I haven’t seen an analog computer since my undergraduate days-of course, digital computers are fast and cheap.
But analog is so…elegant! Plus, they solve problems in real time-just set the dials-and you get your answer.
I wonder if advances in micro machining technology might lead to a renaissance in analog computing? Anybody know if analog is a dead end, or are there any prospects for it?
Analog computers can solve only a narrow class of math problems. And they don’t solve them in real time anymore than digital computers solve things in real time. The kind of problems they are good at integration and differential equations are being done more and more by digital computers in communications because the digital computers ad more accurate and more flexible.
Off the top of my head, I can’t think of anything you could calculate with an analog computer that a good fast DSP couldn’t do in the digital world these days. DSPs also have a lot of advantages over old fashioned analog circuits. DSPs don’t have to worry so much about noise problems in their signal paths, have better development tools allowing the circuit to be realized much more quickly, and can be modified very easily in the field (a simple firmware download, vs. a hardware board replacement for an analog computer).
The DSP revolution about 10 to 15 years ago pretty much nailed the coffin shut for analog computing.
Back 20 years ago, you still saw analog computers used in fast feedback control systems because the digital circuits couldn’t keep up. By the late 90s though most of these circuits were being replaced by digital designs. I may be a bit biased because I’ve personally been involved in a lot of designs where we’ve replaced old clunky analog controls and replaced them with digital circuits, but in the projects I have been involved in the digital circuits have outperformed the older analog circuits. It seemed like the analog stuff always had some twitchy behavior that you really had to fight to control, where the digital circuits were much more easily tweaked and didn’t need as much fiddling to make them work right.
The only place I’m aware of analog processing of any type supplanting digital systems is in the pro audio world. Even there, the DSP systems have made enormous strides in emulating audio circuitry.
Nonetheless, the market for pro level analog audio equipment is booming. A single vintage Fairchild 670 will run at least $25k, should you be able to find one. There has never been a better range of high-end analog preamps, compressors and the like available at reasonable ($1-6k) prices. And tape - meaning high-end r2r machines like the Studer A800 - has re-emerged as the recording platform of choice for a large percentage of the professional market.
this is not to say we don’t live and die by Pro Tools and the like…it’s just that we as an industry have seen the complete integration of digital and analog systems and utilize the most sonorous/effective tools from both platforms.
This has NOTHING to do with the end result - mp3, AAC encoding, digital v. analog and the like. This is all in the production and recording stage. No need to go into the A/D debate vis a vis consumer listening. I have mixed feelings about that, as do many producers and engineers. But it is a stone cold fact that high end analog audio - using one or more of the three T’s (tubes, tape, transformers) captures a more expansive range of audio information in a manner which is more pleasing and melodious. Sometimes it’s a pain in the ass, it’s always more expensive, but the difference is real and tangible.
and your firstborn for a set of tubes for it. I find it amusing, the number of LA2 kit builds, and all sorts of other stuff. A 670 is however a design of rare beauty. However, none of these are analog computers.
As engineer_comp_geek writes, for almost anything you can replace it with DSP now, and avoid all the second order weirdness that leads to instabilities. Not that DSP is a cakewalk, you can get into arbitrary trouble there too, but at least the trouble stays where you last left it. You can get gigasample ADCs for not a lot of money, and there is no analog device on the planet that could stay stable integrating something with that sort of bandwidth.
A type of analog computer is still used in automatic transmissions.
Just because there is no necessity doesn’t mean it won’t be used for novelty. Also, I believe analog computers have been designed virtually in many simulation games, like Second Life or Minecraft.
Don’t many applications of op amps amount to analog computers? Also many simple analog filters in signal processing, especially at RF where digital filters don’t go yet? If you accept that there are optical computers, isn’t every lens an analog computer?
In that sense I guess so, but most people would call them controllers, or processors, not computers. They process the actual signals you want, in the domain of those signals, in real time. But analog computers are designed to solve problems. So the whole idea is that the voltages and currents (or speed, torque) are analogues of the values you have. You could be solving a problem where, for instance, the inputs are tide heights versus time, and the output is the coefficients of the Fourier transform that will enable a different analog computer to plot tide heights in the future. Internally the analog computer is representing the physical properties in its own domain.
Analog computers do not have to be electronic, as observed above, they can be hydraulic, and they can, and have been mechanical. One the most awsome mechanical analog computers was built out of Meccano. Things like bomb sight predictors contained analog computers. The example above of the tide prediction system is one of my all time favourite devices, and you can see it in the London Science Museum. Designed by William Thomson aka Lord Kelvin it was a supercomputer of its time.
I’d be interested in whether a digital simulation of an analogue computer would be able to run faster than an actual analogue computer.
Long since made obsolete by electronic calculators, the Curta Calculator was an all-mechanical four-function calculator the size of a salt shaker. The ones still around are coveted collectors items. Here is a page with a Flash animation that allows you to operate a simulated Curta. (irony of ironies): Curta-Simulator (www.curta.de)
But it was digital.
:smack:
Actually, an analog computer. An analogue computer is completely different.
Am I being whooshed here, or are you making an actual distinction? Isn’t it just a spelling variation?
Sorry, making a joke about “analogue” being the adjective form of “analogy”.
But I see now that “analogue” and “analog” spellings can be used somewhat interchangeably.
Analog synthesizers is experiencing some sort of a renaissance, especially in the modular (both sound and video) department, but even small, cheap monosynths are coming back. Moog and DSI have consistently put out models for a long time now, korg came out with the monotron and the monotribe, and both arturia and waldorf will release new analogs this year.
The meaning of analog has moved about. What is becoming common usage, using analog as a synonym for “continuous” is rooted in a mistake. Definitions of analog computers in the 40’s were quite clear that the internal representations were analogues of the numbers/quantities being dealt with.
Devices that crunched numbers digitally were termed computers after the people they replaced. Before then a computer was a person that did computations, often with a mechanical adding machine. The early machines that processed numbers were termed digital computers. This distinguished them from computers that used analogue representations. All of which were continuous. Somehow the term analogue/analog got misconstrued as being the antonym of digital, and hence came to get its additional modern (and really quite wrong) meaning. This seems to have happened sometime in the late 50’s early 60’s, but maybe later. Analog computers are analog because they represent quantities with an analogue - something in perfect proportion - internally. Digital computers are digital because they use digital numbers. Digital is not the antonym of analog, and analog does not mean continuous - it means in proportion to.
In audio, and then video, the same terminology took hold. Before you could buy CDs nobody called tape or vinyl “analog”. But afterwards the term took off. Sadly the new meaning has taken hold, and we will probably just have to accept it.
I wonder why John Campbell chose Analog when he decided to change the name of Astounding SciFi magazine. That was in 1959, at a time when the divide between digital and analog computer was well understood. But he may have been going back to the roots of the word’s meaning.
I took an analog computing course as a senior EE student way back in 1971. I really enjoyed the course. Of course about 20 minutes later it was outmoded by the digital computer.
It’s been a long time so forgive me if I mess up but as I recall from my dif-e course we could only solve linear ordinary differential equations with constant coefficients.
The analog computer could solve them with variable coefficients. That was a wonderful thing back then.
I suppose nowadays the various elements we constructed such as integrators and differentiators using op amps can now be simulated digitally.
As far as speed goes, once it was set up and activated, as I recall it immediately began to plot the solution on an ink pen plotter and only took a couple of minutes to finish. That seemed really fast back then.
If the analogue computer isn’t instantaneous*, then I’d say yes - because a digital simulation can, in general alter the time axis at will (for example, a ballistic trajectory can be computed in less time than it takes for a projectile to travel it in the real world.
*Some analogue computers are more or less instantaneous - for example, a slide rule. The only way we could simulate those to run faster than reality would be to simulate them being made of incompressible materials (the parts of real-world machines transmit force at the speed of sound in the materials of which they are made). This would not be perceptibly faster to the end user.