Here are some suggestions for how to improve the Dell Inspiron 6000d:
There isn't one, hiss, boo. In the modern world it's essential for laptop users to use good cryptographic keys for secure TCP (TLS) connections, for VPNs, and for protecting data when the laptop is stolen. Just today (2005-03-29) a UC Berkeley staff person had his laptop stolen with unencrypted personal information on it about over 90,000 graduate school applicants. Dell should do their part to help users protect their data properly, by including the hardware random number generator on every machine, particularly on laptops. You did it on the Dimension 4600; why not on the Inspiron 6000d?
Given that USB flash memory devices are so cheap, I'm surprised that you used the Ricoh R5c822 rather than a USB interface. With USB you would not need a special driver; both Windows and Linux would be able to use the SD chip with existing drivers and user training. In any case, you should announce the chip type so the Linux driver people can get it supported.
A good feature of the SD slot is that the card stays mostly inside the machine, unlike a USB device which sticks out and is easily damaged. This arrangement should be retained in the future.
The fan tachometer, the thermal management setpoints, and all the thermal sensor readings should be exported in the ACPI DSDT, just like on a server. Users should be allowed to set these within reasonable limits. In particular, for some people even the very quiet low speed fan may be a problem, and they should have the choice of totally passive cooling.
The datasheet for the ICH-6PM chipset says that the internal temperature sensor is accurate to ±13 C, which is not particularly accurate. I would be willing to spend a few bucks extra for a thermistor on the mobo (which doesn't monitor the chip temperature but does know about blocked airflow or high ambients); I know a model that is linear to ±2 C over 0 to 100 C and is not that expensive. And the description suggests that they may be cueing on the raw forward voltage of the sensor diode, which would have major issues of manufacturing tolerances. Why isn't Intel using the voltage difference at two different currents, which is exactly linearly proportional to the absolute temperature, with no adjustable parameters?
It sticks out two inches from the back of the machine, right into this person's knee, so it's definitely vulnerable to putting damaging torque onto the motherboard. Certainly a right-angle connector should be used, and given the possibility of damage to the mobo, you might consider pro-actively replacing the cable or giving everyone a cable extender with a right-angle bend. I'm certainly going to look for one at my local computer store.
The case of a laptop is important both functionally and in selling the machine to someone who sees the machine on another owner's lap. Here is a list of case issues that need to be worked on.
There is sometimes a vibration in the keyboard, possibly against the case edge, or the optical drive may be involved. The keyboard needs not only to be solid, but to feel and sound solid.
While the lid latch seems secure, it should hold the lid positively pressed against the upper deck, i.e. the rubber buttons should be compressed so the lid can't move. As it is, there is about 1/2 mm of jiggle room when the lid is closed, which is not reassuring even if the lid isn't really going to come loose. In reviews of the Inspiron 4100, unkind adjectives were used for a similar situation, and you don't want that kind of advertising.
Obviously the rubber buttons on the lid cover the screws that hold it together. Whenever the buttons are peeled off to expose the screws, the glue degrades, and quickly enough they won't stay on. While exposed screws may not be quite as aesthetic as covering them with the buttons, that is how you should design the machine for maintainability.
The deck and, particularly, the bottom plate could use some stiffening around the optical drive. If the machine is sitting wrong on my leg, the bottom can bind, preventing the drive from opening.
Here's something done right: the keyboard and mouse buttons are even with the deck within 1/2 mm, while the screen is recessed behind the lid frame by 1 mm, and the rubber buttons around the lid frame are 1.5 mm high, so there's generous clearance between the screen and the keyboard. On the Inspiron 4100 the mouse buttons would rub on the screen, damaging it. I ended up sticking a strip of cloth over the offending buttons.
Qualify and offer 533 MHz memory. According to a co-worker, Dell is notorious for offering fast CPUs and chipsets but slow memory. His speculation is that Dell is very value conscious and they need to test the tolerances and error rate of the memory they offer, from whatever vendor they currently are getting the lowest price from. This takes work, and in some cases the test program is not done on time or is never done, so the faster memory cannot be offered. Dell should make a policy that if a particular FSB speed is offered, matching memory should also be available, and people should be encouraged to get it. Of course Dell will charge extra for the faster memory.
The display panel uses 6 watts of power, most of which goes to the lamp. At this power level, when the machine is used outdoors the screen is just barely readable. The human visual system is designed to look at targets illuminated by ambient light, e.g. prey, predators, or other people. It has automatic color and brightness compensation for varying ambient light, such as sky brightness vs. direct sun vs. tungsten lamps. It is not designed to achieve its goals on self-luminous objects, limited in nature to the sun, moon, stars and campfires. PDAs like the Sony Treo have a passive display viewed by the ambient light, so they are equally readable and have equally good color rendition in direct sunlight versus dim artificial illumination, and (important for mobile devices) there is zero power to the lamp unless you turn on the optional edge lighting. Dell should commission a large format passive display panel and use it in their laptops.
The only problem with the display on the Treo is that it has 33% reflectance max (according to the Samsung data sheet for the panel). But a gray-looking readable screen is a whole lot better than 185 nits of active illumination in a full sunlight environment. I'm not sure if 33% is an intrinsic limit; if the panel vendors knew that Dell was getting serious about passive laptop screens, they might have an alternate technology allowing the reflectance to be raised substantially.
According to the ACPI BIOS, when the battery is being charged and it reaches about 95% of nominal charge, the regulator gates the charging current on (2 amps) or off (zero) until the charge is full. According to Sony, the correct way to charge a LiIon battery is to feed current into it so it will charge in the time specified on the data sheet (1 hour for polymer technology), but when the voltage reaches the maximum allowed value, 4.20 volts per cell, which happens at about 80% charge, the voltage should be held constant. The battery will then reach 99% of full charge in another hour. While gated charging is probably not harmful as long as the voltage stays below 4.20 volts, likely it takes longer to charge the battery than voltage regulated charging.
Military users, field repair people, field
biologists, and families with small children and boisterous pets have special
needs in computers. You could probably sell a ton of low-end machines to the
same market segment that buys
Barbie's Computer, and the
Army for the Information Age is going to need a lot of higher-end,
higher-profit machines. Here is a short list of features that would make the
machine more rugged. I recommend many of them for the normal models as
While ABS is very strong, to my mind laptop cases should be fiber reinforced, e.g. epoxy glass. Particularly, make the effort to get glass cloth into interior ribs and stiffening members. Graphite fiber reinforcement is more expensive, but it also is electrically conductive, helping you to meet TEMPEST (electromagnetic radiation) goals.
The laptop's glass screen is its most expensive and most vulnerable single component. It needs to be much better protected both from scratches and from breakage from mechanical deformation. The screen should be constructed as a sandwich like this, from the viewing side:
Is Van Eck phreaking really a threat for flat panels, or is it restricted to CRTs? In any case, TEMPEST goals are a lot easier to meet if the basic model is designed from the ground up for minimum electromagnetic radiation, and particularly, minimum radiation useful to an enemy.
Backlighting the display puts a lot of demands on the lid. The space could be used for strength members if the display were passive only (no backlighting at all). Alternatively, make the lid entirely out of an egg crate construction of transparent polycarbonate strength ribs, painted white on the back surface. Your goal should be that a large St. Bernard dog, child or gunnery sergeant could sit on the lid without cracking anything or breaking the display.
All open holes for the modem, network, PCMCIA, etc. should have elastomeric plugs to keep out sand, liquids and insects. The battery needs a good seal too. I had ants running in and out of my Inspiron 4100 with their dirty little feet, and the Inspiron 6000d will be no better this summer.
It should be possible to seal the keys against liquids and dust by using a membrane, with the springs on the outside. Hall effect (magnetic) keys are probably best; capacitive coupled keys are a second choice due to electromagnetic emission (TEMPEST).
While in a conventional laptop it's a good move to use the fan to draw air under the keyboard and over the memory, ASICs and power converter, that area needs to be sealed in a ruggedized machine. An additional heat pipe dedicated to general cooling, and a fan to circulate air within the sealed compartment, could keep the ruggedized machine cool.