Dan's Data letters #70Publication date: 3 November 2003.
Last modified 03-Dec-2011.
Okay. I get that SFF boxes are low-power computers, and as such don't really require much in the way of power. What I don't understand, however, is how Shuttle's XPCs get away with running cutting edge Athlons and P4s, along with whatever drives and uber video card you manage to stuff in, off something like a 200W power supply? I'd guess they feed less power to rails more commonly loaded by extra drives and crazy expansion, but that's just a guess. Any ideas?
Around two hundred watts really is all you need, if you know you're not going to have to deal with a phalanx of drives, extra expansion cards, 103 fans, blah blah.
Even overclocked, it's hard to find any PC CPU that draws much more than a hundred watts. 80 watts is a more realistic load for a top-end processor that's working hard.
The graphics card's the next biggest current-sucker; 75 watts peak for a high end GeForce FX, maybe 20 watts less for the hottest ATI cards.
Then there are the drives and the motherboard itself, of course; you can count on them to add up to maybe 30 watts more overall.
Add it all together, and you can see that the 200 and 220W PSUs that the XPCs use are running pretty close to capacity (in a stacked XPC; not in a Celeron-powered one using built-in graphics), but that's OK because there's really no way for the system to need more than the PSU can deliver.
A similarly low-rated PSU will be able to run a normal-sized PC with the same hardware. Lots of tweakers install PSUs with massive ratings, but that doesn't mean they ever actually use all of that power.
I have a small grayscale camera mounted on top of a robot. The video signal is then transmitted as a composite video signal to an old Creative Video Spigot card. This card is used to capture the footage and store it as bitmaps for image processing. I want to replace the composite video signal with a wireless system.
I'm going to take the video signal and pass it through an A/D 8 bit converter, then pass that signal through an 8 bit in serial out shift register. This serial data can then be fed into a transmitter. Once at the receiver, the data is acquired by reversing the previous technique (serial in 8 bit out, D/A, composite video). Is this sensible?
Video is just a waveform; you can digitise it and undigitise it like any other waveform, so yes, this would work. You'd need a sample rate well up into the megahertz, though; D2 video is eight or ten bit direct digitisation of a composite signal, and it's sampled at four times the subcarrier rate, which is 14.318MHz (for NTSC, anyway, I think).
Since 2.4GHz wireless video transmitters and receivers are commodity items these days, you might like to look into just grabbing a couple of little black boxes and a pair of suitable antennas and doing it that way instead.
Our DAT tape drive died (again). So, I started looking for backup options, and came up with the new LTO drives, which will hold 100Gb per tape (200Gb if you go the next level up). However, this is going to cost $6K Canadian.
This then led to the next idea. For $6K, I can buy 20 160GB hard drives. Or to be cheap, 10 drives for 1/2 the price. And backup software can be directed to use a disk file instead of tape. So if I can set up a bunch of disk drives as backup devices, preferably swappable, I have a fast, cheap(er) backup solution.
So, I'm looking for a way to swap a hard drive without shutting down the computer. And an easy way to swap the drives.
I suspect that some form of add-on USB drive case, with a bunch of quick change drive sliders might do the trick.
You could use hot-swap-compatible IDE drive cradles and a controller to match. Lots of cheap IDE RAID controllers support hot swap, if you use the right kind of cradles; their RAID functions aren't anything to get excited about, but you don't care about that anyway.
Or you could use a bunch of USB 2 or FireWire external drive boxes, as you say; they're not terribly expensive, they protect drives better than an ejected cradle does, and you can plug them into any PC, not just one with the cradle receiver installed.
Or, if you're happy to just store bare labeled drives, you could do a lot worse than check out Wiebetech's DriveDock products. They only take a second to attach to a drive, and are quite well suited to this application, provided you never plug them in clumsily and break pins off the back of the drive.
I bought a cheap OEM internal 6-in-1 memory card reader from Jaycar. It fits in a 3.5 inch bay and connects directly to the motherboard via a USB header.
However, the damn thing doesn't, work and after many an hour Googling for an answer as to why, the best answer I could come up with was that the pin alignment/configuration of the USB cable on the reader and the pin alignment/configuration of the USB header on my motherboard don't match up.
The computer is a Shuttle SN41G2, and although I have pin information for my mobo's USB headers, the 6-in-1 card reader didn't come with any manuals, so I have nothing to compare it to.
Also, I have an LG TV that has composite video inputs and a component (RGB) input.
However most of my devices only have S-Video as their best video output method.
Do you know if I can buy an adapter that will convert the RGB input on the back of my TV into an S-Video input?
My guess is that I would need to pay big bucks for a video signal transformer, but I m hoping something cheap exists out there.
Yes, there's a good chance that your USB problem does stem from a pinout mismatch. This is why many cases with front USB ports have cables for those ports that terminate in tiny single-pin plugs.
The only way to solve the problem (if this is in fact the problem, of course; you could just have a broken reader) is to chop up the connector and re-connect it properly. This is safe with data-only cables, but not with cables like USB, which has only four conductors, but one of 'em carries +5V and another one's a ground, and you don't want to get those wrong.
Given this problem, and the fact that chopping up the connector will obviously void your warranty, I'd just take the thing back and get a different one - maybe one that uses a nice simple loopback external-connector cable, like this.
Regarding your video question - yes, there are cheap(ish) S-Video-to-RGB converters, but there's not a lot of point to using them; low quality converters just aren't worth using unless there's no other way to connect the devices together. You'd be wiser to apply the price of a better quality converter to the purchase of a better TV.
When I turn on off a light or turn on a fan my digital TV Set Top Box's output blips for a while and then comes back Is this is a power thing or an antenna thing?
Antenna. The thing that switched created a bit of radio frequency noise - probably from a tiny electric arc - which munged a bit of the MPEG-2 data stream being decoded by the STB, and that created enough of a hiccup that the box needed a moment to pick up the pieces. The same thing happens when you play a damaged or incorrectly encoded MPEG file on your computer; a very small data error can create a quite noticeable artifact.
The same noise will also probably be visible on an analogue TV, but only as a tiny glitch; the analogue signal takes up a lot more space, but it doesn't depend on the last line having been received correctly for the next one to be displayed, so brief noise bursts cause much less damage.
My parents are building a house, and I've scored the job of "technical adviser" to work out a way to install (and have work) some security cameras and possibly a server cupboard.
It struck me that I could have the electrician install RJ45 sockets all over the place and use them as a cure-all. It would be easy to create weird cables such as USB-to-RJ45, and a corresponding cable to plug the front door cam into my USB port.
The whole scheme sounds fine to me, except for the possibility of somehow running too much current through one of the wires and causing a fire inside my walls. Would you have any idea of the current that would encourage a single UTP wire to turn into an incandescent death trap?
This is generally not a bad idea, but USB is not a good example. It's got a five metre cable length limit (if you don't use repeaters), and it doesn't matter what kind of cable you use - the limit is timing-related, not loss-related.
For various other reasonably undemanding applications, though - low current DC power, moderate quality analogue video and audio, various data applications including networking - your idea would be fine.
Expect Category 5 UTP cable to have 24AWG wire inside it; Cat6 is likely to give you 23AWG. Both of these gauges are pretty darn thin, which means significant resistance (and thus voltage drop), and relatively low current handling.
The approximate current capacity ("ampacity") figure for 24AWG is only 1.3 amps. This can vary substantially with the cooling the wire gets, though. For your purposes, I definitely wouldn't want to try pushing more than one amp even through 23AWG.
Regarding the laserdisc question in letters 69 - I believe the Constant Angular Velocity (CAV) laserdisc format actually gives a bit of an advantage (picture quality wise, not disc flipping wise) over the DVD format. I did a little digging, and it looks like it has slightly lower numbers of horizontal lines. When you factor in the fact that DVDs use a lossy compression algorithm against the laserdiscs' actual frame for frame recording, though, I think the CAV laserdisc pulls ahead in visual quality. That and it's always looked better to me, which is the real subjective test, I suppose.
Yes, laserdisc can look better than DVD. It's not fair to compare a top-grade laserdisc with a crappy DVD, of course; there were some early DVDs that looked really awful, thanks to poor (or no) human intervention in the compression process. The canonical example of awful compression is the system that looks at a tennis game, and decides that the players are big enough that they need to be displayed, but that the ball can be deleted.
DVD and laserdisc are quite different formats, even if they do both come on optical discs. DVD's digital; laserdisc is analogue. Laserdisc has higher vertical chrominance (colour) resolution; DVD has higher horizontal luminance (brightness) resolution. Horizontal resolution and luminance resolution are both more visible to the human eye and brain. This is why oblong-pixel video displays, like a lot of headset systems, have much higher horizontal than vertical resolution, and why pretty much every video format has higher luma than chroma resolution. So this works out as an overall resolution win for DVD, though the difference isn't huge.
Laserdisc has uncompressed PCM audio, which can sound better than lossily compressed MPEG2-Layer-Whatever; laserdisc also supports AC3 5.1 channel audio just like DVD, though DVD supports a (pretty trivially) higher AC3 bit rate. DVDs can have uncompressed audio, but not many of them do (DTS is lossy compression too; it's just less lossy than the usual MPEG or AC3 audio).
Even well-compressed DVDs can show artifacting, when awful things like high-contrast Venetian blinds and low-contrast cigarette smoke come along. Then again, laserdiscs have analogue chrominance noise issues, which vary in severity from disc to disc and player to player.
It's all a bit of a moot point, of course, since laserdisc ain't comin' back no matter how good it is. You're quite right, though; good laserdiscs on a decent player do, indeed, match good DVDs for video and audio quality.
The format revolution here isn't from laserdisc to DVD, though. Laserdisc was always a lukewarm, half-successful format that not many people bought. It hung on because it gave true home theatre enthusiasts in the pre-DVD days an option short of buying a 35mm projector.
The real revolution was from VHS to DVD, and the difference there is unquestionably chalk and cheese.
In my former employment I used Micro-Scope and found it to be decent, but since then I've found a better program that costs $US29.95 and lets you make unlimited floppy copies.
It boots off a proprietary OS just like Micro-Scope and has most, if not all, of the same features.
It's called #1-TuffTEST-Pro:
I have used Micro-Scope in the past, and was successful in creating an image of it so that extra copies could be made without using the "handy dandy I-only-make-10-copies-and-then-die" master floppy from Micro2000.
The program I used was WinImage 5.0. Apparently it is still available and is currently up to version 6.10, although I haven't tried it out at all.
I found it quite helpful in making extra copies for when the copy we did have putzed itself.
Also, as a comment on the usefulness of the program - it used to be good. MemTest86 now not only does a better job testing memory, but is many times faster. There were times when I was testing a system with 64MB of PC1xx RAM, and the test literally had to be left overnight to complete one pass - and then it was a "Pass" verdict, even though the RAM was obviously the culprit. MemTest will typically tell you within a couple of hours, which is helpful.
(Some other readers suggested other copying strategies, among them plain old DD in Linux, and CopyQM - including the original shareware versions distributed by the original-but-now-defunct creators, Sydex.)
Have you seen alexchiu.com? I'm sure you have better things to do with your time than read crap like this, but nevertheless, it may tickle you and lighten up your day at Alex's expense. Keep up the good work.
Alex is a net.loon from way back. I mention him in passing in my big magnets piece.
Regarding this column, I thought you might be interested in reading this thread. A user on my forum says he's seen the Adams Platform video technology running for real. I know this guy in real life, he works for a networking company of some description, but I seriously, seriously doubt the possibility of 1000:1 compression ratios, so I'm thinking he's been duped somehow. How, I don't know.
He's seen the Adams Platform working, and I've seen David Copperfield make the Statue of Liberty disappear.
The Adams nuts have had many years to just patent their damn algorithm. They haven't, because there is no algorithm.
I don't know how this witness has been duped. I also don't know how Penn and Teller do their double bullet catch trick.
(Well, I didn't until I did a Google search, anyway; if you want to spoil all the fun, look here.)
Just because you don't know how a trick is done doesn't mean it is not a trick, and you must have a very high opinion of yourself if you don't think you can be fooled. This doesn't mean people should grouch around all day being massively skeptical of everything suggested to them by everyone, but there is good reason to demand extraordinary evidence for extraordinary claims, especially when those claims are being made by people whose credibility is suspect for reasons other than their claims.
I've just this past May gotten a degree in mechanical engineering, and am now at grad school, doing a master's in aerospace engineering in Ottawa, Canada. I've been at it for 6 months, and I hate every goddamn second. I don't dare take any more student loans, so I'm getting $CA850 per month for the duration of my masters. Not enough. My question to you is: Should I holler "bugger all this for a lark!" and head back to beautiful Vancouver? Is being an engineer enough?
Also, in response to the question which is no doubt on your mind: "Well, why the hell not ask you?"
This could be my chance to break into the lucrative Tony Robbins market segment - telling people what they already know, for $25,000 an hour!
Hating school is not sufficient reason to give up (says the man who didn't take any mathematics in his last year of high school). Feeling reasonably confident that you're going to hate the career that school is aiming you at, on the other hand, is a sufficient reason. So is a similar confidence that you can make it through, but only at the cost of acquiring a serious booze/dope/Yu-Gi-Oh habit.
You've got a pretty marketable degree already; provided you can handle the pain of making only four times as much money as you're living on at the moment, you can probably find yourself a gig somewhere. A fun gig, even, if you're willing to stomach a starting salary of only three times your current income, for a 50 hour week.
I don't know how lousy the job market is for freshly minted engineers where you are, but very often there are jobs out there, even in awful markets, if you can handle making substantially less money than the high school careers advisor promised you.
("You're a mechanical engineer? Not any more! Now you're an X Prize rocket scientist, making almost $CA20,000 a year provided you pretty much live in our chilly warehouse assembly facility and don't sue us about any H2O2 burns!")
At this point, let's both pretend I wrote an uplifting but unhelpful paragraph saying find-a-job-you-like-and-you'll-never-work-a-day-in-your-life, and a-job-well-done-is-its-own-reward, and some-of-the-best-people-never-figure-out-what-they-want-to-do-with-their-lives.
Bottom line: I'm no career guidance authority. I just did what I was good at and enjoyed, starting when I was in high school, and never went to university, and found to my surprise that after a few years I was making a decent living as a know-it-all Internet picocelebrity. I didn't pull myself up from abject poverty or invent transparent aluminium or anything.
(Of course, you're not in a particularly awful situation either, unless there's some dread disease or eight dependent children or whatever that you haven't told me about.)
But I think it is sound advice to make a list of the things you think you would like to do for the next few years of your life (at least), in order of preference and regardless of whether they line up with wherever the heck you think your career path so far is pointing, then work your way down that list and figure out whether each option in turn will let you make enough money to get by, and sock a bit away, without requiring more than one meal a day to be breakfast cereal.
Yes, the above is not exactly Crosslegged Dude On A Mountaintop Wisdom. But like I said, this schtick seems to work for Tony Robbins.