Stephen Bigelow
$54.95 0-07-913732-6
|
Troubleshooting, Maintaining & Repairing PCs Stephen Bigelow $54.95 0-07-913732-6 |
|
| Chapter: 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 | 41 | 42 | 43 | 44 | 45 | 46 | 47 | 48 | 49 | 50 | 51 | 52 | 53 |
| Reserve your copy at a Beta Bookstore near you! |
Contact Bet@books © 1998 The McGraw-Hill Companies, Inc. All rights reserved. Any use of this Beta Book is subject to the rules stated in the Terms of Use. |
CHAPTER 34
Motherboard troubleshooting
The motherboard is the heart of any personal computer. It is the motherboard that provides system resources (i.e. IRQ lines, DMA channels, I/O locations) as well as "core" components such as the CPU, chipset(s), real-tome clock (RTC), and all system memory - including RAM, BIOS ROM, and CMOS RAM. Indeed, most of a PC’s capabilities are defined by motherboard components. This chapter is intended to provide a guided tour of contemporary motherboards, and show you how to translate error information and symptoms into motherboard repairs.
Active, passive, and modular
Before going any further, you should understand the difference between a motherboard and a backplane. For the purposes of this book, a motherboard is a printed circuit board containing most of the processing components required by the computer. PC purists often refer to a motherboard as an active backplane. The term "active" is used because there are ICs running on the board. The advantage of a motherboard is its simplicity - the motherboard virtually is the PC. Unfortunately, the motherboard has disadvantages. Namely, It is difficult to upgrade. Aside from plugging in an upgraded CPU or adding RAM, the only real way to upgrade a motherboard is to replace it outright with a newer one. For example, the only way to add PCI bus slots to an all-ISA motherboard is to replace the motherboard with one containing PCI slots.
On the other hand, a backplane (also referred to as a passive backplane) is little more than a board containing interconnecting slots - there are no ICs on the backplane (except perhaps some power supply regulating circuitry). The CPU, DRAM, BIOS ROM, and other central processing components are fabricated onto a board which simply plugs into one of the backplane slots. Other expansion devices (i.e. video board, drive controller, sound board, and so on) just plug into adjacent slots. The PS/2 was one of the first PCs to use a backplane design. Backplane systems are easy to troubleshoot - unlike traditional motherboards which require the entire system to be disassembled, a processor board can be removed and replaced as easily as any other board, so it is also a simple matter to upgrade the PC by installing a new processor board. The great limitation to backplanes is the bus. Where traditional motherboards can optimize a system with different busses, the backplane is limited to a single bus style (usually ISA or MCA). High-performance bus architectures like VL or PCI are not readily available.
In an effort to provide a motherboard which is more upgradeable and serviceable, manufacturers are experimenting with modular motherboards. The modular motherboard places the CPU, math co-processor, and key support ICs on a replaceable card which plugs into a motherboard which holds BIOS ROM, CMOS RAM, DRAM, other system controllers, and bus interfaces. The modular approach allows a motherboard to be upgraded far more than a traditional motherboard without having to replace it outright - the replacement processing card is then much cheaper than a new motherboard. However, today’s PC architectures can usually support a variety of CPU versions and an extensive amount of RAM on the original motherboard, so "modularity" has never become a very popular approach.
Contrary to popular belief, expansion bus connectors are not needed to make a motherboard. You can see this in any laptop or notebook computer motherboard (Fig. 34-1). The devices which traditionally demanded expansion slots (video and drive controllers) are easily fabricated directly onto the motherboard. Even the motherboards used in most desktop and tower PCs over the last few years integrate video and drive controller circuits. If upgrades are needed in the future, the motherboard-based circuits can be disabled with jumpers, and replacement sub-systems are plugged into expansion slots.
Understanding the motherboard
Before you can troubleshoot a motherboard effectively, it is important that you know your way around and be able to identify at least most of the available components. Although each motherboard is designed differently, this process of identification is not nearly as difficult as it might sound. This part of the chapter will familiarize you with the essential functions and components that you’ll find on a modern motherboard.
Socket 7, Socket 8, or Slot 1
When examining a motherboard, you’ll probably find it designated as "Socket 7", "Socket 8", or "Slot 1". These classifications refer to the type of CPU which the motherboard can support; Socket 7 motherboards are generally designed for Pentium and Pentium MMX CPUs, Socket 8 motherboards are made for PentiumPro CPUs, and Slot 1 motherboards are slated for Pentium II systems. This does not mean that a motherboard can support ANY such processor, only that the motherboard supports a given "class" of processor. For example, older Pentium motherboards may only support Pentium CPUs up to 120MHz, while newer Pentium motherboards can support Pentium (or Pentium MMX) processors up to 200MHz - but all would be categorized as "Socket 7" motherboards. You can find a complete breakdown of socket/slot designations in Chapter 14.
AT, ATX, and NLX
Another important classification that you must be familiar with is the motherboard’s "form factor". In simplest terms, the "form factor" is little more than the dimensions of the board and its mounting hole positions, as well as the general layout and placement of key components such as the CPU, SIMMs, and expansion slots. Today, there are three major form factors to consider; AT, ATX, and NLX. It is important for you to understand that form factors do not directly influence performance - a "baby AT" motherboard and an NLX motherboard can offer exactly the same performance characteristics. Form factor is most important in system assembly and access for service.
AT-style motherboards - the "AT-style" motherboards really represent the classic approach to component placement as shown in Fig. 34-2. AT-style motherboards are typically available in two variations; the "Baby AT", and the "Full AT" - both variations simply effect the overall dimensions of the motherboard (Full AT motherboards are larger). You can usually identify an AT-style motherboard based upon two distinctions. First, look at the power connectors where the power supply attaches. An AT-style motherboard uses two sets of 6-pin in-line connectors. Second, the CPU is usually positioned in line with the ISA bus slots (almost always obstructing full-length ISA cards).
NOTE: The Socket 5 connector shown in Fig. 34-2 indicates an i486 motherboard.
ATX-style motherboards - the "ATX-style" motherboards are the result of the first serious industry push to "standardize" the dimensions and connections of a PC motherboard such as the ATX Slot 1 motherboard shown in Fig. 34-3. An ATX motherboard is distinguished by three points. First, all I/O port connectors are concentrated into a single "I/O panel" at the rear of the motherboard. Second, the ATX motherboard uses a 20-pin PS/2 style power connection from the power supply. Third, the CPU is located clear and away from all expansion bus slots - eliminating any interference with full-slot expansion cards. ATX motherboards can be found supporting Socket 7, Socket 8, and Slot 1 CPUs.
NLX-style motherboards - while ATX motherboards represented a good effort at standardization, they still retain all the assembly problems of AT-style motherboards - namely that the motherboard is cumbersome to install, and time-consuming to upgrade or replace. The "NLX-style" motherboards (Fig. 34-4) overcome this disadvantage by making the motherboard a replaceable "card", and moving all expansion slots and connection headers (i.e. speaker connector, power switch connector, and so on) to a "riser card". The NLX motherboard itself then plugs into the riser card. In this fashion, the motherboard can quickly and easily be removed from the system to change jumpers, add memory, or install a replacement motherboard.
Learning your way around
Now that you’ve seen some essential motherboard classifications, it’s time to actually look at a motherboard up-close, and identify the critical parts that you should expect to find. For the purposes of this book, we’ll use the Intel PD440FX ATX Slot 1 motherboard shown in Fig. 34-5. Other motherboards and form factors will appear a bit different, but the basic key parts are all the same. The pinouts of each major connector are illustrated in Fig. 34-6.
NOTE: The chipset components illustrated below are presented for example purposes only. Your motherboard will undoubtedly use different chips (and chipsets) - each offering their own set of characteristics.
Troubleshooting a motherboard
Since motherboards contain the majority of system processing components, it is likely that you will encounter a faulty motherboard sooner or later. The BIOS POST is written to test each sub-section of the motherboard each time the PC is powered up, so most problems are detected well before you ever see the DOS prompt. Errors are reported in a myriad of ways. Beep codes and POST codes (Chapter 19) provide indications of fatal errors that occur before the video system is initialized. Still, there are plenty of symptoms that can elude the initial testing at start time. This part of the chapter digs in and presents a lengthy selection of motherboard symptoms for you to reference.
Repair vs. replace
This is the perennial troubleshooting dilemma. The problem with motherboard repair is not so much the availability of replacement parts (although that can be a challenge) as it is the use of surface-mount soldering (SMT). You see, a surface-mounted IC cannot be desoldered with a conventional tools. To successfully desolder a surface-mounted IC, you need to heat each of the IC's pins (often in excess of 100) simultaneously, then lift the IC off the board. It is then a simple matter to clean up any residual solder. Unfortunately, specialized surface-mount soldering equipment is required to do this. The equipment is readily available commercially, so it is easy to buy - but you can invest $1000 to $2000 to equip your work bench properly.
As you can imagine, the "repair vs. replace" decision is an economic one. It makes little sense for the part-time PC enthusiast to make such a substantial investment to exchange a defective IC (which are usually under $30). It is generally better to replace the motherboard outright, which is only a fraction of the cost of such SMT equipment. On the other hand, professionals who intend to pursue PC repair as a living are well served with surface-mount equipment. The customer's cost for labor, the part(s), and markup is typically much less than purchasing a new motherboard (especially the high-end boards such as i486/66 and Pentium motherboards).
Start with the basics
Since motherboard troubleshooting does represent a significant expense, you should be sure to start any motherboard repair by inspecting the following points in the PC. Remember to turn all power off before performing these inspections:
Symptoms
Symptom 34-1. A motherboard failure is reported, but goes away when the PC's outer cover is removed. There is likely to be an intermittent connection on the motherboard. When the housing is secured, the PC chassis warps just slightly - this may be enough to precipitate an intermittent contact. When the housing is removed, the chassis relaxes and hides the intermittent connection. Replace the outer cover and gently re-tighten each screw with the system running. Chances are that you will find one screw that triggers the problem. You can leave that screw out, but it is advisable to replace the motherboard as a long-term fix.
Symptom 34-2. The POST (or your software diagnostic) reports a CPU fault. This is a fatal error, and chances are that system initialization has halted. CPU problems are generally reported when one or more CPU registers do not respond as expected, or has trouble switching to the protected-mode. In either case, the CPU is probably at fault. Fortunately, the CPU is socket mounted, and should be very straightforward to replace. Be sure to remove all power to the PC, and make careful use of static controls when replacing a CPU. Mark the questionable CPU with indelible ink before replacing it.
Zero-insertion force (ZIF) sockets are easiest, since the IC will be released simply by lifting the metal lever at the socket's side. Slide out the original CPU and insert a new one. Secure the metal lever, and try the PC again. However, many CPUs are mounted in pin grid array (PGA) sockets, and a specialized PGA removal tool is strongly suggested for proper removal. You should also be able to use a small, regular screwdriver to gently pry up each of the four sides of the CPU, but be very careful to avoid cracking the IC, the socket, or the motherboard - never use excessive force. When the IC is free, install the new CPU with close attention to pin alignment, then gently press the new CPU into place.
A word about heat sink/fans. Most i486 and later CPUs are equipped with a metal heat sink (or heat sink/fan) assembly. It is vital to the proper operation of your system that the heat sink be re-installed correctly - otherwise, the new CPU will eventually overheat and lock up or fail. Be sure to use good-quality thermal compound to ensure proper heat transfer to the heat sink (remember that a sound mechanical connection does not guarantee a good thermal connection).
Symptom 34-3. The POST (or your software diagnostic) reports a problem with the floating point unit. Math co-processor (also called the Floating Point Unit or FPU) problems are generally reported when one or more MPC registers do not respond as expected. Fortunately, MCP faults are not always fatal. It is often possible to remove the MCP or disable the MCP availability through the CMOS Setup. Of course, programs that depend on the MCP will no longer run, but at least the system can be used until a new one is installed. On older systems that use separate MCPs, the device is socket mounted, and should be very straightforward to replace. Be sure to remove all power to the PC, and make careful use of static controls when replacing an MPC. Mark the questionable MPC with indelible ink before replacing it. If the MCP is integrated into the CPU (i386DX, i486DX, Pentium, and later CPUs) are a bit more expensive because you’ll need to replace the entire CPU, but the replacement process is no more difficult (remember to remount any heat sink/fan assembly properly).
Symptom 34-4. The POST (or your software diagnostic) reports a BIOS ROM checksum error. The integrity of your system BIOS ROM is verified after the CPU is tested. This is necessary to ensure that there are no unwanted instructions or data that might easily crash the system during POST or normal operation. A checksum is performed on the ROM contents, and that value is compared with the value stored in the ROM itself. If the two values are equal, the ROM is considered good and initialization continues - otherwise, the BIOS is considered defective and should be replaced. Chapter 9 provides an index of major BIOS manufacturers.
Traditionally, BIOS ROM is implemented as one or two ICs which are plugged into DIP sockets. They can be removed easily with the blade of a regular screwdriver, as long as you pry the IC up slowly and gently (be sure to pry the IC evenly from both ends). When installing new DIP ICs, you may have to straighten their pins against the surface of a table, or use a DIP pin straightening tool. Ultimately, the IC pins will fit nicely into each receptacle in the DIP socket. You can then ease the IC evenly down into the socket. Alignment is critical to ensure that all pins are inserted. If not, one or more pins may be bent under the IC and ruin the new ROM. Also, be sure to insert the new IC(s) in the proper orientation. If they are accidentally installed backward, they may be damaged.
Newer BIOS ICs use flash EEPROM technology which allows the device to be erased and reprogrammed in the field without having to replace the entire BIOS ROM IC. When a flash BIOS fails its checksum test, it also has probably failed. Since flash BIOS devices are often fabricated as PLCC ICs, it is a bit easier to replace them, but you will need a PLCC removal tool to take the original IC out of its socket - there simply is not enough room for a screwdriver.
Symptom 34-5. The POST (or software diagnostic) reports a timer (PIT) failure, an RTC update problem, or a refresh failure. The PIT is often an 8254 or compatible device. Ultimately, one or more of the its three channels may have failed, and the PIT should be replaced. It is important to realize that many modern motherboards incorporate the PIT functions into a system controller or other chipset IC (refer to Chapter 11 for a listing of chipsets and functions). Since the PIT is typically surface-mounted, you can attempt to replace the device, or replace the motherboard entirely.
Symptom 34-6. The POST (or software diagnostic) reports an interrupt controller (PIC) failure. The PIC is often an 8259 or compatible device, and there are two PICs on the typical AT motherboard (PIC#1 handles IRQ0 through IRQ7, and PIC#2 handles IRQ8 through IRQ15). Of the two, PIC#1 is more important since the lower interrupts have a higher priority, and the lowest channels handle critical low-level functions such as the system timer and keyboard interface. Generally, a diagnostic will reveal which of the two PICs have failed. Make sure that there are no interrupt conflicts between two or more system devices. You can then replace the defective PIC. In many current systems, both PICs are integrated into a system controller or chipset IC. You can replace the defective IC if you have the appropriate surface-mount equipment available, or replace the motherboard entirely.
Symptom 34-7. The POST (or software diagnostic) reports a DMA controller (DMAC) failure. The DMAC is often an 8237 or compatible device, and there are two DMACs on the typical AT motherboard (DMAC#1 handles channel 0 through channel 3, and DMAC#2 handles channel 4 through channel 7). Of the two, DMAC#1 is more important since channel 2 runs the floppy disk controller. Generally, a diagnostic will reveal which of the two DMACs have failed. Make sure that there are no DMA conflicts between two or more system devices. You can then replace the defective DMAC. In many current systems, both DMACs are integrated into a system controller or chipset IC. You can replace the defective IC if you have the appropriate surface-mount equipment available, or replace the motherboard entirely.
Symptom 34-8. The POST (or software diagnostic) reports a KBC fault. The keyboard controller (KBC) is often either an 8042 or an 8742. Since the KBC is a microcontroller in its own right, diagnostics can usually detect a KBC fault with great accuracy. The KBC may either be a socket-mounted PLCC device, or (in rare cases) a surface-mounted IC. Remember remove all power and mark the old KBC before you remove it from the PC. You will probably need a PLCC removal tool to take out the old KBC. If you cannot exchange a defective KBC, you’ll need to replace the motherboard.
Symptom 34-9. A keyboard error is reported, but a new keyboard has no effect. The keyboard fuse on the motherboard may have failed. Many motherboard designs incorporate a small fuse (called a pico-fuse) in the +5Vdc line that drives the keyboard. If this fuse fails, the keyboard will be dead. Use your multimeter and measure the +5Vdc line at the keyboard connector. If this reads 0Vdc, locate the keyboard fuse on the motherboard and replace it (you may have to trace the line back to the fuse which looks almost exactly like a resistor).
Symptom 34-10. The POST (or software diagnostic) reports a CMOS or RTC fault. With either error, it is the same device that is usually at fault. The CMOS RAM and RTC are generally fabricated onto the sane device. RTC problems indicate that the real-time clock portion of the IC has failed, or is not being updated. CMOS RAM failure can be due to a dead backup battery, or a failure of the IC itself. When dealing with a CMOS or Setup problem, try the following protocol. First, try a new backup battery and reload the CMOS Setup variables. If a new battery does not resolve the problem, the CMOS/RTC IC should be replaced. Often, the CMOS/RTC IC is surface-mounted, and will have to be replaced (or the motherboard will have to be replaced). However, there is a growing trend toward making the IC socket-mounted and including the battery into a single replaceable module (such as the Dallas Semiconductor-type devices). Modules are typically replaceable DIP devices.
Symptom 34-11. The POST (or software diagnostic) reports a fault in the first 64KB of RAM. The first RAM page is important since it holds the BIOS data area (BDA) and interrupt vectors - the system will not work without it. When a RAM error is indicated, your only real recourse is to replace the motherboard RAM. On older motherboards, if the diagnostic indicates which bit has failed and you can correlate the bit to a specific IC, you can sometimes replace the defective IC (typically surface-mounted). Otherwise, you will need to locate and replace all of the motherboard RAM, or replace the motherboard entirely. Newer motherboards utilize SIMMs or DIMMs for ALL system memory, so it should be a relatively simple matter to cycle through each SIMM or DIMM with a known-good unit and isolate the defective memory.
Symptom 34-12. The MCP does not work properly when installed on a motherboard when external caching is enabled. Some non-Intel math co-processors (or Floating Point Units) work in areas which must be non-cached. For example, a Cyrix EMC87 MCP with an AMI Mark IV i386 motherboard has been known to cause these types of problems. When MCP problems arise (especially during upgrades), try disabling the external cache through CMOS Setup. As another alternative, try a different math co-processor.
Symptom 34-13. A "jumperless motherboard" receives incorrect CPU Soft Menu settings, and now refuses to boot. This may occur on a motherboard such as the Abit IT5V, and is usually due to accidental settings during system configuration. Fortunately, this type of problem can be corrected by removing power from the motherboard - try turning off the system and unplugging it for several minutes. When you restore power to the system, the CPU Soft Menu will automatically reset the CPU frequency for the lowest setting and allow the motherboard to boot. You can then go back into the CPU Soft Menu and correct any speed setting errors. If this were a jumpered motherboard, you would need to find the CPU speed jumper and set it correctly.
Symptom 34-14. When I install two 64MB SIMMs, only 32MB of RAM are displayed when I turn on the computer. The motherboard is probably using a 430VX chipset which (though supporting 128MB of RAM), will not support 64MB memory devices. The 430VX only supports the following memory devices:
The layout for a 64MB SIMM is 16Mx32-bit which isn't in the list shown above. When you install two 64MB SIMMs, the system will use the 4Mx32-bit specification to calculate the memory, thus displaying 32MB. Unfortunately, this is a limitation of the motherboard, and cannot be corrected without upgrading the motherboard.
Symptom 34-15. A Creative Labs PnP sound board refuses to work on one motherboard, but the board works just fine on another motherboard. This is an issue where the PnP BIOS is usually at fault. Check with the motherboard manufacturer to see if there is a BIOS update to correct PnP problems.
Symptom 34-16. The system CD-ROM drive refuses to work once an IDE bus master driver is installed. This is almost always a driver problem which is not interacting properly with the IDE/EIDE bus controller on the motherboard. In almost all cases, you should update the IDE bus master driver, or disable bus mastering completely.
Symptom 34-17. You cannot get an AMD 5x86 133MHz CPU to run on your motherboard. Check your voltage first. The AMD 5x86 runs on 3.3V, so you may need a voltage regulator in the CPU socket (the AMD CPU may already be damaged). Also check your BIOS version - you may need an updated BIOS to support the AMD CPU properly. Check your jumper settings next - the speed or CPU type selection is almost always set wrong. If you cannot jumper the motherboard correctly (i.e. 33MHz bus speed), then the motherboard itself is limited - it cannot enable the 4x internal CPU clock for the AMD 5x86. In this case, you will need to use a different CPU, or replace the motherboard outright.
Symptom 34-18. You cannot get a Cyrix 5x86 CPU to run on your motherboard. Check your voltage first. The Cyrix 5x86 uses 3.3V, so you may need a voltage regulator in the CPU socket (the Cyrix CPU may already be damaged). Also check your BIOS version - you may need an updated BIOS to support the Cyrix CPU properly. Check your jumper settings next - the speed (33MHz) or CPU type selection is almost always set wrong. If problems persist, you may need a different CPU or motherboard.
Symptom 34-19. You see the error message "System Resource Conflict" on the AMI BIOS POST display. This is an error generated by AMI PnP BIOS (though other PnP BIOS may produce similar errors), and is generated when the BIOS detects a resource conflict during initialization. You may try to force the BIOS to reconfigure the conflicting resource by pressing the <Insert> key during POST. If problems continue, you may need a BIOS update which may be able to resolve assignment conflicts more intelligently. Otherwise, you may need to try reconfigure the conflicting resource manually (disabling its PnP support), or remove the offending device entirely.
Symptom 34-20. The system hangs after using MEMMAKER under DOS. This is most prevalent with AMI’s WinBIOS which cannot support the "highscan" option used with EMM386.EXE. Make sure to disable the "highscan" option from EMM386 before running MEMMAKER. You may also choose to upgrade the system BIOS to a more recent version which may be more robust when testing memory.
Symptom 34-21. Your "Power Management" icon does not appear in the Windows 95 Control Panel. This occurs even though the APM parameter under the BIOS Power Management Setup is enabled. This problem occurs if you do not enable the APM function before you install Windows 95. If you have already installed Windows 95, you must re-install it. Before doing so, however, make sure that the APM function is enabled.
Symptom 34-22. Systems with a Western Digital 1.6GB HDD fail to boot even through BIOS recognized the presence of HDD. This is a typical problem with large hard drives which often need additional time to start up after powering the system. Check your BIOS Advanced Setup and increase the "Power-on Delay" time. This should correct the problem. This problem may reoccur if CMOS default values are reloaded, or CMOS contents are lost.
Symptom 34-23. After installing Windows 95, the system can no longer find the CD-ROM drive on the secondary IDE channel. You may also find that the IDE drives are running in MS-DOS "compatibility mode". This problem occurs often with motherboards using the Intel 430HX chipset - Windows 95 is not recognizing the Intel 82371SB drive controller on the motherboard, and this causes BIOS to disable the secondary IDE channel - devices on the secondary channel are not being detected after the system is rebooted. In most cases, you can upgrade the BIOS to correct this problem, or move the IDE devices to a separate IDE controller. You may also be able to find an update to the MSHDC.INF file which will force Windows 95 to recognize the 82371SB controller.
Symptom 34-24. The system hangs up or crashes when the chipset-specific PCI-IDE DOS driver is loaded. This is a known problem with Micro-Star motherboards using a VIA VP1 chipset and Award BIOS 4.50PG. The problem is with the BIOS version and its interaction with the PCI controller portion of the VIA chipset. Upgrading the BIOS version should resolve the problem.
Symptom 34-25. You notice that your Pentium motherboard is unusually picky about which SIMMs it will accept. This occurs even though the SIMMs are all within the proper type and rating. There are several possible problems to consider. First, Intel chipsets are very discriminating when it comes to memory speed, so make sure that the memory speed is well within the required range (usually 70nS or faster). Second, try changing the Wait States in the CMOS Setup to a lower speed (i.e. 4-4-4-4). If your system works under this low speed, then increase the speed (i.e. 3-3-3-3, 3-2-2-2, 3-1-1-1, and so on) and keep trying till the best number has reached. Finally, the memory itself may be of questionable quality - try good-quality memory bought from a reputable vendor. Make sure the vendor offers a liberal return policy so that you can return questionable memory easily.
Symptom 34-26. You experience a problem with pipeline burst cache. This is a recognized problem with UMC pipeline burst cache (especially on an Amptron motherboard). The problem can usually be solved by adjusting the cache control to 4-4-4-4 (the default in CMOS is typically 2-3-3-3). This will reduce performance, but should stabilize cache operations.
Symptom 34-27. You get no display, or the system refuses to boot because of the keyboard controller. Note that the video adapter proves out fine in another system. This is a problem with the VIA 82C41 24-pin keyboard controller (especially on the Amptron PM-7600 motherboard). A fault with the KBC may cause a "no display" or "fail to boot" condition. The VIA 82C41 is extremely sensitive to damage from power supply surges/spikes, and ESD damage. Replace the KBC, or replace the motherboard with a more robust model.
Symptom 34-28. Your customer forgets their password. The PC password is stored in the CMOS RAM which is located in either the motherboard chipset or the Real Time Clock chip. If it is stored in the chipset, the CMOS memory is backed up by a coin-shaped lithium battery (or other battery). If it is stored in the RTC chip, it has an internal battery to back up the CMOS RAM. For the external battery, follow these steps: First, make a complete backup of the CMOS settings. Turn off the system, and then remove the battery for a period of at least two hours. This should clear the CMOS setting and erase the password. For the RTC battery, follow these steps: Determine which RTC chip you have - there are five different kinds of Real-Time Clock CMOS chips:
For the Dallas DS 12887 and Benchmarc RTC chips, if you can boot to the A: prompt, flash the BIOS chip with the same boot block record, but different BIOS revision. For example, if you have a P/I P55TP4XE motherboard with BIOS revision 0202, flash the BIOS chip to BIOS revision 0115. A BIOS checksum error will be generated. Enter the CMOS setup screen, reload setup defaults, then save and exit. At this point, the password has been cleared. You can flash the BIOS back to the original revision. If you can't boot to the A: prompt, turn off the system, remove the BIOS chip, and insert another with the same boot block record but different BIOS revision. Power on the system. A BIOS checksum error will be generated. Turn off the system. Reinstall the original BIOS. Power on the system again, and hit <Del> to enter the BIOS setup screen. Reload the setup defaults, then save and exit.
For the Dallas DS 12887A, there is a jumper on the motherboard which clears the CMOS. Please check your manual for the location of this jumper (it will vary between motherboards). Shorting this jumper should erase the system configuration information (including password) stored in the CMOS. To clear the CMOS, make sure the system is off. Short the jumper for a moment and then remove it. Do not leave this jumper shorted. After clearing the CMOS, the password should be erased.
For the BQ3287A and Dallas DS12B887 RTC chips, short the same jumper as in the previous section, but make sure to power the system on and off before removing the jumper.
Symptom 34-29. You encounter problems with Western Digital hard drives (the drives work on other systems). This type of problem has been identified with Asus motherboards using Award BIOS with older Western Digital (~1.6GB) drives. Note that problems do not appear in newer Western Digital drives. There are several means of addressing the problems: First, disable the "Quick Power-on Self Test" in your CMOS Setup, and enable the "floppy seek" option. This will increase the time that the drive gets to spinup. If your CMOS offers a "Power-on Delay Time" instead, try increasing that time. Also avoid using DEFRAG, or the "disk surface scan" feature of ScanDisk with Western Digital drives - both have been reported to increase the number of bad blocks on the disk.
Next, consider a BIOS upgrade (especially if you’re using a motherboard with the Intel 430FX chipset). Some BIOS versions use a "park head" command that can cause problems with Western Digital hard drives. Finally, check the Western Digital web site (www.wdc.com) for any drive patches that might be currently available. If all else fails, you might replace the drive outright.
Symptom 34-30. You encounter memory parity errors at bootup. If you’re using non-parity memory devices (i.e. a 32-bit device instead of a 36-bit device), you will need to disable DRAM ECC or Parity Checking through the CMOS Chipset Features settings. This problem can occur if you reload default CMOS settings which restores parity/ECC on a system with non-parity memory. Also keep in mind that the Triton chipset does not support parity, so even if you use parity RAM, you should try disabling parity checking. If the system is configured properly, you may actually have a memory failure, and you’ll need to isolate the memory fault.
Symptom 34-31. You flash a BIOS, but now you get no video. When you flash a BIOS, the CMOS settings are left useless. This means you will have to restore the proper CMOS settings before the system may run properly. Clear your CMOS and reload the proper settings. The BIOS IC itself may also be troublesome. There are some problems when flashing an Intel flash ROM IC. Make sure that there are no warnings or cautions in the system documentation or from the manufacturer’s web site before flashing a particular BIOS IC. Try restoring the original BIOS if possible, or contact the manufacturer for a replacement BIOS.
Symptom 34-32. You are trying to use a PnP sound card and PnP modem together on the same system, but you’re getting hardware conflicts. This is an all-too-common problem with PnP systems. In general, the modem should take COM2 (2F8h and IRQ3), and the sound card should take 220h, IRQ5, and DMA 1. Try adding the cards one at a time - install the sound card first and let Windows 95 detect it. Add the modem next. If problems persist, configure the cards manually (disable their PnP support) if possible.
Symptom 34-33. After setting the DRAM speed to 70nS in the Advanced Chipset Setup, the system crashes or refuses to boot. Chances are that you have the incorrect number of wait states set for your memory configuration - 70nS RAM typically requires at least one wait state. Disable any "Auto Configure DRAM Timing" feature, then set the number of Wait States to 1. That should clear up the problem.
Symptom 34-34. There is 32MB (or more) of memory, and the BIOS counts it all during POST, but you only see 16MB in the CMOS Setup screen. This is a problem which has been identified with some Award BIOS versions. To correct the problem, make sure that the "memory hole" option in the Advanced Chipset Setup area is disabled. The "memory hole" option assumes a maximum of 16MB of physical RAM in the system.
Symptom 34-35. You move a working IDE drive from an older 386/486 system to your new Pentium system, but the system no longer works. In most cases, the data transfer mode is set improperly for the old IDE hard drive (i.e. using LBA mode when the IDE drive requires CHS mode). Find the Peripheral Setup screen in your CMOS Setup and make sure to change all the PIO mode settings to Mode 0 (chances are the settings are currently at Automatic, and are configuring the data transfer incorrectly). In some cases, you may need to repartition and reformat the drive in order to use it on a different (older) controller.
Symptom 34-36. Windows 95 locks up when you install a Diamond Stealth Video 3200 board and an Intel EtherExpress Pro 10/100 network card. However, you verify that both cards work fine on other systems. Problems begin when you load the Intel network driver. This is a problem that has been identified with Premio motherboards and is due to a problem in the system BIOS. Upgrade the Premio BIOS to the latest version.
Symptom 34-37. You install an Intel Pentium P55C (MMX) 200MHz CPU, and you set the CPU speed jumper(s) for 200MHz, but the system still reports 166MHz. In virtually all cases, you have set the speed jumper(s) incorrectly. Take another look at the documentation for your motherboard, and see that the speed is indeed set correctly. If problems persist, the BIOS may not recognize the higher CPU speed correctly, so try upgrading the motherboard BIOS.
Symptom 34-38. The system frequently locks up or crashes after installing a Cyrix 6x86 CPU. In most cases, the Cyrix 6x86 is not being cooled properly, and is overheating. Make sure that you have a heat sink/fan assembly attached properly to the Cyrix, and see that the fan is running. Also, the Cyrix 6x86 P166+ is a 3.52V CPU. Check your voltage regulator and see that it is set to provide 3.45 to 3.6 volts.
Symptom 34-39. After installing a Pentium 120MHz motherboard, you get Registry corruption or "out of memory" errors from Windows 95. This happens most often with slightly older Pentium motherboards (~100-120MHz), and is almost always a BIOS version problem which causes the motherboard to misbehave under Windows 95. You will need to update the BIOS version for your particular motherboard.
Symptom 34-40. The motherboard fails to "auto-detect" the hard drive parameters. This is a known problem on Dataexpert EXP8551S motherboards, and is due to a problem with Windows 95 in recognizing the PCI/ISA/I/O controller portion of the chipset. You can use the following procedure to force Windows 95 to recognize the chipset properly:
NOTE: You should make a backup copy of the MSHDC.INF file before proceeding to edit the file. That way, you can easily restore the original file if necessary.
If the problem persists, you should try entering the specific hard drive parameters for your particular drive into the CMOS Setup.
Symptom 34-41. The motherboard refuses to detect the SCSI controller during bootup. This problem has been identified with the Dataexpert EXP8551 motherboard, but may occur on many different types of PCI motherboards. In most cases, you will have to change the configuration of your PCI slots on the motherboard. For example, if the SCSI controller is installed on slot 2, you will need to configure the PCI slot 2 in CMOS Setup.
Symptom 34-42. You find that you cannot run a Cyrix 6x86 CPU on a particular motherboard. This is a problem that has been identified on Eurone/Matsonic motherboards, and is usually the result of an incompatible motherboard clock generator. Some clock generators support the Cyrix 120, 133, and 166MHz models, but exempt the 200MHz model - other clock generators support the 120, 150, 166, and 200MHz models, but exempt the 133MHz model. So if you’re using a 133MHz or 200MHz Cyrix CPU, you may be using the "wrong" clock generator. You will have to replace the CPU with a speed suitable to the particular clock generator, or change the motherboard to one which will accommodate the particular CPU speed.
Symptom 34-43. The system can only count up to and recognize 8MB of RAM though the system can accommodate even more. This is often a problem identified with Freetech 586F61x motherboards using Award BIOS version D or earlier. You can duplicate the problem by initiating a software reset with <Ctrl>+<Alt>+<Del>, then hitting the hardware reset - BIOS will only count memory up to 8MB. You will need to update the Award BIOS to version E or later. Freetech provides the BIOS patch on their web site.
Symptom 34-44. When four 8MB SIMMs are installed in the system (32MB), the system only counts up to 24MB. This is a known problem with Gigabyte motherboards (typically the GA-586ATE, ATM, and AP ver 1.x). The motherboard does not support "double sided" SIMMs (i.e. 2MB, 8MB, 32MB, or 128MB) in the center bank. Install the SIMMs in bank 0 and bank 2 - leaving bank 1 empty.
NOTE: Some motherboards require the banks to be filled in sequential order, or allow you to change the bank order with jumpers.
Symptom 34-45. Gold-plated SIMMs do not work properly in tin-plated sockets. As a general rule, you should avoid mixing metal types when choosing SIMMs - the metal in the SIMM socket must be the same as the metal on the SIMM itself. Otherwise, tin debris will transfer to the gold surface and oxidize. This will eventually result in memory failures which suggest faulty SIMMs.
Symptom 34-46. Even though all peripherals in the system are SCSI, Windows 95 will continue to detect the PCI IDE controller. You notice that this occurs even though the controller was disabled in CMOS. This is a known problem with the Iwill P54TS motherboard. Normally, Windows 95 will try to recognize and try to enable I/O devices, but should not enable devices that are deliberately disabled in CMOS. This is typically a BIOS problem, so try upgrading your BIOS to the latest version.
Symptom 34-47. You get an "EISA CMOS Configuration Error" when the system starts up. For EISA systems, you MUST run the EISA configuration utility in order to properly setup the system. Without this step, the system will not be able to detect any possible resource conflicts. This type of problem is most common when installing a new EISA motherboard, when CMOS contents are lost, or when devices (such as memory) are added or removed.
Symptom 34-48. The SMP (dual processor) mode refuses to run in Windows NT. The most common problem is an incompatibility with the SMP HAL shipped with Windows NT (versions prior to 3.51) and the motherboard's chipset. If you are upgrading from an older version of NT (prior to 3.51), first install NT as a standard PC (single processor kernel), then install NT with the default multi-processor kernel it provides (NT will not recognize your dual CPUs if you upgrade straight to a multi-processor configuration).
Symptom 34-49. When attempting to upgrade your flash BIOS, you encounter an "insufficient memory" error. In most cases, you simply don’t have enough conventional memory available to execute the flash program. Most flash programs require about 560KB or conventional RAM. Try booting "clean" with a DOS diskette, then run the flash upgrade.
Symptom 34-50. You see a prolonged system message saying "Updating ESCD…" each time the system boots. The Extended System Configuration Data (ESCD) area is part of a PnP system. One or more PnP devices are attempting to update your BIOS settings. To stop this from occurring, set the BIOS to "program" mode.
Symptom 34-51. You notice a yellow (!) sign over your USB port in the Device Manager. Windows 95 indicates that it has detected an unknown PCI device. In virtually all cases, the proper driver for the USB on your system has not been installed, and Windows 95 cannot recognize the USB hardware. You can usually correct this problem by updating your system BIOS to a newer version which supports the USB under Windows 95.
Symptom 34-52. The Device Manager under Windows 95 indicates four COM ports (at unusual IRQs and I/O addresses), but there are only two physical ports on the motherboard. This problem has been identified with the Ocean Rhino motherboard which is running a very old Award BIOS. The Award BIOS has since been upgraded to provide full support for Windows 95, so download the newest BIOS version from the motherboard manufacturer.
Symptom 34-53. The performance of a motherboard with an AMD K5 CPU seems extremely poor. This is almost always because of the motherboard BIOS - chances are the BIOS was released before the AMD K5 was widely introduced, so there may be problems providing proper AMD support. Make sure that you are using the very latest BIOS which supplies adequate AMD support.
Symptom 34-54. The system hangs up after installing a Cyrix 6x86 CPU. There is probably a problem with the utilization of system cache which is causing the system to hang up. Try disabling the internal (L1) and external (L2) cache.
Symptom 34-55. When attempting to upgrade the BIOS version, you cannot use a key sequence such as <Ctrl>+<Home> to reboot the PC in order to start the flash process. The current BIOS version does not support such key sequences. To flash the BIOS, start the flash program manually from the DOS prompt. For example:
A:\> AMIFL PAIV17.ROM <Enter>
Symptom 34-56. You find that a particular SVGA board refuses to work on a particular motherboard. However, the video board proves out fine on other systems. In most cases, this is a compatibility problem between the video chipset and the motherboard. There may be a BIOS upgrade for the motherboard or video board that can overcome the problem. You may simply have to use a different video board.
Symptom 34-57. When the on-board printer port is set to 3BCh (and EPP/SPP mode) and another parallel port add-on card is set to 378h or 278h, the BIOS only recognizes the add-on card. Port 3BCh seems to disappear. This may be a configuration problem with the Winbond chipset which specifies that LPT1 on the motherboard should be set at 378h (EPP or SPP), while add-on parallel ports should be set at 278h or 3BCh. The Winbond chip was designed this way for Windows 95. Check with the motherboard manufacturer for any available BIOS upgrades which can correct this issue.
Symptom 34-58. With 32MB of RAM on the motherboard, Checkit 3.0 causes the system to reboot when performing DRAM tests. This is because Checkit 3.0 will not perform memory testing over 16MB. This is an issue with Checkit - not the motherboard. Upgrade to a later version of Checkit.
Symptom 34-59. The IBM Blue Lighting CPU will not run on a motherboard which should support it. In most cases, the problem is an older BIOS version. Make sure that you are running the latest version of BIOS before installing the IBM Blue Lightning. Also check to make sure that any CPU type and speed jumpers are set properly for the CPU.
Symptom 34-60. When using a benchmark program such as SYSINFO, the "Overall Performance" rating of a Pentium 100 system marks better than a Pentium 120 system. This is because of the PCI bus speed. For a 100MHz system, the PCI bus speed is 33MHz. For a 120MHz system, the bus speed is 30MHz. The slightly faster PCI system will register a bit better performance.
NOTE: Always make sure that your benchmark and diagnostic programs are updated for the CPUs and other hardware that you are testing.
Symptom 34-61. You cannot get parallel port devices to work on your motherboard. In most cases, you must set the proper parallel port mode (i.e. SPP/ECP/EPP) for the particular device you plan to use. Often, setting the port to "Compatibility Mode" will work for many common peripherals. Parallel port modes are selected through the CMOS Setup - usually under "Integrated Peripherals" or some similar heading.
Symptom 34-62. You notice that some configurations of memory provide less performance than others. This type of problem is most noted on motherboards with 440FX chipsets, and is usually the result of a BIOS problem. Try updating your BIOS to the latest available version.
Symptom 34-63. You see no performance improvement when enabling PCI/IDE bus mastering. The problem is often that you are using an older (or buggy) driver. Make sure that you have installed the most recent bus mastering driver file (Triton I, Triton II and Natoma chipsets may use the same driver).
Symptom 34-64. The BIOS banner displayed on power-on is showing the wrong motherboard model. In virtually all cases, this is a problem with the BIOS version. Get the latest update for your motherboard BIOS.
Symptom 34-65. The Pentium P55CM BIOS shows a 150MHz CPU even though the CPU is a 166MHz model. This is almost always due to a BIOS fault. You should upgrade to the very latest BIOS version for your particular motherboard. If you cannot flash the BIOS, replace the BIOS IC outright.
Symptom 34-66. You see "Static Device Resource Conflict" error message after the system memory count when using the P55CM CPU. This is usually a problem with the PCI bus system. Press and hold the <Insert> key before turning on the computer. Release the <Insert> key when the video comes up. This forces the system to reassign PCI resources. If the error message still appears, remove all PCI cards (except for the video card) and try again. Reinsert one PCI card at a time until the problem returns - that is where the problem is.
Further Study
That’s all for Chapter 34. Be sure to review the glossary and chapter questions on the accompanying CD. If you have access to the Internet, take a look at some of the motherboard resources listed below:
Abit Computer Corp.: http://www.abit.com.tw/html/emain.htm
Acer America Corp.: http://www.acer.com
American Megatrends (AMI): http://www.megatrends.com
American Predator Corp.: http://www.americanpredator.com
ASUS: http://www.asus.com
Biostar Microtech Intl.: http://www.biostar.net
CompuTrend Systems, Inc. (Premio): http://www.premiopc.com
Data Expert Corp.: http://www.dataexpert.com
Diamond Flower, Inc. (DFI): http://www.dfiusa.com
Elitegroup Computers, Ltd. (ECS): http://www.ecs.com.tw
Famous Technology Co., Ltd.: http://www1.magic-pro.com.hk/famous/index.html
First International Computer, Inc. (FIC): http://www.fica.com
Fong Kai Industrial Co. (FKI): http://www.fkusa.com
Gemlight Computer Ltd.: http://www.gemlight.com.hk
Genoa Systems Corp.: http://www.genoasys.com
Giga-Byte Technology Co., Ltd.: http://www.giga-byte.com
Intel Corp.: http://www.intel.com
Iwill Computer: http://www.iwill.com.tw
Jbond: http://www.jbond.com
J-Mark Computer Corp.: http://www.j-mark.com
Kam-Tronic Computer Co., Ltd.: http://megastar.kamtronic.com
Micronics Computers, Inc.: http://www.micronics.com
Microway: http://www.microway.com
Micro Star International Co., Ltd. (MSI): http://www.msi.com.tw
PC Chips Manufacturing Ltd.: http://www.pcchips.com
Pine Technology Ltd.: http://www.pinegroup.com
Shuttle Computer International: http://www.shuttlegroup.com
Soyo Computer Inc.: http://www.soyo.com.tw
Supermicro Computer Inc.: http://www.supermicro.com
Tekram Technology: http://www.tekram.com
Tyan Computer: http://www.tyan.com
| Chapter: 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 | 41 | 42 | 43 | 44 | 45 | 46 | 47 | 48 | 49 | 50 | 51 | 52 | 53 |
| Reserve your copy at a Beta Bookstore near you! |
Contact Bet@books © 1998 The McGraw-Hill Companies, Inc. All rights reserved. Any use of this Beta Book is subject to the rules stated in the Terms of Use. |
