Dan's Data letters #52
Publication date: July 2003.Last modified 23-Oct-2012.
Flashing from CD
I recently put together a new computer. The motherboard is an Abit IS7.
Abit just released a new BIOS, which you're supposed to copy to a bootable floppy so you can flash it to the BIOS. Except I just threw out the floppy on my old PC and I didn't bother to buy one for the new PC. I do have a CD burner, though.
Before I started experimenting, I figured I would ask you - can you flash the BIOS off of a CD-ROM, provided you change the boot order to CD before HD?
Tom
Answer:
I wrote an answer to this.
Then someone pointed me to this page, which makes things much simpler. You want the "Clean Bootable CD-Rom (for BIOS upgrade)" part.
Or, for an even simpler solution, you can flash the BIOS of various Abit boards from Windows, using this utility.
My completely redundant original answer follows. O the embarrassment.
You can, but you need a floppy to make the bootable CD from!
This problem should, however, be surmountable without finding a PC with a floppy drive and a CD writer to make your disk.
Any half-decent CD writing program these days can make bootable CDs. Regular "El Torito" bootable CDs contain a floppy disk image file of up to 2.78Mb in size (the actual size of a "2.88Mb" floppy, thanks to the usual 1000-versus-1024 confusion among storage device makers), and all you need to make one is a CD writing program and a bootable floppy disk image, or an actual bootable floppy (in this case, one with the BIOS flashing program and data file on it as well as the boot files) and a drive to read it.
To make a floppy image on a machine with no floppy drive, you'll need something that can emulate a floppy from a disk image, or a RAM-drive that does the same thing. FileDisk, from here, is a fairly simple command line utility that can emulate drives of all sizes and should work for this purpose; you'll just have to make it pretend to be a floppy drive. The easiest way to do that is to start point it to a floppy image on startup. You could use the RomFs image file from the same site, I think. Then you ought to be able to format that "drive" as a bootable floppy as normal, write stuff to it, and either use it directly as a source drive for making a bootable floppy or, if your CD writing software doesn't understand that a "floppy" drive can be something other than A: or B:, just use its image file. Or, if you want to be kinky, use some other image file creation utility to create an image file from a "drive" that's actually an image file already.
Here's another possible disk image/emulation option. I don't know how happy it'll be on XP, though. This one ought to work.
Here's a decent guide to making bootable CDs.
Cramped computing
I'm installing a computer in some small spaces in a boat, and my mission is to mount my DVD+RW/+R drive remotely from the case.
What is the distance limitation with IDE?
There seems to be conflicting information, one source says 18" (circa 1998), yet 36" cables are produced now. I need at least 6 feet of cable.
Richard
Answer:
Sorry, but you can't have it. You might be able to manage three feet, but
that's already well outside the spec.
I talk about this in my cable review here.
Single-ended Fast or Ultra SCSI (what you'll be using, if you buy consumer SCSI gear) allows a three metre cable length. To get three metres with Ultra SCSI you need to have fewer than five active IDs (one of which will be taken up by the controller card), but that won't be a problem for you if you only want to connect one or two drives.
SCSI DVD writers are thrillingly expensive, though, so it'd be a better idea to use USB 2.0 (five metre cables) or FireWire (four metres). You can get external boxes with USB and/or FireWire sockets that accept IDE devices, but CD and DVD writers may or may not work properly in them; it's a good idea to make very sure that the combination you're considering will work before you buy it. Pre-built external FireWire and USB optical drives also very probably use IDE drives, and cost more, but are much more likely to actually perform as advertised.
Another option - you could use Small Form Factor PC components and see if you can get the system board close enough to the drives, despite the cramped confines of the boat, that IDE cables will be fine after all.
Gameless desperation
I recently bought an Albatron KX400-8XV motherboard, and 512Mb PC2700 DDR RAM to go with it, with my existing Gainward Geforce3 Ti200 128Mb card and 1GHz Duron.
Now, after reinstalling windows, it works OK... EXCEPT.... Anything 3D-ish. OpenGL tends to crash or reboot the machine, Direct3D enjoys bluescreening or dumping me back to the desktop. I've tried a new power supply (old one was 300W, new one 350W - real watts too I think).
Curiously, the DirectDraw/Direct3D tests from dxdiag don't crash.
I have tested the RAM with Memtest for a few hours, no errors there. I've tried new and old 3D drivers, new and old motherboard drivers, I even bumped the AGP voltage up to 1.6V, no help there. Even software rendered mode in Half-Life gives me some funky purple textured squished image (it's the actual game, you can still run around), before it crashes.
I'm at my wits' end. What could it be? If I can't think of anything more, I'm going to return the motherboard and swap it for a replacement, but I'd like to avoid that.
Lord Booga
Answer:
I can't suggest anything more elegant than shotgun debugging, but you haven't
ruled out the video card itself as a suspect. You could have inadvertently
static-zapped it in the process of transferring it to the new board. The
screwed up display in software mode points to a video card issue.
If you can lay your hands on some other 3D-capable card for an afternoon (preferably something power-hungry - an original GeForce would be great!), try swapping it in and see if the problem's still there. If it is, then I'd guess it's indeed the motherboard at fault, since you've tried swapping PSUs already.
If the problem goes away with a different video card, then it could still be the motherboard just not liking your GeForce3, but I doubt it.
Battery balderdash
I've recently acquired myself a RC car (Tamiya TL-01 based, kinda crap, I know, but gonna be replacing it for a Yokomo MR4TC soon, now that I'm more comfortable with the running of these machines).
Anyway, I've been doing a lot of reading on the matter, and a few things are confusing me. At www.balakracing.com/batteryfaqs.htm , there is this statement:
"Can I charge the same pack more than once a day? Here is where the NiCD and NiMH cells differ again. Most people will tell you that you should not use your NiCD packs more than once a day."
Is this true or just a ploy to make you buy more batteries?
Secondly, seeing as I can get Sanyo 2100mAh AA NiMH cells for a very good price, would there be any major disadvantages to creating a 7.2V battery out of those? The only thing I can think about is that they might not be able to provide a high current as compared to sub-C cells, but why not make a battery pack out of six pairs (in parallel) of AA cells (ending up with a battery of about 4200mAh), for a weight and size similar to that of a normal set of sub Cs? I've given it a lot of thought and it ought to be feasible (whether it is actually allowed in a tournament would be a different story, but for everyday running it should not be a problem). What do you think?
Thomas
Answer:
Only charging batteries once a day makes some amount of sense for people
who've dropped big bucks on fancy high-capacity matched R/C hobby packs
and want them to last as long as possible, despite discharging them in four
or five minutes in a modified-motor R/C car or even faster in a plane or
heli.
Restricting recharges to once a day is a bit excessive, but the packs will come out of the toy piping hot (pretty warm, even if you're using a stock-class motor), and should be allowed to cool before you recharge them. If you've got no option, though, then recharging a still-warm pack will only shorten its life, not instantly kill it.
I'd ignore anything this Balak guy says about batteries. I started reading his page, but stopped pretty quickly, when he said that "NiCD cells use Nickel Cadmium as the electrolyte and NiMH cells use Nickel Metal Hydride". He links to a dictionary definition of "electrolyte", but hasn't bothered to actually wonder how this supposedly solid metallic electrolyte is doing anything in the batteries.
(He's updated the page now and fixed that error; the old version's here, for those foolish enough to doubt me. The rest of the page contains a mixture of sensible advice and "common knowledge" among R/Cers, including the mysterious alleged benefits of reverse pulse charging; many R/C racers say that the super-hard discharges of R/C use do something that the battery manufacturers never thought of that makes reverse pulse and various other voodoo useful, but I'm inclined to think that the manufacturers might have noticed it by now, after decades, if it were true.)
NiCd cells have, of course, nickel and cadmium electrodes; the positive electrode is nickel, and the negative one is cadmium. NiMH cells also have a nickel positive electrode, but the negative electrode is a complicated alloy of manganese, nickel, magnesium, cobalt, aluminium and secret rare earth herbs and spices. Both kinds of cells have an alkaline electrolyte.
Your brilliant idea about using NiMH AAs in place of NiMH or NiCd sub-Cs for hobby applications won't work. The little cells would drop dead very quickly indeed. You might be able to get one or two good launches out of them before they got too hot to deliver respectable peak current.
Even with two packs in parallel, you'd have nothing like the five-amp-average, 20-amp-peak current capacity that even a basic "sport" class NiCd stick pack can easily deliver. Sub-C NiMH cells made for R/C use (Sanyo's RC3000 cells, for instance) have much better high current performance than regular cheap NiMH cells.
PS/2 power
I just wondered if you knew how much current capacity a PS/2 port has, either per port or between two?
Matt
Answer:
In theory, I think, 100mA at 5V per port. In practice, it can be considerably
higher; some motherboards just connect the 5V PSU rail directly to the PS/2
5V Vcc pin.
Here's a page about the PS/2 spec, though it has little to say about current capacity.
Heater wars
I'm interested in hearing your opinion on some of the things mentioned in this thread in the overclockers.com.au forums.
Quote:
"If you know anything about resistors also you will know it is not as accurate as you think because you need it to travel the heat in a particular way, shape and direction as to exactly match that of , say, an AMD core. Heat is actually produced unbalanced across the core. This actually matches the unbalanced nature of the core thus giving more accuracy than anything produced from a web site to date."
It seems to me that how the heat load is distributed across a small area would mean bugger all to the performance of the sink.
Quote:
"Another observation we have made is on some simulators (even our own) the heatsink may
1. Perform great on the simulator and poor on the processor.
2. Perform poor on the simulator but great on the processor."I can't see how this could be possible, provided that the heat loads are similar.
Chris
Answer:
The conductivity of the heat sink base matters if it's having to distribute
heat injected into it in a really tiny spot, but this isn't that
much of an issue - especially not with CPUs that have integrated heat spreaders,
like P4s and the upcoming 64 bit Athlons.
If you want to perfectly model a CPU, no, you can't do it with a resistive heater. If you want to make something that lets you compare CPU coolers better than any real CPU will, though, then a resistive heater is the way to go. It's not perfect, but like democracy and the scientific method, it's the best option available. CPUs just aren't consistent heat sources, and if you test dozens of heat sinks on the one motherboard, you'll need a new motherboard soon enough (or a surface mount rework station, to replace the CPU socket).
And then bingo - your CPU voltage changes slightly, because seldom do two motherboards deliver exactly the same voltage from a given Vcore setting, and everything gets even fuzzier.
I'm inclined to think that oddities in CPU performance when tested on real CPUs and different heaters arise when people screw up the installation of the cooler on the simulator, or, possibly, on the CPU. I do not consider myself immune from such mistakes, though I do multiple test runs and regularly re-check with previously tested coolers to try to make sure nothing wiggy is happening. I've tested so darn many coolers, though, that I wouldn't bet my life that I haven't screwed up royally on one of them!
Another possibility is that someone's simulator may, like mine, have its heater top higher, or lower, than the top of a real CPU. This difference will affect the strength of the clip clamping; a large positive clamping force difference will also probably cause the clip to "take a set" and not hold as strongly when transferred from the over-tall heater to a real CPU, thereby making it seem that the heater overstated its performance. Which it did, somewhat, but it also damaged the cooler so that it'd never work as well again on a proper CPU socket. Clip strength can make a quite large difference to performance.
(It should also be noted that 5% is considered "quite large" by many CPU cooler reviewers. :-)
My CPU simulating heater - which isn't nearly as fancy as the one being raved about in the OCAU forum thread - is indeed rather harder to clip a cooler onto than is a real CPU; it's not a clip-wrecker, but it does make them clamp a bit harder than they normally would.
Card shifting
Memory cards is a new business line for my company. What are your thoughts on the memory card market? Is it price price price?
Cris
Answer:
Yes, memory cards - in popular formats, at least - are pretty much just
a commodity these days. Digital cameras and USB 2.0 are now fast enough
that regular consumers have gear that can tell the difference between a
slow card and a fast one, but the fast cards aren't necessarily the expensive
ones. Sandisk's top-priced cards have,
for some time, been miserably slow - I think their current Super Ultra Hyper
We Really Mean It models are OK, but the dirt cheap Ridata cards that're
easy to buy here in Australia
are just as good, and so are the similarly off-brand cards that Aus PC Market
stock.
If you haven't read this, you may find it interesting.
Goop removal
I need to clean some Arctic Silver off an Intel 875 Northbridge Chip. Is a combination of acetone and denatured alcohol safe to use on it?
Jacek
Answer:
Yes, it should be fine, but I'm not a fan of acetone in general. It's very
bad for you (both by inhalation and by absorption through the skin; your
liver gets to clean it out of your bloodstream, and does not enjoy that
job), and it's also just more ferocious than is necessary for the task.
Why use something that'll eat most plastics if it falls on them, when there
are less horrible solvents that work just as well for cleaning CPUs and
heat sinks?
My favourite solvent for this task is ordinary naphtha lighter fluid ("Zippo" brand, or any other), available from any tobacconist. It doesn't have to be the brand name stuff; anything in one of those little steel bottles with the flip-up nozzle on the top will do.
Naphtha still isn't what you'd call non-toxic, of course, but it's less toxic than acetone, an excellent cleaner, pretty cheap, safe to use on most plastics, and also useful for lighting barbecues.
(OK, technically, you can also light a barbecue with acetone. Make sure to use lots, and have someone videotape the event.)
It should also be noted that if someone's "acetone" is actually just "nail polish remover containing acetone", then it'll probably have oil and fragrance (as if there's something you can add to acetone that'll actually make it smell nice...) and other stuff in there, which'll spoil a clean-up job.
Glowing heads
Have you ever tested any caving headlamps? My friend has a Spe-Lu 26 LED that is blinding for 8 hours, while my Petzl Duo 24 LED is crap. Both cost $US300.00. Is there anything out there brighter than the Spe-Lu?
John
Answer:
I presume your Petzl light's fitted with an after-market 24 LED lamp, like
the ones on show at here?
If it's still running from the little AA-cell battery pack, then yes, it'll die pretty fast. It'll do better with a C-cell pack (which is an option, I understand), particularly if you use rechargeables.
The Spe-Lu light has a 3Ah 7.2V NiMH pack (which makes me think they're using six cell R/C model car stick packs, something worth bearing in mind if your friend wants some extra batteries cheap), which has rather more capacity.
I don't know of any production LED headlamps that're brighter than the Spe-Lu, at the moment.
The Spe-Lu's 26 LEDs are only running at about 150 milliwatts each when it's in its four watt maximum brightness mode, so it's not beating them to death; a light with two or three white Luxeon Stars in it would be brighter and a bit more efficient, but I don't think anyone makes such a thing. Hobbyists have made all sorts of giant LED arrays, but none of them have made it into a commercial product yet.
WOMP-tisha-WOMP-tisha
I'm in the process of shopping for a subwoofer for my car (a reasonably small sub to boost the low frequencies over the road noise in my car, not something that's going to annoy other drivers by boosting my low frequencies over the road noise in their cars). Anyhow, while shopping around, I've been told a couple of things that I'm not sure about.
First, why does a sealed box need to be smaller than a ported box? What happens if a sealed box is too big? I usually prefer the less-boomy sound of a sealed box, and I've already got a box, but it's quite large and I'm not sure if I should put a port in it or just get a really big subwoofer (or two) - but then I'm contradicting the intent I weakly stated above. (Aw hell, it probably doesn't matter, because my wife and her family are going to call me a wanker for wasting money on a subwoofer anyway. One strong motive for trying to put a smaller subwoofer in the big box is that it will cost a lot less. At least then the level of wanker-ness will be lower, because I won't have wasted so much money.)
Second, what happens when you try to drive a subwoofer with an amp that has a lower power rating than the subwoofer? I've read that you can ruin the tweeters in 2- and 3-way speakers by driving an underpowered amp too hard, because the amp clips the low end and sends too much power to the high end. But a subwoofer has only the one big driver. Wouldn't using a weaker amp simply give you less sound power output? And what happens when you overextend the amp?
Derek
Answer:
A sealed box doesn't need to be smaller than a ported box. Actually, sealed
boxes need to be bigger, generally speaking, for a given bass response,
at least when you're talking about normal speakers that don't get driven
around by someone.
Ported boxes can be tuned for a hump in their response at some reasonably low frequency, below which their bass response drops off like a cliff. Sealed boxes tail off more gently and naturally as the frequency drops, but the flipside of this is that to get the -3db point in the response curve (where useful bass response pretty much stops) as low with a sealed box as you can with a ported one, you have to make the sealed box bigger.
Things get weird when you're installing subs in cars, though, because you're effectively making a complex multi-ported bandpass enclosure. The car itself works as another enclosure around the subwoofer box, and the box isn't necessarily even a box, as such; if you install speakers on the parcel shelf or behind the rear seat or in some other similar situation where they've got the car's boot (trunk) behind them, they'll be in an "infinite baffle" setup where you're aiming to make the installation feel to the speaker as if it's mounted in a hole in a board of infinite size.
Some of the drivers used in car audio are very peculiar. Their impedance may be ultra-low (to get maximum power out of amplifiers whose output voltage is limited by their DC-to-DC converters), and they may have other parameters that are very different from those of home audio drivers. Some subs work fine in lots of different applications - like the one used in this sub, for instance - but some are very specialised.
The weirdness of some automotive subs is generally the result of the limited maximum box size people can use in cars, and the limited care factor of many boom car owners for fidelity, as opposed to bulk bass.
If you're building a sealed box for use with a driver that's suitable for infinite baffle installation, the box can't be too big. As the size of an infinite baffle enclosure approaches infinity, its effect on the driver installed in it approaches zero. This is good, provided your driver's expecting it; loosely sprung drivers that're expecting to be installed in a small sealed box ("acoustic suspension") will not be happy with infinite baffle.
In infinite baffle, the bigger the box is, the less it'll restrict the driver's motion, and the closer your bass response will be to the driver's own resonant frequency, below which you'll never have much response. In acoustic suspension, the springiness of the air inside the box is necessary to restrict the driver's motion. It depends on the driver.
There are diminishing returns involved here, though. Take an unremarkable 12 inch bass driver and install it in an infinite baffle box the size of a refrigerator and you'll probably have a speaker with a resonant frequency a few Hertz above the driver's own resonance. Switch to a sealed box the size of a house, and you'll only shift the -3dB point a couple of Hertz lower. Start with a (probably more expensive) driver that wants a small acoustic suspension enclosure, and you may get better bass response from a box that you couldn't fit a 17 inch monitor inside.
Your question about what driver to put in your box cannot be easily answered, if you're thinking of making it into a ported enclosure. To do that, you need to know the Thiele-Small parameters of the driver(s) you're considering, the box volume, and the bass response you're looking for. Then you have to do Equations.
No such mathematical exertion is needed when designing an infinite baffle system. You just pick a driver that's OK for that kind of box, then you make the box as big as you can, and you're done.
Acoustic suspension is a bit more complex, but not a lot; it's not nearly as difficult to design as a ported enclosure, and there's a range of allowable enclosure sizes for a given driver.
Driving a sub with an underpowered amp is perfectly fine, as long as you don't wind the amp up too far, cause it to emit lots of nasty DC, and fry your driver. AC makes a speaker act like a motor; DC makes a speaker act like a toaster. And yes, tweeters will fry first, followed by midranges, followed by woofers and subs, but you still don't want to try it with a sub.
This is not something to worry about unless persons in a state of extreme musical apathy get control of the volume knob. Amps driven into clipping (which is what you have to do to get significant DC in the output signal) sound absolutely awful, and in all but the most outrageously terrible cases (volume all the way up, bass knob all the way up, treble knob all the way up, loudness button pressed, all graphic equaliser sliders all the way up, "Billy Thorpe and the Aztecs Live on PCP" in the CD player) you'll get at least a few minutes of horrible sounding music before you do permanent damage. Even a car sub should sound bad enough when driven from a clipping amp that you'll know you've gone too far.
