标签: Computers

Several years ago, at the start of my engineering career, I was involved in the testing of a three-axis, cryo-cooled IR tracking system for the Navy. The thing looked like a long, cylindrical missile, with a round nose and a golden lens. We had to bring the tracking system and its control electronics to a GE test centre in Philadelphia, Pennsylvania, US where we would bolt it up to a hydraulically driven servo-controlled three-axis gimbal system for tracking tests. As I walked into the test centre, I passed two gigantic consoles, standing side-by-side. I remember them standing over 10-12 feet tall, and at least as wide. A desk-like area jutted out at waist height, and above that was what looked like a giant breadboard. I was intrigued. I had played with Heathkits when I was a kid ("101 Electronic Projects"!). I asked our host, "Is this a giant analogue computer?" He answered that it was. The year was 1984. The three-axis gimbal was probably no more than 50 feet away from the analogue computers (this becomes important later in my story). Now, since the tracking system was eventually to mount up to a Navy fighter jet, it needed to run on 440 V, 400Hz, 1 phase. The electrical engineers with us had this cute little aluminium block that they said was a DC to 400Hz converter. It was about 2 × 3 × 4 inches size, had an orange stripe around it, pigtail contacts on the bottom, and a warning label. I forget, now, exactly what the warning label said, but I do recall quite specifically (even to this day) that it warned the user that the converter had to be mounted up to an aluminium heat sink, and went as far as to specify the dimensions, and provide a few tapped holes on the bottom to make such mounting simple and straightforward. But we didn't have a heat sink. I asked the lead electrical engineer if he had brought a heat sink with him, but he said that it really didn't need one. I pointed out the warning label, but he insisted that no heat sink was required. I was just out of college—what did I know?! We did need a DC power supply, though, and we needed one with a pretty good current capacity. The lab had such a power supply, over there on the other side of the room, on the bottom of a 19-inch rack system, but we needed such a high current that we had to bypass the front, fused contacts and tie directly into the output of the supply. I say "we" here, to be polite, but I wasn't the guy who did the tie-in. Maybe it was at that point that I asked about the heat sink, I can't exactly remember. Anyhow, we shored up a few mechanical issues and started running our initial tests. The tests were all going well when, without warning, a shower of sparks erupted and a voluminous cloud of pure white smoke billowed from the DC power supply at the far end of the room. Remember the scene in the original Star Trek when they would fire a phaser at some poor, unsuspecting computer, and it would erupt in sparks and white smoke? It was that. Exactly that! Who knew? But it wasn't the fault of the DC power supply. It was the fault of the DC to 400Hz converter. Remember the orange stripe? You probably realised right off that it was a thermal paint stripe. It had turned to whatever colour it wasn't ever supposed to turn to. So, our 400Hz converter was toast, as was the facility's DC power supply. And the client—the Navy—was due in a few hours. After waiting for someone else to raise their hand (one of the lead electrical engineers, perhaps?) I realised that no one else knew what to do. For some strange reason, I did. From my Heathkit days, the term "Multistable Multivibrator" was stuck in my head. I asked our host if anyone knew how to program and run the analogue computers. We got lucky: someone did. I asked their expert to make me two circuits. The first circuit needed to be a 400Hz signal generator. The second circuit needed to be a power amplifier. He hooked up his wires and blocks and connectors, and turned what were probably a few million dollars worth of cutting edge computers (well, what might once have been cutting edge) into a Navy power supply. He tied an oscilloscope to the terminals to check his work, and we were good. I think we had a little clipping at the top, but it was good enough to run our equipment for the test and eventual Navy demo later that afternoon.   Not what you want to see just before the client shows up! Steven Swyak submitted this article as part of Frankenstein's Fix, a design contest hosted by EE Times (US).  

If you're writing code for an 8051, trust me, you're not a dinosaur. There's a new study of the MCU market. The forecast is that 6.7 billion 4- and 8bit MCUs will ship this year, up 6% over last year. Considering that so many feel 8 bits is dead that's pretty astonishing growth. Those parts are up 40% since 2009. 16 bit MCU shipments have almost doubled since 2009. And 32 bitters have gone up by a factor of four over the same period, though one wonders how many 32 MCUs really existed five years ago. ARM's 2012 annual report indicates that some 26 billion cores were shipped last year. This number presumably includes the 19 billion 4/8/16/32 bit MCUs mentioned above. What about PCs? Something like 350 million were shipped last year. This source pegs 2013 Q1 tablet sales at 49 million units, or about 200 million per year. To a first approximation, it doesn't really matter how many cores made it into each PC and tablet; together they represent just a tiny sliver of all CPUs and MCUs produced last year. Twenty-six billion cores is an astonishing number. Seventy years ago there were zero programmable electronic digital computers in the world. Zero cores. All computation used mechanical calculators and some speciality electronics. Even by the early 1970s the notion of ubiquitous computing wasn't considered absurd; it simply wasn't considered. Ken Olson, head of Digital Equipment Corporation, said in 1977 (six years after the first commercially-successful microprocessor was introduced) "There is no reason anyone would want a computer in their home." A drop in the bucket But 26 billion is a pittance. That's just a handful per capita. So many toys have CPUs. Medical gear. Cars. Calculators. I bought a digital oral thermometer recently, with LCD display and some sort of MCU under a blob of epoxy. For $4.99. This stuff surrounds us, and it's clear that it won't be long before these numbers explode. The data shows total MCU shipments doubled in the last five years. If that rate of growth continues, figure on around 80 billion MCUs/year in a decade. But another factor will made even that prediction low: prices are collapsing. Today one can get a 32bit MCU for $0.39. The semiconductor industry has taught us that falling prices drives increased consumption; that cheap computers create entirely new application areas. A young insurance salesman who stopped by here recently was interested in the stuff on my lab bench. I showed him some simple disassembled products: a TV remote control. That oral thermometer. An opened cell phone. An old GPS. He had no idea there were computers in these things. Computers are everywhere. Apparently they are invisible.

Are you writing code for an 8051? If so, believe me, you're not outdated. There's a new study of the MCU market. The forecast is that 6.7 billion 4- and 8bit MCUs will ship this year, up 6% over last year. Considering that so many feel 8 bits is dead that's pretty astonishing growth. Those parts are up 40% since 2009. 16 bit MCU shipments have almost doubled since 2009. And 32 bitters have gone up by a factor of four over the same period, though one wonders how many 32 MCUs really existed five years ago. ARM's 2012 annual report indicates that some 26 billion cores were shipped last year. This number presumably includes the 19 billion 4/8/16/32 bit MCUs mentioned above. What about PCs? Something like 350 million were shipped last year. This source pegs 2013 Q1 tablet sales at 49 million units, or about 200 million per year. To a first approximation, it doesn't really matter how many cores made it into each PC and tablet; together they represent just a tiny sliver of all CPUs and MCUs produced last year. Twenty-six billion cores is an astonishing number. Seventy years ago there were zero programmable electronic digital computers in the world. Zero cores. All computation used mechanical calculators and some speciality electronics. Even by the early 1970s the notion of ubiquitous computing wasn't considered absurd; it simply wasn't considered. Ken Olson, head of Digital Equipment Corporation, said in 1977 (six years after the first commercially-successful microprocessor was introduced) "There is no reason anyone would want a computer in their home." A drop in the bucket But 26 billion is a pittance. That's just a handful per capita. So many toys have CPUs. Medical gear. Cars. Calculators. I bought a digital oral thermometer recently, with LCD display and some sort of MCU under a blob of epoxy. For $4.99. This stuff surrounds us, and it's clear that it won't be long before these numbers explode. The data shows total MCU shipments doubled in the last five years. If that rate of growth continues, figure on around 80 billion MCUs/year in a decade. But another factor will made even that prediction low: prices are collapsing. Today one can get a 32bit MCU for $0.39. The semiconductor industry has taught us that falling prices drives increased consumption; that cheap computers create entirely new application areas. A young insurance salesman who stopped by here recently was interested in the stuff on my lab bench. I showed him some simple disassembled products: a TV remote control. That oral thermometer. An opened cell phone. An old GPS. He had no idea there were computers in these things. Computers are everywhere. Apparently they are invisible.