Dan's Data letters #57Publication date: August 2003.
Last modified 12-Sep-2012.
I have a Lian Li PC-690 case that I have been extremely happy with up to a few days ago. I turned it on and there was a puff of smoke and light. I turned it off and took the side panel off. Turned it back on and was amazed to see this great light show. It looked like the insulation on the wires coming from the case failed, because they were all on fire and were even melting the connector. I managed to not damage anything else - just those wires. I was lucky that my daughter was not the one to turn this on, who knows what might have happened.
Do you know of any prior issues with Lian Li's case wiring doing this after some use?
Do you mean the fan speed controller wires? Because, apart from that, Lian Li cases don't have any wiring; they don't come with PSUs, or anything.
Anyway, any PC PSU is perfectly capable of melting down its wiring if something shorts out. Usually, shorts are caused by someone closing a case side on a wire and cutting through the insulation, or some similar user-related mishap, but sometimes components fail in such a way that they go short circuit. If that happens, and the short isn't enough of a short to pop a fuse in the PSU, then smoke and light and similar excitement can occur. This is not peculiar to Lian Li cases; it can happen in any PC. It usually doesn't, though. Most components go open circuit when they fail, which is much less entertaining.
I've been reading a lot about the Digital Doc 5 including the review you did on it, and my plans for a new case fan setup revolve around it.
The main thing I want to accomplish is to quell the excessive noise coming from my processor fan. Now it's not as bad as a lot of people's horrid setups, but it still is noisy - a 4-5000RPM Thermaltake Volcano 9. My thought is to remove the fan from the heatsink and instead put 2 smaller 80mm fans on it, doubled up, and program the second to come on only when temperature reaches a limit. Most of the time, for surfing the web and doing pretty much anything that does not involve 3D, encoding or anything intensive, the CPU should stay cool with a relatively small amount of air running through its heat sink, no?
The biggest problem is if I have two 35CFM (say) fans say working in parallel, will they, combined, produce 70CFM? I would say not a chance. But how much over 35 will they produce?
Another thought is to have the primary fan low, around 35, and have the second fan blow much harder.
Fans on top of each other (in series, not in parallel) won't achieve much. They'll have more static pressure, so they'll be able to blow an amount of air closer to their free-air rating through a given resistance (in this case, the heat sink), but there's not a whole lot of resistance there in the first place. Also, two fans spinning on top of each other without a spacer tube will interfere in all kinds of chaotic ways to reduce each other's efficiency. If they're spinning at different speeds, one of them will be trying to drive the other.
A better idea would be to get a thermally controlled fan that can run from pretty-low to pretty-high power, then play around with the thermal probe location until it works as you like. One of Thermaltake's orange "Smart Case Fan II" fans, which they're supplying on various of their current CPU coolers, would be a good choice.
Remember how you said CPUs usually die at 80-90c? Well, funnily enough, the dual thermal probe I had installed was going off the scales when I came back to find the fan had died on my Volcano 7+. For the record, I had the probe taped to the bottom of the heatsink, as close to the core as I could get it.
I waited for it to kick back in, and saw the probe finally read 120c. Strangely, the chip still works (it's an Athlon XP 1600+). And at the time, I was using Arctic Silver 3 paste, which hasn't caked or gone dry. It's still the same consistency as when I put it on the first time.
There's a very good reason to be using good quality thermal paste, and it justifies voiding your warranty (AMD say that any thermal paste other than the stuff they’ve specified voids your warranty).
CPUs don't die when they get up around the 80 to 90 degree C mark - they just stop working until they cool off again. They may keep producing heat - and may eventually manage to actually burn themselves into broken-ness - but they don't keep computing.
What AMD actually say about the thermal solutions they want you to use (which isn't necessarily what's been reported), and what actually happens in the real world, are two different things, as has been explored in some detail.
People who use thermal paste on top of the pre-applied thermal goop on a heat sink, as per page 6 of this amusing PDF file, may well get their warranty request refused. But this depends on the policy of whatever AMD or retail entity the user's dealing with.
Warranty returns are refused on the grounds of user damage to the processor, among other things. AMD may not assure users that thermal pastes other than Shin Etsu G 749 (or whatever) are OK, because some thermal pastes really are awful and shouldn't be used for a moderately high performance task like CPU cooling. Also, some pastes are sufficiently electrically conductive that you need to use them very carefully, if at all, on CPUs with components and bridges on top.
Use very bad, or very conductive, paste and your warranty should be void.
But the friendly store you bought your CPU from, and to which you attempt to return it, may well give you a trade-in anyway (well, assuming you're not two years into your three year AMD retail CPU warranty period, anyway...). The AMD distributor the retailer then deals with to get restitution may well be perfectly cool about it too. It depends on the retailer, and it depends on the distributor, and it depends on how long you've had the CPU and what signs of abuse it shows.
You can try to play hardball about this and wave around copies of your country's consumer protection laws, but if you do that, the retailer will probably just affix a "murdered by owner" sticker to the processor and smile sweetly at you. You won't be able to make your claims stand up in court.
If the only visible difference between your dead CPU and a brand new live one is that your processor's got some silvery grease on it, then I think you'll find that someone in the RMA chain will just wipe the darn grease off. That someone might as well be you.
I just got a 50 inch Yamaha PDM-1 plasma screen, and was wondering if watching a widescreen picture with bars on the top and bottom of screen will burn the screen after a few hours?
Not after just a few hours, but eventually, yes.
Plasma screens wear out after a while, but so does every screen technology; a plasma screen actually ought to last longer than a CRT. Plasma pixels that aren't displaying anything, though, will wear out and go dim slower than pixels that are displaying something, so the black-bars area of the screen when you're displaying a letterboxed image will, in time, end up brighter than the rest of the panel. This is an "anti-burn-in" issue, though, not really a "burn-in" one; the black bars are aging slower, and normal video displayed on the rest of the screen shouldn't age it any faster than usual. You ought therefore to be able to get away with playing lots of letterboxed video without seeing anything nasty.
Different plasma subpixel phosphor colours will burn in at different rates, too; blue will probably burn in quicker, giving a yellow colour cast to the burned areas.
Current high quality plasma screens compensate for burn-in by monitoring their pixel brightness and restricting it so that new, un-faded subpixels can't go to full brightness. That way, it'll take a while for faded subpixels to become so faded that unfaded subpixels look noticeably brighter. Eventually, though, burn-in will still be noticeable, if some part of the screen is black (or white) most of the time. Any time you switch to full-screen video, you'll see brighter bars at the top and bottom. This may not be for a long time, though, depending on how much you use the screen.
Lately, I've come across cable select on HDDs. For years I've just ignored that little jumper on the drive labeled CS and just set the drives to master or slave. For the last few months I've been working as a PC assembler building about 30 PCs a day and have noticed that most (all?) hard drives I install come with the jumper set to CS.
I've noticed some UDMA/66 IDE cables have a cut in one of the wires. After a few enquiries, it turns out that these are "cable select" cables. I've read that it's the standard for all 80 wire cables to allow cable select (as opposed to 40 wire UDMA/33 cables, which had to be special to allow it). Now, not all 80 wire cables that I use have that cut, but I'm assuming that the conductor (wire 28) just isn't connected electrically to the IDE connector. Is this the case?
I could write an answer here, but this page says it better.
I have windows 2000 on Primary IDE master using NTFS.
I installed Windows 98 on Primary IDE slave.
If I boot using the boot menu of the BIOS, I can pick the master or the slave, and get Win2k or Win98. But I want to use the BOOT.INI file of Win2k to control this, for the pretty text menu goodness.
In Linux terms,
/dev/hda1 = WIN2k NTFS
/dev/hdb1 = win98 fat32
How is it done in Windows?
I have some appliances which run on that wonderful old Ni-Cad technology (cordless 12V drill, Dustbuster, etc). I know that Ni-Cad charging is fairly dumb - i.e. a simple unregulated transformer plugged straight into the battery for a manually-determined period of time.
My problem is that I am not sure what the right amount of time is. The Dustbuster manual implies that you can leave it in all the time, and the drill manual mentions something vague about a 12 hour charge the first time and doesn't give you much guidance after that. What's the story with Ni-Cad charging? Is there some kind of plateau point on the charge vs. time curve, and can you just leave these things plugged in all the time?
The dumb chargers you mention are basic cheap trickle chargers. Fast NiCd chargers can throw a full charge into high quality cells in a lot less than an hour. They generally use delta-peak detection, when the battery voltage drops by a few hundredths of a volt, to pick the end of the charge cycle. They may also have battery temperature monitors; when the battery temperature starts climbing rapidly, it's done, or you're charging it too hard.
Unlike NiMH cells, NiCds under gentle charge only get warm when they're full. If the battery portion of a gadget feels warm, it's definitely charged. If it's not under hard enough charge to get noticeably warm even when full, then it does no real harm to leave it connected, at least for a while.
Dumb trickle chargers usually get the job done in 12 to 14 hours, but they may be faster or slower. You can't tell without measuring. Many cheap drill chargers take only a few hours to do the job, but aren't smart about it; if you leave batteries in them too long, they'll bake them.
I have one of the Marui Airsoft tanks, but one of the motors at the back has broken so it only goes round in circles. Can you give me any advice on how I can get the motor fixed? I have asked the same question to a lot of other people, but they say to get a replacement of the tank or the part from the place where I bought it. That's all very well, but I bought it from an American site, and I live in England, and they said that they can't send me anything that will help.
You can't get it fixed, but replacing it shouldn't be too difficult, if the motor is in fact the problem.
If the problem is actually a blown drive channel in the speed controller or a receiver fault of some kind, then you'll need the help of an electronics tech to fix it, if it can be economically fixed at all. The problem may be as simple as a bad solder joint, though; one of the motor wires may have come adrift.
If the problem is the motor, then you can replace it. Practically every toy with little DC motors in it uses one or another Mabuchi motor; if you take the tank apart and disassemble the drive motor box at the back, you should find it quite easy to get the motor out and determine what Mabuchi model (as per these listings) it is. A replacement should be quite cheap from a hobby shop, or very cheap from an electronics place that stocks these little "hobby" motors.
Isn't this exciting? Can't wait to go out and buy mine!
Couldn't you just paint the APs and NICs black?
(I wrote that piece more than eight years ago, now. When I were a lad, it were all fields around here.)
A couple of years ago, I remember everybody linking to this page where a guy uses his K6 as the heat source for a still. One quick question - can't you go blind if you drink booze that's been made this way (I mean with a home still in general, not with a crazy one using a CPU...)?
Yes, badly made moonshine can contain methanol ("wood alcohol"), which is Bad News in general terms, though its effects are greatly mitigated by ethanol.
Denatured alcohol (called "methylated spirits", here in Australia) used to be almost all ethanol with just a bit of methanol in it, which was why people who drank it could live for rather a while, albeit not in the best of health. I'm indebted to a reader for the information that Australian methylated spirits is now, actually, not "methylated" at all; it contains ethanol, some awful-tasting chemicals (including this one) to stop all but the most determined drinkers from imbibing it, and a near-zero amount of fluorescent dye, but no wood alcohol.
Anyway, methanol metabolises to formaldehyde and then to formic acid, and neither of those compounds are something you want in your body in large quantities.
But the still doesn't make the methanol. The methanol is made by whatever your alcohol-production system is - generally yeast acting on some fermentable substance or other (for instance, grape juice, potatoes, molasses or Mexican bandit urine, for wine, vodka, rum and Corona, respectively). All sorts of nasty extra compounds are created by fermentation of many source materials, including the grains used for beers and whiskies.
To make spirits, you distil your yeast-fermented beverage (the yeast can only give you a few per cent alcohol before it's killed - alcohol is yeast waste product), and, if done correctly, you get rid of the nasties, most of which have noticeably higher or lower boiling points than ethanol and are thus easy enough to avoid.
Methanol is chemically very similar to ethanol - looks the same (different refractive index, but that's hard to see), smells the same, burns the same, tastes the same. But it's quite easy to separate methanol from ethanol by distillation, because methanol's boiling point is lower.
The early, low-boiling-point output of a still resembles paint stripper, as will crappy moonshine; if you screw up the distillation and leave the methanol in, you'll probably also leave in acetone and lots of other odious compounds that ought to tell anyone that Something Ain't Right.
Those who're not too picky about their tipple, though, may drink this early output - and all of the output from a still may be awful if it's not being operated properly, ramping up the temperature carefully to separate the components.
You can only make poisonous moonshine, though, if your source material actually has methanol in it. In the famous CPU still, all that was being distilled was rum. If you start out with some commercial beverage that doesn't have methanol in it, distilling it further won't introduce any.