| Previous | Table of Contents | Next |
Of course, each PC also has pluggable components in the form of expansion cards (modems, network cards, video cards, and so on). There are also pluggable microchips that usually have a two rows of 8 to 16 pins that fit into a similarly sized socket. Each of these types of components is susceptible to physical movement or “chip creep” due to the metal expansion and contraction that happens during heating and cooling. What’s more, contacts that aren’t made of gold are also susceptible to oxidation, much the same way iron rusts.
Any pluggable component (even a cable) is susceptible to these two problems. Fortunately, two simple solutions exist. Many times, pressing down on chips or boards can make an unreliable or intermittent problem go away. If that fails, you can always reseat the component by pulling it out and putting it back in. Chips don’t usually need reseating; typically, pushing down on them will do the trick.
If you choose to reseat a chip, be very careful when pulling it out. Otherwise, you can ruin it by bending the small pins. Also, be careful not to allow the pins to fold underneath the chip. You’ll want to make certain that each pin aligns correctly with its socket hole.This problem isn’t as bad with modern CPUs because their ZIF (zero insertion force) design does not require you to exert any force on the chip. You’ll want to make sure the lever next to the CPU gets raised before you start yanking on the chip—this releases the tension on the CPU.
Therefore, if your PC seems totally dead, you can probably reseat your CPU without worrying too much about hurting anything.
You can cause serious damage to electronic components if you work on them while they’re turned on. Make sure any device you’re about to work on is powered down.
The interface cards that connect to your motherboard aren’t as fragile. You can pull them out and push them back in; this will sometimes fix an intermittent connection or knock dirt away from the pins.
Electronic components are extremely sensitive to even the smallest amount of static electricity. You must “ground” yourself before touching any components. By touching a metallic ground, you discharge any static electricity that may have built up on your body while walking around. Most computers have a three-prong outlet that includes a ground. If you leave your computer plugged in (but turned off) while working on it, you can ground yourself by touching its internal metal chassis.
Cleaning a component or PC can also help clear up odd hardware problems. Dust and dirt can seriously impede the performance and/or reliability of a component. If you’re not lucky enough to have a climate-controlled data center, dust will likely be one of your enemies.
Some people get good results from using a vacuum cleaner to remove excess dust and dirt from the inside of their machines’ cases. However, an overly powerful vacuum cleaner can suck up loosely connected jumpers, so be careful. Also a vacuum can create static electricity, which can ruin your whole day.Although buying canned air is more expensive than using an existing vacuum cleaner, using canned air can give better results because it’s portable and you can aim it with pretty good accuracy. However, you don’t get rid of the dirt this way; you merely relocate it.
Some people who work with electronics like to use electronic cleaner spray on just about any dusty component. I don’t like to use this kind of cleaner on network or computer components—maybe I’m squeamish, but apart from the fact that this stuff tears a hole in the ozone the size of Kentucky, it seems like overkill to me to spray a volatile compound all over your gear. Most times, the aforementioned methods work just fine.
You can use a soft pencil eraser on the gold contacts of your interface cards to clean oxidation off them.
You would be amazed at what can be fixed via a simple hardware swap. As a perfect example of black box troubleshooting, most hardware repair is accomplished by swapping circuit boards and other distinct and separate components. Even pros don’t engage in component-level repair anymore; the days of a field technician soldering a new capacitor or microchip onto a broken circuit board have been gone for some time now. To save time, people in the field repair business simply swap out a suspect component for a new, presumably working component.
It always bothered me that people never repaired these types of things—until I looked at the costs involved. At anywhere from $65–$150 an hour for repairs, it’s insanely expensive to repair cheap ($10–$50) electronic gear rather than replace it. Why spend $100 to fix your $40 Ethernet card? You can buy two for the price of fixing it. This probably doesn’t make Greenpeace happy, but it sure makes sense economically.
This is even something you can do yourself. Suppose, for example, you’ve troubleshot a network application’s problem to a particular workstation, and you’ve ruled out the user login and network files.
You’ve tried to run the Microsoft Exchange client for a user on her workstation, but got a fatal exception error. You logged her into a couple of other workstations, and she worked fine every time.
Therefore, you know that it’s a hardware problem, but you don’t pass the buck—you’re interested in finding out what the hardware problem is.
This problem is a good candidate for swapping components. Sometimes it can be obvious what to swap first—for example, if someone’s Microsoft Word is repainting the screen badly on one workstation, but not another, you would probably swap the video card first. In this example, it’s hard to say what might be causing the fatal exception; fatal exceptions are caused by bad operating system components, local applications, faulty hard drives, bad video cards or drivers, and so on.
While noting whether the problem remains, you swap the following items:
When you swap the memory, the fatal exception error goes away. Furthermore, when you install this memory into a known good PC, that PC starts exhibiting the same lockup problem when you run the network application. It doesn’t matter how the memory is broken—that is, whether it has bad circuitry on it somewhere or is simply somewhat incompatible with this brand of PC/motherboard. It only matters that it is broken.
| Previous | Table of Contents | Next |