Dan's Data letters #31
Publication date: 23-Feb-2003.Last modified 03-Dec-2011.
XP flavours
WinXP Home, I have heard, cannot be client/server networked. For that you need XP Pro. Having not yet worked seriously with anything beyond 2K, I dunno whether this is true, though being Micro$oft it would make sense. I need to access DHCP on a W2K server, POP3, Web, Database, print and file servers.
Also, some late model PC133 SDRAM is apparently notso-hotso with older (PC100 and PC66) chipsets. I was under the impression that the standards were backward-compatible, but I read about single-sided SDRAM good for only P4s and late P3s, whereas double-sided SDRAM is better for older systems. Seeing as I'm a lazy b*stard and RAM specs are MEGO reading matter, could you, if possible, please clarify a little bit for me?
Pat
Answer:
The Home version of XP can't log on to
Active Directory domains, and you can't use various Microsoft network
management stuff on XP Home boxes. So if you plug a Home box into a corporate
network, the admin won't be able to use Microsoft's central software management
features - though some other system would work.
Home also lacks a selection of other networking buzzword-stuff - an IPsec interface, Simple Network Management Protocol, "Simple TCP/IP services", an SAP database agent, the NetWare network client, Network Monitor, and the Multiple Roaming feature. This is all stuff that corporate networks often like to have on their client machines, but which isn't relevant for ordinary LAN tasks.
An XP Home machine can connect to all of the server types you list - file servers, print servers, mail servers, Internet servers, whatever. It's just the Microsoft-brand central admin stuff that it can't do, and XP Pro can.
With regard to your RAM question - yes, some older boards don't like single sided RAM. Actually, what they don't like is single banked RAM - RAM that has all of its chips electrically arranged as one bank. This often means that the chips will all be physically on one side of the RAM module, but that's not always the case.
Problems are more likely to arise if you install different kinds of memory modules, though. If all of your RAM is of the same type, it's more likely to work.
More LED twiddling
I'm in the process of designing some mountain bike lighting, and while trying to get the most battery life possible, and have ended up with a couple of simple questions that have been tricky to find rock solid answers to.
First off, if a variable voltage regulator circuit (based on the LM317T) is set to supply, say, 6 volts, and the circuit is connected to a 12V power source, will the circuit draw only half as much current from its 12V source as the 6V device is drawing from the circuit? If not, then roughly how efficient can I expect the regulator circuit to be?
Also, this raises a similar question to do with LEDs. If an LED is connected to a higher voltage supply with the correct resistor, say a 4V 30mA white LED running from 12V, is it going to be highly efficient, and draw little more than 10mA (which I doubt) or is it still going to draw 30mA and just give off the excess energy (two thirds of it) as heat?
If all of the excess energy is being dissipated as heat from the resistor, then wouldn't the required resistor-wattage increase (perhaps linearly?) as the difference between the LED's forward voltage and the supply voltage increased? If so, then how would the resistor wattage be calculated?
Also, using the formula [Resistance = (DC Supply - LED forward voltage) x 1000/LED current in mA] with a supply voltage very close to the LED's forward voltage, say within 0.1 volts or even less, you end up with a very low resistor value, like less than 5 ohms. This could be minimising the energy wasted from the resistor as heat quite significantly.
So long as I select a resistor value slightly higher to allow for grunty batteries, can I use a very low value resistor (with a suitable lower supply voltage to match) without a higher risk of thermal runaway than if the supply voltage and resistor value, were higher?
Matt
Answer:
The LM317 devices are plain old series regulators; they just dump what they
drop as heat. Series regulators behave like a little magic goblin swapping
series resistors in and out of the circuit as needed to keep the voltage
across the load to whatever you requested. If the circuit current is 100mA,
and input is 12V, and the regulator's set to 6V, then the regulator will
dissipate the same 600mW as the load. DC-to-DC converters are much more
complex devices, but they can do this job efficiently; plain regulators
can't.
This does indeed mean that it pays to have a supply voltage fairly close to whatever load voltage you want. It's not too hard to pull that off, with series/parallel LED arrays. Using a high current capacity power supply (non-rechargeable batteries have considerable internal resistance, and will limit current somewhat all by themselves; common rechargeable chemistries have low internal resistance, and are more dangerous) with a very small series resistance is not a great idea, because as resistor value approaches zero, the usefulness of the resistor in preventing runaway also approaches zero.
In the real world, dropping only a small fraction of the supply voltage is likely to be just fine. How small "a small fraction" can be though, is open to debate.
The "four volt, 30mA" rating for your LED is just a spec-sheet figure; the LED's behaviour will vary with its temperature. If you're using a simple series resistor, you have to pick a resistor value which will limit the circuit current appropriately even when the LED is trying really hard to go into thermal runaway. At any given second, the LED behaves like a resistor - but its value changes constantly.
To make the circuit work, you can either err on the side of caution all the way down your back-of-an-envelope calculations, which is the way most of us do it; that gives you less light than you might otherwise have, but doesn't abuse the LED(s). Or you can muck about with alligator clip leads and a couple of multimeters and a handful of resistors, and see how the LED actually behaves, and try to push the envelope and get everything you can out of it.
Since LED efficiency falls as output power increases, the visible difference between the results of these two strategies can be depressingly small.
As you guessed, you'll waste more and more power as your resistor (or regulator) drops more and more volts, at a given current. Fortunately, though, LEDs do not thermally run away forever, unless they have no in-line resistance; as the LED warms up, the voltage across it will fall and the circuit current will thus rise, but it takes more and more power to warm the LED up more as it gets hotter and hotter compared with the ambient temperature. The unchanging inline resistance is there to stop the LED's impedance change from being sufficient to keep the vicious circle going.
The back-of-an-envelope way to calculate resistor wattage is to take the volts it's dropping times the current flow, as per normal, and then add as large a fudge factor on top to allow for LED runaway as feels good to you. Or, of course, you can do a proper test using a couple of meters and, preferably, a bench power supply.
Hijacked boot
I have a Asus P4B533-E, which has a onboard Promise FastTrak Lite RAID controller. Attached to this are dual Western Digital 400BB hard drives in an array. This board also has an onboard ATA133 controller which I have connected to Zip drives, DVD-ROM and a CD writer. Do you know of any controller card that I can add to the system so I can add more devices (Sony DRU500AX and WD1200JB?). I want more space for video editing.
I have tried an ATA100 controller card, and this installs OK until I connect a HDD, and then the system wants to boot from that disk. Are there any controller cards available on which you can control the BIOS (I do not need to boot from this device) and work with ATAPI devices like DVD writers?
Greg
Answer:
The problem people usually have in this situation is that they can't
boot from a newly installed controller card, no matter how hard they try.
Your problem is backwards.
Any ordinary add-on IDE host adapter card should work fine with any drives you like, and many of them do have an accessible setup BIOS, but I think the BIOS setup you should look at is your motherboard's, not the controller card's.
In this case, what you apparently have to do is the normal press-delete-on-startup routine to get into the BIOS setup, then select "SCSI" as the first boot device, since the motherboard believes its IDE RAID controller is a SCSI device. You may or may not need to set the other boot devices in the list to "no device".
Misplaced probe?
Being in the market for a new CPU cooler, I of course checked out your excellent cooler comparison, and of the units on offer I have decided on the Thermaltake Volcano 9.
In my travels I happened to peruse the Thermaltake website's relevant section and noticed what seems to be an incongruity with their information, here.
In the "Functional Description" in the section marked "Function 2", the first photo depicts what appears to be the fan speed temp sensor being fixed to the underside of a CPU. Now, unless I am mistaken, the leads from the probe would have to get in the way of the insertion into the ZIF socket.
I know this is only a minor thing, and most would never have intentions of trying to fit the sensor in this way, something I seem to remember you mentioning in your review, and I don't want to embarrass anyone at TT (or myself if I'm wrong!) but if this is indeed the case I thought you might like to know about it in case some folks may find it useful.
Stewart
Answer:
The probe wires are very thin; you can thread them between the pins and
the CPU will still go into the socket. It'll be slightly tilted because
there are sensor wires under it on one side, but the cooler clip lets the
heat sink tilt on one axis to compensate for that, so you shouldn't need
to muck around with shims. It's not the most elegant solution in the world,
but it does work.
Unlettered drive
I have just fitted a second hard drive (Fujitsu 13.4gb) to the double IDE cable that also feeds my CD-RW. The CD-RW is configured as slave and my second hard drive as master. Even though I have asked the BIOS to auto-detect the drive, which it has done, I have not been given a drive letter, thus it does not appear on my system for me to format. I guess I am missing something really simple.
Kevin
Answer:
It sounds as if you haven't partitioned it. I talk about how to do this,
using the FDISK program, here.
Magic sugar?
In your valiant efforts to rid the world of quacks, shysters and charlatans, have you ever come across a company called Mannatech?
A close relative recently showed us number of products (Ambrotose being the most common apparently). These claim to cure all the bodies ills using 8 different sugars, or something along those lines.
Of course it seems these products are only available in a multi-level marketing scam ... er I mean scheme, to protect their patents. What really got me suspicious were a number of phrases obviously recited by rote including such gems as "improves cell-to-cell communications" and "doctors cant keep track of all discoveries every year".
Of course these things are ridiculously expensive, so you are expected to recruit new customers to offset your own costs.
I thought you might be interested to hear of this as it appears to be doing the rounds in Oz at the moment.
Alex
Answer:
I've not personally had the pleasure of meeting Mannatech representatives,
but the company pops up like clockwork on the
Healthfraud
list.
And your relative might have said that Mannatech products cure all ills, but Mannatech certainly didn't. Oh, no. Absolutely not. Their products are nutritional in nature and not promoted for the treatment of any particular disease, don't you know. Their stuff helps you achieve optimal health in a general, overall, you-can-neither-sue-nor-regulate-us kind of way.
Since, as far as I know, Mannatech have never presented credible evidence to the contrary, I don't think there's any reason to believe "Ambrotose" doesn't contain anything that your body can't make for itself, or get from a normal diet.
Mannatech used to sell a diet aid that actually worked pretty well, because it contained ephedra. Ephedra works well for getting thin; so does methamphetamine, and for the same reason, and with much the same risks per kilogram lost, except of course for the risk of getting yo' ass arrested. Spend your days speeding and you won't eat much. Lots of people use mild stimulants for slimming with great effect; I'm not a drugs-are-bad-m'kay extremist who believes anyone who gets a bit tweaky in the course of their weight loss regimen is certain to have a heart attack tomorrow. But, in most countries at least, if you want to go that way then you can get no-brand ephedra/Ma Huang/whatever pills much more cheaply than from any trapezoid-marketing operation.
Now, I think Mannatech's slim-aid stuff is just vitamins and caffeine. Oh, sorry, not caffeine, "guarana"; active ingredient "guaranine". Guaranine, in case you're wondering, means "caffeine extracted from guarana berries instead of from coffee beans". Caffeine in tea leaves can, similarly, be called "theine", though people don't generally bother to do so. Anyhow, caffeine and vitamin slim-pills won't do anything that a No-Doz and a multivitamin pill won't. And you can get generic caffeine pills that're cheaper than No-Doz, too!
I might also venture the opinion that whoever's responsible for this should be taken out to the woodshed and soundly thrashed.
Mannatech, by the way, listed on the NASDAQ (symbol: MTEX) in 1999, not long after their little Darryl See embarrassment. The stock peaked (immediately...) at more than $US40, then started trundling downwards, as people realised that this was just another company whose product was the opportunity to sell products (more or less as per HerbaLife's model). They dribbled around below five bucks for years, but have recently fought their way back up to about $US10.
In the USA, the Georgia Council Against Health Fraud (archived page) and the associated National Council Against Health Fraud both have things to say about them, as does MLM Watch (a spin-off project from Quackwatch).
