Dan's Data letters #75
Publication date: 24 November 2003.Last modified 03-Dec-2011.
CPUs are forever
One of my friends was telling me about diamonds that can be manufactured for about $US10 each which are being used in CPUs. They said there could be CPU speeds of about 50GHz at the same cost as today's CPUs. As I had no idea that diamonds could be manufactured, much less that cheaply, I was wondering - are these diamonds as in "a girl's best friend", or are these industrial diamonds, as are used on the tips of drills?
Tammy
Answer:
Diamond isn't yet being used in processors, and we're not talking
about solid lumps of pure diamond, either.
Pure diamond is an excellent thermal conductor, beating copper at room temperature by a factor of about five, and bulk diamond creation could make it useful as a thermal transfer medium in chip packages. The big news, though, is the creation of diamond transistors. Heat is a major limiting factor for modern CPUs, and diamond transistors can get rid of it more easily.
At the moment, though, we're still several years away from practical diamond semiconductor products.
Bulk manufacture of gem-quality diamonds isn't quite ready for prime time yet, but industrial diamonds have been made in factories for ages now. It's not surprising that you didn't know that any diamonds could be fabricated; it's also not common knowledge that natural diamonds are not actually particularly rare. Rubies are something like 50 times harder to find than diamonds, but their price certainly doesn't reflect it (mind you, you can buy huge synthetic rubies dirt cheap). All this is pretty much exactly the way De Beers like it.
Here's an article about this. Here's another one about future processors and limiting factors. And while we're at it, here's a piece about De Beers, which did not greatly please them; here's Wired's more recent piece.
Movement Avoidance Device #3882
Have you ever considered reviewing the keyring remote controls that can be matched with a receiver in your computer to turn it on and off? I saw these around a couple of years ago, but have no idea if they even exist any more, where to find them or if any are any good.
Know of any that exist and are available in Australia?
Nick
Answer:
I reviewed one a while ago. Aussies can still
buy it, from
Aus PC Market.
Here, doccy-doccy-doccy...
Where the hell can I find a user manual for my ASUS P2-99 motherboard? I just want to know where I can connect the extra USB cables that came with my new case... I'm already using the PCI slots.
Ruben
Answer:
You can dig up PDF-format manuals for pretty much every Asus mobo from
their download
page. The P2-99 page is
here.
I can save you the trouble, though - the P2-99 does not have any USB headers. This is normal for mobos of this vintage (I reviewed it back in 1999). It was a while later that mobos started coming with the capability to run more than two root ports.
Zapper surgery
I have an LG air conditioner. I went to use it last night and would you believe it, when I switched it on nothing appeared on the LCD of the remote control.
The controller was still sending signals OK; I just couldn't see my settings. I tried changing batteries, but it had no effect. The controller hasn't been abused (yet).
Someone at work suggested that it's perhaps a dry solder joint, or something. OK, but the next trick is how to get the controller apart without destroying it. I can't locate ANY screws (under labels or anywhere else), and there doesn't appear to be any obvious plastic clips. I'm suspecting it's glued somehow and pulling it apart will kill it. I wondered if you had any generic advice on pulling things apart that contain "no user serviceable parts"?
Mark
Answer:
Clips can be hard to spot and harder to release without breaking. Even if
you manage to break every clip that holds a remote control together,
though (just one or two broken clips are seldom a big deal), you can always
use a few small dabs of hot-melt glue, or some other weak adhesive, to hold
it together afterwards. Hot-melt is weak enough in this application that
you'll be able to get back into the case again if necessary.
If the thing actually is glued together, head off to a hobby store or unusually well equipped electronics or hardware store and get yourself a razor saw. Most razor saws are basically miniature tenon saws; they have very thin, fine-toothed blades, stiffened with a thicker metal brace along the top of the blade, and they'll fit into the groove between the two halves of most remotes. This allows you to slice the casing apart without losing a lot of material or, as long as you're careful, chopping into the circuit board.
There are fancy ultra-hyper-thin razor saws for the connoisseur out there, but you're not making a 1/500th scale model of the Mary Rose here, so any old piece of knock-off crap should do. You can get cheap hobby knife sets that come with a razor saw that slots into the standard handle; that'll be fine, and also give you a collection of other blades.
But, our signal!
Is traditional electric power "contaminated" by wireless transmissions in Australia? This article says it is.
I've never heard of such a thing here in the Midwest United States. How big a problem is it?
I live in Chicago, a city of approximately 3-5 million in the city and roughly 9-10 million in the metro area. Would it be reasonable to assume we have a great deal of "contamination" in our power lines?
Matthew
Answer:
No, radio frequency and microwave interference is not a cause of "dirty"
power - at least, not as far as I know.
It's a bit tricky to winkle actual definite points out of Peter Familari's article (written for the Melbourne Herald Sun), since he (or his subeditor...) weasels around by saying that there's "a perception that Australia's mains electricity supply is contaminated", rather than coming right out and saying that it is. There's also a "perception" that water swirls down the drain in different directions in the northern and southern hemispheres, and if you ask the right people you can find a bunch of other "perceptions".
And what do you know - "in other parts of the state [Victoria], the power supply is so bad that computers are known to crash often". Fortunately, I live in New South Wales, where computers never crash.
Actually, most computer PSUs are quite tolerant of lousy power of all kinds.
Familari then quotes an engineer as saying that there are "known problems with powerlines", without actually explicitly saying that the engineer's talking about what appeared in the preceding paragraphs. The engineer could mean the known problem of increasing power line capacity without spending fortunes on end-to-end upgrades, for instance (see here). Or he could be talking about the power "contamination" that actually happens, when low power factor equipment like many PC PSUs (they tolerate dirty power, but they cause it too) pollutes the mains waveform with its reactance, or when heavy power users (usually commercial) introduce spikes and dips and surges and sags when they switch their manufacturing plants, welding gear, aluminium smelters and what have you.
Radio frequency "contamination" of the mains is unimportant to all gear with a vaguely well designed power supply. Heck, it's harmless to most gimcrack unapproved gear as well. Apart from the fact that there's bugger-all energy to be had from RF contamination compared with the amount of regular low frequency current being drawn by appliances (close-range illumination from a military search radar won't be enough to create power "surges and troughs"; "radio and microwaves and mobile-phone services" don't have a hope), RF also doesn't pass well through distribution transformers. Those transformers are highly resonant around the 50/60Hz that the mains supply uses, amazingly enough; RF is a long way from that resonance. This means that even zillions of miles of high tension distribution cables won't deliver nearly as much RF power to your wall sockets as you might expect. And even if RF does make it to your wall socket, it then gets eaten by the smoothing caps inside your appliances' PSUs.
Familari may think that RF or other contamination affects the performance of audio gear; this is a popular contention among audiophiles, some of whom then spend hilarious amounts of money on special power cables that're meant to undo in three feet whatever evil has entered the cable in the many kilometres of wire that lead back to the generating utility. Other golden-ears run their hi-fi gear from battery-powered inverters, not necessarily even charging the batteries while they're listening. The concept of the blinded test is, generally, alien to them, despite the fact that this is one of the easiest things to do such a test with; you don't need A/B switchers or level matching or any such stuff to see whether a cheap jug-plug lead to an unfiltered wall socket sounds as good as a bank of truck batteries and a sine wave inverter - all you need is someone to switch it without you seeing. Double blind (where neither the tester nor the testee can tell which thing's being tested while the test's in progress) is better, but single blind would be a good start.
If the mains supply really does suck where you live - which it may; brownouts, spikes and surges are the major problems in the happy parts of the world where the mains doesn't just drop dead entirely for many hours each day - then the best solution is often a proper dual conversion Uninterruptible Power Supply. Dual conversion UPSes are much more expensive than the cheap "Standby Power Supplies" that usually get sold as a "UPS", but they run their inverter all the time, so no mains oddities can make it to their output and there's zero "cut-over time" when the power fails.
"Line interactive" UPSes are also good enough for most purposes. They run their inverter all the time too, but it only supplies a small portion of their output when the power hasn't failed.
Even a cheap SPS may contain good filtering components, though; if your power keeps sagging and resetting your PC, then a basic SPS with an adequate volt-amp rating is all you need.
If you don't need to be able to ride out power interruptions of more than a few seconds, then a line conditioner is a better idea than any kind of UPS/SPS. It'll give you top-class power filtering and higher capacity than a more expensive dual conversion UPS, but it has no battery and inverter. Line conditioners are a good choice for hi-fi and home theatre gear.
(I ramble on about UPSes and power filtering here and here.)
Almost all "surge filter" power boards, by the way, are close to worthless for dealing with any real power problems. But if you carve one out of mahogany and give it a one foot long cable made from Kevlar-wrapped carbon-conductor one-gauge wire terminating in a plug with .999 fine silver pins that need to be re-straightened at the factory if you ever unplug it, then you'll probably be able to name your price in the audiophile market!
The plenum's connected to my wrist-watch... uh oh...
Howdy Dan, currently there's this discussion [registration required] going on about how you can increase throttle response, horsepower, and possibly even MPG by simply grounding a bunch of things together on your car. This sounds like a giant goblet of snake oil to me, but I figured with your mastery of all things electric you might be able to set the record straight.
There's a guide on how to do this "Ground Net" here.
Is this actually going to do anything for my car?
Jeff
Answer:
Surprisingly enough - maybe.
It is, indeed, possible for multiple sensors in a car that're meant to share the same ground to actually be separated from ground by varying amounts of resistance. In AC systems this can get quite complicated, but car power is (for now, at least...) a simple 12VDC setup where the negative terminal is the ground (and isn't really grounded, seeing as the car's separated from the road by its insulative tyres - this is why cars can accumulate so much static electricity as you drive around on a dry day). In DC circuits, there's not going to be any weird dynamic reactance stuff going on; everything's just simple volts, amps and ohms, and you can do the calculations on the back of one envelope.
If a sensor doesn't see as many volts between positive and ground as it ought to, its readings may be off. And sensors that share supply wires can (in theory at least) float each others' ground up and down depending on how much current they're passing. Solidly link all of the ground points together with nice straightforward cable, rather than relying on possibly corroded, possibly painted, possibly heavily lubricated, possibly loosely connected structural members and mechanical components, and you remove such awkward variables.
Whether there's actually any need to do this, though, can be quite easily determined. Take your faithful multimeter and see if, while the car's running, there's a potential difference between the locations the Ground Net instructions tell you to hook together. If the meter settles to something very close to 0.000 volts, as it will if you touch the probes to two ends of a good conductor that's not passing much current, then there's no need for extra grounding.
If you want to fiddle some more, then try the resistance range, with or without the car running; if the resistance between the points-to-be-grounded is trivially higher than the fraction of an ohm you read when you just touch the probes together (multimeter leads are not superconductors), then once again, all signs point to the ground-net just being decorative.
Someone will notice a difference from any voodoo car modification, computer tweak or medical treatment, so the fact that there are people in the thread you linked to reporting noticeable results from the Ground Net does not surprise me. Note that there's also someone reporting that his lights now don't dim when he starts his car, despite the fact that I don't think this mod has anything to do with the electron path to or from the lights or the starter motor.
Only 30 sleeps...
I want to give out key ring LED lights for Christmas (white LEDs preferably). What would be your opinion on the best key ring LED light?
Rudolf
Answer:
Probably the Photon Micro-Light 3, though you'll suffer less bank balance
damage if you give people Photon Is or IIs.
I talk about all three Photons here.
If you use this link to go to the Photon site (their direct prices are pretty good, since they include US shipping - I don't know where you're located, though) and buy stuff, then I'll get a cut. It won't cost you any more, so you might as well.
And yes, I'd still recommend Photon lights even if I didn't get a cut; there are lots of key-ring light options, including countless Photon knock-offs (so many that the Photon people themselves have gotten in on the act), but the Photon lights hold the crown for reliability, which is what you want in a light that's going to be bouncing around in a pocket all day.
Teeny tanks v2.0
I'm probably one of the first in Holland to get Combat Digi-Qs with rotatable turrets. Thanks to all your info I got myself into buying a set. Man, this is really fun. I work at a help desk, so any thing with an EPROM is interesting, but this beats it all. As long as the chief isn't in the house, lots of desktop battles occur here.
The only thing we can't figure out is how to change the ID. The button (to start the change ID sequence) you refer to in your review seems to not exist on the controller. Can you shine a light on this matter?
Marcel
Answer:
It looks as if you've got the new "Q
private controller".
These controllers apparently automatically set IDs when you just turn them all on near each other. I don't know how this works for team setup, exactly, but Babel Fish at least makes it perfectly clear that "In the manual ID register type controller, selects ID number itself until recently sort prayer itself."
