Vous êtes sur la page 1sur 26




CONTENTS AT A GLANCE The Universal Troubleshooting Process

Define your symptoms Identify and isolate Replace Re-test

Viruses and Computer Service

Computer viruses explained The tell-tale signs Anti-virus software Sterilizing your shop

The Spare Parts Dilemma

Parts are always changing Inventory costs money A better Strategy

Quick-Start Bench Testing

The symptom doesnt start at all The system starts but wont initialize The system starts but crashes/ reboots intermittently After an upgrade Windows 95 boot symptoms

Benchmarking the PC
Avoiding benchmark problems Obtaining benchmarks

Further Study

As a PC technician, you must understand a basic rule of businesstime is money.

Whether you are the boss or work for someone else, the ability to identify and isolate a PC or peripheral fault quickly and decisively is a crucial element to your success. It requires a keen eye, some common sense, and a little bit of intuition. It also requires an understanding



of the troubleshooting process, and a reliable plan of action. Even though the number of PC configurations and setups are virtually unlimited, the methodology used to approach each repair is always about the same. This chapter is intended to illustrate the concepts of basic troubleshooting and show you how to apply a suite of cause-and-effect relationships that will help you narrow the problem down before you even take a screwdriver to the enclosure. By applying a consistent technique, you can shave precious time from every repair.

The Universal Troubleshooting Process

Regardless of how complex your particular computer or peripheral device might be, a dependable troubleshooting procedure can be broken down into four basic steps (Fig. 4-1): define your symptoms, identify and isolate the potential source (or location) of your problem, replace the suspected sub-assembly, and re-test the unit thoroughly to be sure that you have solved the problem. If you have not solved the problem, start again from Step #1. This is a universal procedure that you can apply to any sort of troubleshootingnot just for personal computer equipment.


When a PC breaks down, the cause might be as simple as a loose wire or connector, or as complicated as an IC or sub-assembly failure. Before you open your tool box, you must have a firm understanding of all the symptoms. Think about the symptoms carefullyfor example:
s Is the disk or tape inserted properly? s Is the power or activity LED lit? s Does this problem occur only when the computer is tapped or moved?

Define your symptoms Identify and isolate Repair or replace


Problem solved

Return to service


The universal troubleshooting procedure.


By recognizing and understanding your symptoms, it can be much easier to trace a problem to the appropriate assembly or component. Take the time to write down as many symptoms as you can. This note-taking might seem tedious now, but once you have begun your repair, a written record of symptoms and circumstances will help to keep you focused on the task at hand. It will also help to jog your memory if you must explain the symptoms to someone else at a later date. As a professional troubleshooter, you must often log problems or otherwise document your activities anyway.



Before you try to isolate a problem within a piece of computer hardware, you must first be sure that the equipment itself is causing the problem. In many circumstances, this will be fairly obvious, but some situations might appear ambiguous (i.e., there is no power, no DOS prompt, etc.). Always remember that a PC works because of an intimate mingling of hardware and software. A faulty or improperly configured piece of software can cause confusing system errors. Chapter 3 touched on some of the problems that operating systems can encounter. When you are confident that the failure lies in your systems hardware, you can begin to identify possible problem areas. Because this book is designed to deal with sub-assembly troubleshooting, start your diagnostics there. The troubleshooting procedures throughout this book will guide you through the major sections of todays popular PC components and peripherals, and aid you in deciding which sub-assembly might be at fault. When you have identified a potential problem area, you can begin the actual repair process and swap the suspect sub-assembly.

Because computers and their peripherals are designed as collections of sub-assemblies, it is almost always easier to replace a sub-assembly outright, rather than attempt to troubleshoot the sub-assembly to its component level. Even if you had the time, documentation, and test equipment to isolate a defective component, many complex parts are proprietary, so it is highly unlikely that you would be able to obtain replacement components without a significant hassle. The labor and frustration factor involved in such an endeavor is often just as expensive as replacing the entire sub-assembly to begin with (perhaps even more expensive). On the other hand, manufacturers and their distributors often stock a selection of sub-assemblies and supplies. You might need to know the manufacturers part number for the sub-assembly to obtain a new one. During a repair, you might reach a roadblock that requires you to leave your equipment for a day or two, or maybe longer. This generally happens after an order has been placed for new parts, and you are waiting for those parts to come in. Make it a point to reassemble your system as much as possible before leaving it. Gather any loose parts in plastic bags, seal them shut, and mark them clearly. If you are working with electronic circuitry, be sure to use good-quality anti-static boxes or bags for storage. Partial re-assembly (combined with careful notes) will help you remember how the unit goes together later on. Another problem with the fast technological progress we enjoy is that parts rarely stay on the shelf long. That video board you bought last year is no longer available, is it? How



about that 4 CD-ROM drive you put in some time back? Today, theres something newer and faster in its place. When a PC fails and you need to replace a broken device, chances are that youll need to upgrade simply because you cannot obtain an identical replacement device. From this standpoint, upgrading is often a proxy of troubleshooting and repair.

When a repair is finally complete, the system must be reassembled carefully before testing it. All guards, housings, cables, and shields must be replaced before final testing. If symptoms persist, you will have to reevaluate the symptoms and narrow the problem to another part of the equipment. If normal operation is restored (or greatly improved), test the computers various functions. When you can verify that the symptoms have stopped during actual operation, the equipment can be returned to service. As a general rule, it is wise to let the system run for at least 24 hours to ensure that the replacement sub-assembly will not fail prematurely. This is known as letting the system burn in.
Check your CD for the burn-in utilities 486TST.ZIP and BURNIN43.ZIP.

Do not be discouraged if the equipment still malfunctions. Perhaps you missed a jumper setting or DIP switch, or maybe software settings and device drivers need to be updated to accommodate the replacement sub-assembly. If you get stuck, simply walk away, clear your head, and start again by defining the current symptoms. Never continue with a repair if you are tired or frustratedtomorrow is another day. Even the most experienced troubleshooters get overwhelmed from time to time. You should also realize that there might be more than one bad assembly to deal with. Remember that a PC is just a collection of assemblies, and each assembly is a collection of parts. Normally, everything works together, but when one assembly fails, it might cause one or more interconnected assemblies to fail as well.

The Spare Parts Dilemma

Once a problem is isolated, technicians face another problem: the availability of spare parts. Novice technicians often ask what kinds and quantity of spare parts they should keep on hand. The best answer to give here is simply: none at all. The reason for this somewhat drastic answer is best explained by the two realities of PC service:


After only 15 years or so, the PC is in its sixth CPU generation (with such devices as the AMD K6 and Intel Pentium II). As a result, a new generation matures every 24 to 36 months (although the newer generations have been arriving in 18 to 24 months). Even the standardized products, such as CD-ROM drives, have proliferated in different speeds and versions (8, 10, 12, 16, and even 20 speeds). Once production stops for a drive or board, stock rarely remains for very long. You see, even if you know what the problem is, the chances of your locating an exact replacement part are often quite slim if the part is more than two years old. Notice the word exactthis is the key word in PC repair. This


is the reason why so many repairs involve an upgrade. For example, why replace a failed EGA board with another EGA board when you can install an SVGA board (which is typically EGA compatible) for the same price or less? Choosing the right parts to stock is like hitting a moving target, so dont bother.



Financial considerations also play a big role in choosing parts. For computer enthusiasts or novice technicians just tinkering in their spare time, the expense and space demands required for inventory are simply out of the question. Even for more serious businesses, inventory can burden the bottom line.

Unless you are in the business of selling replacement parts and upgrade components yourself, dont waste your money and space stocking parts that are going to be obsolete in less than 24 months. Rather than worry about stocking parts yourself, work to develop your contacts with computer parts stores and superstores that specialize in PC parts and sub-assemblieslet them stock the parts for you. Because parts stores generally have an inside line with distributors and manufacturers, parts that they do not stock can often be ordered for you. Even many reputable mail-order firms can provide parts in under 48 hours with todays delivery services.

Benchmarking the PC
We all know that todays personal computers are capable of astounding performance. If you doubt that, consider any of the current 3D games, such as Quake II or Monster Truck Madness. However, it is often important to quantify the performance of a system. Just saying that a PC is faster than another system is simply not enoughwe must often apply a number to that performance to measure the improvements offered by an upgrade, or to objectively compare the performance of various systems. Benchmarks are used to test and report the performance of a PC by running a set of well-defined tasks on the system. A benchmark program has several different uses in the PC industry depending on what youre needs are:
s System comparisons Benchmarks are often used to compare a system to one or more

competing machines (or to compare a newer system to older machines). Just flip through any issue of PC Magazine or Byte, and youll see a flurry of PC ads all quoting numerical performance numbers backed up by benchmarks. You might also run a benchmark to establish the overall performance of a new system before making a purchase decision. s Upgrade improvements Benchmarks are frequently used to gauge the value of an upgrade. By running the benchmark before and after the upgrade process, you can get a numerical assessment of just how much that new CPU, RAM, drive, or motherboard might have improved (or hindered) system performance. s Diagnostics Benchmarks sometimes have role in system diagnostics. Systems that are performing poorly can be benchmarked as key components are checked or reconfigured. This helps the technician isolate and correct performance problems far more reliably than simple visual observations.




One of the most serious problems encountered with benchmarks is the integrity of their numbers. Youve probably heard that statistics can lie, and the same thing is true of benchmarks. In order for benchmarks to provide you with reliable results, you must take some precautions:
s Note the complete system configuration When you run a benchmark and achieve a re-

sult, be sure to note the entire system configuration (i.e., CPU, RAM, cache, OS version, etc.). s Run the same benchmark on every system Benchmarks are still software, and the way in which benchmark code is written can impact the way it produces results on a given computer. Often, two different versions of the same benchmark will yield two different results. When you use benchmarks for comparisons between systems, be sure to use the same program and version number. s Minimize hardware differences between hardware platforms A computer is an assembly of many interdependent sub-assemblies (i.e., motherboard, drive controllers, drives, CPU, etc.), but when a benchmark is run to compare a difference between systems, that difference can be masked by other elements in the system. For example, suppose youre using a benchmark to test the hard-drive data transfer on two systems. Different hard drives and drive controllers will yield different results (thats expected). However, even if youre using identical drives and controllers, other differences between the systems (such as BIOS versions, TSRs, OS differences, or motherboard chipsets) can also influence different results. s Run the benchmarks under the same load The results generated by a benchmark do not guarantee that same level of performance under real-world applications. This was one of the flaws of early computer benchmarkingsmall, tightly written benchmark code resulted in artificially high performance, but the system still performed poorly when real applications were used. Use benchmarks that make use of (or simulate) actual programs, or otherwise simulate your true workload.

Benchmarks have been around since the earliest computers, and there are now a vast array of benchmark products to measure all aspects of the PCas well as measure more specialized issues, such as networking, real-time systems, and UNIX (or other operating system) platforms. Table 4-1 highlights a cross-section of computer benchmarks for your reference. In many cases, the table includes a URL or FTP site where you can obtain source code for the benchmark, or download the complete benchmark program. Today, Ziff Davis and CMP publish a suite of freeware benchmark utilities that have become standard tools for end users and technicians alike.
Use caution when exploring the benchmarks of Table 7-1. Many classic benchmarks will no longer work on todays PC platforms or might require specialized hardware configurations. Stay with the end-user benchmarks that are highlighted with an asterisk(*).

BENCHMARKING THE PC 77 TABLE 4-1 AN INDEX OF COMPUTER BENCHMARKS GETTING STARTED SOURCE/APPLET AVAILABILITY ftp.cs.wisc.edu/007 http://www8.zdnet.com/ pcmag/pclabs/bench/ bench3d.htm http://www.sysopt.com/ 3dbench.html http://www.sysopt.com/ cbench.html http://www8.zdnet.com/ pcmag/pclabs/bench/ benchbm.htm http://www.sysopt.com/ pub/busperf.zip http://www.sysopt.com/ cachk4.html ftp://oak.oakland.edu/Sim Tel/msdos/sysinfo/ ctest259.zip ftp://swedishchef.lerc.nasa. gov/drlabs/cpu/ (select cpu2.unix.tar.Z or cpu2.vms.tar.Z) ftp.nosc.mil/pub/aburto ftp.nosc.mil:pub/aburto/ c4.shar ftp.nosc.mil/pub/aburto ftp.nosc.mil/pub/aburto ftp.sei.cmu.edu/pub/hartstone ftp.nosc.mil/pub/aburto http://www8.zdnet.com/ pcmag/pclabs/bench/ benchjm.htm netlib.att.com/netlib/ benchmark/livermore netlib.att.com/netlib/ benchmark/linpack ftp nosc mil/pub/aburto

BENCHMARK 007 (ODBMS) 3D WinBench 97*

DESCRIPTION/PURPOSE designed to simulate a CAD/CAM environment PC 3D system benchmark


PC 3D graphics benchmark

AIM BatteryMark*

overall performance and multitasking throughput mobile PC battery benchmark

Bonnie Busperf Byte CacheChk CompTest

bottleneck checking PC bus-performance benchmark UNIX performance PC cache-checking benchmark general PC benchmark


floating-point benchmark (UNIX/VMS)

Dhrystone MIPS Fhourstones Flops Hanoi Hartstone Heapsort IOBENCH IOZONE JMark*

short test for system programming integer-only benchmark MFLOP rating benchmark recursive-function benchmark real-time benchmark array-sorting benchmark multistream benchmark read/write test PC Java virtual machine benchmark multipurpose benchmark general-performance testing algebraic processing matrix multiplication benchmark

Khornerstone LFK (Livermore Loops) LINPACK Matrix Multiply



TABLE 4-1 AN INDEX OF COMPUTER BENCHMARKS (CONTINUED) SOURCE/APPLET AVAILABILITY ftp.nosc.mil/pub/aburto monu1.cc.monash.edu.au/ pub/musbus.sh ftp.cs.wisc.edu/wwt/ Misc/NAS http://www8.zdnet.com/ pcmag/pclabs/bench/ benchnb.htm ftp://ftp.cup.hp.com/dist/ networking/benchmarks eos.arc.nasa.gov http://www8.zdnet.com. pcmag/pclabs/bench/ benchsb.htm otis.stanford.edu/pub/ benchmarks/c/small/sieve.c tantalus.al.iastate.edu/pub/ Slalom/ ftp.nosc.mil/pub/aburto ftp.inria.fr:/system/benchmark/SSBA/ssba1.22F.tar.Z ftp.nosc.mil/pub/aburto http://www.sysopt.com/pub/ vidspd40.zip http://www8.zdnet.com/ pcmag/pclabs/bench/ benchweb.htm netlib.att.com/netlib/benchmark/whetstone http://www8.zdnet.com/ pcmag/pclabs/bench/ benchwb.htm http://www8.zdnet.com/ pcmag/pclabs/bench/ benchw97.htm ftp://ftp.winsite.com/pub/pc/ win3/util/wintch12.zip http://www winmag com/soft

BENCHMARK Matrix Multiply MUSBUS NAS Kernels NetBench*

DESCRIPTION/PURPOSE matrix multiplication benchmark computational fluid dynamics PC networking benchmark

Netperf Nettest Nhfsstone PERFECT RhosettaStone ServerBench*

networking performance networking performance file-server performance PC network-server benchmark

Sieve of Eratosthenes SLALOM SPEC SSBA Stanford SYSmark TFFTDP TPC A/B/C ttcp VidSpeed WebBench*

integer testing CPU-intensive benchmarks UNIX performance compares RISC/CISC app/OS-based benchmark FFT benchmark POS benchmark TCP/UDP performance PC video benchmark PC Internet-browser benchmark

Whetstone WinBench 97*

floating-point benchmark PC subsystem benchmarks


overall Windows 95/NT benchmark Windows benchmark overall PC Windows 95

Wintach Wint ne 97*

BENCHMARKING THE PC 79 TABLE 4-1 AN INDEX OF COMPUTER BENCHMARKS (CONTINUED) GETTING STARTED SOURCE/APPLET AVAILABILITY ftp.cs.wisc.edu/wwt/ http://www.winmag.com/softMi /NAS ware/wt97.htm wpi.wpi.edu netcom.com/pub/micromed/ uploads/xstones.summary.z

BENCHMARK NAS Kernels Wintune 97* WPI Benchmarks Xstone

DESCRIPTION/PURPOSE computational fluid dynamics overall PC Windows 95 benchmark general benchmarks general benchmarks

*See below for a more detailed description

Winstone 97 Winstone 97 is a 32-bit Windows benchmark test used to gauge a PCs

overall performance under Windows 95 or Windows NT. Winstone has two components: business and high-end. The business Winstone test measures the time a PC requires to execute a set of application scripts that exercise eight best-selling applicationsit then weights the test timings and converts it to a relative score (the score is relative to the performance of a Gateway 486DX2/66 with 16MB of RAM, whose score is defined as 10). The high-end Winstone measures the time a PC takes to execute a set of application scripts that exercise six applications in CAD, applications development, image editing, and 3-D visualization. These tests weight a given machines timings equally, derive a composite number, and convert this number to a relative score (the score is relative to the performance of a Dell Pentium/100 system with 32MB of RAM, whose score is defined as 10).
WinBench 97 WinBench 97 is a more hardware-oriented utility that provides a detailed

measure of graphics, disk, processor, CD-ROM, and video-playback performance under Windows 95 and Windows NT. WinBench produces two graphics. WinMark 97 scores that reflect performance of a machines graphics subsystem: the business graphics WinMark 97 score reflects performance when running the typical business applications. The high-end graphics WinMark 97 score reflects performance when running the corresponding test applications. WinBench also tests disk-subsystem performance, producing a business-disk WinMark 97 score, and a high-end disk WinMark 97 score. WinBench 97 also provides two scores indicating the speed of a processor subsystem (including CPU, secondary cache, and system RAM). The CPUmark32 score reflects the speed of a PCs processor subsystem under a 32-bit operating system, and the CPUmark16 score reflects the speed of this subsystem under a 16-bit operating system. WinBench 97 includes CD-ROM tests based on a profile of six of todays most popular Windows CDROMs (including sequential-read tests, access-time tests, and CPU utilization tests). A variety of CD video-playback tests report performance replaying a number of video clips.
3D WinBench 97 3D WinBench 97 is a specialized benchmarking utility that measures

the performance of a 3D graphics subsystem (including the Direct3D software, the monitor, the graphics adapter, the graphics driver, and the bus used to carry information from the graphics adapter to and from the processor subsystem). You can use 3D WinBench 97 to test hardware graphics adapters, drivers, and the value of such processor-enhancement technologies as MMX.



BatteryMark BatteryMark 2.0 uses a combination of hardware and software to measure

the battery life of notebook computers under real-world conditions (the hardware used in BatteryMark 2.0 is the same ZDigit II device required by version 1.0). BatteryMark exercises a different 32-bit software workload engines for processor, disk, and graphics tasks. BatteryMark mixes these workloads together and adds periodic breaks in the work that reflect the way users pause while working. BatteryMark 2.0 works with Advanced Power Management (APM) under Windows 95.
NetBench NetBench 5.01 is our benchmark test for checking the performance of net-

work file servers. NetBench provides a way to measure, analyze, and predict how a file server will handle network file I/O requests. It monitors the response of the server as multiple clients request data, and reports the servers total throughput. To test application servers, you should use the ServerBench utility instead.
ServerBench ServerBench 4.0 is the latest version of Ziff-Davis standard benchmark

for measuring the performance of servers in a true client/server environment. ServerBench clients make requests of an application that runs on the serverthe servers ability to service those requests is reported in transactions per second. ServerBench 4.0 runs on IBMs OS/2 Warp Server, Microsofts Windows NT Server 4.0 (for both Digital Alpha and x86-compatible processors), Novells NetWare 4.11, Suns Solaris 2.5 on SPARC, and SCOs OpenServer Release 5 and UnixWare 2.1. To test network file servers, use the NetBench utility instead.
WebBench WebBench 1.1 is the Ziff Davis benchmark test for checking performance

of Web-server hardware and software. Standard test suites produce two overall scores for the server: requests per second and throughput (as measured in bytes per second). WebBench includes static testing (which involves only HTML pages), and dynamic testing (including CGI executables, Internet Server API libraries, and Netscape Server API dynamic link libraries.
JMark JMark 1.01 is a suite of 11 synthetic benchmark tests for evaluating the perfor-

mance of Java virtual machines. The JMark 1.01 suite simulates a number of important tests of Java functionality. It includes Java versions of a number of classic benchmark test algorithms, as well as tests designed to measure graphics performance in a GUI environment. You can download JMark 1.01 from Ziff Davis, or run the tests online within your browser.
Wintune 97 Wintune 1.0 for Windows 95/NT is a recent benchmark entry from CMP,

the publishers of Windows Magazine. Wintune 97 is an overall benchmark to measure Windows 95/NT performance. It has a fast user interface that allows the program to load much faster than the earlier Wintune 95, and will now support testing of the latest Pentium II systems. Wintune 97 tests video systems on the fastest new computers at full-screen resolution.
Check your CD for the benchmarking utility JBENCH.EXE.


Viruses and Computer Service

Few developments in the personal computer field have caused more concern and alarm than the computer virus. Although viruses do not physically damage computer hardware, they can irrevocably destroy vital data, disable your PC (or shutdown a network), and propagate to other systems through networks, disk swapping, and on-line services. Even though virus infiltration is generally regarded as rare, a good PC technician will always protect themselves (and their customers) by checking the system for viruses before and after using their diagnostic disks on the PC. A careful process of virus isolation can detect viruses on the customers system before any hardware-level work is done. Virus-isolation tactics also prevent your diagnostic disks from becoming infectedand subsequently transferring the virus to other systems (for which you might be legally liable). This section of the chapter outlines a virus-screening procedure for PCs. Chapter 45 covers the symptoms and countermeasures for viruses in more detail.



There have been many attempts to define a computer virus, and most definitions have a great deal of technical merit. For the purposes of this book, however, you can consider a virus to be some length of computer code (a program or program fragment) that performs one or moreoften destructivefunctions and replicates itself wherever possible to other disks and systems. Because viruses generally want to escape detection, they might often hide by copying themselves as hidden, system, or read-only files. However, this only prevents casual detection. More-elaborate viruses affect the boot sector code on floppy and hard disks, or attach themselves to other executable programs. Each time that the infected program is executed, the virus has a chance to wreak havoc. Still other viruses infect the partition table. Most viruses exhibit a code sequence that can be detected. Many virus scanners work by checking the contents of memory and disk files for such virus signatures. As viruses become more complex, however, viruses are using encryption techniques to escape detection. Encryption changes the virus signature each time the virus replicates itselffor a well-designed virus, this can make detection extremely difficult. Just as a biological virus is an unwanted (and sometimes deadly) organism in a body, viral code in software can lead to a slow, agonizing death for your customers data. In actuality, few viruses immediately crash a system (with notable exceptions, such as the much-publicized Michealangelo virus). Most viruses make only small changes each time they are executed and create a pattern of chronic problems. This slow manifestation gives viruses a chance to replicateinfecting backups and floppy disks, which are frequently swapped to infect other systems.
Frequent system backups are an effective protection against computer viruses because you can restore files damaged by viruses. Even if the backup is infected, the infected files can often be cleaned once they are restored from the backup.




Viruses are especially dangerous because you are rarely aware of their presence until it is too late and the damage is already done. However, there are a number of behaviors that might suggest the presence of a virus in your system. Once again, remember that one of the best protections against viruses (or other drive failures) is to maintain regular backups of your data. None of these symptoms alone guarantee the presence of a virus (there are other reasons why such symptoms can occur), but when symptoms do surface, it is always worth running an anti-virus checker just to be safe. The following symptoms are typical of virus activity:
s The hard drive is running out of disk space for no apparent reason Some viruses mul-

tiply by attaching copies of themselves to .EXE and .COM filesoften multiple times. This increases the file size of infected files (sometimes dramatically) and consumes more disk space. If left unchecked, files can grow until the disk runs short of space. However, disk space can also be gobbled up by many CAD, graphics, and multimedia applications, such as video-capture systems. Be aware of what kind of applications are on the disk. You notice that various .EXE and .COM programs have increased in size for no reason This is a classic indicator of a virus at work. Few rational people make keep track of file sizes, but dates can be a giveaway. For example, if most of the files in a sub-directory are dated six months ago when the package was installed, but the main .EXE file is dated yesterday, its time to run that virus checker. You notice substantial hard-drive activity, but were not expecting it It is hardly unusual to see the drive-indicator LED register activity when programs are loaded and run. In disk-intensive systems, such as Windows 95, you should expect to see extensive drive activity because of swap-file operation. However, you should not expect to see regular or substantial disk activity when the system is idle. If the drive runs for no apparent reasonespecially under MS-DOSrun the virus checker. System performance has slowed down noticeably This symptom is usually coupled with low drive space, and it might very well be the result of a filled and fragmented disk, such as those found in systems that deal with CAD and multimedia applications. Run the virus checker first. If no virus is detected, try eliminating any un-needed files and defragment the drive completely. Files have been lost or corrupted for no apparent reason, or there are an unusual number of access problems Under ordinary circumstances, files should not be lost or corrupted on a hard drive. Even though bad sectors will crop up on extremely rare occasions, you should expect the drive to run properly. Virus infiltration can interrupt the flow of data to and from the drives and result in file errors. Such errors might occur randomly or they might be quite consistent. You might see error messages, such as Error in .EXE file. Regular errors might even simulate a drive failure. Try running a virus checker before running a diagnostic, such as ScanDisk. Inadequate power problems can also have an effect on drive reliability. The system locks up frequently or without explanation Faulty applications and corrupted files can freeze a system. Memory and motherboard problems can also result in system lockups. Although viruses rarely manifest themselves in this fashion, it is pos-


sible that random or consistent system lockups might suggest a virus (or virus damage to key files). s There are unexplained problems with system memory or memory allocation Although there might be one or more memory defects, it is quite common for viruses to exist in memory, where other files can be infected. In some cases, this can affect the amount of free memory available to other applications. You might see error messages such as: Program too big to fit in memory. If you are having trouble with free memory or memory allocation, run a virus checker that performs a thorough memory check. If the system checks clear of viruses, you can run diagnostics to check the memory.


In the race between good and evil, evil usually has the head start. As a result, anti-virus detection and elimination packages are constantly trying to keep up with new viruses and their variations (in addition to dealing with more than 2000 virus strains that have already been identified). This leads to an important conclusion about anti-virus softwarethey all quickly become obsolete. Even though first-class shareware and commercial packages can be quite comprehensive, they must all be updated frequently. Some of the most notable anti-virus products are found in Symantecs Norton Anti-Virus and VirusScan from McAfee and Associates. If you use MS-DOS 6.0 or later, you already own Microsoft Anti-Virus (MSAV). Another important factor in anti-virus programs is their inability to successfully remove all viruses from executable (.EXE) files. Files with a .COM extension are simply reflections of memory, but .EXE files contain header information that is easily damaged by a virus (and are subsequently unrecoverable). It is always worth trying to eliminate the virusif the .EXE header is damaged, youve lost nothing in the attempt, and you can reload the damaged .EXE file from a backup or its original distribution disks, if necessary. Remember that there is no better protection against viruses and other hardware faults than keeping regular backups. It is better to restore an infected backup and clean it, than to forego backups entirely.


Sterilization starts by assuming that all machines coming in for service are infected with a virus. You should assume the possibility of an infectioneven if the complaint is something innocent (i.e., the keyboard is acting up). This section of the chapter shows you how to create anti-virus work disks that will be used to boot and check the systems brought in for service. Guard your master anti-virus disks by placing them somewhere ELSE besides the shop. That way they wont be infected accidentally. Immediately write-protect your work disks! Also, be ready to discard your work disks frequently. Replacing a 50-cent work disk is much cheaper than having to scan and clean every disk in your shop! If possible, writeprotect work disks, too. If the anti-virus software, DOS, and the DISKCOPY program can all fit, you should use double-density disks, rather than high-density disks. Double-density disks can be used in high-density drives (but not vice versa). Routine, pre-service virus scanning makes good sense. It will save time by detecting virus-related problems right away. You wont waste time disassembling cabinets and



troubleshooting hardware. Also, eliminating viruses are much easier than reformatting or replacing the hard drive (a devastating choice if your customer has no current backup). Reformatting a hard drive on a system with a virus might not solve the problem and result in a callback. On the other hand, not wiping out your customers entire drive is a sure way to make a friend. Finally, pre-service virus checking is quickthe computer is on the bench anyway. Sticking in a disk and turning on the computer is all the labor required.
In the procedure, it is assumed that your floppy disk is A:, your main hard drive is C:, and your CD-ROM (if installed) is D:. If your particular system is configured differently, please substitute the correct drive letters. 1 Start at the DOS command line. You should exit Windows or Windows 95 before pro-

2 Ensure that your system is virus-free. Run a current virus checker, which checks for the

most important types of virusesincluding memory-resident viruses. Once the system is clean, you can proceed. 3 Format 10 floppy disks as bootable (system) disks. If your diskettes are totally blank, use the FORMAT command, such as:
C:\DOS\> format a: <Enter>

Next, make the diskettes bootable by transferring system files. Use the SYS command to make the diskettes bootable, such as:
<Enter> 3. C:\DOS\> sys a: 3. If you purchase your diskettes pre-formatted, simply use the SYS command. 4 Test a diskette. Reboot your computer and see that the system will boot successfully to the A: DOS prompt. If so, you have created simple boot disks (you need only test one disk), but other steps are required to complete a virus-checking disk. 5 Copy the virus checker to your first bootable floppy disk. Virus checkers are typically self-contained, single-file tools, such as Nortons NAV.EXE, Microsofts MSAV.EXE, or the shareware tool FPROT.EXE. Copy the necessary executable file(s) to your diskette. 6 Create an AUTOEXEC.BAT file that will start the virus checker. Ideally, you want the virus checker to start automatically, so create a simple AUTOEXEC.BAT file that will start the virus checker. For example, MSAV.EXE could use a command line such as: a:\msav.exe

3. You might also add command line arguments to streamline the virus checker even further. Save the AUTOEXEC.BAT file to your floppy disk. 7 Test the diskette again. Reboot the system with your master anti-virus floppy disk. The system should boot cleanwith no drivers or TSRs loaded that might confuse the virus checkerand the anti-virus program should load. Depending on exactly which virus checker and command line options you choose, the checker might run through a complete scan automatically, or you might have to manually start testing from the programs menu. 8 Duplicate the original disk to the other work disks. Use the DOS DISKCOPY command to duplicate your original virus-checking diskette to the other nine diskettes you


have prepared. You might have to swap back and forth between the source (original) and target (new) diskettes several times. When the new diskette is done, DISKCOPY will ask if you want to repeat the procedure. 9 Mark the diskettes carefully. You have just created a batch of anti-virus work disks. They should be immediately write-protected, and kept together as a set.
Step 8 instructs you to create 10 copies of the virus-checking software. Even though the disks are exclusively for your use and you will only use one disk at a time, this kind of multiple duplication might violate the license agreement for your anti-virus software. Be sure that your license allows multiple copies of the software before proceeding. Using the virus work disks Whenever a PC comes in for service, use one of your


anti-virus work disks to boot and check the system first before trying a boot disk or diagnostic disk. Professionals always create anti-virus diskettes in batches because the diskettes are disposable. That is, if a virus is detected and cleaned, the diskette that detected the infection should be destroyed, and you should boot the system with a new work disk to locate any other instances of the same virus or any different viruses. This might seem radical, but it is cheap insurance against cross-contamination of the diskette. Once a system is booted with a work disk and checks clean, you can put that work disk away, and boot the system again with a diagnostic or boot disk as required. It is also advisable to check the PC for viruses again once the repair is complete.
Problems with anti-virus tools The protocol outlined should help to protect you (and your customer) from virus attacks. Still, there are two situations where trouble can occur. First, viruses are proliferating with the aid of powerful new programming languages and vast avenues of distribution, such as the Internet. You will need to update your virus work disks regularly with the very latest anti-virus software. Too often, technicians buy an anti-virus package and continue to use it for years. The software certainly remains adept at detecting the viruses that it was designed for, but it does not take into account the many new strains that crop up regularly. As a result, older virus checkers might allow newer viruses to pass undetected. Second, technicians tend to get cheap with their floppy disks. If a work disk detects and eliminates a virus, it should be considered contaminated, and you should throw it away. Start again with a fresh work diskette. Continue checking and eradicating viruses until the system checks clean. The 50 cents or so that the diskette is worth is not worth the risk of contracting the virus.

Quick-Start Bench Testing

Many problems can plague the PC, but perhaps the most troubling problems occur during startupwhen the computer fails to start at all or does not start completely. Startup problems make it almost impossible to use diagnostics or other utilities to help isolate problems. Since Windows 95, even more difficulties can develop. This part of the chapter offers you a series of possible quick start explanations for full and partial system failures.




Symptom 4-1. There is no power light, and you cannot hear any cooling fan Chances are that there is insufficient power to the computer. Use a voltmeter and

confirm that there is adequate ac voltage at the wall outlet. Check the ac cord nextit might be loose or disconnected. See that the power switch is turned on and connected properly. Check the power-supply fuse(s). The main fuse might have opened. Replace any failed fuse.
If you replace a main fuse and the fuse continues to fail, you might have a serious fault in the power supply. Try replacing the power supply.

Symptom 4-2. There is no power light, but you hear the cooling fan running This usually means that some level of ac power is reaching the system. Use a volt-

meter and confirm that there is adequate ac voltage at the wall outlet. Unusually low ac voltages (such as during brownout conditions) can cause the power supply to malfunction. Verify that the power-supply cables are attached properly and securely to the motherboard. Use a voltmeter to verify that each output from the power supply is correct. Table 4-2 illustrates the proper voltage for each wire/color. If any output is very low or absent (especially the +5-volt output), replace the power supply. Finally, use a voltmeter and verify that the Power good (or PwrOK) signal is +5 V. If this signal is below 1.0 V, it might inhibit the CPU from running by forcing a Reset condition. Because the Power good signal is generated by the power supply, try replacing the power supply.
Symptom 4-3. The power light is on, but there is no apparent system activity Check the power-supply voltages. Use a voltmeter to verify that each output from

the power supply is correct. Table 4-2 lists the proper voltage for each wire color. If any output is very low or absent (especially the 5-V output), replace the power supply. Use a voltmeter and verify that the Power good (or PwrOK) signal is +5 V. If this signal is below 1.0 V, it might inhibit the CPU from running by forcing a continuous Reset condition. Because the Power good signal is generated by the power supply, try replacing the power supply. Check to see that the CPU is cool, that the heatsink/fan assembly is fitted on correctly, and that the CPU itself is inserted properly and completely into its socket. Check the CPU socketif the CPU is seated in a Zero Insertion Force (ZIF) socket, be sure that the sockets tension lever is closed and locked into place. If there is a separate math coprocessor on the motherboard (i286 and i386 systems), be sure that the MCP is inserted properly and completely into its socket. Next, check the expansion boards and be sure that all expansion boards are seated properly. Any boards that are not secured properly, or that are inserted unevenly, can short bus signals and prevent the PC from starting. Check the motherboard for shorts. Inspect the motherboard at every metal standoff and see that no metal traces are being shorted against a standoff or screw. You might want to free the motherboard and see if the system starts. If it does, use non-conductive spacers (such as a small piece of manila folder) to insulate the motherboard from each metal standoff. If the system still fails to start (and all voltages from the power supply are correct), replace the motherboard.

QUICK-START BENCH TESTING 87 TABLE 4-2 PINOUTS OF ATX AND BABY AT POWER CONNECTORS ATX POWER CONNECTOR COLOR Orange Orange Black Red Black Red Black Gray Purple Yellow Orange (22AWG) Brown (22AWG) Blue Black Green Black Black Black White Red Red VOLTAGE +3.3 Vdc +3.3 Vdc GND +5 Vdc GND +5 Vdc GND PwrOK +5V standby +12 Vdc +3.3 Vdc 3.3 V sense 12 Vdc GND PS-ON GND GND GND 5 Vdc +5 Vdc +5 Vdc PIN 1 2 3 4 5 6 7 8 9 10 11 11 12 13 14 15 16 17 18 19 20


BABY AT POWER CONNECTORS COLOR Orange Red Yellow Blue Black Black Black Black White Red Red Red VOLTAGE PwrOK +5 Vdc +12 dc 12 Vdc GND GND GND GND 5 Vdc +5 Vdc +5 Vdc +5 Vdc PIN 1 (P8) 2 (P8) 3 (P8) 4 (P8) 5 (P8) 6 (P8) 1 (P9) 2 (P9) 3 (P9) 4 (P9) 5 (P9) 6 (P9)




Symptom 4-4. The power light is on, but you hear two or more beeps There is no video. Check the video board first. Video problems can easily halt the initialization process. Turn off and unplug the PC, then be sure that your video board is inserted completely into its expansion slot. Consider the beep code itselfa catastrophic fault has been detected in the Power On Self-Test (POST) before the video system could be initialized. BIOS makers use different numbers and patterns of beeps to indicate failures. You can determine the exact failure by finding the BIOS maker (usually marked on the motherboard BIOS IC), then finding the error message in Chapter 15. In the vast majority of cases, the fault will be traced to the CPU, RAM, motherboard circuitry, video controller, or drive controller. Symptom 4-5. The power light is on, but the system hangs during initialization Video might be active, but there might be no text in the display. The Power On

Self-Test (POST) has detected a fault, and is unable to continue with the initialization process. BIOS makers mark the completion of each POST step by writing single-byte hexadecimal completion codes to port 80h. Turn off and unplug the PC, then insert a POST board to read the completion codes. Reboot the computer and find the last code to be written before the initialization stopsthat is the likely point of failure. You can determine the meaning of that POST code by finding the BIOS maker (usually displayed in the initial moments of power-up), then locating the corresponding error message in Chapter 15. Without a POST board available, it will be extremely difficult to identify the problem.
Symptom 4-6. You see a message indicating a CMOS setup problem The system parameters entered into CMOS RAM do not match the hardware configuration found during the POST. Enter your setup routine. If you are working on an older system (early i386 and i286 systems), you will probably need to boot the PC from a setup disk. If there is no setup disk available, you might be able to find a suitable routine at one of the sites at: oak.oakland.edu:/SimTel/msdos/at or ftp.uu.net:/systems/msdos/simtel/at. If you cant get a CMOS setup utility online, try CMOSER11.ZIP or GSETUP31.ZIP on the CD.

Review each entry in the CMOS setupespecially things like drive parameters and installed memoryand be sure that the CMOS entries accurately reflect the actual hardware installed on your system. If not, correct the error(s), save your changes, and reboot the system. Finally, test the CMOS battery. See if CMOS RAM will hold its contents by turning off the PC, waiting several minutes, then rebooting the PC. If setup problems persist, and you find that the values you entered have been lost, change the CMOS backup battery.
Symptom 4-7. You see no drive light activity The boot drive cannot be located.

The most frequent cause of drive problems is power connections. Inspect the 4-pin power cable and see that it is attached properly and completely to the drive. Check the powersupply voltages next. Use a voltmeter and verify that the +5- and +12-V levels (especially +12 V) are correct at the 4-pin connector. If either voltage is low or absent, replace the power supply. Locate the wide ribbon cable that connects to the drive and be sure that it


is attached correctly and completely at the drive and controller ends. Look for any scrapes or nicks along the cable that might cause problems. Start the CMOS setup. If you are working on an older system (early i386 and i286 systems), you will probably need to boot the PC from a setup disk. If no setup disk is available, you might be able to find a suitable routine at: oak.oakland.edu:/SimTel/msdos/at or ftp.uu.net:/systems/msdos/simtel/at. Check the CMOS setup next. Review the drive parameters entered in the CMOS setup, and be sure that the CMOS entries accurately reflect the actual boot drive installed on your system. If not, correct the error(s), save your changes, and reboot the system. Also, be sure that the drive-controller board is installed properly and completely in its expansion slot, and see that any jumpers are set correctly. Try booting the system from your boot floppy. If the system successfully boots to the A: prompt, your problem is limited to the hard-drive system. Now try switching to the C: drive. If the drive responds (and you can access its information), there might be a problem with the boot sector. Try a package, such as PC Tools or Norton Utilities, to try and fix the boot sector. If you cant access the hard drive, try a diagnostic to check the drive controller and drive. Check for boot-sector viruses. A boot-sector virus can render the hard drive unbootable. If you havent checked for viruses yet, use your anti-virus work disk now, and focus on boot-sector problems. If you cannot determine the problem at this point, try replacing the drive with a known-good working drive. Remember that you will have to change the CMOS setup parameters to accommodate the new drive. If all else fails, try a new drive-controller board.
Symptom 4-8. The drive light remains on continuously The boot drive cannot


be located. This typically happens if the signal cable is inserted backwards at one end. In most cases, this type of problem happens after replacing a drive or upgrading a controller. Be sure that the cable is inserted in the correct orientation at both the drive and controller ends. If you cannot determine the problem at this point, try replacing the drive with a known-good working drive. Remember that you will have to change the CMOS setup parameters to accommodate the new drive. If all else fails, try a new drive-controller board.
Symptom 4-9. You see normal system activity, but there is no video Be

sure that the monitor is plugged in and turned on. This type of oversight is really more common than you might think. Be sure that the monitor works (you might want to try the monitor on a known-good system). If the monitor fails on a known-good system, replace the monitor. Next, trace the monitor cable to its connection at the video board and verify that the connector is inserted securely. Check the video board. It is possible that the video board has failed. If the problem persists, replace the video board.


Symptom 4-10. The system randomly crashes/reboots for no apparent reason Check for viruses first. Some viruses (especially memory-resident viruses) can

cause the PC to crash or reboot unexpectedly. If you havent run your virus checker yet, do so now. Check the power-supply cables and verify that they are attached properly and securely to the motherboard. Use a voltmeter to verify that each output from the power supply is correct, as outlined in Table 4-2. If any output is low (especially the +5-V output), replace the power supply.



With all power off, check to see that the CPU is cool, that the heatsink/fan assembly is fitted on correctly, and that the CPU itself is inserted properly and completely into its socket. If the CPU overheats, it will stalltaking the entire system with it. If the CPU is seated in a Zero Insertion Force (ZIF) socket, be sure that the sockets tension lever is closed and locked into place. Also be sure that all SIMMs are seated properly in their holders and locked into place. You might try removing each SIMM, cleaning the contacts, and re-installing the SIMMs. Be sure that all expansion boards are seated properly. Any boards that are not secured properly or that are inserted unevenly, can short bus signals and cause spurious reboots. If youve recently installed new expansion hardware, be sure that there are no hardware conflicts between interrupts, DMA channels, or I/O addresses. Inspect the motherboard at every metal standoff and see that no metal traces are being shorted against a standoff or screw. You might want to free the motherboard and see if the crashes or reboots go away. If so, use non-conductive spacers (such as a small piece of manila folder) to insulate the motherboard from each metal standoff. If the system continues to crash or reboot (and all voltages from the power supply are correct), replace the motherboard.

Symptom 4-11. The system fails to boot, freezes during boot, or freezes during operation for no apparent reason This is the classic sign of a hardware

conflict. A PC is designed with a limited number of resources (i.e., memory, I/O addresses, IRQ lines, DMA channels, etc.). For the PC to function properly, each device added to the system must use its own unique resources. For example, no two devices can use the same IRQ, DMA, or I/O resources. When such an overlap of resources occurs, the PC can easily malfunction and freeze. Unfortunately, it is virtually impossible to predict when the malfunction will occur, so a conflict can manifest itself early (any time during the boot process) or later on (after DOS is loaded) while an application is running. Resolving a conflict is not difficult, but it requires patience and attention to detail. Examine the upgrade and its adapter board, and check the IRQ, DMA, and I/O address settings of other boards in the system. Make sure that the upgrade hardware is set to use resources that are not in use by other devices already in the system. For example, some motherboards offer built-in video-controller circuits. Before another video adapter can be added to the system, the motherboard video adapter must be disabledusually with a single motherboard jumper. Some sophisticated adapter boards (especially high-end video adapters and video-capture boards) require the use of extra memory space. If memory exclusions are needed, be sure that the appropriate entries are made in CONFIG.SYS and AUTOEXEC.BAT files. If memory exclusions are not followed, multiple devices might attempt to use the same memory space and result in a conflict.
Symptom 4-12. The system fails to recognize its upgrade device Even if

the hardware is installed in a system correctly, the PC might not recognize the upgrade device(s) without the proper software loaded. A great example of this is the CD-ROM drive. It is a simple matter to install the drive and its adapter board, but the PC will not even recognize the drive unless the low-level CD-ROM device driver is added to CONFIG.SYS and the MS-DOS CD-ROM driver (MSCDEX) is included in AUTOEXEC.BAT. If the


PC is running in a stable fashion, but it does not recognize the expansion hardware, be sure that you have loaded all required software correctly. If you are mixing and matching existing sub-assemblies from new and old systems, be sure that each device is fully compatible with the PC. Incompatibilities between vintages and manufacturers can lead to operational problems. For example, adding a 3.5" floppy drive to an i286 AT system can result in problems because the older BIOS could not format 3.5" high-density (1.44MB) floppy disks. A DOS utility (such as DRIVER.SYS) is needed to correct this deficiency. It is also possible that the upgrade device might simply be defective or installed incorrectly. Open the system and doublecheck your installation. Pay particular attention to any cables, connectors, or drive jumpers. When you confirm that the hardware and software installation is correct, suspect a hardware defect. Try the upgrade in another system if possible. If the problem persists when you attempt the upgrade on another PC, one or more elements of the upgrade hardware are probably defective. Return it to the vendor for a prompt refund or replacement. If the upgrade works on another system, the original system might be incompatible with the upgrade or you might have missed a jumper or DIP switch setting on the motherboard.
Symptom 4-13. One or more applications fail to function as expected after an upgrade This is not uncommon among video adapter and sound board upgrades.


Often, applications are configured to work with various sets of hardware. When that hardware is altered, the particular application(s) might no longer run properly (this is especially true under Windows). The best way to address this problem is to check and change the hardware configuration for each affected application. Most DOS applications come with a setup utility. You can adjust most Windows configurations under the Control panel icon. Under Windows 95, you can access system-configuration settings under the System icon under the Control panel.


Symptom 4-14. The Windows 95 boot drive is no longer bootable after restoring data with the DOS backup utility This happens frequently when a re-

placement drive is installed, and you attempt to restore the Windows 95 backup data. Unfortunately, the DOS version of backup is not configured to restore system files. Start backup and restore your root directory with System Files, Hidden Files, and Read-Only Files checked. Next, boot the system from an MS-DOS 6.x upgrade setup disk #1 or a Windows 95 startup disk, then use the SYS command to make the hard drive bootable such as:
A:\> sys c: <Enter>

You should then be able to restore the remainder of your files. When backing up a Windows 95 system, your best approach is to use the Windows 95 Backup program. Once the new drive is installed, partitioned, and formatted, install a new copy of Windows 95, start Windows 95 backup, then restore the remaining files to the drive.
Symptom 4-15. Windows 95 will not boot and ScanDisk reports bad clusters that it cannot repair This is a problem encountered with Western Digital hard



drives. If your WD drive fails in this way, you can recover the drive, but you will lose all information on it. Backup as much information from the drive as possible before proceeding:
s Download the Western Digital service files WDATIDE.EXE and WD_CLEAR.EXE

s s s s

from WD at: http://www.wdc.com/. You can also get these files from AOL by typing keyword WDC. Copy these files to a clean boot floppy diskette. Boot to DOS from a clean diskette (no CONFIG.SYS or AUTOEXEC.BAT files) and run WD_CLEAR.EXE. This utility clears all data on the media (and destroys all data). Next, run the WDATIDE.EXE utility to perform a comprehensive surface scan. Repartition and reformat the drive, then restore your data.

Symptom 4-16. You see a Bad or missing <filename> error on startup A file used by Windows 95 during startup has probably become corrupt. Locate the file mentioned in the error message. If you can find the file, erase it and try re-installing it from original Windows 95 disks or CD. Symptom 4-17. Windows 95 reports damaged or missing files, or a VxD error During startup, Windows 95 depends on several key files being available. If a key

file is damaged or missing, Windows 95 will not function properly (if it loads at all). Run Windows 95 setup again and select the Verify option in Safe recovery to replace the missing or damaged file(s).
Symptom 4-18. After installing Windows 95, you cant boot from a different drive The Windows 95 setup program checks all hard disks to find just one that con-

tains the 80h designator in the DriveNumber field of a boot sector. Windows 95 will typically force the first drive to be bootable and prevent other drives from booting. However, there are two ways to correct the problem after Windows 95 is installed:
s Use the version of FDISK included with Windows 95 to set the primary active partition. s Use a disk-editor utility to change a disks DriveNumber field so that you can boot from

that hard disk.

Symptom 4-19. Windows 95 Registry files are missing There are two registry

files: USER.DAT and SYSTEM.DAT. They are also backed up automatically as USER.DA0 and SYSTEM.DA0. If a .DAT file is missing, Windows 95 will automatically load the corresponding .DA0 file. If both the .DAT and .DA0 registry files are missing or corrupt, Windows 95 will start in the Safe mode offering to restore the Registry. However, this cannot be accomplished without a backup. Either restore the Registry files from a tape or diskette backup, or run Windows 95 Setup to create a new Registry. Unfortunately, restoring an old registry or creating a new registry from scratch will reload programs and re-add hardware to restore the system to its original statea long and difficult procedure. Use the following DOS procedure to backup the Registry files to a floppy disk:
attrib r s h system.da? attrib r s h user.da? copy system.da? A:\

QUICK-START BENCH TESTING 93 copy user.da? A:\ attrib +r +s +h system.da? attrib +r +s +h user.da?


Symptom 4-20. During the Windows 95 boot, I get an Invalid System Disk error This often happens during the first reboot during Windows 95 setup, or

when you boot from the startup disk. When you see a message such as Invalid system disk. Replace the disk, and then press any key. There might be several possible problems. First, your disk might be infected with a boot-sector virus. Run your anti-virus work disk and check closely for boot sector viruses. Windows 95 setup might also fail if there is antivirus software running as a TSR, or your BIOS has enabled boot-sector protection. Be sure that any boot-sector protection is turned off before installing Windows 95. Check for disk-overlay softwareWindows 95 might not detect overlay software such as Disk Manager, EZ-Drive, or DrivePro, and overwrite the master boot record (MBR). See the documentation that accompanies your particular management software for recovering the MBR. To re-install the Windows 95 system files, follow these steps:
1 Boot the system using the Windows 95 Emergency Boot Disk. 2 At the MS-DOS command prompt, type the following lines:
c: cd\windows\command attrib c:\msdos.sys -h -s -r ren c:\msdos.sys c:\msdos.xxx a: sys c: del c:\msdos.sys ren c:\msdos.xxx c:\msdos.sys attrib c:\msdos.sys +r +s +h

3 Remove the Emergency Boot Disk and reboot the system. Symptom 4-21. Windows 95 will not install on a compressed drive You are

probably using an old version of the compression software, which Windows 95 does not recognize. Although Windows 95 should be compatible with all versions of SuperStor, it does require version 2.0 or later of Stacker. Be sure that your compression software is recent and see that there is enough free space on the host drive to support Windows 95 installation. If you have the PlusPack for Windows 95, you should be able to install DriveSpace 3 for best Windows 95 support.
Symptom 4-22. The drive indicates that it is in MS-DOS compatibility mode For some reason, Windows 95 is using a real-mode (DOS) driver instead of a pro-

tected-mode (32-bit) driver. Be sure that any software related to the hard drive (especially hard-disk drivers) are using the protected-mode versions. Windows 95 should install equivalent protected-mode software, but you might need to contact the drive manufacturer and obtain the latest Windows 95 drivers. If you are using Disk Manager, be sure that youre using version 6.0 or later. You can get the latest patch (DMPATCH.EXE) from the Ontrack web site at: http://www.ontrack.com/. Finally, check your motherboard BIOSWindows 95 might use DOS-compatibility mode on large EIDE hard disks (hard disks with more than 1024 cylinders) in some computers. This might occur because of an invalid drive



geometry translation in the system ROM BIOS that prevents the protected-mode IDE device driver from being loaded. Contact your system manufacturer for information about obtaining an updated BIOS.
Symptom 4-23. Disabling protected-mode disk driver(s), hides the partition table when FDISK is used As with Symptom 4-22, there are problems prevent-

ing 32-bit operation of your hard drive(s). Do NOT use the Disable all 32-bit protected-mode disk drivers option. Instead, upgrade your motherboard BIOS to a later version.
Symptom 4-24. You cannot achieve 32-bit disk access under Windows 95

If the Windows 95 system refuses to allow 32-bit disk access, there might be a conflict between the motherboard CMOS setup entries and the BIOS on your EIDE controller. For example, if both BIOS have settings for Logical Block Addressing (LBA), be sure that only one entry is in use.
Symptom 4-25. Windows 95 does not recognize a new device In some cases, Windows 95 is unable to recognize a new device. When this happens, check to see if there is a hardware conflict between the device and other devices in the system (you can see conflicts represented in the Device Manager with small yellow exclamation marks). Also be sure that any necessary drivers have been installed properly. If problems continue, remove the new device through your Device Manager, and reinstall it through the Add new hardware wizard. Symptom 4-26. Windows 95 malfunctions when installed over Disk Manager Disk Manager should typically be compatible with Windows 95, but there are some

points to remember. Check your Disk Manager version first. If you are using Disk Manager, be sure that youre using version 6.0 or later. You can get the latest patch (DMPATCH.EXE) from the Ontrack Web site at: http://www.ontrack.com/. Check the slave drive with Disk Manager. Although the Windows 95 file system is supposed to work properly with a slave drive only using Disk Manager, there are some circumstances where problems can occur:
s When a Windows 3.1x virtual driver replaces the Windows 95 protected-mode driver

(such as WDCDRV.386).
s When the cylinder count in CMOS for the slave drive is greater than 1024 cylinders. s When the motherboard CMOS settings for the slave drive are set to Auto-detect. Symptom 4-27. You have problems using a manufacturer-specific harddisk driver (such as Western Digitals FastTrack driver WDCDRV.386) for 32-bit access under Windows 95 Generally speaking, Windows 95 has 32-bit pro-

tected-mode drivers for a wide variety of EIDE devicesin actuality, you should not need a manufacturer-specific driver. If Windows 95 has not removed all references to the driver from SYSTEM.INI, you should edit the file and remove those references manually, then reboot the system. Be sure to make a backup copy of SYSTEM.INI before editing it.


Further Study
Thats it for Chapter 4. Because this is the last chapter of this section, take some time to review the glossary and chapter questions on the accompanying CD. If you have access to the Internet, set your Web browser to some of the following contacts: Symantec: http://www.symantec.com McAfee and Associates: http://www.mcafee.com IBM Setup routines: oak.oakland.edu:/SimTel/msdos/at, or http.uu.net:/systems/ msdos/simtel/at Ontrack Software: http://www.ontrack.com/ Ziff Davis benchmark site: http://www8.zdnet.com/pcmag/pclabs/bench/ Benchmarking news group: comp.benchmarks