Atomic I/O letters column #64Originally published in Atomic: Maximum Power Computing Reprinted here December 2006.
Last modified 16-Jan-2015.
I've been toying with the idea of using a low boiling point liquid in a water cooling system. I then noticed how a power plant works by boiling water, creating pressurised steam (vapour), which then drives turbines to generate electricity.
So, what if the water in a water-cooling rig was replaced with a heat pipe fluid that boiled into a pressurized vapour at around 20 to 30 degrees, and could be used to drive a mini turbine, and generate electricity?
Of course, you would need a special CPU water block and radiator (condenser or heat exchanger) for it to work, but if the system had an efficiency of 25% (typical of steam engines with a condenser), it could output around 20 watts of power when used with an 80 watt CPU. That could easily run a few extra fans, and keep the system cooler.
Do you think this is possible and/or practical?
Steam engines are heat engines, and the efficiency of heat engines depends on the temperature difference between the hot end and the cold end. A closed turbine of the kind you're describing uses the Rankine cycle, and can indeed achieve quite impressive efficiency. But to do that it needs high pressure, superheated steam - water vapour heated up as hot as the engine designers can possibly make it. An unremarkable steam locomotive could be running on 200 degree C steam at 15 atmosphere pressure, and the big turbines run much, much hotter.
Bigger temperature differential means better efficiency, for all heat engines, so you want to ramp them suckers up until boiler rivets start pinging off the walls.
If you're running a heat engine with the cold end at room temperature and the hot end only as hot as a CPU can make it - say, 80 degrees C - its efficiency will be miserable. Assuming your masterful mini-lathe skills leave you with the world's first 100% efficient turbine and pump (you need the pump to return the condensate in the low pressure condenser to the high pressure boiler), you may be visiting the lofty heights of, oh, 0.95% efficiency.
In other words, I'm afraid your turbine in the real world won't make enough power to run the pump that feeds its boiler, much less turn a generator.
Can you please tell me how you can make copy protection on a CD/DVD?
As far as I know, when you or I write data to a CD or DVD the data gets stored as dips and bumps. If this is correct, then how can you possible protect that? I mean, all you would have to do is create a program that can read the dips and bumps exactly as it is on the disc, and you should be able to replicate anything!
Obviously I am missing something. If everything is just 1s and 0s then it should be simple to copy anything.
But the important part here is that there's more data on the disc than just the data the user wants to write there. There's various framing and table-of-contents and error-correction data, for instance, and people creating copy-protection schemes can screw with all of it.
Deliberate errors, files that appear to be much bigger than the disc, weird abuses of the Red Book audio format for real audio tracks or tiny pseudo-tracks; all kinds of stuff that, technically, means the disc in question no longer qualifies as a CD or DVD, since it's deliberately out of spec.
Back in the floppy-disk days, game companies did conceptually similar things, which could often be defeated by using a block copying program that did exactly what you'd like to do - read the raw data from the disk at the lowest possible level, then write that same data to the other disk, without trying to figure out whether it makes any sense.
Unfortunately, the panoply of CD and DVD formats out there, and the very nature of the "fuzzy" encoding of data on the physical discs themselves, makes it impossible to bit-copy optical discs.
Find more info about this problem here.
I recently put a DVD of important data into my drive. I pulled it straight back out when it started making rude noises. A little sticky piece of a discarded envelope had managed to attach itself to the surface of the drive. Unfortunately some glue residue remained on the disk surface rendering it unreadable. As expected, a wash with soapy water did not remove the glue.
My usual method of removing glue residue is to use eucalyptus oil. However, I am concerned the oil may decide to have my data for lunch as well as the glue.
Do you know if the eucalyptus oil will eat my data or be content with the glue?
You'd probably be OK, since polycarbonate is pretty tough stuff, but the oil might cloud the surface and, as you say, stuff the disc.
So burn yourself another disc full of whatever and try some different solvents on that. Meths/isopropanol/vodka would be my first choice; not very good at removing adhesive, but very plastic-safe. Spray silicone lubricant can be useful, too - for temporarily filling scratches on a disc you're copying (see also this column), as well as for removing gunk.
I read your old piece about the DIY Ghetto UPS with interest.
Do you know how long a typical deep cycle marine battery would power a computer through an inverter?
How long is a piece of string?
Deep cycle batteries come in a variety of capacities and voltages. Even if you stick with 12 volt batteries, mainstream dealers commonly stock batteries ranging from less than 100 to more than 200 amp-hours (Ah), and the same places will also have batteries not specifically made for the marine market that range from only a few amp-hours (starter batteries for small motorcycles) way up into the hundreds.
Here are some typical examples. As you can see, there's a lot of variation.
A modern low power PC (desktop Core Duo/Solo motherboards and CPUs, for instance, give the same low-power, high-performance deal as the older desktop Pentium M machines) with an LCD monitor could be expected to only draw around eight amps, total, from a 12 volt battery through an inverter - less when doing nothing, more when working hard. So you could be pretty confident about getting three hours of run time from a tiddly 25 amp-hour (Ah) battery.
These kinds of batteries, as I explain in my car-battery-UPS article, are commonly specified with a "20 hour" capacity figure - how much capacity you get if you only load them hard enough to flatten them over 20 hours. The heavier the load, the lower the real capacity you get will be. Even a mere 20 kilo "80Ah" marine battery, though, should deliver most of that capacity into an 8A load. That's only around twice the 20 hour load for a battery of that size, not some huge bank-of-spotlights kind of figure. So eight hours of 8A run time shouldn't be a problem.
Note that even "deep cycle" batteries don't actually enjoy being run flat. As many an irritated motorist knows, if you run a lead acid battery flat it'll suffer damage, thanks to a process known as "sulfation". But "desulfator" gadgets are now becoming more commonly available - and, surprisingly enough, seem to work - and there are also electrolyte additives that reduce the sulfation problem. So there's something to be said for using cheap car batteries instead of fancy sulfation-resistant deep cycle models.
It's still not a bad idea to put automotive batteries things in a plastic tray full of acid-neutralising sodium bicarbonate if you're using them indoors. I've haven't bothered doing that with the two in the corner of my office, though, and they're still pristine. Not being bounced around in a car really helps them stay tidy.
The PC version of Oblivion says "2GHz processor required, or equivalent".
I've got an Intel Centrino with Core Solo Processor T1300, Intel 945GM Express chipset, 1Gb of RAM, 667MHz system bus. It's only 1.66GHz - would that be "equivalent"?
That CPU is quite fast enough. The Core Solo/Duo CPU core is the successor to the Pentium M. They get a lot of work done per clock tick.
(Intel are currently suffering from Processor Numbering Disease, where every chip has a number attached to it that tells you something about how fast it is compared with other chips in the same line, but which tells you nothing about how fast it is in general. Historically, processor manufacturers do this when they've got good chips that run at relatively low clock speeds, then stop doing it when they've got fast-clocked ones again.)
I'm presuming you're talking about a laptop here, though (it's possible to run Core chips in desktop machines now, but they're not terribly popular yet), and so the 945GM is all you've got for graphics, and all you'll ever have in that computer.
This means you are screwed, because Intel's integrated graphics adapters are always slow as a wet week for 3D, and Oblivion is a big graphics-card hog.
Intel's strategy for improving the 3D performance of their integrated graphics adapters has, traditionally, been to give them exciting names that suggest great speed. The chips themselves remain simple low-power cheap units for the business and non-gamer home markets. The 945GM's graphics component is called the Graphics Media Accelerator 950, and it's the fastest Intel integrated adapter ever, which puts it only a bit below the minimum specification for Oblivion.
This means you could probably get Oblivion running on your computer, with all of the graphics sliders wound all the way down and all of the boxes unchecked, a la Doom 3 on a Voodoo 2, and possibly play an enjoyable game. I wouldn't get my hopes up, though.
Minimum specifications are usually pretty darn optimistic, because the game makers know that minimum specs sell the game, and then people who actually have that minimum hardware try to play the game, and then they buy better hardware.
If you've got a laptop that can't be upgraded, it's best not to taunt yourself with overly demanding software.