Atomic I/O letters column #95
Originally published 2008, in Atomic: Maximum Power ComputingReprinted here June 21, 2009 Last modified 16-Jan-2015.
Enclosure d'Elegance
I'm about to take the plunge and buy a Lian Li PC-P80 case (I know it's excessive, but I feel the need). I don't intend to do overclocking or water cooling. Someone suggested that I might be better off buying a server case for this amount of money ("how about a proper server case where the HDDs are hot-swappable and in a separate chamber from the m/b?"). I find the idea of hot-swap drives appealing, but from what I've seen so far those cases are very expensive.
Do you agree that a server case would be a better choice? Do you know of any server cases that are in the same price bracket and would be a better buy than the PC-80?
John
Answer:
If you get a case with hot-swappable drive bays, you'd better buy a motherboard or controller
card that supports them, or you're going to get a nasty surprise when you pull one out.
(A lot of mobos now come with SATA controllers that do support hot-swap. You just have to avoid pulling drives that contain swap files, or your operating system.)
You can get perfectly good server cases with the same basic specs as the basic black Lian Li "Armorsuit" PC-P80 for less than the $AU500 to $AU550 (or less than $US400, in the States) that the Armorsuit costs. They won't be all aluminium and smooth and nice, though; you really do get something extra for your money when you buy the P80.
Those server cases won't necessarily come with hot-swap drive bays, either. The usual server version of these is a caddy arrangement where you need one caddy per drive you want to swap. SATA drives make it easier, since the standard SATA connectors are physically well-suited to easy plugging and unplugging without any extra hardware, and you can get 5.25-inch-bay and outboard USB doodads that let you use bare SATA drives like giant floppy disks. This sort of setup is not robust, though; if you knock a drive off a desk, or possibly even if you just don't take proper anti-static precautions, you can lose a terabyte of data in a hundredth of a second.
There are all sorts of server cases with umpteen SATA and/or SCSI caddies built in, but they can cost quite a bit more than the PC-P80, and normal computer stores don't stock them.
(There more options at the $AU800-odd price point of the PC-P80R, the bright red Armorsuit with "ATI Cross Fire X" cut into the side panel. That's right - advertising ATI on your case will cost you $300 more than the price of the ad-free Armorsuit!)
If you're actually going to be swapping hard drives regularly then one or another hot-swap solution is a great idea, but as I mention in this column, it's now very easy to add a swappable bay or three to any normal case. All of the other stuff in server cases - dozens of drive bays, space for more than one motherboard, dual redundant PSUs - is unnecessary for desktop PCs (and many servers, actually). All that stuff just gives you a computer that'll herniate you if you have to lift it, and that'll probably consume more power too (for extra fans, and for dual PSUs if one of them's just ticking over waiting for the first one to die).
There's a lot to be said for 19-inch rack server hardware if you want to install a bunch of stuff in your computer room - normal PCs, networking gear, music production gear - and keep it tidy.
Again, though, this is pointless for most users.
Solid-state fan!
I thought Inventgeek's Ion Cooled Computer might interest you.
Russell
Answer:
When that story first popped up and got mentioned on
Slashdot
and such, there was considerable criticism based on the bit at the end where the inventor
just arbitrarily decided that his ionic blower moves 325 cubic feet per minute, yet
makes no noise.
But the chap responsible should be congratulated for sticking with the idea and improving it. He made a version 2.0, and a CPU cooler, too.
(Russell actually sent me this letter a million years ago, when only the first project had been completed. I've updated my e-mail reply from back then to take the new projects into account.)
Inventgeek tested at least the CPU cooler properly. It didn't work as well as the stock fan, but it did work well enough to be useful. Fair enough. Hats off to the guy for making a no-moving-parts air-shifter that really does, to some extent, work.
Ionic air movement clearly isn't a great solution for PC cooling, or for many other applications. A big part of the reason why "Ionic Breeze"-type air filters, famously, don't bloody work, is that they don't move very much air at all. That's why they're so quiet; you can't move 325cfm through a computer case, or any other small housing, without making noise.
A good home vacuum cleaner will have a free-air flow rate of only about 90CFM. It makes a lot more noise than a computer fan with a similar CFM rating because of its high-suction motor, but imagine extending the hose far enough that you couldn't hear the motor any more, then hooking it up to the back of a computer case. Do you reckon you might hear just a little wind noise - not to mention the rattle of computer components going down the tube - if you jacked even that air-flow rate up by a factor of more than three?
(There are plenty of electrostatic air cleaners that do actually work, by the way. Their electrostatic components are just there to trap dust; they have conventional fans to move air through them.)
A computer that's cooled by an electrostatic blower is still a neat contraption, but the blower's probably not going to give you a lot more air-flow than you'd get from completely passive convective cooling in a suitably designed case (or if you just left the side off the enclosure...). I'd be nervous about using any such machine if the ambient temperature was high. I'd also be annoyed about how often I had to clean off the fans and grille inside - electrostatic precipitation will stick every incoming dust particle to the fins and grille, and thus block the grille up quite quickly.
Electrostatic air movement does have applications on very small scales, where you're trying to cool tiny integrated-circuit components and want to be able to fit the air-mover on the chip too.
In the macroscopic world, though, boring old fans work way better.
(Given the first letter on this page, by the way, I know somebody's about to e-mail me about how an ionic air mover could charge up the computer it's blowing air into, and thereby create an electrostatic-discharge death trap for many pieces of hardware. Fortunately, even if you make a blower that lets plenty of charge pass by the grilles that give the air its impetus, the grounded chassis of a PC makes it quite hard to electrostatically charge it. You could probably do it with one of those oddball all-plastic see-through cases, but in general electrostatic damage to assembled PCs happens when a charged person touches the grounded computer, not the other way around.)
De-blinging
I really want to buy a Thermaltake A2309 iCage for mounting 3.5-inch drives in 5.25-inch bays.
However, they bling-ified it with an LED fan.
Is there some way to disable the LED in the fan without destroying the fan or causing a fire hazard?
William
Answer:
The fan in the iCage, judging by the pictures on
the product
page and in the OCAU
review, looks like the kind that has LEDs wired up around the edge of the fan under
some tape:
It's easy to snip the whole LED assembly off one of those fans, or at least cut the wires that feed them. Make sure the cut wires are insulated and not touching, and you're done.
Wall wart or black art?
I've had occasion to play around with several laptops, sometimes having to jury-rig or swap power adapters for them. I always wondered how the power adapters are specced. I would believe it would be the second of these two hypotheses, but I would like your opinion.
A: Their output power corresponds to the worst-case scenario power requirements for the components of the most powerful laptop in the series for which the adapter is built. So for example if the top-of-the-line ThinkPad 770 series with DVD and hard drive running screen at full brightness draws 60W, the 770-series adapter would be, say, 65W. Because manufacturers prefer to ship many different models with a single adapter model, they usually have a large headroom.
B: The most draining component might actually be the battery as it recharges. Because battery voltages are by laws of physics always a multiple of a fixed number depending on their type (3.6 for lithium ion, I guess), there are actually few different adapter specs out there. The only difference is the amperage, which would govern the time it takes to charge the battery (something consumers rarely bother about), depending on the capacity.
I would tend to believe this second version, because laptops get hotter when charging and I know using an underpowered adapter usually results in a slow charging or barely charging battery.
Manufacturers and resellers worldwide like to issue dire warnings about using replacement adapters, and often play the "fancy adapter plug" game (in which Apple seems to be a winner). Am I wrong in saying that if B is right then it's typical marketing crap and that basically all laptops using a given battery voltage (which is to say a couple generations of models) could share a common adapter?
(That is, in the universe where razor blades are cheap and all oil filters are the same.)
Florian
Answer:
Yes, power adapters are specced to cover the highest-powered laptop in the line working
as hard as it can, plus overhead to deal with battery charging and extras like
all of the USB ports outputting their full 2.5 watts. (Actually, many USB ports today
can deliver more than the specified half-amp at five volts, in order to work
with similarly out-of-spec high-power-draw peripherals.)
Battery charging power can be quite substantial, though. If it only takes a couple of hours to more or less fully charge a 75 watt-hour battery, then the charge power pretty much has to be up around 50 watts (not the 37.5 simple mathematics gives you, because considerable power is lost in the charging process).
Manufacturers often change the power plug specifically to stop people accidentally plugging old power supplies into new laptops and overloading the PSU, but no, they don't always do it for that reason.
You can tell, because it's generally quite safe to use off-brand one-size-fits-several mains (or car) laptop power supplies, as long as they come with the right plug for your laptop and their ratings match (or for current, exceed) the ones on the stock power adapter's sticker. I'm sure there are some cheap Taiwanese mains or car adapters that blow up laptops, themselves, or both, but there certainly doesn't seem to be a plague of them.
That's mainly because modern laptop power adapters really are just giant wall-warts. The bulk of the regulating and all of the charging hardware is built into the laptop, not the power supply. In the olden days external power bricks for portable (and some less-portable) computers commonly had multiple-voltage output and sometimes even weirder stuff like clock signals, but modern laptop adapters are just simple high-current one-voltage DC devices.
The only real complexity in modern laptop PSUs comes from their switchmode circuitry - thank goodness, or they'd all weigh five kilos. Switchmode power supplies were fairly black magic 20 years ago, but are thoroughly understood by even the dodgiest Chinese manufacturers today.
(Some of those manufacturers make awful PSUs that don't come anywhere near meeting their specifications, but even those manufacturers all know how to make a proper PSU - they just choose not to!)
