For us here at RuggedPCReview.com, this global marketplace often means a good deal of detective work when trying to figure out who actually makes a machine. You could argue that a computer is a computer and it's not really important who designed and manufactured it. That may be so for some, but I really like to know who did the design, who specified the features, and where manufacturing took place. It'd be silly to praise a company for their excellent design when, in fact, all they did was strike a deal with a Chinese manufacturer and put their label on the machine. There's nothing wrong with that, and many companies do a great job searching for good products that they then sell and service in the US. But it'd still be good to know the actual origin and background of a machine.

What are some of the different business models?

• There are resellers that sell machines from other companies.

• There are distributors which carry machines from a variety of sources and often put their own names on the machines.

• There are vendors and system integrators that sell value-added third party machines under their own name. They may or may not have exclusive arrangements with their supplies.

• There are companies that have their own engineering resources and jointly develop machines with Taiwanese or Chinese manufacturers.

• There are companies that design their own machines, but have them built by a Taiwanese or Chinese contract manufacturer.

• And finally, there are those who still design and manufacture their own machines.
  

However, it doesn't end there. Some of the Asian manufacturers have their own relationships and interconnections. As a result, we've seen machines where the top part came from one Asian company and the bottom part from another. We've seen machines seemingly made by Taiwanese manufacturers also being marketed by Chinese companies, apparently under reseller agreements (by and large we assume that machines are made in countries with lower manufacturing costs and marketed or re-sold in countries with higher costs). It can get really confusing.

There are also an awful lot of vendors out there, some of which we never heard from. This morning, for example, I came across Chinese Evoc Group, which has been around since 1993 and makes a large variety of rugged, embedded and industrial computers and components, including some interesting looking panel PCs and rugged notebooks (check the Evoc JNB-1404 and Evoc JNB-1502 rugged notebooks).

Does it even matter where all those computers come from? Probably not to consumers. Whether the Dell or HP notebook at OfficeMax is actually made by Quanta or by Wistron hardly matters (though it really concerns me that apart from CPUs, some other chips and software, almost nothing is made in the US anymore). All those Taiwanese OEMs are top notch, and an increasing number of the Chinese ones as well. It does matter to us, though.

Knowing, and reporting on, all those lesser known Asian OEMs means finding the hidden gems, the companies whose products we'd love to see on the US market. Covering them may lead to OEM deals with US and European companies, and such relationships can be win-win arrangements for all involved. Our feedback may also help them adjust their products for the US and other Western markets that often have different values, priorities and expectations. In that sense, I hope that we at RuggedPCReview.com can be a clearinghouse and conduit of information.

Disregarding some smaller players and initiatives, here's the big picture: In 1993, the Apple Newton made news when then Apple CEO John Sculley pushed it hard and predicted that such devices and their infrastructure would one day be a trillion dollar industry. Sculley was scorned for that remark, as was the Newton for its various shortcomings. But the Newton, way ahead of its time, was still good enough to get Microsoft to respond with its own mobile platform, just as a few years prior Microsoft had responded when pen computing with its PenPoint operating system threatened to compete with Windows.

So Windows CE was introduced in 1996, together with a lineup of little clamshells handhelds. The same year, Palm Computing released the little Palm Pilot that no one thought was going to be successful because it neither had a keyboard (considered mandatory after the Newton handwriting recognition fiasco) nor an expansion slot. But much to everyone's surprise, the Palm Pilot took off while Windows CE devices quickly garnered a reputation for being clumsy and underpowered.

Microsoft's approach was to reluctantly add features and gradually allowing more powerful hardware, always concerned that devices might eat into the much more lucrative low-end notebook market, just as they are now worried about netbooks. Microsoft's hardware partners played along and came up with some amazingly innovative devices (yes, you could get a Windows CE-based "netbook" with a 10-inch display and 800 x 600 resolution ten years ago), but even that didn't work against Palm who sold handhelds by the millions and adeptly crafted a "Palm economy" and thriving develop community that quickly dwarfed Microsoft's tentative and fragmented efforts.

At some point, Microsoft had the chutzpah to steal from Palm by trying to launch a handheld platform called the "Palm PC," but Palm's lawyers quickly nixed that, and their ho-hum handheld PC platform went nowhere. In a last ditch attempt, Microsoft nuked its multiple processor architecture approach around the turn of the millennium and tried again with the "Pocket PC," a markedly improved platform that has survived, in almost unchanged form, to this day.

Palm, in the meantime, thrived and reached a 75% global marketshare. When I gave a keynote presentation at the Taipei International Convention Center in 2001 on the future of pen computing and PDAs, I noted that Palm's OS was aging and Windows CE was gaining market share and might catch Palm within four or five years, but no one really believed that. Yet, it happened in a remarkable, unlikely succession of events that saw Palm fumble its leading position away and sink into virtual irrelevance while Microsoft, hardly more adept with its own mobile efforts repositioned Windows CE as, essentially, an embedded platform for the vertical market.

That approach, while it made sense, wasn't actually one that I thought was automatically going to be successful. In the late 1990s, Symbol Technologies, now part of Motorola, had been one of the first to adopt non-proprietary operating systems into its products. At some point, they offered both a Palm OS product and a very similar one powered by Windows CE, and at the time we were told that the Palm device did far better. Yet, Symbol was one of the very few vertical market companies that chose Palm, whereas Microsoft was remarkably successful in quietly positioning Windows CE as sort of a low-cost subset of Windows that would leverage corporate IT expertise and investments.

So while a lot of people wondered why Microsoft couldn't do any better in the mobile space, it as probably because they didn't want to. In 2002 I reviewed the T-Mobile Pocket PC Phone, an early smartphone that was amazingly good and would still fit right into the smartphone landscape of today, both in terms of looks and performance. Yet, not much happened after that. HP pretty much gambled away the "iPAQ" brand that came into its possession when they took over Compaq. Taiwanese and Korean companies became the new driving force, with the likes of HTC and Samsung settings trends and directions. And somehow the notion took hold that every handheld had to be a phone, which, in the US at least, meant being forced into overpriced 2-year contracts with telcos that couldn't care less about anything other than profit.

The reason why Windows CE became so successful is not because it's so good. It's a nice workmanlike effort, to be sure, but it's clumsy, sluggish and about as agile as a riverboat. But it only took over because a) the proprietary computing platforms of earlier handhelds were no longer acceptable, b) Palm let it by self-destructing, and c) because IT uses Windows and Windows CE sort of fits in. So there. It works, but it's ugly, really ugly.

It took Apple with the iPhone to demonstrate just how ugly Windows CE was. Unlike the Newton, the iPhone was right from the start, and it totally redefined how a mobile device should work. Its effortless elegance is exactly what people want, and Apple made it look natural and easy. The iPhone is human interface engineering at its very best. It may not meet all the IT-mandated checkmarks (yet) and thus earned the stern finger-wagging from some corporate types, but even they probably have an iPhone in their pockets. Once you know how simply and beautifully things can work, you never want to go back.

In a sense it's deja-vue all over again. Apple has a better product and a better idea, but Microsoft still dominates the desktop. Palm, back from the pretty-much-dead, tries again with a slick little box, just like the Palm Pilot once was, only this time they're copying Apple. The question in my mind is how long even workers and industrial users are willing to put up with klutzy, clumsy Windows CE now that almost everyone knows how well handheld electronics can work.

For example, "Atom" has from the start referred to two very different processor families.

The initial generation of Atom processor was the Z5X0 that was codenamed "Silverthorne" with a tiny 13 x 14 mm package footprint. They were targeted at mobile internet devices (MIDs) and used the also entirely new "Poulsbo" System Controller Hub. The processor has about 47 million transistors, which is more than the Pentium 4 had. Bus frequency is 400 or 533MHz (which support Intel's HyperThreading). Thermal Design Power is between 0.85 watts for a low-end 800MHz version without HyperThreading, and 2.65 watts for a 1.86GHz verison with HyperThreading. The chipset uses about 2.3 watts, which means total CPU and chipset consumption isn't even 5 watts. And the chipset has hardware support for H.264 and other HD decoding. However, as a the combo is targeted for internet devices, there is PATA but no SATA support.

A second family of Atom processors, the N2X0 that was codenamed "Diamondville," was meant for standard low-cost PCs and netbook type of devices. The N2X0 is similar in many ways to the 5XX platform, but used a somewhat larger 22 x 22 mm package. The N270 has a TDP of 2 watts and costs less than US$44, the same speed N230 4 watts and US$29. As of now, the N2X0 processor generally uses a version of the older i945 chipset. In order to reduce its power consumption down to 5.5 watts, its frequency (and performance) have been lowered as well and the chipset is called the i945GSE. This is used in the N270. The N230 chip, geared towards desktops, uses the i945GC that is quicker, but also uses 18 watts! Note that the i945's GMA 950 IGP is not able to decode HD signals. The N2X0 can be used with SiS chipsets. though I haven't seen any such systems.

From the looks of it, system designers have been struggling in figuring out whether to use the Z5xx or the N2xx chip. In netbooks it was a slamdunk for Diamondville as almost all netbooks use the 1.6GHz N270. However, there are exceptions. When Panasonic introduced its Toughbook CF-H1 Mobile Clinical Assistant, it came with the 1.86GHz Atom Z540 processor. And when Samwell, one of Taiwan's major OEMs in the semi-rugged and rugged space, introduced what is essentially a rugged tablet version of a netbook, they also picked a "Silverthorne" processor, in this case the Z530P.

I am not sure what drives the decision to go with a Atom N270 versus a Atom Z530. On the surface, they seem to have about the same performance and use about the same amount of power. One glaring difference in their specification is that the N2XX series supports the ever-important SATA (serial ATA) disk interface whereas the Z5XX does not and needs to use PATA drives. On the other hand, the technically inclined point out that the N2XX's use of a very slow version of the already dated i945 chipset makes for sluggish graphics performance and that the i945's GMA 950 IGP is not able to decode HD signals. Anyone who has tried playing back high-def video on a N270-based netbooks knows the pain. However, both versions of the Atom score about the same on the two benchmark systems we use here at RuggedPCReview (PassMark 6.1 and CrystalMark 2004). The Z5xx, in fact, scored very low in 3D graphics, which one would assume are at least somewhat of importance in any "mobile internet device."

But things are getting more interesting yet. Despite what on the surface appears to be the more lucrative "Diamondville" market with its many millions of N270 chips, on April 8, 2009, Intel announced the expansion of the Z5xx platform with a new high-end version, the 2GHz Z515, and a new gas miser version, the "up-to-1.2GHz" Z515. At the same time, Intel spoke of an entirely new Atom platform called "Moorestown" that combines the "Lincroft" system-on-chip with the "Langwell" hub of which as of now all I know is that it uses a lot of acronyms and is still based on the 45nm manufacturing technology.

On the N2xx horizon, there is the N280 processor, and apparently also a dual core Atom chip. There is not much material out there on those, and I need to look more into it.

There was another development. For embedded computing Intel quietly expanded the Z5XX platform with larger form factor versions that carry a "P" in their name, and then special "large form factor with industrial temperature options" versions marked with a "PT." I was aware that Intel would release a "large package" version of the Atom, but not the timing and the purpose. Well, this happened in March of 2009 when Intel added the "large form factor" Atom 1.1GHz Z510P and 1.6GHz Z530P as well as the "large form factor with industrial temperature option" 1.1GHz Z510PT and 1.33GHz Z520PT. What does that mean? In essence, the P and PT versions look like larger chips. Instead of the tiny 13x14mm package of the original Z5xx chips, they use a 22x22mm package, which is actually the same size as the N2xx chips. As far as temperature range goes, 0 to 70 degrees Celsius (32 to 158 degrees Fahrenheit) is considered "commercial," whereas -40 to 85 degrees Celsius (-40 to 185 degrees Fahrenheit) is considered "industrial." Interestingly, only the "PT" series processors support the industrial temperature range; the "P" series versions are listed with the same commercial temperature range as the initial chips.

Intel's updated Z5xx product brief now stresses fairly strongly that there are industrial as well as commercal temperature range packages for both the Z5xx processors as well as for their complementing US15W system controller hubs (GMA 500 graphics, I/O controller and memory controller). The brief also stresses that the small footprint versions are for space-constrained handheld and embedded devices whereas the large form factor is pitched for designs without small space restrictions but industrial temperature requirements. So why then do the "P" processors still have the same commercial temperature rating? Probably because the large package also includes "an integrated heat spreader" that "further contributes to its value for thermally constrained, fanless applications." Since the thermal design power of these chips was already tiny, I am not sure what the integrated heat spreader does, or why it was necessary.

In terms of performance, the "P" large form factor and "PT" large form factor/industrial temperature range chips appear unchanged, though the TDP is up a bit from 2.0 to 2.2 watts. However, if you compare the Intel's summary sheets for the Z530 and the Z530P it looks like the 530P chip is missing Intel Virtualization Technology as well as Demand Based Switching. Virtualization technology, according to Intel, allows "consolidating multiple environments into a single server or PC" which I believe means the CPU acts as if it were multiple CPUs operating independently so you can run different operating systems at the same time. Demand Based Switching was described as an enhanced version of Intel's SpeedStep technology (see description) that is available in both versions of the Z530. These are generally fairly involved server-based issues and I am not sure what the relevance to the new "large package" Atom processors is.

In any case, the "large package" also has a different "ball pitch," which refers to the spacing of the little balls of solder that replace pins on the underside of these tiny processor packages. From what I can tell, the 0.6mm ball pitch of the original Z5xx series requires high density interconnects (HDI) on the printed circuit boards, and those are more difficult to do and also more finicky--not what one would want in a rugged product (for an example of these issues, read this). So the "P" series would address that issue with its larger package size whereas the "PT" series would appeal to automotive and other transportation and industrial applications that often have a -40 to 185 degrees Fahrenheit requirement.

Now add to this that Atom chips, despite all the hoopla and market acceptance, are pretty poor performers, benchmarking no better than the lowly original Core Solos. Graphics performance, especially, is weak (what's considered weak in one device can be more than adequate in another, of course). There's the low power consumption, of course, but even that is not a given. We've benchmarked exceedingly thrifty Core 2 Duo machines as well as power-guzzling Atom systems, so proper setup and configuration are an issue.

Sometimes it almost seems like the Atom is sort of a trial balloon, one where Intel very successfully created an attractive image of a hip processor, but is also somewhat aimlessly trying out various applications to see where the Atom will fit and stick.

Which gets us to the processor. There was a time when processors cost next to nothing and the mere thought of needing to cool them with a big fan would have been preposterous. When I bought my first IBM PC in 1981, it cost US$4,000, in 1981 dollars. It was powered by a 4.77MHz Intel 8088 processor that you could be at any electronics store for about six dollars (the folks who proclaim that ALL electronics components have become so much cheaper obviously weren't around in 1981). Intel managed to parlay the processor business into a near monopoly, with Microsoft and Intel going lock-step in a mutually advantageous game of creating ever more resource intensive software. Microsoft made Windows bigger and bigger, and Intel delivered the processors needed to run it. That's what got us to a point where software needs minutes to boot, and the processor, chipset and graphics card all need big fans for cooling. Oh, and while the cost of computers has come way down, the cost of Intel processors has gone way, way up. A big new one can cost a thousand dollars, and even more modest ones approach the cost of low-end notebooks. A halfway decent Core 2 Duo costs more than a little Acer Aspire One netbook.

How can Acer, and everyone else who makes small, inexpensive computers do it? Increasingly by using the Intel Atom processor, which is smaller, uses less power, and costs relatively little. Why did Intel do it? Because they found themselves in a predicament. Microsoft increasingly insists that every computer must run Windows proper. The 1990s experiment with Pocket PCs is essentially over. By insisting that small platforms had to be compatible with Windows, yet making sure they didn't get powerful enough to be a threat to the Windows business, Microsoft successfully kept the wings of mobile devices clipped, to the extent where they eventually disappeared as viable platforms. Just the other day I came across a press release from a major manufacturer of rugged handheld computers that said its customers increasingly demanded full Windows even on handheld devices. And that gets us right back to the Atom processor.

Now cost isn't as much of a factor in the vertical marketplace as it is in the consumer market. I am not saying cost doesn't matter, but a market where a device may cost US$4,000 has a bit more leeway than one where customers expect US$800 pricing. What does matter, though, is size, weight and battery life. So what Intel did with the Atom processors is essentially remove the processor as a major power consumption factor. What do I mean by that? Well, an average Core 2 Duo desktop processor uses around 65 watts, a mobile version about 35 watts. There are chips that use considerably more or a bit less, but those are the rough numbers.

Now how do we know how much power a processor uses? After all, Intel sells them using a weird nomenclature that, unlike light bulbs that have a watt rating, seems unrelated to performance. Instead, Intel usually provides what they call the "Thermal Design Power," or TDP. TDP is described as "The maximum amount of heat which a thermal solution must be able to dissipate from the processor so that the processor will operate under normal operating conditions." There's a good deal of debate as to what TDP actually means and how it relates to real world power consumption of a processor. But for the sake of the argument, let's assume we're talking watt-hours and the processor is in a battery-powered computer. We can easily compute the battery's watt-hours by multiplying volt and amp ratings. A powerful notebook computer battery may provide 75 watt-hour, just enough to run a typical desktop processor for an hour (and that's without the power needed for the display and everything else in the notebook). A frugal notebook processor with a TDP of 25 watts would run three hours, and that sounds about right (in the real world, the processor uses power conservation modes most of the time, but you have to add in the power used by all the other computer components).

Now what does an Atom processor use? Between 0.6 and 4 watts. There are two different families of Atom chips, one geared towards mobile Internet devices (MIDs) and one towards netbooks and other low-cost PCs. The most popular chip in mobile computing is probably the 1.6GHz Atom N270, which has a TDP of 2.5 watts. That's the chip you find in almost all current (early 2009) netbooks and in many embedded components. Why two families? Because MIDs and PCs have different feature requirements. MIDs are usually multimedia-oriented and power consumption is totally crucial because the devices are so small. Netbooks and similar generally rely more on compatibility and standard PC interfaces (like SATA).

So where do the Atom processors fit in as far as power consumption goes? Well, 2.5 watts is sensationally low compared to just about anything else available. The generally unloved Intel Core Solo uses about 5.5 watts in its ultra-low power version (U1300/1400/1500), the Core 2 Solo (U2100/2200) about the same, the mobile Core 2 Duos between 10 watts (U7500) and 45 watts (Q9100/9300). So the most popular Atom processor uses less than half the power of a Core Solo and only a small fraction of that of the Core 2 Duo chips.

Now keep in mind that processors need corresponding chipsets, and those use power, too. Intel designed a super-efficient chipset to go with the MID-oriented Z5xx series of Atom chips that was once codenamed Silverthorne. That chipset, the "Poulsbo System Controller Hub," can do high definition video decoding and other neat stuff required in consumer multimedia devices, and it only uses about 2.3 watts. However, it does not support serial ATA and some other essentials, which rules it out for many computing applications. The N2x0 series of Atom chips uses the i945GSE, which is a slowed-down version of an older Intel chipset, the i945. That's good as far as compatibility goes, but there is no high-def decoding and 3D performance is low. The i945GSE uses about 5.5 watts, so overall consumption of the N270 and the chipset is still only about 8 watts, but it's not exactly a state-of-the-art solution.

How about performance? This is where it gets a bit complicated because overall "performance" of a computer depends not only on the CPU, but also the chipset, the memory, the hard disk or SSD, overall system configuration and -- very important -- the OS platform and software loaded. That said, we run fairly extensive benchmarks on all systems that come to our lab, and so far we've found that an average Atom N270 device scores roughly one third of that of a 2.5GHz Core 2 Duo T9400, about 30% less than that of a 1.2GHz Core Duo U2500, about the same as a 1.2GHz Intel Core Solo U1400, and about 50% better than that of a 1GHz Celeron M 373. So we're talking decent, but certainly not blazing speed.

As far as architecture goes, the Atom is an interesting mix of old and new technologies. It's definitely state-of-the=art in terms of miniaturization, using Intel's hafnium-based high-k manufacturing. That is a fancy terminology describing the use of different conductor materials to make even tinier transistors possible. The architecture of the chips is less advanced. There's only a single core, though Intel uses the old HyperThreading technology known from as far back as the Pentium 4. There are also advanced new power savings technologies.

Overall, the Atom is certainly an interesting marketing phenomenon. At this point, everyone is clambering to get onboard the Atom bandwagon, and somehow Intel managed to stay clear of the nuclear power connotation though one would expect that from a name like "Atom." Intel, though, stresses the hafnium-based manufacturing process, and hafnium's primary use is in control rods in nuclear power plants, so that may be the "Atom" connection. In any case, even with the sub-optimal chipset situation, the lack of some features, and only moderate performance, Atom is hot. And in the new Intel world order of massively expensive processors, Atom is cheap, too, with prices of well under US$100 depending on the type and version. I've seen $44 for the N270 mentioned, and about the same for some of the low-end Z5x0 chips plus their Poulsbo chipset. Oh, and if you wonder what the difference is between the N270 and the 230, there is a N270 and a 230 that run at the same speed, the N270 is for mobile applications whereas the 230 uses a bit more power (4 watts) and is used with a considerably more power-hungry version of the of the i945 chipset, making the Atom 230 more suitable for desktop use.

As usual, there are numerous expert opinions out there, and the overall consensus seems to be that, for now, the Atoms just represent Intel's first step into the small form factor embedded and a MID market that is pretty much dominated by ARM-based designs.

With Intel's resources and marketing savvy, Atom as a "low power" processor platform may well be here to stay. As is, they are off to an amazingly good start.

For much more information on the Silverthorne platform, check Intel's Intel Atom processor Z5xx Series.

Nor are they new. There have been numerous attempts at selling downsized miniature laptops over the years, going back to the early 1990s and before. None were ever successful. People simply did not want an underpowered mini version of a notebook with a small screen and a keyboard that was not full size. Apparently that's changed and "netbooks" sell by the millions. Go figure.

One difference perhaps is that technology has come a long way. Even an underpowered mini notebook can do just about anything anyone would ever need in terms of computing. Standard wordprocessing, scheduling, spreadsheets, presentations, email and internet access tasks can all be done on a mini notebook. Let's take a look at what "netbooks" offer:

For the most part they are clamshells measuring about 10 x 6.5 inches and weighing between two and three pounds. They have displays measuring between 7 and 10 inches diagonally and they usually offer WSVGA resolution, which means 1024 x 600 pixels. Their keyboards are usually around 90%-scale, which is infuriating because that makes touch-typing a pain and also because there'd actually be enough room for a full QWERTY layout by making punctuation keys smaller, but apparently Taiwanese and Chinese ODMs and OEMS do not realize that. Memory is usually limited to a gigabyte, though some can be expanded to a gig and a half. Storage is either via Flash for Linux-based netbooks or generously-sized hard disk for Windows-based units. Most come with a rudimentary onboard cam, SD card or multi-card slots and, of course, Bluetooth and WiFi. And most are powered by Atom chips, generally the 1.6GHz N270.

How do they work? It depends on your expectations. Benchmark performance is about a third of that of a modern notebook, so routine stuff can take much longer than you're used to. The biggest limitation is the small screen. My Acer Aspire One, one of the most popular netbooks, has a 8.9-inch screen which is bright and sharp, but 1024 x 600 pixels simply isn't enough for anything these days. Working with it becomes a continuous for screen real estate, which means turning off unneeded toolbars and a lot of scrolling, scrolling, scrolling. The term "netbook" is also a total misnomer as the one thing where the current generation of netbooks falls way behind is fast web access. Pages take forever to load.

If they are such a pain to use, why do I have a netbook? Because they have a lot going for themselves, too. My Acer One runs Windows XP speedily on 1.5GB of RAM, and the 160GB hard disk is both quick and large enough. With its 6-cell battery the little Acer can run as long as six hours on a charge, and sometimes more. I like its dual SD card slots. I occasionally miss an optical drive, but have my office network set up so I can access the DVD drive of a desktop. Most of all, I like the Acer's small and handy size. Packing and transporting even a compact notebook is usually a pain, but the little Acer netbook fits absolutely anywhere. Even its power supply is tiny. In my office, I hook it up to a 20-inch LCD and a full-size keyboard and mouse. I get full 1600 x 1200 pixel resolution, which makes working on the little Acer feel like working on a "real" computer.

So, "netbooks" they are not. But there does seem to be a good-size niche for surprisingly competent little notebooks that go for less than US$400. Price is definitely an issue. I'd rather have a more rugged device with a touch screen. Fujitsu and Panasonic and others make them, but for several times the money. Why not a rugged netbook with a very small price? It might sell in large quantities.

So why hasn't Linux taken over? Because it's too complex. Sure, there are distributions that install simply and easily, but you can also spend hours trying to get one little thing to work right. Linux is a giant patchwork of code from all over the world. Perhaps the biggest challenge is that almost all Linux developers think Linux is so simple that absolutely everyone should be able to perform arcane steps and procedures.

Linux suffers from the expert syndrome. The expert syndrome is what makes academics speak in nearly incomprehensible language. It makes them look and sound important and, in their minds, is a reflection of their superior intellect and knowledge. Coders, likewise, revel in acting as if their most complex systems were child's play and anyone who does not master it must be an idiot. Some of the instructions for Linux are so complex and incomplete as to make it impossible for anyone who does not already know the systems to install things or make them work.

In all my time of working with Linux I've found perhaps a handful of truly useful tutorials and instructions. Sadly, this pits an incredibly productive global community of Linux coders and developers squarely at odds with the rest of humanity who can no more compile a kernel than split an atom. The rest of humanity also does not appreciate being talked down to when it comes to doing simple things like properly extracting a file, making a wireless connection work, or numerous other things that should be simple and self-explanatory but, in Linux, are not.

Unfortunately, I do not see a solution to this problem. You either have tightly controlled empires like Microsoft or Apple where things are centrally controlled and packaged, or you have loosely knit global communities of techies with all their human brilliance and flaws. So things will likely continue the way they have for decades, with Linux being both a a terrific solution but also one that can be endlessly frustrating.

Having founded and run several print magazines myself, I know all about the work and hardship that goes into creating a quality magazine, and how things are all different in this age and day of the Internet and web. Advertising dollars are increasingly going away from print, and people no longer want to wait for information to appear in print. Everything is available instantly. Yet, the information on the web is .... different. In a way it almost does not compete with print. How else would one explain the fact that there appear to be more magazines on newsstands than ever? I myself absolutely cannot imagine life without computers and the web, yet I have a good dozen print magazine subscriptions that I never intend to give up. Magazines and the web are as different as radio and TV -- both convey information and entertain, but in different ways. Unfortunately, tech companies like Microsoft do not seem to understand that, and the phone companies who have taken over the smartphone business are clueless about the market that has fallen into their laps.

Fact is, online is becoming much like TV -- far too many channels and nothing to watch. It's all commercials and infomercials. You have to channel-flip not because you can, but because you're constantly avoiding commercials and seeking something, anything, meaningful to watch. And quality is getting lost in a vast sea of drivel. You can google a particular product and instantly get 10,000 references to it, mostly junk. By now the web is jam-packed with virtually content-free sites that are just landing pages for ads and more ads. Even reputable sites are doing it: two paragraphs of content and then commercial bombardment. The ever more popular "customer reviews" are often little more than "this product sucks!", "no, this product is the best ever" slugfests, and the same goes for bulletin boards where there is endless posting and almost no factual information. With the exception of the by now almost suffocating commercialization it's all worth it, of course. But it is NOT a replacement for a good print magazine.

When I look at the final copy of Smartphone & Pocket PC Magazine (the Smartphone & Pocket PC Super Resource Guide Dec/Jan 2009) I see a hundred pages of superb, comprehensive information, a reference guide I am certain to keep around for years. You'd have to visit literally thousands of websites to get that amount of good information, and even then you would not get the quality. A complete and total spec list of ALL smartphones with touch screens? Check. A complete and total spec list of ALL PDAs? Check. Reviews and ratings of hundreds of the best software apps? Check. A complete analysis of GPS on Windows Mobile, including product reviews and comprehensive comparison charts? Check. Detailed reviews of the leading and upcoming smartphone platforms? Check. And that is just a small part of it. If a consultant were given the task of compiling the huge wealth of information contained in just one issue of Smartphone & Pocket PC Magazine, it'd cost many tens of thousands of dollars, and probably hundreds of thousands. For a company like Microsoft to let such an incredible resource die -- a resource that does nothing but promote Microsoft's mobile embedded platform -- is simply unimaginable. Spending millions on nonsensical commercials and sitting on billions, yet not support real, quality, serious information, it just does not compute. The cost of supporting a resource like Smartphone & Pocket PC Magazine that provides real information is absolutely minuscule compared to the billion here, billion there mentality of big business.

Lacking any meaningful support from the Windows Mobile side of things, Thaddeus Computing is now going on to cover the iPhone platform with their new iPhone Life magazine. It'll be an uphill battle as now they'll be dealing with one single hardware and software vendor (Apple), one single service provider (AT&T), and application software vendors who do all of their selling through Apple's App Store, so the impact of print advertising will be less traceable than ever. The iPhone is hugely popular, of course, but neither will people buy another iPhone (they're locked into a 2-year contract) nor can they buy another model (there's only one). The phone companies have historically not supported enthusiast magazines and there is no indication they ever will. They also don't "get it," something at least the Microsoft field people certainly did.

But won't Apple be thrilled to see one of the most respected niche and enthusiast publishers switch allegiance? Likely not, if they even notice. Apple is sitting on its own billions of cash, but I am fairly certain none of it will go to supporting a small magazine that could spread high quality news and real information at an annual cost that's a tiny fraction of the interest on Apple's cash reserves alone. And AT&T, which in the U.S. has a service monopoly on the iPhone? Hah.

So best of luck to the folks at Thaddeus Computing. It's an absolute crying shame to see Smartphone & Pocket PC Magazine die, and those in the Windows Mobile industry who let that happen deserve to be accused of colossal, inexcusable shortsightedness. Maybe someone will come to their senses and buy Thaddeus. 25 years of experience and commanding knowledge of the major serious mobile platform in the world AND they know how to compile and present information AND they have all the magazine distribution channels in place AND running them for a year probably costs peanuts? No brainer if you ask me.

All of this needs to be tested and certified, and the way it is usually done is by following standard procedures that describe a controlled lab testing setup, like document 60529 issued by the International Electrotechnical Commission (IEC).

The problem is that lab tests do not always accurately predict what may happen in real life. In that respect the ratings should really be considered guidelines rather than hard data. Consider, for example, two devices that both carry an IP67 rating. One of them has no external ports other than a single surface mount connector used to provide interfacing via a port replicator or dock. The other has a variety of commonly used ports, all protected by individual rubber plugs. One machine may also have an externally accessible expansion slot and an easily replaceable battery, each nicely sealed via o-rings and other high quality seals. Which device do you think is more at risk for leaking?

I'd say the second as it has multiple areas of entry as opposed to just one. No matter how well engineered the device may be, the probability of something going wrong is higher. A protective cover may not be pushed in all the way. A seal may have shrunk or gotten broken. A door was inadvertantly left open. It can happen.

A compromised seal may not necessarily mean a leak into the inside of the device. The port itself may carry enough sealing in addition to the protection provided by its cover to ward off damage. Then again, it may not. Bottomline is that the simplest and most foolproof protection is best.

Anything mission-critical should be failsafe. Failsafe means that if a system fails, it must fail in its safe state. A relay that snaps closed when it loses power is an example. The problem with protective rubber and other seals I'd that none are fail-safe. They are all fail-fail. So the best way to proceed is to have as few potential points of failure as possible.

What that means is that, all else being equal, a device with fewer possible points of failure will almost always be a better choice as far as protection us concerned.

The N270 is a single-core processor that runs at 1.6GHz and has a TDP of 2.5 watts -- significantly less than even an ultra-low voltage Intel Core Solo and only a small fraction of the power consumption of your average consumer notebook. There are other system parts that use power, and for now Intel doesn't offer Atom-compatible chipsets that are nearly as miserly as the processor itself. Further, a lot of the advanced features we've come to take for granted in Intel Core processors are simply not part of the Atom. Instead, Intel resorted to the hyper-threading technology from its past. It's all quite complex and it probably takes a chip design experts to tell how various Intel technologies impact performance.

What we can do is run benchmarks, and that's what we did on an Atom N270-powered Acer Aspire One netbook, an exceedingly handy little clamshell computer with an WXGA 8.9-inch display and a weight of just over two pounds. The tiny Acer came with a gigabyte of RAM, a 160GB 5400rpm disk, and ran Windows XP. Our standard benchmark suite, PassMark, did not complete and so we switched to CrystalMark 2004R2. Here are the results:

These figures suggest that systems equipped with the Atom N270 are quite a bit quicker than machines with the Atom's predecessor chip, the A110, but only a bit faster than the first-gen Intel Core Solo. The 1.6GHz Atom N270 is no match for the 1.2GHz Core Duo U2500 that's used in a number of high-performance Tablet PC slates. The high clock speed of the single core N270 is therefore a bit misleading. Clock cycle for clock cycle, the unloved Core Solo is more powerful.

However, in a lean, smartly designed system with enough RAM and a speedy disk, such as the Acer One netbook, the N270 can deliver both power and economy. The Acer feels fairly quick, and it runs about 2-1/2 to three hours on a small 24 watt-hour 3-cell battery and 5-1/2 to six hours on a 49 watt-hour 6-cell battery.

I was reminded of that when I came across some very interesting display products from a company called Digital Systems Engineering, located in Scottsdale, Arizona. They have the DVE Raptor display where DVE stands for "Driver Vision Enhancement." It is a ruggedized LCD display designed to operate under the kind of extreme environmental conditions encountered in tactical wheeled and tracked vehicles. The 10.4-inch SVGA display is sunlight readable with a super-strong 1,000 nits backlight (standard notebooks have less than 200 nits), good vertical an horizontal viewing angles, and zero color shift.

What's most amazing, though, is the Raptor display's environmental specs. It carries an IP67 rating, which means it is not only totally sealed against dust, but it is also waterproof to the extent where it is submersible. Hopefully that won't happen in a tactical vehicle, but this display will continue to operate under water. It can also operate in an extremely wide temperature range of -40 to 158 degrees Fahrenheit, handle any degree of humidity, and operate at 45,000 feet of altitude. Needless to say, the milled aluminum and heavily sealed and protected display has been shock and vibration tested to MIL-STD-810F specs.

The screen, which only weighs a bit over eight pounds, is also MIL-STD-3009 compliant. MIL-STD-3009 (also referenced as DOD-STD-3009) sets requirements for aircraft display equipment for use with night vision imaging systems. For mobile computers that generally means they must not interfere with night vision equipment in a cockpit. Part of this document is the U.S. Navy MIL-HDBK-87213 Revision A (Electronically/Optically Generated Airborne Displays) that describes, among other, criteria for legibility of electro-optical display equipment and daylight readability in bright environments, which is a military requirement. This can be an issue with daylight readable displays marketed to the govenment and armed forces.

If the indestructible Raptor is overkill, Digital Systems Engineering has a line of MSM monitors where MSM stands for Mil Spec Monitor. These come in various display sizes (8, 10, 12, 15) and are lighter than the Raptor. Despite IP67 sealing, they only weigh between 3.5 (8.4-inch display) and 6.9 pounds (15 inch display). Yet, the MSMs are MIL-STD-3009, MIL-L-85762A and MIL-PRF-22885 compliant and have an incredibly bright 1,400 nit backlight in addition to anti-reflective and anti-glare surface treatment, making them viewable under any lighting conditions.

To learn more about those super-rugged monitors, check Digital Systems Engineering's website at http://www.digitalsys.com.

In fact, the chip performed so well in the Nomad that I was certain other manufacturers would quickly follow suit and use the formidable PXA320 chip as well. Interestingly, that didn't happen. If I remember correctly, the only other product I've come across that uses the PXA320 is the Aceeca Meazura MEZ2000, which I think is still in the planning stage. Everyone else still seems to be using the older PXA27x, even in new designs. The PXA27x is certainly a good and time-proven processor, but it is no match for the PXA320 when it comes to performance.

Maybe something is going on that I am not aware of. Maybe Marvell isn't pushing the chip and it's such a secret that no one realizes technology has advanced. Maybe it's too expensive, or has some drawbacks I am not aware of. As is, the Nomad with its powerhouse PXA320 chip appears to continue to enjoy a significant performance edge over anyone else out there.

Except in one area.

Digitizers.

How much progress has there been since I began reviewing pen computers back in 1993? Basically none. And as far as I can tell, that sad situation sits squarely in Wacom's court. Wacom's patented digitizer technologies have resulted in Wacom having almost 96% market share in Japan, and a good 70% in the rest of the world. The Wacom digitizers I used on 1993 pen computers worked, sort of, but were hugely frustrating because it was essentially impossible to calibrate them. The Wacom digitizes I have used in vastly better and more powerful computers in 2008 worked, sort of, but were hugely frustrating because it's essentially impossible to calibrate them. I mean, there are any number of touch screens where you can calibrate 25 points or more, do edge compensation, and all sorts of other cool stuff geared towards enhancing precision and improving the user experience. A Wacom digitizer calibration? Four points, and that's it. Along the edge of the screen, the digitizer is often so badly off that it becomes frustrating to use it.

I've complained about this for pretty much as long as I can remember, and there hasn't been any change. Anything else in computing has improved dramatically. What gives? Is Wacom's technology inherently incapable of working better? Is no one else able to come up with a better alternative because of patent blocks? I don't know, but between Microsoft's marginal handling of the Tablet PC and the dismal performance of the Wacom digitizer, pen computing is where it is.

There. End of sermon. I just had to say it.

I remember when Panasonic showed me the results of their Toughbook corrosion testing on an invitational tour of their facilities in Osaka back in 2002. Without special consideration of salt water and salt fog exposure, there could quickly be appalling damage as shown on the picture to the right (click on it for a larger version). Panasonic explained how they had been approached with requests for such testing, performed the salt water and salt fog tests, and were surprised to see the extent of the damage. They then systematically changed design and materials to ward off or minimize the effects of salt. This benefitted all subsequent Toughbooks, and also showed Panasonic how to develop special solutions for customers who use their products in environments where they are exposed to salt fog and water.

When you look at these pictures it becomes obvious that sealing alone is not enough when it comes to salt water exposure. Sealing standards only tell how well a product keeps dust and water out of the inside of the unit. They don't tell what salt can do to components that lay outside of the sealing barriers. What can salt do when it gets under a keyboard? Inside a hinge? Underneath protective doors? The result can be ugly. Nothing can ever ward off salt entirely when a product is used in marine environments. Users need to keep computers away from excess exposure as much as possible, and equipment needs to be cleaned meticulously after any exposure. That means that cleaning must be possible in the first place, which means that places that are potentially expose to salt water and fog must be accessible. There are just a whole bunch of additional considerations.

This is why the famous MIL-STD-810F (Department of Defense Test Method Standard for Environmental Engineering Considerations and Laboratory Tests) document includes a 9-page section on Salt Fog testing.

MIL-STD-810F Method 509.4 describes testing methods to determine the effectiveness of protective coatings and finishes on materials for corrosion, electrical effect and physical effects. The tests can also determine the effects of salt deposits on the physical and electrical aspects of materiel. The product is exposed to salt fog mist from a 5% salt solution via atomizers at about 95 degrees Fahrenheit for a minimum of four alternating 24-hour periods, two wet and two dry. The product is then examined for salt deposits that can clog or bind components, electrical malfunction, and potential short and long-term impact of any observed corrosion.

The reason why I am writing this all down is because my return coincided with an announcement from GETAC that its impressive B300 rugged notebook had received Salt Fog certification. Here's part of their press release:

LAKE FOREST, CA. – September 2, 2008 – GETAC Inc., a leading innovator and manufacturer of rugged computers that meet the demands of field-based applications, announced today that its B300 ruggedized notebook PC received full Salt Fog certification based on testing standards set by the Department of Defense (MIL-STD-810F – 509.4). Salt Fog is a specialized test used to evaluate and determine the effectiveness of protective coatings and finishes on materials to repel salt corrosion and may also be applied to determine the effects of salt deposits on the physical and electrical aspects of materials. Adding the Salt Fog certification to an already robust and rugged notebook PC makes the GETAC B300 the ideal choice for military installations, marine applications such as the Coastguard and other industries where salt or salt air can impact equipment performance.

“Salt is one of the most aggressive chemical compounds in the world,” said Jim Rimay, president, GETAC. “Salt will quickly corrode a computer’s exterior, impair vital electrical system functions through salt deposits and have a physical impact by restricting free movement of its mechanical components. The B300 addresses these issues with its Salt Fog certification and elevates it to an elite status among ruggedized computers for safe and uninterrupted operation in any location, especially in coastal regions of the world.”

We recently did a detailed hands-on test of the Getac B300 and found it to be a very impressive machine full of clever engineering and innovation. A combination of optical coatings and superbright backlight make the screen readable in the brightest sunlight, and amazing power conservation methods can extend battery life to a stunning 12 hours. It's good to see that the company also invests in testing against one of the less-often mentioned environmental threats to mobile computers -- salt fog exposure. While most specs include resistance to drops and vibration, salt fog/water exposure can destroy a piece of equipment just as surely. Once the corrosion is detected, it's usually too late, so it's nice to see Getac take proactive steps.

MIL-STD-810F, however, only describes testing methods, and not the criteria that determine passing tests. It would therefore be nice to know what Getac found during its tests, and what the company did to make the B300 as immune to salt fog damage as possible.

I don't know what the thought process was that led to the design of the original C5 medical tablet, but it was certainly a smart decision to go after the medical market. It's a tough one to break into for a variety of reasons, but also one where mobile systems can make a huge impact. At Kaiser, the HMO I use, they finally have terminals in almost every examination room so they can call up patient info, and they can now also call up x-rays onscreen, but it took them forever, and I still see no portable electronics. I suppose it's the same elsewhere.

The Motion C5 was an attempt to provide a portable computer that could do more and was easier to integrate into the daily workflow of medical people. So they made it small and light and gave it an integrated handle to easily carry it around. They integrated an RFID reader and a bar code reader and also a camera. They also made it white so it fits in with all the other medical equipment, and it's easy to wash and disinfect. Motion also created a small, handy dock for it. So the overall idea was the provide a small computer that was easy to carry around and that included all sorts of data capture methods. It all still depended on systems integrators to package the hardware with medical systems software, and then have hospitals actually pick it up and use it. I am not sure how many did, but the Motion C5 was, and currently still is, probably the best mobile hardware for such projects.

When I first looked at the C5 I wondered why Motion limited the platform to just one market. True, it's a potentially huge market, but the C5 seemed sturdy enough to be used in other mobile applications, and it already carried IP54 sealing, which means it was didn't mind a bit of rain and some spills. Motion apparently agreed and created a second version of the C5, the F5. They called this one a "Field Tool," -- not the greatest of names, but obviously an attempt at communicating that this computer should be seen as a tool for jobs rather than a conventional computer.

I must admit, I had a bit of a hard time with the F5. When I wrote about the C5, I had no problem seeing the design decisions that had been made to make this computer just right for the medical market. The size, the shape, the features, the color and so on. The F5 is gray instead of white, but other than that, it's the same computer. It does include Motion's "View Anywhere" display because unlike the C5, the F5 would probably be used outdoors where sunlight viewability counts. So there wasn't any additional thought on how to make a computer best suited for use in the field.

The way I see it, the field IS different from a hospital. You won't always have a dock to charge a computer, and so the fairly small battery of the C5 may not be enough. And in the field it does come in handy to have a USB port or two and perhaps even an old serial port for some arcane instrument or measuring tool you need to hook up. And having some sort of expansion slot also comes in handy. Wireless communication is great and we can't do without, but it's been my experience that even with Bluetooth and WiFi, there are times when it's a lot simpler to just copy files onto a USB key or a SD card than to send them. The F5 can't do that as it doesn't have any ports or slots and totally relies on wireless or the dock.

All of this made it a bit more difficult to review the product. I am used to Motion having a very clear rationale for a machine, and in this case the rationale seemed to be that the healthcare C5 was good enough to be offered for other markets. That was probably a good idea, but something still doesn't feel quite right. Even the "View Anywhere" display that I remember as effective from previous reviews of Motion tablets seemed rather low-contrast compared to other sunlight-viewable technologies on the market.

The F5 is also one of the few machines that uses the Intel Core Solo processor. The Solo is essentially a Core Duo with one core not used, sort of like an 8-cylinder engine with only four of them running to conserve fuel. It is an economical chip, with a thermal design power of just 5.5 watts, which is only a bit more than half of what a Core Duo chip running at the same clock speed uses. Problem is that benchmark performance is much lower, too, and generally closer to the lowly Intel A110 than even an ultra-low-power Core Duo. The F5 is no slug at all, at least with Windows XP, but with Motion always being at the forefront of technology I wonder why they didn't just use an Atom processor instead. They did switch from the Core Solo U1400 to a Core 2 Solo U2200 which is said to include better caching and even more power-saving technologies, so perhaps that was the right move for now.

Anyway, just a few thoughts on what is, in fact, an interesting and welcome addition to the hardware alternatives available to those who need to implement computing solutions in the field. The official review of the Motion Computing F5, with pics and specs and all is here.

Usually, rugged equipment is dropped or exposed to water to show that it can survive the kind of punishment encountered in the field. MobileDemand's earlier videos pretty much followed that tradition. xTablets were dropped, exposed to showers, rolled down a hill and so on. But soon the videos showed drops more extreme than anything that would likely happen in the real world. And instead of being exposed to a showerhead, the computer was strapped to the top of a car and run through a car wash five times, with the computer running and its display on camera during the whole ordeal.

And now the "We could use a hammer..." video. It's very smart. No one would actually use a computer as a hammer (though, come to think of it, I've used a variety of objects as hammers when none was handy), but the image of using that sophisticated piece of electronic equipment as a hammer certainly drives the point home, no pun intended.

Using the xTablet Tablet PC computer as a hammer really means to illustrate a point: shock and vibration do happen in the field. If you use a machine in a truck or as a data capture device you do not intend to damage it, but sooner or later it will fall. And constant vibration is affecting the computer. Eventually things can happen. Electrical parts may touch and short-circuit. Fasteners may come loose. Structural pieces may crack. Seals may deform and begin leaking. Electrical contacts may become unreliable. The display panel may become get out of alignment. Fasteners and ties may get loose. Wiring may chafe. Materials may fatigue and then break. Parts may deform or crack. And so on. At best, sealing may be compromised, electrical noise may be introduced, and individual parts are headed for failure. At worst, the computer fails.

This is why manufacturers usually provide test data, usually how a product performed when using the procedures described in MIL-STD-810F. Those procedures try to replicate conditions actually encountered in the field during transportation and operation. That makes sense, but the testing is quite involved and not very easy to interpret. Witness the following caution regrading acceleration testing found in MIL-STD-810F 514.5:

Care must be taken to examine field measured response probability density information for non-Gaussian behavior. In particular, determine the relationship between the measured field response data and the laboratory replicated data relative to three sigma peak height limiting that may be introduced in the laboratory test.

That's a mouthful, and the results are even more difficult to read. General integrity test conducted may then yield results such as, say, a power spectral density of 0.04G²/Hz, 20 to 1000Hz, descending 6dB/oct to 2000Hz. MobileDemand, like all the other serious rugged equipment vendors and manufacturers, has its gear tested in accordance with the MIL-STD-810F (and other) procedures, but what has more impact, some tech specs comprehensible only to engineers or a video of a man using the that rugged Tablet PC as a hammer and it still works?

"We could use a hammer..."

Brilliant.

To see the "We could use a hammer..." video, click this Blip.tv link.