In article <1991Feb13.030614.24117@ddsw1.MCS.COM>, whos@ddsw1.MCS.COM (Ben Feen) writes: >While competing in an academic tournament, our team buzzed in on the >following question: > >NAME THE TWO TYPES OF COMPUTERS. > > ...The answer? "Analog and Digital". >Now, WHAT IS AN ANALOG COMPUTER? I have never seen a computer that adds 2+2 >as 2+2... Have you? Anyone? I've programmed them. I recall doing a "lunar lander" simulation on one, probably about 1965 or so. An analog computer represents numbers as a quantity of something that is proportional to the thing being represented. The "something" is usually a voltage, but it could be a distance. For example, a slide rule is an analog computer; the logarithm of your "number" becomes a "distance" on the scale. The electronic analog computer uses a set of amplifiers called "operational amplifiers"; these are DC amplifiers with very high gain and very high input impedance. Therefore their transfer characteristic is determined (almost) entirely by the feedback network. This network consists of plug-in resistors and capacitors. The wired network represents the physical representation of a differential equation. For example, if the input impedance of the opamp is a resistor, and the feedback impedance is a capacitor, you have an integrator. Back in the days when a computer with all the power of a TRS-80 filled a large room and cost megabucks, analog computers were a cost effective way of dealing with many types of problems, especially simulations (e.g. lunar lander). They gave you a real-time solution, too. I suspect that these things are extinct by now. CPU power and memory are so cheap that digital solutions are much cheaper to build. whos@ddsw1.MCS.COM (Ben Feen) writes: >Now, WHAT IS AN ANALOG COMPUTER? I have never seen a computer that adds 2+2 >as 2+2... Have you? Anyone? Read about Vannevar Bush's differential analyzer. This was a fully mechanical (at least initially) analog computer from the 1930's. It could solve differential equations. The people at Cambridge built one using Meccano (the British equivalent of Erector Sets), you can read about it in Maurice Wilkes' autobiography, Memoirs of a Computer Pioneer (which is good reading for more than one reason). Some years ago, Scientific American published a piece in their Amateur Scientist column on how to build a differential analyzer from Erector Set parts. This column is apparently one of the ones that made it into their Amateur Scientist anthology. As others have noted, electronic analog computers are still used for some purposes, and they were quite common in the 1950's and '60s. A slide rule is a mechanical analog multiply/divide/log/antilog arithmetic unit. Doug Jones nraoaoc@nmt.edu (Daniel Briggs) writes: >My boss had just shown me some pictures of the radio telescope that he >had done his dissertation with. (Not all that many years ago, >either.) The conversation eventually turned to how this thing was >pointed, as it had an azimuth-elevation mount. I believe the Parks radiotelescope in Australia works on this principle. There is a master equatorial that is linked to the alt-az mechanisms that drive the real (210 foot?) telescope. Anyone who uses DDAs (digital differential analyser) to do graphics today are in a sense doing analogue computation without realising it. Mechanical differential analysers made up of wheels and gears for solving Stieltjes type integrals are extremely ingenious. I believe a lot of work was done on that during the last world war at Harvard and MIT (before my time :-). I think Vannevar Bush was one of the pricipals. Anyone remember the term "Bush Analyser?" I'll bet there is still one of those in some dingy basement at MIT. Electronic analogue computers are basically a bunch of interconnected integrators made up of operational amplifiers with a capacitor on the feedback path (gak, you mean no one knows of them anymore? Even the EE Ph.D qualifying exam at Stanford in 1970 had Profs asking questions on them :-). In the good old days, the op-amps were made from a couple of long-tailed-paired 6SN7 octal valves (oops, I mean "a couple of differential amplifiers using tubes" :-). Later, hi-mu miniature valves like the 12AT7 and 12AX7 were used instead of the 6SN7s. Early digital computers also used 6SN7s for the Eccles-Jordan (capacitor-coupled, diode-steered) bistable multivibrator elements. What is a "bistable multivibrator?" We now call them "flip-flops." :-) :-) Later, in the mid-'60s, the opamps became solid-state encapsulated blocks and were built by the likes of Philbrick and Analog Devices. Today, you can probably use LM311s; but, alas, the art of solving integral and differential equations using analogue computers have died away to be replaced by numerical integration using behemoth binary devices :-). In the late '60s, "hybrid computers" became popular. These were the days when numerical integration using digital computers is still too slow/expensive in the lab. Do people remember EAI analogue computers coupled to Burroughs(?) digital computers? Purdue had one of those in the EE department around 1967. I wasn't too happy when that happened because they had to junk an RPC-4000 digital computer to make room for the EAI. The RPC-4000 was manufactured by Royal Precision Company, germanium transistors, a Flexowriter and a Drum memory (not drum secondary storage -- the instructions are executed straight off the drum!). Finally, until the mid-'70s Heathkit used to sell an Analog Computer kit. Them were fun days when only real men (oops, real people) programmed computers :-) :-). Kok Chen, AA6TY kchen@apple.com Apple Computer, Inc.