Dan's Data letters #4Publication date: 22-Oct-2002.
Last modified 03-Dec-2011.
I'm about to put lights inside my computer and I'm having trouble figuring out which would provide more light: a blue 20cm (tube size) neon from Brisbane Car Sound which runs from a 12V cigarette lighter plug, or a few red or yellow LEDs?
With the neon I'll obviously have to convert the cigarette lighter plug to a Molex plug, but would a neon of this type be safe to run from my PSU without blowing it up? The neon only costs about $AU20.
Also, if I were to use LEDs, which colour out of yellow and red would be the brightest for my case? At Jaycar there are yellow 5mm 5000mcd LEDs and red 5mm 6500mcd LEDs. Also, I was wondering how many I would need to supply sufficient lighting to the inside of my case?
The tube (which, being blue, doesn't actually contain neon; actual neon glows orange-red) should be much brighter than an array of 5mm LEDs, unless you put a few dozen LEDs in the array. Some cheap tubes are miserably dim, but most aren't.
The tube should also be perfectly safe to use in your PC. Sometimes RF from cheap tubes' power supply can cause issues, but that happens with the ones that are allegedly meant for use in PCs as well. LEDs have no such problems, since they're low voltage all the way. LEDs, if not severely overdriven, will also greatly outlast cheap tube lights. The nastier tubes have been known to drop dead after a couple of weeks; you can generally expect a much better lifespan than that from a half-decent tube, but LEDs will last effectively forever; don't overdrive them and, according to the most frequently quoted manufacturers' spec, they'll still be at 80% of their original brightness after ten straight years of service.
Regarding brightness - the higher up the spectrum you go towards green, the brighter an LED with a given output power will look, all things being equal. 100mW of red LED light looks dimmer than 100mW of yellow, which looks dimmer than 100mW of green, assuming they're all pushing the same number of actual photons. 100mW of blue looks dimmer than the same amount of green, because now you're past the eye's peak sensitivity point, but it'll still look much brighter than 100mW of red.
Then again, red and yellow LEDs are cheaper than green and blue ones.
Millicandela (mcd) ratings are affected by focus. The tighter the LED's beam, the more millicandelas it'll score for a given actual power. Most Jaycar stores have demo breadboards with LEDs on them so you can see how bright they really are; if the LED you want isn't on show, pester someone to set it up.
How many you'll need for sufficient illumination depends on your definition of "sufficient". Many people find cold cathode fluorescent lights (CCFLs - the super-thin gas discharge tubes) are too bright. Generally speaking, I'd say 16 recent-model high intensity yellow LEDs would give you enough light for a case. Just a few of them would still be noticeable, though.
Given how cheap LEDs are (in small quantities, at least), you could easily hack up a circuit-board-less mini-array of two or three LEDs running through a resistor from 12V, and see what you thought. That'd still be useful as an accent light for your CPU or something, even if you decided to go with a tube for the main light.
I plugged the floppy power cable into my new Quantum 3D Voodoo 5 6000 card wrong (the power connector was specially installed) and blew the card up. The wire burnt in seconds. Is this normal, I thought? Hasn't my PSU got a cutoff circuit? Why can't they put a simple regulator on the input line to stop surges? I must get this card working!
Um, yeah, that'll happen. If you get a four pin drive-type power connector backwards, you'll apply 12V to components expecting 5V; result, toast. If you manage to plug it in off-by-one, then you'll apply voltage to an earth pin, which could blow a track off the board. Result, also toast.
Is this normal? Well, yes. Not that I've ever made this particular mistake myself, but people who manage to reverse plugs with power supply wires in them commonly blow stuff up and/or set cables on fire.
Your PSU will blow a fuse if you ask for way too much current; it takes considerably less than its maximum current capacity to blow up a component if you plug stuff in wrong.
A regulator wouldn't help you, by the way. Regulators regulate voltage, not current. A current limiter could have saved the day, if it clamped down fast enough; it's possible to put some kind of current limiter (a simple fuse would qualify, in this case) on the supply lines for any piece of hardware. But it's not often done, because it's not normally assumed that people will manage to connect the power plug wrongly.
Fuses are used on some hardware, though; many motherboards, for instance, have little surface mount fuses to protect the board from mis-plugged expansion port devices. The classic example is older PS/2 keyboards with large smoothing capacitors that charge too quickly and can, in theory, damage the motherboard if you hot-plug the keyboard (which you're not meant to do). Plug such a keyboard in when the computer's turned on and you may blow the protection fuse, which'll zonk the PS/2 port. Just bridging the blown fuse with a shred of wire, mind you, will bring the port back to life, and hot-plugging the same keyboard in the future probably won't damage the motherboard. The reason for this is that the value of the fuse has been deliberately set rather low, to give it the best chance of actually doing its job by blowing before damage has been done, and avoiding the situation where a $500 piece of hardware manfully sacrifices itself to protect a $0.05 fuse. The low value, however, means the fuse can and will blow when no damage would have been done - so-called "nuisance triggering". Nuisance triggering is a big reason why designers don't bother with protection hardware like this.
Most hardware just relies on the PSU's own circuitry to keep it safe, and that generally works very well. Provided stuff is plugged in properly.
To be honest, I'm not sure why you're so hot to run this weird old video card. I can understand someone wanting to buy a V5 6000 to put in a cupboard and sell in ten years, when it'll probably be worth quite a bit of money; they're certainly a rarity. But there's really no practical reason to bother with a V5 6000 any more. A GeForce4 MX isn't far behind it in performance, and any full GeForce4 beats the pants off it in every department.
I have a... friend... who recently got a burnt copy of Unreal Tournament 2003. The game asked for a CD key so he downloaded a key generator, and within seconds he was fragging away. My question is, how on earth are people able to make these generators? CD keys for games are often more than 20 characters long, with millions of different combinations, and I'm assuming that game publishers invest a lot of money into making the whole CD key system.
Also, when a cd key is generated, does it mean that a legitimate user is having his key stolen?
Mr Mc Chicken
There are two kinds of software key.
First up, consider software that lets you convert an unregistered version into a registered one with a serial number that you have to pay for. That sort of key is generated by algorithmic twiddling of the user name (and/or company name, and/or e-mail address) you enter when you register. The name-to-serial algorithm may or may not be difficult to work out (some software has really long keys; the key for MailWasher is 294 characters!), but all it takes is one coder somewhere to figure it out and create a keygen and bing, it's cracked forever. Or at least until the next version of the software changes the algorithm. Even if the key algorithm is fiendishly complex, people can still tell other people their paid-for username and key.
This key-generation strategy explains why ripped-off keys for various shareware apps come with both a username (generally something rather whimsical) and a key; without the right username, the key won't work.
The other kind of software key system works like credit card numbers. Credit card numbers are not random; they obey algorithmic rules. The initial digits tell you what kind of card it is and who issued it, and the whole number has to pass a Luhn Algorithm test.
If you generate a credit card number in a truly random way - just roll a ten-sided die 16 times - you're unlikely to come up with a number that passes the Luhn test (only one in ten numbers will), very unlikely to come up with a number with starting digits that correspond to a real card type and issuing institution, and very very unlikely to come up with the number of an actual card that exists, and its expiry date, and the extra three digits on the end of the number on the signature side of current Visas and MasterCards that're used by some vendors for extra verification.
It's easy to generate a credit card number that passes the first two tests, though; the Luhn algorithm's simple enough that you can do it with pencil and paper, and there are lists of tons of valid card prefixes. There are lots of freeware utilities that'll spew out dozens of card numbers whenever you like.
There's no way for a merchant, by himself, to tell that one of these fake numbers is, actually, fake. All online shopping cart software worth its salt does at least a Luhn check on numbers to make sure they've been typed in properly, but that only tells you that the number could be real, not that it is. The next step is to use the card issuers' verification services; the merchant, or more commonly the merchant's card-handling hardware and software, calls up Visa or Mastercard or American Express or whoever, and checks to see that the card details are the details of a real card, and that the card isn't lost or stolen or overdrawn.
Games with online play that come with a CD key work the same way. The game software on every CD is exactly the same; it has no idea what serial number's on the sticker on the jewel case. It, therefore, does basically the same thing to the key you type in when you install the software as a simple Luhn-checker does to a credit card number; it checks the serial against some algorithm or other to see if it matches, and if it does, it lets you play the game locally. If someone cracks the key-check algorithm, then they can generate arbitrary quantities of keys that'll pass the game's built in test.
When you try to connect to an online game server, though, the game checks with an authentication server run by the publishers, which knows all of the real keys - the keys that have actually been printed on a sticker, of which there will be far fewer than the number of keys that pass the algorithmic test, just as most Luhn-passing 16-digit numbers aren't real credit card numbers. Your generated key might be a real key (in which case yes, the faked-key player can take the place of a legitimate buyer of the game), but it very probably isn't. If you fail to provide the authentication server with a key that it accepts (and known-to-be-warezed serials are likely to be regularly deleted from the authentication database), then you won't be able to play online.
Well, not without further hacking, anyway. It's apparently possible to play UT2003 online with a generated key, after a bit of config fiddling to input a hard-coded server list; presumably an outboard utility could update that list from the standard server list and make the game work exactly like a registered version. But, to be perfectly frank, I really think buying the game would be less trouble.
Copy protection schemes that actually do make every game CD different, so only the key provided with that CD will work with it, have been suggested. But anybody who thinks that'd stop crackers is, themselves, on crack, and it'd do nothing more to stop people playing online with fake keys than the simpler, all-CDs-alike system does.
I can scarcely believe my eyes. Tubes! Tubes! (I think you Commonwealth folks call them valves?). On a new AMD Hammer motherboard of all things (AOpen AK86)!
And here we're worrying about the thermal characteristics of the CPU - those tubes are bound to cook the innards of the average PC case. The 80-100 watts of heat from the CPU look paltry compared to the heat output from real, live, glowing filaments!
I can't wait to see the wanky coolers that'll get designed for the tubes (electronics newbies have no idea that a certain amount of heat is required for these things to work, hence the filament called the heater!).
Well, at least the 5.1 sound system will be resistant to EMP effects from the next atmospheric nuclear detonation!
Yeah. Weird. AOpen produced a one-valve board a while ago, but apparently that wasn't enough for them :-).
The tubes aren't a thermal problem, though. They're not power valves, just dual triode pre-amps, probably 6DJ8/6922/ECC88 type. Their full rated heater power isn't much more than two watts per unit, their total aggregate power isn't much more, and many applications run them well below rated power. Check out my review here for a bit more info on this subject.