One American sub lost in the 60's is believed to have tried to launch a (regular) torpedo with the outer tube door closed, so even though this has got to be one of the stupidest things you can do on a submarine, it can happen... I doubt that the supercavitation torpedos are inherently more dangerous than regular ones -- except that I think they are still experimental. One other thing may have worsened the odds for the Kursk -- when you can't pay your crew regularly, they are apt to get sloppy.
Re:Toms Hardware - Sentionalistic journalism!
on
AMD And THG update
·
· Score: 2
I've seen 3 or 4 fans fall off, always when the computer was shipped. That's out of hundreds of computers. It's not a real big problem, but it's certainly a problem. Most mobos nowadays don't go to hobbyists or techies that build their own boxen, they go to corporate desktops or naive home users that are NOT GOING TO OPEN THE BOX BEFORE PLUGGING IT IN. So if their heatsink/fan fell off, the Athlon will melt down.
The fact that most MB manufactures doesn't use the diode might tell something about the likelihood of the cpufan falling off!! The mobo in THG's article did use the diode -- just not with a circuit that would react instantly. With a dead fan it would run for several minutes, then shut down; with no heatsink, the CPU melted before the mobo reacted. Apparently all unmodified mobos are the same, which leads me to think that AMD recommended the too-slow circuitry.
I'm still wondering why THG didn't get AMD's response before posting their original article, but it does look like they've persuaded AMD to recommend some changes in future mobos...
Yes, it sounds like Siemens blamed the AMD chip when it was really a deficiency in their motherboard design. It ALSO sounds like AMD recommended motherboard circuits to support their chip that are not adequate if the heatsink falls off, therefore most or all mobos for the Athlon presently in production can smoke the CPU in this case, and that AMD's video used a modified mobo.
In short, neither Siemens nor AMD was entirely truthful, and if you buy a complete system, you need to open the case and make sure everything is where it belongs before turning it on. That's good advice whether or you've got AMD or Intel. We techies already know that, but Dell, etc., are shipping a lot of computers to people who DON'T know it, and might not be able to check for things like this even if the computer mfgs sent along instructions...
I don't know why THG didn't get AMD's input before publishing the smoking Athlon pictures. Maybe they didn't ask, maybe AMD didn't answer. But now AMD knows there is a slight problem, they should put the fast-acting shutdown circuit in their recommendations for the mobo mfgs, and quality mobos for Athlon will soon have it. And Dell, etc., will evaluate whether it costs more to require that on Athlon mobos, rather than occasionally having to replace a burned-up CPU and possibly the whole mobo.
a) EE Times is naturally going to concentrate on the EE jobs, but it's not only EE's. Go to uspto.gov and look at the job listings.
b) I think engineers would generally make better patent examiners (in their particular field) than physicists because engineers have a better grasp of the existing practice. When it comes to the worst patents the hard part evidently isn't understanding the application or searching the patent database, it is recognizing when the patent claims to cover an idea that is never-patented but already in use (and therefore public domain). E.g., to check out an application concerning communications between computers, familiarity with the rarefied mathematics of communications theory is less useful than the ability to recognize an attempt to patent the IP protocol...
(Maybe I'm biased. I switched from physics to EE. But I remember one physics prof who might not have realized that a patent application for a "circular transportation facilitation device" had prior art called the "wheel". No decent engineer would ever make that mistake.)
Familiarity with "hard" science fiction might help too./. once brought up a patent which claimed to cover the very idea of PDA's -- it's not going to hold up, because Arthur C. Clarke described a PDA in Imperial Earth, written in the 70's.
What I do worry about though, is:
1. Will they get good engineers with working knowledge in their fields, or just people with EE degrees? Starting pay of $50K isn't much for a good engineer, especially if the work is non-creative, involves excessive paperwork, and is located in the only city run by a congressional committee...
2. How will their engineers stay current in their fields when they aren't actually designing anything?
3. The bureaucratic motivations are still all wrong. Reject an application and the applicant can, and often will, sue the bureau. Rubberstamp the application, and there may be worse lawsuits, but the bureau isn't involved. Now, if the bureau got to pay the legal fees & court costs whenever a patent was successfully challenged...
Are we giving "same as the last design, just add this component" electrical device patents? I think they are, and what's wrong with that? Most patents simply protect one particular device from near-exact copying, and don't try to cover all ways of doing the same thing. Such patents protect step-by-step improvements on old designs, without hindering anyone who designs a different competitive product. Narrow patents like these meet the reason for patents expressed in the Constitution: "To promote the progress of science and the useful arts".
Patents become counter-productive (block progress) when they are too broad and claim to cover all ways of accomplishing something. This can only occur if either someone is remarkably innovative, or if they are claiming more than they actually originated. I suspect it is almost always the latter case, because when "great inventions" are placed in their historical context, nearly always the inventor was either pulling together existing ideas (Edison, the Wright Brothers), or was unable to make his idea work well enough in practice and had to leave it to a later generation with better materials (Da Vinci, Babbage).
The problem is, when it comes to software and high-tech "business methods", the patent office is often so confused as to grant ridiculously overbroad patents. Sometimes it allows very wide-ranging claims on an idea that is at best a slight improvement upon existing practice -- because the existing practice was never thought patentable, so searching their database doesn't turn it up, and the examiners are insufficiently familiar with the ideas that are in the public domain. An extreme example of this is the recently granted Aussie patent for the wheel; since patent databases don't stretch back to 5,000BC you aren't going to find the prior art there, but you think someone _should_ have noticed. Or in many cases, there is some small idea that may actually be innovative, but the patent claims far more than that -- analogous to claiming to have invented the wheel when actually you just invented a better kind of cotter pin to hold the wheel on the axle... You'll get clobbered if you go to trial against anyone who isn't using your new cotter pin, but often they'll pay modest royalties instead of bearing the expense of going to trial.
The most damaging patents of all are extremely broad ones filed by people who never created a successful product, but who simply guessed which way technology was headed, filed a vague patent application, and kept amending it for many years. About 1960 someone filed a patent on a block of silicon containing six transistors, interconnected by soldered wires; a long series of amendments and legal arguments kept this from being either issued or rejected until the 1980's, when it finally issued as a patent on integrated circuits! An IC of the period (say an 80286 CPU) resembled the original device like an aircraft carrier resembles a dugout canoe. The "inventor" hadn't done the 20-some years of work that advanced IC manufacture from six disconnected transistors to about a million connected transistors, but he wanted to cash in on it.
The level of insight offered in this speech is outstanding and thoroughly depressing when compared to the level of debate offered by parliament or congress.
When you collect the dozen or so best speeches of a century and ignore the rest, it's bound to look pretty good. I'm insufficiently familiar with McCauley's contemporaries in Parliament to comment on them, except to note that you can see the effects of their less inspired decisions abroad in the Crimean War and their colonies, and at home in anything by Carles Dickens.
In an area I am more familiar with, 19th century America had a few great men in Congress, and more than a few blackguards. One Senator ran short on words, so he beat an elderly Senator unconscious on the Senate floor. A Senator was publicly caught buying votes in his home state; the Senate promptly formed a committee to investigate, not the (ex)Senator, but the man who exposed him. Overall, American politics was cleaner in the 20th century than the 19th. Sometimes their language was more high-faluting than any modern American would use, but that doesn't make it more expressive, or less likely to have cloaked evil intentions or plain pigheadedness.
I don't know how the British Parliament compared with that. Certainly they weren't crude enough to commit physical assaults, nor do I think any MP would have been as badly educated as Tennessee congressman David Crockett, but probably many of them weren't as eloquent as Davy Crockett...
I'm not sure if that was a serious question, but... Metal fatigue is a big concern at NASA because it can cause spacecraft to come apart in flight. A piece of metal under stress can develop tiny cracks. (If you designed the metal as thin as possible to save wait, substitute "will" for "can".) When the stress is relieved and then applied again, the cracks may grow. After a certain number of cycles, the cracks get too big and the metal breaks.
The classic case of this was a very early jet airliner called something like Comet. Because early jet engines were pretty inefficient, they made the skin exceptionally thin so as to have more weight-carrying capacity available for fuel. When it climbed to altitude, the pressurized cabin would slightly bulge the skin outwards around the windows. When it landed, the skin would pop back. After a few months, planes started coming apart in mid-air. Microscopic examination (of not yet crashed planes) found patterns of cracks in the skin near the windows. They had to scrap the entire fleet...
The same thing could quite easily happen with the space shuttle -- not just from cabin pressure, but also from high-stress launches and landings. Or the skin on a Mars probe may expand and contract thousands of times due to sun heating as it rotates in space. Making the metal thicker will prevent this, but every ounce of structural metal takes away an ounce of payload. So NASA has to design right to the edge of initiating metal fatigue for repeated-use items (the space shuttle), and for some probes it may accept that metal fatigue will happen, but the cracks will grow so slowly that the mission is finished before it fails. This requires very good software for simulating crack growth and analyzing metal fatigue.
I wonder how many of those "error" messages really indicate errors? When I am programming, I will put in lots of messages to make debugging easier later on. I will disable them on the final compile, but there have been times when this got forgotten in the rush to release on the deadline. I wonder how often that happens with OSS -- especially since OSS releases are usually not the end of the project.
Also, messages that were intended simply to show the progress of the program or confirm it went down the correct path often inadvertently sound threatening: "Cannot find file xxxx.xxx", when what you really meant was "No initialization file xxxx.xxx found, using defaults."
Of course, as the author said, the problem isn't that OSS is worse than commercial software, but that it should be better. Is there anything in OSS as bad as the error message I sometimes got from Win95, "Cannot find file", without the file name and path? Not to mention how Windows allows an application to silently malloc some memory, forget to free it, and repeat until it crashes a different application or the OS itself...
Certainly I am not the first to notice that Industrial-Age schools and Industrial-Age prisons are very similar to factories in terms of set times and hierarchy. You forgot industrial-age armies... I agree about the resemblance. But prisoners and (most) schoolchildren don't have to get themselves up and ready to work on time.
From my own experience, basic stuff like showing up to work on time can be a very big problem with people who were raised on welfare, without any example in the family of an adult that got up every morning and went to work. I do know guys that seemed to really want to hold down a job, but couldn't come in on time three days in a row. Skilled white-collar workers often have some latitude in their schedule, because the work can get done whenever they show up, but the jobs an unskilled 18-year old can get are utterly unforgiving. If the restaurant is going to start serving breakfast at 6:30, someone's got to be there starting to set up at 5:30 -- and if they can't count on you to be there every morning, they will find a way to get along without you entirely...
ubernostrum, you are biasing the test. You started one example with "Double-click an icon on the desktop to open a folder" and the other with "type tail myfile". You aren't comparing equivalent actions. A fair comparison is something like:
Starting conditions: Target folder opened, target file located/spelling known, skilled at use of the OS and the mouse/keyboard.
GUI (Windows):
1. Doubleclick the myfile icon.
2. Wait for it to open
3. Click and drag the scroll button to the bottom.
CLI (Unix?)
1. Type "tail myfile"
2. Wait for it to open
I get 4 seconds in the GUI, or 4 seconds just to type two words on the command line. However, if I miss the right spot with the mouse, recovering from that is going to at least double the time. This is much more common (for me at least) than hitting the wrong key. The keys are big, but the scroll bar is narrow... So I'm not sure which would actually win on the average -- except that DOS has no "tail" command, so on my system I would still need to hit Ctrl-End after opening the file in Edit.
How about this test: "Start from c:, open a folder named something like my documents (GO), now open a file in it called something like my file." The Windows user spends 5 or 10 seconds locating the "My Documents" icon. The CLI user either already knows how it is spelled and types it in about 5 seconds, or spends about 30 seconds listing folders and scanning the list to find out. Etc. So the CLI will usually be faster if you've memorized the exact spellings, otherwise it's slower at simple jobs.
For the simple jobs, it isn't the interface, it's how familiar you are with the task. The GUI is certainly faster when you don't know what you are doing.
Or consider a hard job: rename all 50 files in a folder according to certain rules. In Windows, you do them one at a time: right click on each file, select "rename", switch hands to the keyboard,... If I knew the CLI command to do it all at once it would be much, much faster -- even including listing and reading the manual page to get the switches right. But I wouldn't know where to start in Unix. I still know all the DOS commands pretty well, but I can't think of a set of DOS commands that would do that. And printing out the directory, then going down it and typing a rename command for each file is going to take a very long time.
I printed that out and passed it on -- I can follow it's progress through the cubes by the gasps as people reach the last paragraph...
What is really strange here is that HP grew big by building very good electronic instruments, selling them high, and supporting them very, very well. An HP scope or meter might cost twice as much as the competition, but it would quite likely go 20 years with no service other than calibration, and even after 20 years you could still get it fixed if you were still using it. But I have heard many such stories about terrible tech support on computers, printers, etc. (On-site repairs here are pretty good, paid for by a corporate budget, but for printer drivers and software setup issues I've learned that the downloads on the web site are all you are going to get...) So WTF happened to their computer & printer divisions?
Of course, now the computer & printer division _is_ HP. The instrumentation company was split off and is called "Agilent." (Yucch!) Agilent tech support is decent as far as listening to what you are saying and trying to solve the actual problem. But sometimes the problem is just that the HP/Agilent software is so crappy it seriously limits what you can do with their excellent hardware...
"Bazaar" is a rather distorted view of how open-source projects run. If you try to start one that way, you will get an endless discussion of specs and no code. And if you tried to run an on-going project that way, you'd either get the same endless discussion, or infinite forking and no progess.
Open source projects start with one or two people defining the core and writing some code, then recruiting others to add to the code base. They progress if the founder is good enough at dealing with people to be able to hand out assignments to volunteers and to decide which improvements do or don't get into the code base. That is, they are run as mild dictatorships: Linus Torvalds welcomes suggestions but ultimately decides what becomes part of Linux. Anyone could go make his own changes no matter what Linus thinks, but given the respect Linus has earned, if you fork the code you will probably be working alone. Linux coders voluntarily submit to Linus's decisions because it's better to be working with Linus and 10,000 others than to be going it alone... It's a dictatorship that is benign enough that people volunteer to come under it, and dissenters can safely be ignored rather than shot.
What the "eyes and brains" can sometimes show is how the age of one site compares to another one. E.g., on one site you find iron tools in the upper layers, bronze lower down, and stone tools in the lowest level, the sequence is pretty clear. The actual ages aren't, but the order is, and depth of the layers can give some hints as to the length of each era. They when you find the same style of bronze tools somewhere else with just a few iron tools you know they were the same people, and that it was contemporary with a particular era on your main site.
But this does not give you real dates, so you cannot compare the ages of sites left by different, non-interacting cultures. The technological level alone does not limit the possibilities much -- stone axes are probably still in use today in New Guinea.
The only way you get real dates is when they start leaving written records. E.g., Egyptian civilizations are datable because they etched-in-stone a nearly continuous record of the years from their first cities up through Roman times, and the Roman Catholic church has kept a reasonably accurate tally of the years since. Mesopotamian civilizations would keep records for a while, but then collapse leaving gaps of unknown length, but because they interacted with Egypt you can often associate the date. But if you want to figure out when some European hunter lived, without radio-dating your only chance is to find some definite link (same kind of clothes, pottery, or equipment) as was used by people that interacted with one of the record-keeping civilizations. (The illiterate barbarian tribes really are important to history even though they don't write any -- they keep overruning civilizations grown too soft or too complacent.)
What I want to know is: why solar cells? It seems to me it would be much easier to make a space-based heat engine.
You forgot one more advantage: steam power plants routinely run at over 30% efficiency, higher than solar cells. All you need is a big mirror to focus light on the boiler, a big radiator in the shadow of the mirror, and the steam plant in between. And you can make the mirror extremely thin, there is no wind and very little gravitational/accelerational forces, and you just keep the mirror stationary (or stabilized by spinning about the optical axis) to keep it pointed at the sun. On Earth, a solar steam plant requires bracing the mirror against wind and gravity, and then mounting the whole shebang on a giant turntable to follow the sun. Or maybe you have a giant array of small steerable mirrors, but it's still very expensive, possibly more than an immoveable array of solar panels that would collect the same average power. In space: it's lighter than the solar panels (real thin mirrors...) so if you are lifting it from earth it's much, much cheaper. If you are building from space-mined materials, you've got a lot fewer factories to lift into orbit to build the steam plant.
Shaunuk, you mixed so much good and bad physics mixed together it's hard to tell where to begin. Individual photons do get more energetic with shorter wavelength (blue-er light), and if you could build solar cells tailored to each wavelength, the higher the energy the higher the voltage per cell.
But the number of photons in sunlight at each wavelength drops with increasing wavelength through most of the visible spectrum. So the red light has the most available power. Don't even consider the cosmic rays and other radiation in space, for power they compare to sunlight like a gnat to a supertanker.
But the second problem is the way solar cells work. They are semiconductor diodes, arranged so that an incoming photon of sufficient energy will knock an electron from the positive (anode) to the negative (cathode) end. All the electrons put together form a current flowing from the cathode. (Of course, they better go through something and come back to the anode for you to get any current or power...). Each color or wavelength can push an electron across a different voltage, but the photocell only works at one voltage. So photons that are too red are wasted completely because they lack the energy to make the electron go, and ones that are too blue have excess energy beyond what the electron absorbs. The result is that you have to design your photocell for some sort of compromise voltage, probably about 1.5V corresponding to a particular wavelenght of red. (Of course, they put a bunch of cells in series to make 12V or higher outputs from the panels). Even if everything else worked at 100% efficiency the inherent losses from energy mismatch will never allow a cell like this to get better than maybe 30% efficient.
UV light would just waste more of it's energy. One UV photon would at best impart to one electron the same energy as a red photon does, with the rest of the energy wasted. It's more likely that big a mismatch would just keep the photon from being captured at all--it would reflect or be absorbed as heat only.
I can think of two ways to work around this. One is to use a prism or diffraction grating to split the light out into different wavelengths, illuminating a set of photocells with different working voltages. This would certainly raise the efficiency with respect to the light entering the device, but it would be expensive, and I'm not sure you could make the separation work without restricting the light entry to a slit -- and blocking light outside the slit certainly won't give you much efficiency overall. Or maybe you could make a stack of solar cells, each one transparent to the wavelengths preferred by the cells below. Assuming this transparency is possible, there's a problem with cost -- as long as the semiconductor is the biggest cost of the panels, it will be far cheaper to just cover more square feet with inefficient cells than to stack up cells for higher efficiency.
The space junk is spread out pretty thinly. You'll have less trouble with it than with keeping earthside solar panels clean, especially in cities and deserts. (Of course, in northern Michigan where I live, the solar panel will be perfectly clean, it just has a foot of snow on top. 8-)
Once it a while, something will hit at 10,000 kph and blast a hole right through a solar panel or a mirror. This happens because to make this worthwhile, you'd have miles of solar collectors. But it won't affect your generating capacity noticeably, because the rest of the panels or the mirror will still be working (mirrors would be very thin metal, or aluminized mylar, not glass), and it's miles across. 20 years of this isn't going to reduce your capacity as much as a few days dust accumulation in Nevada.
One thing I don't know about though: what is the effect of hard radiation (cosmic rays, solar wind, etc.) on solar panels? I hear that solar panels are expected to last 20 years on Earth; is the much greater exposure to radiation hard enough to re-arrange the crystal matrix a bit going to make space-borne panels shorter lived, or will the lack of rain, wind, and crud going to make them longer lived?
However, I think that an SPS built now would use the much cheaper technology of using a mirror to focus sunlight on a boiler, which runs a steam turbine. Cosmic rays won't hurt a steam plant, although the crew is going to need good medical insurance...
Actually, the best orbit for beaming to one spot on the ground is not just geosynchronous but geostationary: above the equator & 24 hour period. Because the Earth's axis is tilted, the orbit is at 40,000 miles radius, and Earth is only 8,000 miles in diameter, the satellite rarely passes into the Earth's shadow. At the equinoxes, it will be shadowed when it goes around the far side of the Earth, but that's only a couple of hours a day or less for a few days. So for that couple of hours (which would probably be around midnight), you have to draw power from some other satellite in a slightly different orbit, turn on your hydrogen-burning back-up power turbines, or simply declare a "blackout holiday". The rest of the year, the satellite is "above" or "below" the poles while transiting the far side.
Now, if they wanted to set up a "Cyber Court" which was like a normal court only with clueful judges, it would be a good thing. But apparently they want this special court just because it's too hard for the poor incompetents in the FBI and Justice Dept to have to actually show probable cause... AAAGHHH!
The one good thing about this -- after the Supreme Court tosses this out, they might take a good hard look at the FISA too...
I forgot flywheels, yes those could be effective energy storage for a household. There are some dangers with the high-energy flywheels (having one crack is like having a little accident with a lot of TNT), but in a fixed installation you just put it underground with a vertical spin axis and if it breaks the ground absorbs the fragments. Wonder if they have the frictional losses low enough to hold through one of our five-day storms? How would flywheels scale up to city-size storage?
I mentioned hydrogen for large-scale systems, from water electrolysis of course. Turning it back into electricity isn't cheap on a household scale. AFAIK, fuel cells cost something like $20K minimum, and a motor-generator big enough to let you run your house normally is probably over $1K, under 30% efficient, and requires repairs frequently. On a city-wide scale, I think the water/gravity system can be 80% efficient, better than batteries and much better than fuel cells. But that is only if you generate and use the power within a few miles. If San Diego wanted to sell excess solar power to Seattle, electrolyzing hydrogen and sending it by pipeline would be best
a new government rule that says if your energy source doesn't produce a reliable, steady amount of energy, you cannot sell it at the same rate as a reliable steady one. In a free market, you CAN'T sell something that is available now and then at the same rate as if it was assuredly available when needed. It's worth more if it's there when I want it, than if it's only there when you happen to have it. The only reason for such a gov't regulation would be that they were already regulating the market too tightly.
If we _are_ going to solar power collected on Earth, then covering rooftops is definitely the way to go. It's been a long time since I did the calculations, but the way I figured it (1) we'd need less than 50% of the total roof area in the US, and (2) blocking the sun from that much undeveloped land would be a massive ecological disaster... (There is life even in the Nevada desert.)
But it's very expensive. Solar cells cost over $1/watt the last time I looked. And 1KW of solar cells gives you far less than 1KW of delivered power most of the time -- they are rated for peak power, which is aimed directly at the sun at noon on a clear day on a mountaintop in the tropics. I did some datalogging with a small solar panel where I live this summer; in a Michigan summer, clear days give you the equivalent of 2 hrs/day at full power. Many days aren't clear, some are so cloudy that I never got enough current to measure. Overage, I think a 1KW panel around here will collect 1KW-hour per day in the summer. Winter is going to be a lot worse. I'm also testing a small windcharger -- it didn't collect enough energy to notice in September, October is shaping up a little better, and maybe it will give a decent power output when the winter storms start hitting...
California would be better, and Tucson AZ might get 5KW-hour/day. Most American homes use considerably more than that, so you need several KW of collector. Then you also have to store the energy for night-time use. In a house-sized system, that storage is batteries, so you also need a batter charger and inverter to convert from and to AC. I'm told that the overall cost of a solar power system (panels, batteries, electronics, and wiring) for one house is $20K to $60K, and that is if you cut your power usage well below average and use either a back-up generator or a connection to the power grid for prolonged storms. (In northern Michigan, we'd probably be on back-up power Oct-March, unless another $20K into windchargers would give us winter power.) Or you can pay the power company about $100/month. It only makes sense if (1) you are an eco-freak, or (2) your house is so far back in the woods you have to pay $10K or more to get a power line hooked up. But then you also have to buy the massive inverter needed to power a well. (You can't run a 4 inch submerged pump from 24 or 48VDC -- the wiring needed would be too thick. You need 240V to bring the current down.)
If you are putting solar panels on city/suburban rooftops and connecting them to the grid, then the costs are probably lower. Each place needs an inverter that will sync to AC already on the lines -- in mass production that's not much more expensive than the inverter needed for stand-alone operation. There has to be one big energy storage system, but on a large scale there are cheaper options than batteries. For instance two ponds, one on top of a hill, one at the bottom. Pump water to the top in the daytime, and let it flow back through a turbine at night. Or convert excess electricity to hydrogen, and store it or pipe it to someplace less blessed with sunlight, then burn it in gas turbines, fuel cells, or even a converted coal plant.
But notice that in any of these cases, the final stage is still as costly to build as a conventional (fossil-fuel or hydro) electric system, plus someone has to pay for all those solar collectors, and inverters. The economics isn't there until either fuel gets more expensive or solar cells get much cheaper.
So how is a solar power satellite going to beat this? It will stay at peak power 24 hours a day, so you get 6-12 times as much energy as from the same solar panel in L.A. and reduce the requirement for night-time storage and backup power, but that's not nearly enough to make up for the launch costs.
However, an SPS does not have to use expensive semiconductors for energy conversion. Use a big mirror (e.g. a balloon with one half clear, the other half aluminized) to focus the sun on a boiler. The mirror stays pointed just by pointing the power plant at the sun and then spinning it around the mirror axis. (Pointing the microwave antenna at a fixed spot on Earth could be a problem -- maybe use phased arrays?) The only heavy parts of this system are the boiler, turbine, and condenser. Big steam plants get well over 30% efficiency, and this is better than any solar cell I have heard of. You could do this on earth too, but you've got to turn that big mirror to follow the sun, brace it against winds, etc., so it's pretty costly, although at this time it would be definitely cheaper than launching a much lighter SPS into space...
What would really make it economical would be to mine the materials and build the SPS in space. And the enthusiast's real goal is to get that mfg capacity up there -- because then they could build pretty much anything needed to colonize space. And if some earth-based gov't thinks they have to pay taxes...oops, lost control of that beam for a few minutes, sorry.
At least the system held off the air strikes for 3 weeks. Maybe slightly better planning and target selection came out of that, or at least we waited long enough for it to become clear that no better ideas were forthcoming... Three weeks was a lot longer than many people considered acceptable for counting the vote last year.
Yeah, Pyrosz it's not that bad an idea, it's just that the mechanical arrangements are quite difficult. For proper heat transfer, surfaces must be flat and touching all over -- thermal grease fills in microscopic valleys, but if you don't clamp things together until only a very thin layer of grease separates the parts, you don't get good heat conduction. So a combined case and heatsink normally means the parts are bolted to the case, and to get it apart they've got to come out of the board. Then there's the problem of tolerance stack-up: nothing is ever exactly the intended size and shape, so when you put it together the pins on the parts miss the sockets...
Another issue for motherboards is that the CPU is on the same side as the cards, which isn't the side you can put close to the case. This is because bus connectors are soldered by wave solder (shooting a wave of liquid solder onto the bottom side of the board). Small capacitors and resistors can be glued onto the bottom and survive this process, but IC's might not, and you certainly don't want to do it to either a CPU or a socket...
If you don't have any bus cards or other plug-in parts taller than the CPU, then you could flip the board over and bolt it down with the CPU touching the case. They should be clamped together fairly hard, so you'd have to put holes in the board right around the CPU for bolts, or else put a brace behind it to support that area. And the case has to be unusually thick (at least near the CPU) so it's heat conductivity is enough to spread the heat out.
Insane as all this sounds, the standard cooling method is a little odd too. We use a good system for cooling a number of warm parts scattered all over (air circulation) and try to make it work to cool one extremely hot part...
The Peltier refrigerator would make the CPU taller, and allow the thermal interfaces to be not quite so perfect -- it would be a help here, but I'm dubious about it being worth the cost.
Finally, remember that other parts generate heat too. Not as much as the CPU, but it still has to be removed.
Slashdot Mirror
One American sub lost in the 60's is believed to have tried to launch a (regular) torpedo with the outer tube door closed, so even though this has got to be one of the stupidest things you can do on a submarine, it can happen... I doubt that the supercavitation torpedos are inherently more dangerous than regular ones -- except that I think they are still experimental. One other thing may have worsened the odds for the Kursk -- when you can't pay your crew regularly, they are apt to get sloppy.
I've seen 3 or 4 fans fall off, always when the computer was shipped. That's out of hundreds of computers. It's not a real big problem, but it's certainly a problem. Most mobos nowadays don't go to hobbyists or techies that build their own boxen, they go to corporate desktops or naive home users that are NOT GOING TO OPEN THE BOX BEFORE PLUGGING IT IN. So if their heatsink/fan fell off, the Athlon will melt down.
The fact that most MB manufactures doesn't use the diode might tell something about the likelihood of the cpufan falling off!! The mobo in THG's article did use the diode -- just not with a circuit that would react instantly. With a dead fan it would run for several minutes, then shut down; with no heatsink, the CPU melted before the mobo reacted. Apparently all unmodified mobos are the same, which leads me to think that AMD recommended the too-slow circuitry.
I'm still wondering why THG didn't get AMD's response before posting their original article, but it does look like they've persuaded AMD to recommend some changes in future mobos...
Yes, it sounds like Siemens blamed the AMD chip when it was really a deficiency in their motherboard design. It ALSO sounds like AMD recommended motherboard circuits to support their chip that are not adequate if the heatsink falls off, therefore most or all mobos for the Athlon presently in production can smoke the CPU in this case, and that AMD's video used a modified mobo.
In short, neither Siemens nor AMD was entirely truthful, and if you buy a complete system, you need to open the case and make sure everything is where it belongs before turning it on. That's good advice whether or you've got AMD or Intel. We techies already know that, but Dell, etc., are shipping a lot of computers to people who DON'T know it, and might not be able to check for things like this even if the computer mfgs sent along instructions...
I don't know why THG didn't get AMD's input before publishing the smoking Athlon pictures. Maybe they didn't ask, maybe AMD didn't answer. But now AMD knows there is a slight problem, they should put the fast-acting shutdown circuit in their recommendations for the mobo mfgs, and quality mobos for Athlon will soon have it. And Dell, etc., will evaluate whether it costs more to require that on Athlon mobos, rather than occasionally having to replace a burned-up CPU and possibly the whole mobo.
a) EE Times is naturally going to concentrate on the EE jobs, but it's not only EE's. Go to uspto.gov and look at the job listings.
/. once brought up a patent which claimed to cover the very idea of PDA's -- it's not going to hold up, because Arthur C. Clarke described a PDA in Imperial Earth, written in the 70's.
b) I think engineers would generally make better patent examiners (in their particular field) than physicists because engineers have a better grasp of the existing practice. When it comes to the worst patents the hard part evidently isn't understanding the application or searching the patent database, it is recognizing when the patent claims to cover an idea that is never-patented but already in use (and therefore public domain). E.g., to check out an application concerning communications between computers, familiarity with the rarefied mathematics of communications theory is less useful than the ability to recognize an attempt to patent the IP protocol...
(Maybe I'm biased. I switched from physics to EE. But I remember one physics prof who might not have realized that a patent application for a "circular transportation facilitation device" had prior art called the "wheel". No decent engineer would ever make that mistake.)
Familiarity with "hard" science fiction might help too.
What I do worry about though, is:
1. Will they get good engineers with working knowledge in their fields, or just people with EE degrees? Starting pay of $50K isn't much for a good engineer, especially if the work is non-creative, involves excessive paperwork, and is located in the only city run by a congressional committee...
2. How will their engineers stay current in their fields when they aren't actually designing anything?
3. The bureaucratic motivations are still all wrong. Reject an application and the applicant can, and often will, sue the bureau. Rubberstamp the application, and there may be worse lawsuits, but the bureau isn't involved. Now, if the bureau got to pay the legal fees & court costs whenever a patent was successfully challenged...
Are we giving "same as the last design, just add this component" electrical device patents? I think they are, and what's wrong with that? Most patents simply protect one particular device from near-exact copying, and don't try to cover all ways of doing the same thing. Such patents protect step-by-step improvements on old designs, without hindering anyone who designs a different competitive product. Narrow patents like these meet the reason for patents expressed in the Constitution: "To promote the progress of science and the useful arts".
Patents become counter-productive (block progress) when they are too broad and claim to cover all ways of accomplishing something. This can only occur if either someone is remarkably innovative, or if they are claiming more than they actually originated. I suspect it is almost always the latter case, because when "great inventions" are placed in their historical context, nearly always the inventor was either pulling together existing ideas (Edison, the Wright Brothers), or was unable to make his idea work well enough in practice and had to leave it to a later generation with better materials (Da Vinci, Babbage).
The problem is, when it comes to software and high-tech "business methods", the patent office is often so confused as to grant ridiculously overbroad patents. Sometimes it allows very wide-ranging claims on an idea that is at best a slight improvement upon existing practice -- because the existing practice was never thought patentable, so searching their database doesn't turn it up, and the examiners are insufficiently familiar with the ideas that are in the public domain. An extreme example of this is the recently granted Aussie patent for the wheel; since patent databases don't stretch back to 5,000BC you aren't going to find the prior art there, but you think someone _should_ have noticed. Or in many cases, there is some small idea that may actually be innovative, but the patent claims far more than that -- analogous to claiming to have invented the wheel when actually you just invented a better kind of cotter pin to hold the wheel on the axle... You'll get clobbered if you go to trial against anyone who isn't using your new cotter pin, but often they'll pay modest royalties instead of bearing the expense of going to trial.
The most damaging patents of all are extremely broad ones filed by people who never created a successful product, but who simply guessed which way technology was headed, filed a vague patent application, and kept amending it for many years. About 1960 someone filed a patent on a block of silicon containing six transistors, interconnected by soldered wires; a long series of amendments and legal arguments kept this from being either issued or rejected until the 1980's, when it finally issued as a patent on integrated circuits! An IC of the period (say an 80286 CPU) resembled the original device like an aircraft carrier resembles a dugout canoe. The "inventor" hadn't done the 20-some years of work that advanced IC manufacture from six disconnected transistors to about a million connected transistors, but he wanted to cash in on it.
The level of insight offered in this speech is outstanding and thoroughly depressing when compared to the level of debate offered by parliament or congress.
When you collect the dozen or so best speeches of a century and ignore the rest, it's bound to look pretty good. I'm insufficiently familiar with McCauley's contemporaries in Parliament to comment on them, except to note that you can see the effects of their less inspired decisions abroad in the Crimean War and their colonies, and at home in anything by Carles Dickens.
In an area I am more familiar with, 19th century America had a few great men in Congress, and more than a few blackguards. One Senator ran short on words, so he beat an elderly Senator unconscious on the Senate floor. A Senator was publicly caught buying votes in his home state; the Senate promptly formed a committee to investigate, not the (ex)Senator, but the man who exposed him. Overall, American politics was cleaner in the 20th century than the 19th. Sometimes their language was more high-faluting than any modern American would use, but that doesn't make it more expressive, or less likely to have cloaked evil intentions or plain pigheadedness.
I don't know how the British Parliament compared with that. Certainly they weren't crude enough to commit physical assaults, nor do I think any MP would have been as badly educated as Tennessee congressman David Crockett, but probably many of them weren't as eloquent as Davy Crockett...
I'm not sure if that was a serious question, but... Metal fatigue is a big concern at NASA because it can cause spacecraft to come apart in flight. A piece of metal under stress can develop tiny cracks. (If you designed the metal as thin as possible to save wait, substitute "will" for "can".) When the stress is relieved and then applied again, the cracks may grow. After a certain number of cycles, the cracks get too big and the metal breaks.
The classic case of this was a very early jet airliner called something like Comet. Because early jet engines were pretty inefficient, they made the skin exceptionally thin so as to have more weight-carrying capacity available for fuel. When it climbed to altitude, the pressurized cabin would slightly bulge the skin outwards around the windows. When it landed, the skin would pop back. After a few months, planes started coming apart in mid-air. Microscopic examination (of not yet crashed planes) found patterns of cracks in the skin near the windows. They had to scrap the entire fleet...
The same thing could quite easily happen with the space shuttle -- not just from cabin pressure, but also from high-stress launches and landings. Or the skin on a Mars probe may expand and contract thousands of times due to sun heating as it rotates in space. Making the metal thicker will prevent this, but every ounce of structural metal takes away an ounce of payload. So NASA has to design right to the edge of initiating metal fatigue for repeated-use items (the space shuttle), and for some probes it may accept that metal fatigue will happen, but the cracks will grow so slowly that the mission is finished before it fails. This requires very good software for simulating crack growth and analyzing metal fatigue.
I wonder how many of those "error" messages really indicate errors? When I am programming, I will put in lots of messages to make debugging easier later on. I will disable them on the final compile, but there have been times when this got forgotten in the rush to release on the deadline. I wonder how often that happens with OSS -- especially since OSS releases are usually not the end of the project.
Also, messages that were intended simply to show the progress of the program or confirm it went down the correct path often inadvertently sound threatening: "Cannot find file xxxx.xxx", when what you really meant was "No initialization file xxxx.xxx found, using defaults."
Of course, as the author said, the problem isn't that OSS is worse than commercial software, but that it should be better. Is there anything in OSS as bad as the error message I sometimes got from Win95, "Cannot find file", without the file name and path? Not to mention how Windows allows an application to silently malloc some memory, forget to free it, and repeat until it crashes a different application or the OS itself...
Certainly I am not the first to notice that Industrial-Age schools and Industrial-Age prisons are very similar to factories in terms of set times and hierarchy. You forgot industrial-age armies... I agree about the resemblance. But prisoners and (most) schoolchildren don't have to get themselves up and ready to work on time.
From my own experience, basic stuff like showing up to work on time can be a very big problem with people who were raised on welfare, without any example in the family of an adult that got up every morning and went to work. I do know guys that seemed to really want to hold down a job, but couldn't come in on time three days in a row. Skilled white-collar workers often have some latitude in their schedule, because the work can get done whenever they show up, but the jobs an unskilled 18-year old can get are utterly unforgiving. If the restaurant is going to start serving breakfast at 6:30, someone's got to be there starting to set up at 5:30 -- and if they can't count on you to be there every morning, they will find a way to get along without you entirely...
ubernostrum, you are biasing the test. You started one example with "Double-click an icon on the desktop to open a folder" and the other with "type tail myfile". You aren't comparing equivalent actions. A fair comparison is something like:
... If I knew the CLI command to do it all at once it would be much, much faster -- even including listing and reading the manual page to get the switches right. But I wouldn't know where to start in Unix. I still know all the DOS commands pretty well, but I can't think of a set of DOS commands that would do that. And printing out the directory, then going down it and typing a rename command for each file is going to take a very long time.
Starting conditions: Target folder opened, target file located/spelling known, skilled at use of the OS and the mouse/keyboard.
GUI (Windows):
1. Doubleclick the myfile icon.
2. Wait for it to open
3. Click and drag the scroll button to the bottom.
CLI (Unix?)
1. Type "tail myfile"
2. Wait for it to open
I get 4 seconds in the GUI, or 4 seconds just to type two words on the command line. However, if I miss the right spot with the mouse, recovering from that is going to at least double the time. This is much more common (for me at least) than hitting the wrong key. The keys are big, but the scroll bar is narrow... So I'm not sure which would actually win on the average -- except that DOS has no "tail" command, so on my system I would still need to hit Ctrl-End after opening the file in Edit.
How about this test: "Start from c:, open a folder named something like my documents (GO), now open a file in it called something like my file." The Windows user spends 5 or 10 seconds locating the "My Documents" icon. The CLI user either already knows how it is spelled and types it in about 5 seconds, or spends about 30 seconds listing folders and scanning the list to find out. Etc. So the CLI will usually be faster if you've memorized the exact spellings, otherwise it's slower at simple jobs.
For the simple jobs, it isn't the interface, it's how familiar you are with the task. The GUI is certainly faster when you don't know what you are doing.
Or consider a hard job: rename all 50 files in a folder according to certain rules. In Windows, you do them one at a time: right click on each file, select "rename", switch hands to the keyboard,
I printed that out and passed it on -- I can follow it's progress through the cubes by the gasps as people reach the last paragraph...
What is really strange here is that HP grew big by building very good electronic instruments, selling them high, and supporting them very, very well. An HP scope or meter might cost twice as much as the competition, but it would quite likely go 20 years with no service other than calibration, and even after 20 years you could still get it fixed if you were still using it. But I have heard many such stories about terrible tech support on computers, printers, etc. (On-site repairs here are pretty good, paid for by a corporate budget, but for printer drivers and software setup issues I've learned that the downloads on the web site are all you are going to get...) So WTF happened to their computer & printer divisions?
Of course, now the computer & printer division _is_ HP. The instrumentation company was split off and is called "Agilent." (Yucch!) Agilent tech support is decent as far as listening to what you are saying and trying to solve the actual problem. But sometimes the problem is just that the HP/Agilent software is so crappy it seriously limits what you can do with their excellent hardware...
you don't go to college to learn, you go there to learn how to learn. Very true. But you've still got to pass those damned tests...
In Holling's case, tar and feathers would do nicely.
"Bazaar" is a rather distorted view of how open-source projects run. If you try to start one that way, you will get an endless discussion of specs and no code. And if you tried to run an on-going project that way, you'd either get the same endless discussion, or infinite forking and no progess.
Open source projects start with one or two people defining the core and writing some code, then recruiting others to add to the code base. They progress if the founder is good enough at dealing with people to be able to hand out assignments to volunteers and to decide which improvements do or don't get into the code base. That is, they are run as mild dictatorships: Linus Torvalds welcomes suggestions but ultimately decides what becomes part of Linux. Anyone could go make his own changes no matter what Linus thinks, but given the respect Linus has earned, if you fork the code you will probably be working alone. Linux coders voluntarily submit to Linus's decisions because it's better to be working with Linus and 10,000 others than to be going it alone... It's a dictatorship that is benign enough that people volunteer to come under it, and dissenters can safely be ignored rather than shot.
What the "eyes and brains" can sometimes show is how the age of one site compares to another one. E.g., on one site you find iron tools in the upper layers, bronze lower down, and stone tools in the lowest level, the sequence is pretty clear. The actual ages aren't, but the order is, and depth of the layers can give some hints as to the length of each era. They when you find the same style of bronze tools somewhere else with just a few iron tools you know they were the same people, and that it was contemporary with a particular era on your main site.
But this does not give you real dates, so you cannot compare the ages of sites left by different, non-interacting cultures. The technological level alone does not limit the possibilities much -- stone axes are probably still in use today in New Guinea.
The only way you get real dates is when they start leaving written records. E.g., Egyptian civilizations are datable because they etched-in-stone a nearly continuous record of the years from their first cities up through Roman times, and the Roman Catholic church has kept a reasonably accurate tally of the years since. Mesopotamian civilizations would keep records for a while, but then collapse leaving gaps of unknown length, but because they interacted with Egypt you can often associate the date. But if you want to figure out when some European hunter lived, without radio-dating your only chance is to find some definite link (same kind of clothes, pottery, or equipment) as was used by people that interacted with one of the record-keeping civilizations. (The illiterate barbarian tribes really are important to history even though they don't write any -- they keep overruning civilizations grown too soft or too complacent.)
What I want to know is: why solar cells? It seems to me it would be much easier to make a space-based heat engine.
You forgot one more advantage: steam power plants routinely run at over 30% efficiency, higher than solar cells. All you need is a big mirror to focus light on the boiler, a big radiator in the shadow of the mirror, and the steam plant in between. And you can make the mirror extremely thin, there is no wind and very little gravitational/accelerational forces, and you just keep the mirror stationary (or stabilized by spinning about the optical axis) to keep it pointed at the sun. On Earth, a solar steam plant requires bracing the mirror against wind and gravity, and then mounting the whole shebang on a giant turntable to follow the sun. Or maybe you have a giant array of small steerable mirrors, but it's still very expensive, possibly more than an immoveable array of solar panels that would collect the same average power. In space: it's lighter than the solar panels (real thin mirrors...) so if you are lifting it from earth it's much, much cheaper. If you are building from space-mined materials, you've got a lot fewer factories to lift into orbit to build the steam plant.
Shaunuk, you mixed so much good and bad physics mixed together it's hard to tell where to begin. Individual photons do get more energetic with shorter wavelength (blue-er light), and if you could build solar cells tailored to each wavelength, the higher the energy the higher the voltage per cell.
But the number of photons in sunlight at each wavelength drops with increasing wavelength through most of the visible spectrum. So the red light has the most available power. Don't even consider the cosmic rays and other radiation in space, for power they compare to sunlight like a gnat to a supertanker.
But the second problem is the way solar cells work. They are semiconductor diodes, arranged so that an incoming photon of sufficient energy will knock an electron from the positive (anode) to the negative (cathode) end. All the electrons put together form a current flowing from the cathode. (Of course, they better go through something and come back to the anode for you to get any current or power...). Each color or wavelength can push an electron across a different voltage, but the photocell only works at one voltage. So photons that are too red are wasted completely because they lack the energy to make the electron go, and ones that are too blue have excess energy beyond what the electron absorbs. The result is that you have to design your photocell for some sort of compromise voltage, probably about 1.5V corresponding to a particular wavelenght of red. (Of course, they put a bunch of cells in series to make 12V or higher outputs from the panels). Even if everything else worked at 100% efficiency the inherent losses from energy mismatch will never allow a cell like this to get better than maybe 30% efficient.
UV light would just waste more of it's energy. One UV photon would at best impart to one electron the same energy as a red photon does, with the rest of the energy wasted. It's more likely that big a mismatch would just keep the photon from being captured at all--it would reflect or be absorbed as heat only.
I can think of two ways to work around this. One is to use a prism or diffraction grating to split the light out into different wavelengths, illuminating a set of photocells with different working voltages. This would certainly raise the efficiency with respect to the light entering the device, but it would be expensive, and I'm not sure you could make the separation work without restricting the light entry to a slit -- and blocking light outside the slit certainly won't give you much efficiency overall. Or maybe you could make a stack of solar cells, each one transparent to the wavelengths preferred by the cells below. Assuming this transparency is possible, there's a problem with cost -- as long as the semiconductor is the biggest cost of the panels, it will be far cheaper to just cover more square feet with inefficient cells than to stack up cells for higher efficiency.
The space junk is spread out pretty thinly. You'll have less trouble with it than with keeping earthside solar panels clean, especially in cities and deserts. (Of course, in northern Michigan where I live, the solar panel will be perfectly clean, it just has a foot of snow on top. 8-)
Once it a while, something will hit at 10,000 kph and blast a hole right through a solar panel or a mirror. This happens because to make this worthwhile, you'd have miles of solar collectors. But it won't affect your generating capacity noticeably, because the rest of the panels or the mirror will still be working (mirrors would be very thin metal, or aluminized mylar, not glass), and it's miles across. 20 years of this isn't going to reduce your capacity as much as a few days dust accumulation in Nevada.
One thing I don't know about though: what is the effect of hard radiation (cosmic rays, solar wind, etc.) on solar panels? I hear that solar panels are expected to last 20 years on Earth; is the much greater exposure to radiation hard enough to re-arrange the crystal matrix a bit going to make space-borne panels shorter lived, or will the lack of rain, wind, and crud going to make them longer lived?
However, I think that an SPS built now would use the much cheaper technology of using a mirror to focus sunlight on a boiler, which runs a steam turbine. Cosmic rays won't hurt a steam plant, although the crew is going to need good medical insurance...
Actually, the best orbit for beaming to one spot on the ground is not just geosynchronous but geostationary: above the equator & 24 hour period. Because the Earth's axis is tilted, the orbit is at 40,000 miles radius, and Earth is only 8,000 miles in diameter, the satellite rarely passes into the Earth's shadow. At the equinoxes, it will be shadowed when it goes around the far side of the Earth, but that's only a couple of hours a day or less for a few days. So for that couple of hours (which would probably be around midnight), you have to draw power from some other satellite in a slightly different orbit, turn on your hydrogen-burning back-up power turbines, or simply declare a "blackout holiday". The rest of the year, the satellite is "above" or "below" the poles while transiting the far side.
Now, if they wanted to set up a "Cyber Court" which was like a normal court only with clueful judges, it would be a good thing. But apparently they want this special court just because it's too hard for the poor incompetents in the FBI and Justice Dept to have to actually show probable cause... AAAGHHH!
The one good thing about this -- after the Supreme Court tosses this out, they might take a good hard look at the FISA too...
I forgot flywheels, yes those could be effective energy storage for a household. There are some dangers with the high-energy flywheels (having one crack is like having a little accident with a lot of TNT), but in a fixed installation you just put it underground with a vertical spin axis and if it breaks the ground absorbs the fragments. Wonder if they have the frictional losses low enough to hold through one of our five-day storms? How would flywheels scale up to city-size storage?
I mentioned hydrogen for large-scale systems, from water electrolysis of course. Turning it back into electricity isn't cheap on a household scale. AFAIK, fuel cells cost something like $20K minimum, and a motor-generator big enough to let you run your house normally is probably over $1K, under 30% efficient, and requires repairs frequently. On a city-wide scale, I think the water/gravity system can be 80% efficient, better than batteries and much better than fuel cells. But that is only if you generate and use the power within a few miles. If San Diego wanted to sell excess solar power to Seattle, electrolyzing hydrogen and sending it by pipeline would be best
a new government rule that says if your energy source doesn't produce a reliable, steady amount of energy, you cannot sell it at the same rate as a reliable steady one. In a free market, you CAN'T sell something that is available now and then at the same rate as if it was assuredly available when needed. It's worth more if it's there when I want it, than if it's only there when you happen to have it. The only reason for such a gov't regulation would be that they were already regulating the market too tightly.
If we _are_ going to solar power collected on Earth, then covering rooftops is definitely the way to go. It's been a long time since I did the calculations, but the way I figured it (1) we'd need less than 50% of the total roof area in the US, and (2) blocking the sun from that much undeveloped land would be a massive ecological disaster... (There is life even in the Nevada desert.)
But it's very expensive. Solar cells cost over $1/watt the last time I looked. And 1KW of solar cells gives you far less than 1KW of delivered power most of the time -- they are rated for peak power, which is aimed directly at the sun at noon on a clear day on a mountaintop in the tropics. I did some datalogging with a small solar panel where I live this summer; in a Michigan summer, clear days give you the equivalent of 2 hrs/day at full power. Many days aren't clear, some are so cloudy that I never got enough current to measure. Overage, I think a 1KW panel around here will collect 1KW-hour per day in the summer. Winter is going to be a lot worse. I'm also testing a small windcharger -- it didn't collect enough energy to notice in September, October is shaping up a little better, and maybe it will give a decent power output when the winter storms start hitting...
California would be better, and Tucson AZ might get 5KW-hour/day. Most American homes use considerably more than that, so you need several KW of collector. Then you also have to store the energy for night-time use. In a house-sized system, that storage is batteries, so you also need a batter charger and inverter to convert from and to AC. I'm told that the overall cost of a solar power system (panels, batteries, electronics, and wiring) for one house is $20K to $60K, and that is if you cut your power usage well below average and use either a back-up generator or a connection to the power grid for prolonged storms. (In northern Michigan, we'd probably be on back-up power Oct-March, unless another $20K into windchargers would give us winter power.) Or you can pay the power company about $100/month. It only makes sense if (1) you are an eco-freak, or (2) your house is so far back in the woods you have to pay $10K or more to get a power line hooked up. But then you also have to buy the massive inverter needed to power a well. (You can't run a 4 inch submerged pump from 24 or 48VDC -- the wiring needed would be too thick. You need 240V to bring the current down.)
If you are putting solar panels on city/suburban rooftops and connecting them to the grid, then the costs are probably lower. Each place needs an inverter that will sync to AC already on the lines -- in mass production that's not much more expensive than the inverter needed for stand-alone operation. There has to be one big energy storage system, but on a large scale there are cheaper options than batteries. For instance two ponds, one on top of a hill, one at the bottom. Pump water to the top in the daytime, and let it flow back through a turbine at night. Or convert excess electricity to hydrogen, and store it or pipe it to someplace less blessed with sunlight, then burn it in gas turbines, fuel cells, or even a converted coal plant.
But notice that in any of these cases, the final stage is still as costly to build as a conventional (fossil-fuel or hydro) electric system, plus someone has to pay for all those solar collectors, and inverters. The economics isn't there until either fuel gets more expensive or solar cells get much cheaper.
So how is a solar power satellite going to beat this? It will stay at peak power 24 hours a day, so you get 6-12 times as much energy as from the same solar panel in L.A. and reduce the requirement for night-time storage and backup power, but that's not nearly enough to make up for the launch costs.
However, an SPS does not have to use expensive semiconductors for energy conversion. Use a big mirror (e.g. a balloon with one half clear, the other half aluminized) to focus the sun on a boiler. The mirror stays pointed just by pointing the power plant at the sun and then spinning it around the mirror axis. (Pointing the microwave antenna at a fixed spot on Earth could be a problem -- maybe use phased arrays?) The only heavy parts of this system are the boiler, turbine, and condenser. Big steam plants get well over 30% efficiency, and this is better than any solar cell I have heard of. You could do this on earth too, but you've got to turn that big mirror to follow the sun, brace it against winds, etc., so it's pretty costly, although at this time it would be definitely cheaper than launching a much lighter SPS into space...
What would really make it economical would be to mine the materials and build the SPS in space. And the enthusiast's real goal is to get that mfg capacity up there -- because then they could build pretty much anything needed to colonize space. And if some earth-based gov't thinks they have to pay taxes...oops, lost control of that beam for a few minutes, sorry.
At least the system held off the air strikes for 3 weeks. Maybe slightly better planning and target selection came out of that, or at least we waited long enough for it to become clear that no better ideas were forthcoming... Three weeks was a lot longer than many people considered acceptable for counting the vote last year.
Yeah, Pyrosz it's not that bad an idea, it's just that the mechanical arrangements are quite difficult. For proper heat transfer, surfaces must be flat and touching all over -- thermal grease fills in microscopic valleys, but if you don't clamp things together until only a very thin layer of grease separates the parts, you don't get good heat conduction. So a combined case and heatsink normally means the parts are bolted to the case, and to get it apart they've got to come out of the board. Then there's the problem of tolerance stack-up: nothing is ever exactly the intended size and shape, so when you put it together the pins on the parts miss the sockets...
Another issue for motherboards is that the CPU is on the same side as the cards, which isn't the side you can put close to the case. This is because bus connectors are soldered by wave solder (shooting a wave of liquid solder onto the bottom side of the board). Small capacitors and resistors can be glued onto the bottom and survive this process, but IC's might not, and you certainly don't want to do it to either a CPU or a socket...
If you don't have any bus cards or other plug-in parts taller than the CPU, then you could flip the board over and bolt it down with the CPU touching the case. They should be clamped together fairly hard, so you'd have to put holes in the board right around the CPU for bolts, or else put a brace behind it to support that area. And the case has to be unusually thick (at least near the CPU) so it's heat conductivity is enough to spread the heat out.
Insane as all this sounds, the standard cooling method is a little odd too. We use a good system for cooling a number of warm parts scattered all over (air circulation) and try to make it work to cool one extremely hot part...
The Peltier refrigerator would make the CPU taller, and allow the thermal interfaces to be not quite so perfect -- it would be a help here, but I'm dubious about it being worth the cost.
Finally, remember that other parts generate heat too. Not as much as the CPU, but it still has to be removed.