Dan's Data letters #180Publication date: 24 December 2006.
Last modified 03-Dec-2011.
Insert Fiat joke here
Wind-up cars make some sense to me. What do you think?
Yes, a torsion spring is just another way to store energy, like a battery, but without wasteful mechanical-to-electrical-to-mechanical transformations. But there are a few ways in which it could fall down.
The two obvious ones are safety and energy density.
Safety I'm not going to worry about. Yes, a super-twisted industrial size torsion spring that gets loose could cleave the car in half and then go somersaulting down the road like a caber tossed by The Incredible Hulk, but the safety challenges of jet engines and compressed-flammable-gas-powered cars have been overcome, so I presume this one could be too. The same applies to flywheel energy storage, which can be very dangerous if implemented by a guy in a garage, but can be and has been done safely.
Energy density is the bigger issue. Generally speaking, springs of all kinds have lousy energy density compared with other similarly bulky and heavy storage systems. Springs can accept and deliver energy very quickly, which is good for regenerative braking purposes (and for bows and siege engines...), but they just can't store a lot of energy per kilogram.
A further problem is that springs in general do not have a flat resistance curve. When you start twisting your torsion spring, it's easy to do; it gets harder and harder as the spring twists more. If you used the spring as your only brake, you would therefore have very weak brakes at first, which got stronger and stronger as you approached what seemed likely to be the scene of the accident.
Virtually all regenerative braking systems also have normal friction brakes, to assist if you need more braking than the regenerative system can provide, and to take over if it fails. Making such a system work smoothly when the regenerative brake keeps changing its strength would require, I don't know, torque converters or something, not ordinary clutch and differential gear kinds of arrangements.
Springs aren't the only energy storage system that has this problem - the "air car" regenerative brakes, which use a compressor to pump gas into a cylinder that can drive the compressor as a motor later on, are the same.
These problems also apply to capacitor-based systems, because capacitor voltage changes with charge. Batteries can't accept or deliver power nearly as fast as caps, but they have a reasonably steady terminal voltage.
Oh, and the reverse also applies. Springs want to deliver more energy when they're wound up tight. So you'd need another energy moderating system of some sort (or a very clever one that works both ways) to prevent your car from wanting to take off like a jackrabbit after you've stopped hard, or just braked down a long hill.
Maybe there's some clever transmission that can solve this problem for mechanical systems, but if it's basically a pair of automatic transmissions, one pointing each way, then it ain't gonna be light.
The final problem is that the springs might wear out. I think you'd have a hard time combining reasonable size and weight with high energy density and an acceptably long lifespan - and catastrophic failure of a worn-out spring takes us back to the safety issue. One of the commenters on the Damn Interesting page suggests a hydraulic accumulator, which would probably last much longer than any torsion spring with similar energy capacity. Hydraulic systems have the advantage that they're actually working right now; it's just that the weight/expense/power calculations make them uneconomical for applications other than frequent-stopping delivery vehicles and heavy trucks.
Step 1: Throw PSUs down stairs...
What do you think of this article's suggestion that most PSU reviews are fundamentally wrong, in that they measure voltage fluctuations rather than current fluctuations in the face of dynamic loads?
Well, most PSU reviewers don't measure anything much. I know I don't bother hooking up meters to PSUs when I write about them (OK, except for this one), because there's no point doing so if you can't load the PSU really hard. To do that you need to, at the very least, build a big old resistor board like the one in this old Tom's Hardware piece.
You can get some worthwhile data by just hooking PSUs up to a computer, but you can't specifically probe their weaknesses, and you probably can't take them close enough to their sticker specification to be able to find out if it's realistic, either.
So on the one hand, yes, it's normal for PSUs - even big-brand ones, and definitely cheap ones - to not actually be able to deliver the amount of juice the sticker promises.
But, on the other hand, most PSUs can deliver a lot more power than the computers they're connected to actually need.
Your average home PC builder these days often buys some allegedly-500W-plus PSU, because it doesn't cost significantly more than a mere 350W model. Even when the computer's working as hard as it possibly can, though, it probably won't actually need much more than 200 watts. Maybe 300, for an overclocked multicore SLI machine with lots of drives. Throw in Peltier coolers and scads of huge fans and blah blah blah and I'm sure you can push it higher, but very few people do.
Now, the total draw can be split between the different rails in odd ways. There's often a very heavy load on the 5V rail, for instance, from modern computers.
So it's perfectly possible to put a "500W" PSU in a stacked computer and find that it doesn't work right, despite only having to deliver 280 watts.
And cheap off-brand PSUs may have unusually large Lie Factors in their specifications, and may contain substandard components (especially large electrolytic capacitors) that slide away from their ratings in short order and give you a flaky computer within six months.
But, generally speaking, an off-brand "500W" PSU really ought to be A-OK for practically any computer anybody builds today, unless it's destined to be a high-availability server or something. Your only real concern should be whether the Gilded Flamingo Excellent Power Corporation ticked the more expensive, thus more reliable, boxes in the Power Supply Feature Selection Menu provided by the Original Equipment Manufacturer in Guangzhou that actually fabricates the things for Gilded Flamingo and 23 other "manufacturers".
All of this is really for the best. If we were all building computers that really could stress a genuine, well-engineered 500W power supply, it'd be a minor ecological and home-budget catastrophe waiting to happen. Leave such a computer on all day and, depending on where you live, you could easily end up paying several hundred US dollars extra per year for electricity.
In my reviews I try to explain all this, cut the hype, and tease out any genuinely unusual features (like the few PSUs that still have the -5V rail that some older computers need, or the ability to tie together the two or more separate 12V rails that many PSUs have these days - these Tagan models have both features). If a big beautiful stress-testing rig fell from the sky then I'd use it, but I am insufficiently dedicated to journalistic excellence to build one myself, right now. And even if I had one, I'd have to spend all week testing several randomly selected versions of the same model, to get some idea of the quality control being applied to that model.
If people were having serious problems with PSUs all the time, I'd have to do all that to write a decent review. But since you really can take a short cut to very probable success by (a) taking the sticker ratings with a large grain of salt and (b) buying a second cheap PSU to sit in your cupboard in case the first one drops dead the night before a long weekend, I'm happy to say that such effort on my part appears to be unnecessary.
Stop! Bluescreen time!
My good Sydney friend is encountering mysterious STOP errors in Windows XP. He has not installed any new software or drivers lately. I jumped ship to Macs about three years ago so my knowledge is creaky, but I advised him to repair his Windows installation (he got STOP errors in that process) and then to reinstall (more STOP errors, now getting copying errors and stop errors as soon as Setup starts).
I have now advised him that as I recall, STOP errors are deep hardware problems so he will need to replace his CPU and motherboard (I hope it's not the RAM). I am further convinced it's a CPU thing, as his BIOS has been recently shutting him down because his CPU temperature has been going over 90C.
I have recommended this Aus PC Market bundle, because I know nothing except that AMD CPUs are just as good and cheaper, and $AU264 sounds cheap for a processor and motherboard together. I guess.
I am recommending him Aus PC Market because I like them, and I like them because they support your site. I want to give my Sydney friend your affiliate link to this product, so that when my friend buys it, you get whatever commission that is worth.
STOP errors can be caused by any damn thing, actually. They're a serious error, but they don't have to have a serious cause.
Bad drivers can do it - though if they suddenly start happening without software changes then, as you say, it's not that. A laundry list of defective hardware could also be the culprit, though. Yes, the mobo and/or CPU could be at fault, but so could the hard drive, RAM or power supply - a flaky PSU can cause an incredible number of weird problems.
Now, a cheap CPU and mobo upgrade is still not a bad idea if it'll give your friend a substantial speed boost, but be aware that it may not help with the problem at all. Swap-it-and-see debugging is called for. If you've got multiple RAM modules you should remove them one at a time (disabling dual channel mode if necessary - though the BIOS should do that automatically) and see if that helps. Then the next step is a new PSU (because they're cheap, and, as I mention above, useful to have on the shelf even if you turn out not to need one right now), then perhaps a new hard drive. It's not a bad idea to swap out any consumer hard drive once it reaches its third birthday anyway, if it's holding important data.
If none of the above helps, then the whole CPU and mobo swap becomes necessary.
Of course, a lot of people do just throw up their hands at the horribleness of it all and get a whole new computer, or enough new parts that it's a new computer for all intents and purposes, at the outset.
If the CPU's running hot, then the problem could easily just be a bad CPU cooler. Is the heat sink clogged with dust? Is the fan turning slow? Blow the computer clean (Real Men do that while it's running...), point a desk fan into the works, and see if it comes good. If it does, then it's definitely just a heat problem, which could also result from thermal goop under the CPU heat sink drying out, or a motherboard chipset cooler with a dead fan, clogged heat sink, bad thermal compound, etc.
My own main computer's motherboard chip cooler crapped out months ago - I unscrewed the cheapo custom 40mm fan from the heat sink, Blu-Tacked an 80mm fan to the top of the video card pointing at it, and went about my day.
Regarding CPU choices - yes, on average, in any given month, AMD chips are likely to be better value than Intel ones. In the higher end processors, though, Intel have the edge at the moment. Like you care, though - that $AU264 combo is indeed a good deal.
The combo's strong points:
* Processor quite fast at stock speed, and quite overclockable. A-OK as the basis for a Monster Gaming Rig with the bodacious video card of your choice.
* Processor only single core, but that's fine for the vast majority of users. Uses less electricity that way, too.
* AsRock boards are perfectly fine; it's Asus' budget brand. The 939SLI2 is a proper modern motherboard.
* It has plain DDR RAM, not needlessly more expensive DDR2 (not a big difference these days, but every little bit helps)
The combo's weak point:
* Socket 939 is obsolescent now (replaced by AM2), and processors to suit it won't be on retail shelves for a lot longer. Still, they'll be showing up on eBay forever and a day, so in a couple of years you'll probably be able to buy an Athlon 64 X2 4800+ on eBay for $50.
Regarding affiliate links and such - I don't work on commission. I get a flat rate for clicks on my you-can-buy-this-from links to AusPC. They're the ones that look like http://www.auspcmarket.com.au/adclick.php3?c=103&product_code=SOME_PRODUCT_CODE.
(The product links in this letter reply are not paid links.)
So feel free to tell your friend to click through from one of my reviews and then go on to buy something.
Do not, however, go nuts clicking links all over the place until you feel I've been adequately compensated, because (a) that honks AusPC off and (b) they'll just disregard clicks from your IP addresses, anyway.
Does high-efficiency laundry detergent have a shelf life?
Background: A few months ago, I bought a new front-loading washing machine (they're finally starting to get popular here in Canada). It, of course, should be used with high-efficiency/low-sudsing laundry detergent. As part of a sales promotion, I received coupons for 13 bottles of said detergent, worth a total of roughly $CA150. Of course you can only use one coupon at a time, they expire in a year, and they're banking on your losing them or not using them.
So I thought I'd "stock up", and use one of the coupons every time I go to the store until they're gone, storing them at home until I need them (13 bottles will last me something like three years...).
So: Will they go off/bad? Or is HE detergent likely to remain essentially unchanged after three years' storage at room/cellar temperatures?
There is, you'll be startled to learn, a huge amount of marketing crap in the laundry detergent industry. My favourite example of the contempt with which the companies treat their customers is Raymond Chen's recent blog post here.
Startlingly, though, there actually is a difference between "High Efficiency" detergent and the normal kind - well, there's meant to be, anyway. HE detergent produces less (fewer?) suds for a given amount of cleaning power. People tend to associate suds with cleaning, and assume that more of one mean more of the other (the "scrubbing bubble" fallacy), but the two are usually irrelevant - it's just that water with most surfactants in it will tend to form bubbles readily.
In devices like dishwashers and front-loading washing machines, though, suds get in the way, so they use low-sudsing detergents. Dishwasher detergent can be quite complex but is, essentially, just some fairly strongly alkaline salt or other. In a pinch, you could probably wash your dishes quite well by putting bicarbonate of soda in a dishwasher's detergent dispenser.
(Plain water cleans pretty well, too. That's why people believe that worthless "quantum mechanical" plastic laundry doodads get their clothes clean.)
High efficiency laundry detergent is much closer to ordinary sudsy detergents than it is to dishwashing detergent. I wouldn't be surprised if many HE detergents were actually chemically the same as the regular laundry detergent, but with anti-foaming agents added. There's certainly no huge difference, and no strong justification for HE detergents costing significantly more per wash.
HE detergent is often more concentrated than the regular kind, which justifies a bit of a price hike - but there's no reason why it has to be. And even when it is concentrated, you're still mainly paying for the container, the shipping, the shelving, the advertising and the big fat profit.
There are substances used in household detergents that may go bad on the shelf, but if your detergent doesn't have grease-emulsifying enzymes in it, I bet it'll last forever. Even if it does have enzymes, it'd still probably work well enough without them, 20 years down the track.
The only other cleaning substances that I think are likely to go stale are bleaches, which are common in dishwasher detergent but not, for obvious reasons, in laundry detergent.
I think it's likely that you could formulate your own low-suds detergent much cheaper than the commercial stuff. Start with the old detergent-assisting washing compounds, borax and washing soda (that's sodium tetraborate and sodium carbonate, both still easily available from supermarkets where I live) and add just a little ordinary soap-flake laundry detergent - for, say, a 10:10:1 borax:soda:soap ratio by volume - and you'd probably be almost there right away. Experiment with higher soap ratios for more cleaning power until you start seeing too many suds.
(1:1 borax and washing soda will work fine in dishwashers, too. Sodium carbonate by itself is a decent washing agent; because it's alkaline it saponifies oils on contact, making a little bit of soap as it goes. This is why alkaline solutions feel slippery; they're saponifying the oils on your skin. Strong alkalis do not stop at the skin...)
It's possible that the expensive commercial stuff genuinely does work better for heavily soiled clothes, but I wouldn't be surprised if it didn't, and most clothes aren't filthy when you wash them anyway.
Commercial detergent definitely is loaded with those "optical brighteners" that fluoresce in UV light and so make clothes look cleaner in sunlight. But said clothes are actually now coated with optical brightener particles, which I think, technically, means they're dirtier.
Geek seeks switched-on chick for electronics experiments
Hey, Dan. As a long-time reader and a past donator, I was hoping you could help me out with an electrical circuit I'm trying to come up with. I never made it that far in electrical theory and don't really know how to design what I need so let me describe it. I know you're good with circuits and I wasn't sure who else to approach.
I am needing a circuit to output two DC voltages of varying intensity. The load will be... well, okay. It's for a sex toy I'm trying to build. The load will consist of two standard egg/bullet type vibrators. Each channel would be independent. I need the voltage for each channel to randomly cycle up and down within a specified range.
Better still would be a way to pulse each channel in a somewhat random pattern, maybe with some sort of PWM controller. It would also, of course, need on/off capability and maybe a sort of power dial to limit the applied voltage manually. Max voltage required would probably be in the 3-4.5V range. Most of the eggs run off of 3V, but I've seem some that use three batteries as well. Some way to remote control the setup would be especially keen.
I hope you can help and please don't think I'm too much of a freak. :)
[From someone who's sent me letters before but this time, oddly enough, I'm not going to mention his name.]
I can't help you much with this, but you could probably get a suitably pseudorandom output from the summed output of two or more separate oscillators. You could use that to bias the drive to a conventional power supply (or, since the power levels here aren't very high, just throttle the output from a conventional power supply).
I'm thinking that this is yet another thing you can use 555 timers for, but I could be wrong - two or three simple flip-flop circuits (like this - note particularly appropriate old-fashioned name...) might be more suitable.
Anyhoo, since you don't need cryptographically strong randomness here, this sort of design would probably suffice.