if not impossible, to manipulate and maintain an analogue signal perfectly: gears skip, belts slip, pipes leak, and electrical currents fluctuate. This inherent randomness can [[GoodBadBugs actually be an advantage in some cases]], however, as it allows for rapid, naturalistic "fuzzy math" modeling of inherently chaotic and vaguely defined situations. A common historical usage for electronic analogue computers has been in modeling air circulation and liquid flow through complex pipe systems, for instance. They have also been used to rapidly generate accurate-enough solutions for certain differential equations which are extremely difficult or tedious to solve using digital computation or traditional precision calculation.

In short, analogue electronics are less efficient and harder to program, prone to physical damage, and subject to wearing out but are also more robust and tolerant of errors, and model the world more realistically. In trope terms, an analogue computer embraces "DontThinkFeel".

The defining factor in the conversion is the sampling rate. A few smart people came up with the [[https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem Nyquist-Shannon Sampling Theorem,]] which states that the sampling rate of an ADC must be double that of the highest frequency component of the signal in order to reconstruct it perfectly. It's a very simple explanation, but it works well for most applications. One of them is music. As the upper range of human hearing is 20[=KHz=], this implies that 40[=KHz=] is the most you need to reconstruct any audio signal.[[note]]Most audio streams deliver sound sampled at 44.1[=kHz=] for this reason. The additional 4.1[=kHz=] overhead exists both to deal with the limitations of analog filters, and for compatibility with both PAL and NTSC video signals.[[/note]] To encode and decode signals, there are two major methods used.

The defining characteristic is that digital signal can be copied perfectly. Small bits of noise also don't kill the signal, as long as the value being represented is within tolerance. In regards to electronics, digital signals require less average power, since they can be in a state that's fully off or mostly off.

The defining factor in the conversion is the sampling rate. A few smart people came up with the [[https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem Nyquist-Shannon Sampling Theorem,]] which states that the sampling rate of an ADC must be double that of the highest frequency component of the signal in order to reconstruct it perfectly. It's a very simple explanation, but it works well for most applications. One of them is music. As the upper range of human hearing is 20[=KHz=], this implies that 40[=KHz=] is the most you need to reconstruct any audio signal. [[note]]Most audio streams deliver sound sampled at 44.1[=kHz=] for this reason. The additional 4.1[=kHz=] overhead exists both to deal with the limitations of analog filters, and for compatibility with both PAL and NTSC video signals.[[/note]] To encode and decode signals, there are two major methods used.

The defining factor in the conversion is the sampling rate. A few smart people came up with the [[https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem Nyquist-Shannon Sampling Theorem,]] which states that the sampling rate of an ADC must be double that of the highest frequency component of the signal in order reconstruct it perfectly. It's a very simple explanation, but it works well for most applications. One of them is music. As the upper range of human hearing is 20[=KHz=], this implies that 40[=KHz=] is the most you need to reconstruct any audio signal. [[note]]Most audio streams deliver sound sampled at 44.1[=kHz=] for this reason. The additional 4.1[=kHz=] overhead exists both to deal with the limitations of analog filters, and for compatibility with both PAL and NTSC video signals.[[/note]] To encode and decode signals, there are two major methods used.

In a binary digital system, typically the high voltage is usually 1.5V, 3.3V, or 5V with the low voltage being 0V or the negative of the high voltage.

In short, analogue electronics are less efficient and harder to program, and subject to wearing out but are also more robust and tolerant of errors, and model the world more realistically. In trope terms, an analogue computer embraces "DontThinkFeel".

In short, analogue electronics are less efficient and harder to program, but are also more robust and tolerant of errors, and model the world more realistically. In trope terms, an analogue computer embraces "DontThinkFeel".

The defining factor in the conversion is the sampling rate. A few smart people came up with the [[https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem Nyquist-Shannon Sampling Theorem]], which states that the sampling rate of an ADC must be double that of the highest frequency component of the signal in order reconstruct it perfectly. It's a very simple explanation, but it works well for most applications. One of them is music. As the upper range of human hearing is 20[=KHz=], this implies that 40[=KHz=] is the most you need to reconstruct any audio signal. [[note]]Most audio streams deliver sound sampled at 44.1[=kHz=] for this reason. The additional 4.1[=kHz=] overhead exists both to deal with the limitations of analog filters, and for compatibility with both PAL and NTSC video signals.[[/note]] To encode and decode signals, there are two major methods used.

There have been many attempts to close the analog hole. One of the more popular techniques is to design software that refuses to read certain patterns. A widespread example is the [[http://en.wikipedia.org/wiki/EURion_constellation EURion constellation]], a [[ExactlyWhatItSaysOnTheTin pattern of points]] used to inhibit counterfeiting. It's been a feature of many professional image editing programs, fax machines, and copy machines since the 2000s. Other methods use audio and visual [[http://en.wikipedia.org/wiki/Digital_watermarking digital watermarking]], along with watermark detection software. In theory, this causes a computer attempting to access a re-digitalized file to refuse it (see image); but in practice there are too many programs that lack the (often proprietary) programming necessary for widespread coverage.

The problems with analog signals, in regards to electronics, is that they consume more power as components are rarely completely off, and the maximum value range of their variables is limited by the quality of its components and the amount of power they can safely handle; an analog computer pushed beyond its tolerances can be a real world example of both TimTaylorTechnology and ExplosiveOverclocking. If you're designing or analyzing analog circuits, it also makes for some MindScrew math such as [[http://en.wikipedia.org/wiki/Fourier_series Fourier series]] and [[http://en.wikipedia.org/wiki/Laplace_transform Laplace transforms]].

Digital electronics use a discrete signal, that is, there are hard values with nothing in between them. You couldn't take a pencil and draw a digital signal without lifting it up. Despite what the picture on the right shows, vertical lines are just representations. Though in the real world, the signal is made of a sum of sine waves such that it creates a tiny sine wave that swings wildly periodically. This is known as a [[https://en.m.wikipedia.org/wiki/Fourier_series Fourier series]]. A digital signal can either be a binary signal, in which case an "on or off" state, or a series of defined levels.

One advantage (or problem, depending on which side of the copyright fence you're on) of analog electronics is that analog output cannot be effectively stopped from being copied. Since humans have their own analog processors -- eyes and ears -- and cannot sense digital input by themselves (yet, anyway -- the future [[Franchise/GhostInTheShell might change this]]), the last step of any sort of audio or video interaction -- no matter how digital the whole process has been up to that point -- must by definition be purely analog. And anything that our bodies' organics can process, another analog recorder can as well -- so no matter how much protection, DRM and artificial limitations are imposed on the digital recording and the gear required to play it, an analog copy can always be made. For the reasons explained above such a recording will never be a 1:1 copy and some quality loss is unavoidable, but given good enough recording equipment it can be kept minimal, and once done it can be re-digitalized all over again with none of the original restrictions and no further quality loss, ready to be shared with the world.

This is called the analog hole, and because closing it is effectively impossible it's an ongoing nightmare for every entity concerned with monetizing audiovisual entertainment.

One advantage (or problem, depending on which side of the copyright fence you're on) of analog electronics is that analog output cannot be effectively stopped from being copied. Since humans have their own analog processors -- eyes and ears -- and cannot sense digital input by themselves (yet, anyway -- the future [[Franhcise/GhostInTheShell might change this]]), the last step of any sort of audio or video interaction -- no matter how digital the whole process has been up to that point -- must by definition be purely analog. And anything that our bodies' organics can process, another analog recorder can as well -- so no matter how much protection, DRM and artificial limitations are imposed on the digital recording and the gear required to play it, an analog copy can always be made. For the reasons explained above such a recording will never be a 1:1 copy and some quality loss is unavoidable, but given good enough recording equipment it can be kept minimal, and once done it can be re-digitalized all over again with none of the original restrictions and no further quality loss, ready to be shared with the world.

One advantage (or problem, depending on which side of the copyright fence you're on) of analog electronics is that analog output cannot be effectively stopped from being copied. Since humans have their own analog processors - eyes and ears - and cannot sense digital input by themselves (yet, anyway - the future [[GhostInTheShell might change this]]), the last step of any sort of audio or video interaction - no matter how digital the whole process has been up to that point - must by definition be purely analog. And anything that our bodies' organics can process, another analog recorder can as well - so no matter how much protection, DRM and artificial limitations are imposed on the digital recording and the gear required to play it, an analog copy can always be made. For the reasons explained above such a recording will never be a 1:1 copy and some quality loss is unavoidable, but given good enough recording equipment it can be kept minimal, and once done it can be re-digitalized all over again with none of the original restrictions and no further quality loss, ready to be shared with the world.

Digital electronics use a discrete signal, that is, there are hard values with nothing in between them. You couldn't take a pencil and draw a digital signal without lifting it up. Despite what the picture on the right shows, vertical lines are just representations. [[note]]Actually in the real world, a digital signal is a summation of sinusoidal waves such that the actual sinusoidal wave is very small, but shifts in offset at a periodic rate. This is known as a [[http://en.wikipedia.org/wiki/Fourier_series Fourier Series]][[/note]]A digital signal can either be a binary signal, in which case an "on or off" state, or a series of defined levels.

In a binary digital system, typically 0 volts is the off state, while some other voltage (commonly, 1.5V, 3.3V, 5V, and 12V, usually based on what batteries can provide) are used as the on state. [[note]] Well, mostly. It's poor circuit design to set the threshold at what the operating voltage is supposed to be, and thus you could run most circuits within 20% below their rated operating voltage. Too much voltage though, and it explodes.[[/note]]

The defining characteristic is that digital signal can be copied perfectly, in the sense that small bits of noise won't kill the signal. In regards to electronics, digital signals require less average power, since they can be in a state that's fully off (or mostly off)

The problem with digital signals though is that if part of the signal is trashed, then the entire signal has to be thrown away unless something to correct it is available. This is akin to handwriting, where if someone writes a letter sloppily, forgets a letter, or misspells the word, you may not be able to make out what the word really is. And if you're in a part of the world that has digital TV, you can find it very annoying that poor reception means the channel cuts out completely, rather than just getting staticky like in an analog system.

The defining factor in the conversion is the sampling rate. A few smart people came up with the [[https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem Nyquist-Shannon Sampling Theorem]], which states that the sampling rate of an ADC must be double that of the highest frequency component of the signal in order reconstruct it perfectly. It's a very simple explanation, but it works well for most applications. One of them is music. As the upper range of human hearing is 20[=KHz=], this implies that 40[=KHz=] is the most you need to reconstruct any audio signal. To encode and decode signals, there are two major methods used.

The defining factor in the conversion is the sampling rate. A few smart people came up with the [[https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem Nyquist-Shannon Sampling Theorem]], which states that the sampling rate of an ADC must be double that of the highest frequency component of the signal in order reconstruct it perfectly. It's a very simple explanation, but it works well for most applications. One of them is music. As the upper range of human hearing is 20[=KHz=], this implies that 40[=KHz=] is the most you need to reconstruct any audio signal. To encode and decode audio, there are two major methods used.

An analog to digital converter has two fundamental properties to it: sampling rate and sampling resolution. The sampling rate is how many times per second (in Hertz) the ADC takes a measurement while the resolution is how fine a granularity there is between the maximum and minimum signal amplitude. In general, the higher ''both'' of these are, the more accurate the digital signal is. However, higher is not necessarily better. Two smart people came up with the [[https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem Nyquist-Shannon Sampling Theorem]], which states that the sampling rate of an ADC must be double that of the highest frequency component of the signal in order reconstruct it perfectly. Of course, that's a very simple explanation, but it works very well for most applications. Where this is most applicable is in recording music. As the upper range of human hearing is 20[=KHz=], this implies that 40[=KHz=] is the most you need to reconstruct any audio signal.

A digital to analog converter has the same properties as the ADC. When converting a digital signal into an analog one, it usually doesn't mean it's trying to transplant the stair step looking signal into the analog world. Instead it continuously raises or lowers the amplitude of the signal to that point using two methods. The first is called Pulse Code Modulation (PCM), which is where each sampling point is given some value between a minimum and maximum threshold. Imagine if you were [[https://en.wikipedia.org/wiki/Slalom_skiing Slalom skiing]], the flags are the sampled points and you're recreating the signal. The trail in the snow you leave behind is essentially the reconstructed signal. The other type is Pulse Width Modulation, which requires a much faster sampling rate as the samples are binary (0 or 1). When the sample is 1, the amplitude of the signal raises and if it's 0, it lowers. The amount of time a PWM signal is "1" is referred to as its duty cycle. An example of PWM devices that you encounter are fluorescent and LED lights. The higher the duty cycle, the brighter and less flickering the light appears.

The biggest issue with conversion is something called the quantization error. Because analog signals have infinite subdivisions, it's impossible for any digital system to perfectly reconstruct an analog signal. For instance, if you have a signal level of 3.5 but you can only store 3 or 4 as its value, then you're going to have to pick which value makes more sense. One solution, which takes advantage of our brains constantly interpreting signals based on the previous one, is constantly going back between 3 and 4 which will achieve an effective level of 3.5. [[note]]Our brains are actually creating expected results from sample points. Illusions play with our brain's ability to do this, which is why illusions often confuse us[[/note]]

A digital to analog converter has the same properties as the ADC. However, when converting a digital signal into an analog one, it usually doesn't mean it's trying to transplant the stair-stepping signal world. Instead it continuously raises or lowers the amplitude of the signal to that point using two methods. The first is generally called Pulse Code Modulation (PCM). Imagine if you were [[https://en.wikipedia.org/wiki/Slalom_skiing Slalom skiing]], the flags are the sampled points and you're recreating the signal. The trail in the snow you leave behind is essentially the reconstructed signal. The other type is Pulse Width Modulation, which requires a much faster sampling rate as the samples are binary (0 or 1). When the sample is 1, the amplitude of the signal raises and if it's 0, it lowers. The amount of time a PWM signal is "1" is referred to as its duty cycle. An example of PWM devices that you encounter are fluorescent and LED lights. The higher the duty cycle, the brighter and less flickering the light appears.

The biggest issue with conversion is something called the quantization error. Because analog signals have infinite subdivisions, it's impossible for any digital system to perfectly reconstruct an analog signal. For instance, if you have a signal level of 3.5 but you can only store 3 or 4 as its value, then you're going to have to pick which value makes more sense. One solution, which takes advantage of our brains constantly interpreting signals based on the previous one, constantly going back between 3 and 4 will achieve an effective level of 3.5. [[note]]Our brains are actually creating expected results from sample points. Illusions play with our brain's ability to do this, which is why illusions often confuse us[[/note]]

An analog to digital converter has two fundamental properties to it: sampling rate and sampling resolution. The sampling rate is how many times per second (in Hertz) the ADC takes a measurement while the resolution is how fine a granularity there is between the maximum and minimum signal amplitude. In general, the higher ''both'' of these are, the more accurate the digital signal is. However, higher is not necessarily better. For example, two smart people came up with the [[https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem Nyquist-Shannon Sampling Theorem]], which states that the sampling rate of an ADC must be double that of the highest frequency component of the signal in order reconstruct it perfectly. Of course, that's a very simple explanation, but it works very well for most applications. Where this is most applicable is in recording music. As the upper range of human hearing is 20[=KHz=], this implies that 40[=KHz=] is the most you need to reconstruct any audio signal.

The problem with digital signals though is that if part of the signal is trashed, then the entire signal has to be thrown away unless something to correct it is available. This is akin to handwriting, where if someone writes a letter sloppily, forgets a letter, or misspells the word, you may not be able to make out what the word really is. And if you're in a part of the world that has digital TV, you can find it very annoying that poor reception means the channel cuts out completely, rather than just getting static-y like in an analog system.

The problems with analog signals, in regards to electronics, is that they consume more power as components are rarely completely off. If you're designing or analyzing analog circuits, it also makes for some MindScrew math such as [[http://en.wikipedia.org/wiki/Fourier_series Fourier series]] and [[http://en.wikipedia.org/wiki/Laplace_transform Laplace transforms]]. Another problem is that due to the [[RandomNumberGod random nature of the universe]], it's very hard, if not impossible, to replicate the signal perfectly.

In a binary digital system, typically 0 volts is the off state, while some other voltage (commonly, 1.5V, 3.3V, 5V, and 12V, usually based on what batteries can provide) are used as the on state. Well, mostly. It's poor circuit design to set the threshold at what the operating voltage is supposed to be, and thus you could run most circuits within 20% below their rated operating voltage. Too much voltage though, and it explodes.