Hardware Hurdles

Pen-Based Computer Vendors Face New Twists To Old Challenges

Pen-based system developers must contend with hardware challenges that, at first glance, seem similar to those of laptop or desktop vendors. Under close inspection, however, pen-based systems take these challenges to a new level, demanding innovative solutions to technological barriers that must be hurdled if any vendor’s pen-based system — or the emerging market itself, for that matter — is to succeed.

Anyone familiar with today’s laptop and desktop computers knows there is a never-ending quest to improve the traditional components that make up a computer — higher resolution and more efficient and compact displays, increased processing power, reduced power consumption, increases in memory and storage capacity — all in ever-smaller and more streamlined packaging. Pen-based systems require optimal solutions in all these areas. Essentially, a pen-based computer must deliver the performance and features of a desktop workstation in a machine the size of a notebook that fits in the cradle of your arm.

Processing Power

There’s little doubt that successful pen-based computers will be built on 32-bit architectures. Initially, these architectures will be 80386-based or, in some cases, RISC-based; anything less — and this includes the 80286 — simply won’t provide enough horsepower to drive pen-based systems. High-resolution displays, notebook interfaces, outline fonts, multitasking, power management, and handwriting translation will require higher performance, greater address space, and more precision than 16-bit processors can provide.

The 80386 is the most attractive platform for a number of reasons: it’s available, stable, and familiar to both developers and users alike. Although relatively pricey (from $60 for a 386SX to over $200 for a 386-33 MHz chip or the 386SL chipset — still very competitive with most RISC chips), the CPU will become more affordable as Intel meets increasing demand and as other chip vendors, most notably Advanced Micro Devices, bring their versions of the 80386 to market. And let’s not forget the simple fact that the 80386 is a good chip with a long future.

Particularly important to pen-based systems is the introduction of Intel’s 386SL chipset which, with its special power management circuitry, is expressly designed for notebook systems. (At the heart of the 386SL is the 386SX CPU.) Not to be outdone, AMD has shown a surface-mounted CMOS version of its 80386 that AMD claims is nearly 40 percent smaller in size than standard Intel 386s and consumes almost 70 percent less power than Intel’s 386DX.

Battery Life

Achieving acceptable battery life is perhaps the most important and demanding challenge facing pen-based hardware designers. Users will not accept the two or three hour battery limits of early laptops. In fact, a chief sales pitch for mobile, pen-based computers will be that they’ll be used many more hours per day than desktop or laptop systems, perhaps more than 20 hours before recharging or replacing batteries. As with laptops, the biggest power hog in notebook machines is the display (discussed below). In particular, backlighting of liquid crystal displays (LCD) can require as much as 50 watts. Complicating the problem for manufacturers is that 32-bit CPUs consume more clock cycles and run at higher clock rates than their 16-bit ancestors, thereby consuming even more power. Obviously hardware developers must figure out ways to have their cake and eat it too.

One solution, of course, is to develop more efficient batteries — or more efficient ways of using existing batteries. (Apple opted for a more efficient, albeit “special,” battery for its Portable Macintosh — a battery that, when you need to replace it, can be as hard to find as a gas station on a country road late at night.) For the user, a better solution might be to develop computers that rely upon existing, readily-available batteries and simply use them more efficiently. How can batteries be used better? For one thing, more efficient power supplies can be — and are being — designed. Most computer power supplies, for instance, are only about 50 percent efficient. (Power supply efficiency is measured in terms of the input power lost as it is distributed throughout the system. Assume that 100 percent of the power goes into the power supply from the source; heat dissipation and other factors typically erode this level by about half.)

Although company officials refuse to discuss details, Momenta is designing a power supply the company claims is up to 90 percent efficient. If so, this is an amazing and big step forward. Considering Momenta CEO Kamram Elahian’s background (he was founder of Cirrus Logic, a semiconductor company), we can guess that Momenta has exploited semiconductor technology to perform some of the functions of the power supply. If Momenta in fact has developed a power supply that is 90 percent efficient, the company can probably become wealthy just by licensing the technology alone. For pen-based users, the upshot is that Momenta systems may well be able to use common, off-the-shelf batteries as the power source.

It should be noted that “smart” chipsets like the aforementioned Intel 386SL will also spell r-e-l-i-e-f for power hungry systems. Integrated chips like the 386SL have the built-in ability to, among other things, optimize power management for those subsystems that will be used the most and that draw the most power. In the past, power control for laptops has been limited to shutting down the display or hard disk after periods of non-use. Power-aware logic extends this concept to virtually every subsystem in the computer — and does so without taking over operating system interrupts and service routines, thereby eliminating the problem of conflicts with software that might be accessing these routines.

In addition, operating systems for pen-based systems will have power-management capabilities built in at all levels. Because of its object-oriented nature, for example, GO’s PenPoint allows programs to reuse and share code, thereby conserving memory and power because only a single copy of code needs to be stored in memory. PenPoint also includes more familiar power-management techniques, including features that shut down system components during periods of non-use; future releases of MS-DOS will likely provide limited power-management features as well.

Display Technology and Input Capture

As with most computers, display technology is critical to success — if you can’t see what you’re doing, you won’t use (or buy) the computer. Traditional laptop display challenges include weight, size, speed, resolution, color, durability, power consumption, and readability/lighting. All of these still apply in one way or another to pen-based systems, but when you throw data input and random access to any point on the screen into the equation, things begin to get really complicated. For the time being, LCD technological challenges for pen-based systems have been met and the issues are more of implementation and, for lack of a better word, ergonomics — particularly when it comes to data input. The $64 questions involve simulating real-time “ink flow” and the minimization of parallax.

All of us are used to the natural feel of ink pens smoothly moving across the paper. Users will expect a similar feel when moving the stylus across the glass display — scratchy, grating movement won’t be acceptable, nor will surfaces so slick that the pen slides like on ice.

Parallax is an optical issue. It can be said that a parallax zero condition exists when the “ink” flows directly at the stylus tip and writing takes place exactly where you want it to. A parallax error condition exists when the “ink” flows and writing takes place above or below or adjacent to the stylus tip. We’ve run across this problem in our evaluation of some pen-based computers and, from a user perspective, it can be terribly frustrating — somewhat like losing your depth perception.

Parallax errors arise when there is too great a distance between the stylus tip and the display. Manufacturers are limited to how close the pen tip can actually get to the display because the digitizer (which provides pen point position information) must be layered onto the display. Of course, the digitizer can be layered beneath the display, but then other distortion problems occur. In the short term, vendors will make their individual decisions on how to layer the display and digitizer. In the long term, technology will probably couple the display and digitizer and the problem will go away.

New display technology that seems ideal for notebook computers is rapidly coming along and may, in the future, replace LCDs. Among these technologies are “cold cathode” or “field emission cathodes” displays that require 20 to 30 times less power and can be as little as 3mm thick (Coloray Display Corp. of Fremont, Calif. is pioneering this technology). And we shouldn’t discount twists on LCD either; active matrix LCDs offer promise for color applications. Japanese manufacturers continue to be quite active in LCD research and development.

Memory and Storage

On one hand, the bit-mapped graphics, scalable fonts, and other features mean prodigious storage requirements. Consider that GO’s PenPoint operating system alone is about 3 Mbytes in size while it is expected that most applications written for it will require another 100 to 200 Kbytes of memory. And we haven’t even gotten to the data that needs to be stored!

On the other hand, the size and power limitations of pen-based systems rule out current storage subsystems used with laptop computers. Instead, you can initially expect stylus systems to implement solid-state storage — typically 4 to possibly 32 Mbytes of either static (SRAM) or dynamic (DRAM) main memory (or a combination of the two). SRAM provides more efficient use of battery power since static memory draws little or no power except when it is being accessed. DRAM, on the other hand, can be implemented with fewer chips, important when size is critical.

As mentioned earlier, operating systems for pen-based systems will have built-in power-management and storage conservation capabilities. Code sharing and re-use, and on-the-fly file compression/decompression will make maximum use of whatever solid state storage is available. Over the past year, hard disks small enough to fit into notebook computers have become common. In fact, nearly 200,000 2.5-inch hard disk drives had been shipped by the end of 1989. Areal Technology of San Jose, Calif., for instance, is shipping a 2.5-inch, 60 Mbyte hard disk that is 1/2 inch high, 3 inches in length, and weighs 5 ounces. The drive, which implements a single-platter glass disk, has 22 ms access time and draws only 3 Watts on spin up and 1.5 Watts in idle mode. Later this year, the company will introduce a 125 Mbyte hard disk in the same form factor. Additionally, we should see the adoption of 1.8-inch hard disk drives over the coming months.

Flash memory cards also hold promise. These solid-state devices, which currently provide as much as 10 Mbytes of storage, weigh less than an ounce and draw much less power than any hard disk. You can also assume that systems will have “non-mobile” auxiliary storage such as external 3.5-inch floppy or hard disk drives. These can be connected to the mobile system via standard connectors.

Wireless Networks

An important component to the success of pen-based computing is the emergence of the wireless network. The ability to communicate at network speeds without hard wiring will allow “mobile professionals” to exchange data from the field with desktops or even mainframes at the home office. NCR (who is knee-deep in pen-based system development) recently announced a spread spectrum wireless network that runs on Novell Netware with transmission rates as high as 2 Mbits per second. Other important players include Agilis Systems (another pen-based pioneer) and Telesystems.

An important by-product of wireless network technology will be the introduction of wireless fax transmission. This is technology in its infancy, but wireless fax will be an important component of pen-based computing. (GO’s PenPoint has fax support built into the operating system.)

The hardware challenges of pen-based computing present great opportunities to innovative hardware companies. Although the Japanese have a substantial lead in laptop technologies like flash memory and active matrix displays, there are nevertheless opportunities for new companies to make some big breakthroughs. Momenta’s power supply could be an example. Coloray’s cold cathode tube could be another. And there is still a lot of work to be done to increase the range, reliability, and transmission speeds of wireless networks.

Quotes From The Field: “Our goal is to evolve the existing PC system standards to include new capabilities such as compound documents, object-oriented file systems, distributed file systems, handwriting recognition, and multimedia. A nonstandard implementation might offer a short-term advantage for a particular feature, but a better solution would be to incorporate new technologies directly into the PC architecture or systems software.” — Bill Gates, Microsoft