Libris Britannia 4/science library(b).zip/science library(b)/INFO/DIAGTXT.ZIP/MEM.TXT

Text File | 1986-12-12 | 12KB | 325 lines

                              Diagnosing Memory Problems
                                           by
                                      Arnold Kishi
                                 Hawaii PC Users Group
                         [as updated by Jerry Schneider, CPCUG]

      If you encounter the message "parity error" while operating your PC, a
       malfunction has occurred somewhere in the PC's memory (RAM).  In many
       instances, a parity error can be traced to a defective RAM chip or one
       that is seated poorly in its socket.

      To find the cause of the parity error, turn off the PC, wait about five
       seconds, and turn it on again.  After flipping the power switch on, the
       PC automatically runs a series of diagnostics which tests different
       parts of the system.  If a memory problem exists, a PARITY CHECK 1 or
       PARITY CHECK 2 error message will appear.  This article will help you
       interpret these error messages, isolate defective memory chips, and
       correct the situation.


      System Board Memory

      Early PCs have system boards which hold only 64K of RAM.  These system
       boards have four parallel rows (or banks) of nine 16K 4116 RAM chips.
       Newer PCs and XTs with 256K system boards have four parallel rows of
       nine 64K 4164 RAM chips.  In both cases, each row has eight chips for
       memory (data bits) and one for parity checking (parity bit).

      When facing the PC chassis from the front, the row of chips closest to
       the front is designated row 3, with the next row being row 2, then row
       1, and row 0 next to the expansion slots.  In each row or bank, the
       chip on the extreme left, slightly separated from the data bits, is the
       parity bit chip.


      Expansion Board Memory

      The amount of memory contained on expansion boards varies widely from
       manufacturer to manufacturer.  While the early expansion memory boards
       only contained 64K memory, the current standard seems to be memory
       boards in one of three configurations:  256K, 384K, or 512K.  Most
       memory boards still use the 64K 4164 RAM chip, although use of the
       newer 256K 41256 RAM chip has become more prevalent as its
       availability has increased and unit cost has dropped substantially
       (from $60/chip to $7/chip during the past six months).  The
       introduction of 80286-based computers like the AT has also impacted the
       popularity of mega-memory expansion cards because the 80286 can address
       up to 16 megabytes of RAM, as compared to the 1 megabyte limitation of
       8088-based PCs.








                                         1








      To simplify matters, we will limit our discussion here to those
       expansion memory boards that use the 64K RAM chips.  In general, these
       expansion boards comprise from four (256K) to eight (512K) parallel
       rows of nine 64K memory chips.  The rows are usually arranged
       horizontally, numbered from 1 to 8, and are populated sequentially from
       right to left (row 1 is on the extreme right; row 8 on the extreme
       left).


      Memory Error Messages

      Now that you know which row is where and its "numeric" position
       indicator, you can begin to isolate which chip may be causing the
       parity error.

      If your computer has a memory problem, a memory error message will
       appear on the monitor during the start-up diagnostic (Power-On Self
       Test) or when performing system diagnostics.  This memory error message
       consists of two parts:  a four-digit error code followed by the numbers
       201, e.g. 3040 201; and either a PARITY CHECK 1 or PARITY CHECK 2
       message.

      A PARITY CHECK 1 message indicates that the memory error was detected
       on the system board memory; a PARITY CHECK 2 message identifies a
       memory error on a memory expansion board.


      PARITY CHECK 1 -- System Board Memory Errors

      The first number of the memory error code indicates which 64K bank of
       memory is involved.  On PCs with 256K system boards, this can be rows
       0, 1, 2, or 3.  On 64K system board PCs, the number 0 represents the
       entire 64K bank of 36 16K chips.

      For PCs with 64K system boards, the second digit can be 0, 4, 8, or C,
       and points to the 16K bank within the 64K which is failing; 0 is row
       0, 4 is row 1, 8 is row 2, and C is row 3.  In contrast, for PCs with
       256K system boards, the second digit of the error message identifies
       the 4K page in the memory chip that is failing; this number is not
       needed to identify the problem chip and should be ignored.

      The third and fourth digits represent which bit position (or RAM chip)
       in the 64K row is causing the error condition.  This number represents
       the hexadecimal address of the chip within the row.













                                         2








      Table 1 provides the address for each of the chips on a 64K or 256K
       system board:

        Bank     64K    256K   Parity  Bit  Bit  Bit  Bit  Bit  Bit  Bit  Bit
        Number  System  System   Bit     0    1    2    3    4    5    6    7

          0       00      0x      00     01   02   04   08   10   20   40   80
          1       04      1x      00     01   02   04   08   10   20   40   80
          2       08      2x      00     01   02   04   08   10   20   40   80
          3       0C      3x      00     01   02   04   08   10   20   40   80

                                 M   E   M   O   R   Y      C   H   I   P   S

                                     F R O N T    O F    C O M P U T E R

        Table 1.  System Board Memory Addresses.

      Using the error message example given earlier, 3040  201, we can
       quickly identify the faulty chip as the eighth chip (Bit 6) in the
       first row from the front (Bank 3) of a 256K system board.


      PARITY CHECK 2 -- Expansion Board Memory Errors

      The procedures for diagnosing errors in memory contained on expansion
       boards is similar to that used for system board memory diagnosis.

      The first number of the memory error code indicates which 64K bank of
       memory is involved.  On PCs with 256K system boards, this can be row 4
       or greater; on 64K system board PCs, row 1 or greater.

      For both 64K and 256K system board PCs, the second digit is not used
       and should be ignored.  The third and fourth digits contain the
       hexadecimal address (within the row) of the problem memory chip.

      Because there are numerous third-party manufacturers of expansion
       memory boards and because each manufacturer may use a different
       configuration and/or bank numbering scheme, the rules for
       identifying errant memory chip addresses for your expansion board might
       be different.  You should check the documentation provided with your
       expansion board as reference for proper diagnosis of expansion board
       memory problems.

      For demonstration purposes, however, the following example uses a 384K
       memory expansion board (AST Six Pak Plus), configured with six banks of
       chips, numbered 1 to 6, from right to left.











                                         3








      Table 2 provides the address for each of the chips on the expansion
       memory card.  Notice that the starting address for the rows of memory
       chips is different for the 64K and 256K system boards.  This is due to
       the fact that the expansion board rows begin addressing where the
       system board stops:  for the 64K board, starting address is 1x; the
       starting address for the 256K system board is 4x (x can be any number).


      Bank No.          6    5    4    3    2    1

      64K System       6x   5x   4x   3x   2x   1x

      256K System      9x   8x   7x   6x   5x   4x
       ____________________________________________

      Parity Bit       00   00   00   00   00   00      M

      Bit 7            80   80   80   80   80   80      E

      Bit 6            40   40   40   40   40   40      M

      Bit 5            20   20   20   20   20   20      O

      Bit 4            10   10   10   10   10   10      R

      Bit 3            08   08   08   08   08   08      Y

      Bit 2            04   04   04   04   04   04      C
                                                        H
      Bit 1            02   02   02   02   02   02      I
                                                        P
      Bit 0            01   01   01   01   01   01      S

                        BOTTOM OF EXPANSION BOARD

       Table 2.  Expansion Board Memory Addresses.

      Using the above table, we can see that a 7120 201 error code identifies
       the errant memory chip as the sixth chip from the bottom (Bit 5) in the
       fourth bank from the right (Bank 4) in a 256K system board machine.


      Correcting Memory Errors

      Now that we have identified the problem chip, we should verify it by
       replacing it with a spare chip and run the diagnostics again and see
       whether the error is corrected.  If no spare chip is available,
       exchange the suspect chip with another one in an adjacent bank.  If we
       have correctly identified the problem chip, the diagnostics will
       display a different memory code -- that of the location where we put
       the suspected chip.

      If, however, the error code continues to identify the original
       location, a problem may exist with the socket and you should contact
       your dealer for assistance.è

                                         4









      If the system board or expansion board switches are not set properly or
       a chip is missing, the bit position in the error code may be AA, FF,
       55, or 01.  If you are experiencing problems with more than one memory
       chip, the bit position code displayed will be the sum (in Hexadecimal)
       of the problem chip locations and consequently, will not match any of
       the values in Table 1 or 2.  When this happens, the resultant error
       code could be any number from FF to 00.  Diagnosis of the errant chips
       will involve a trial and error process of switching several chips from
       the identified row to an adjacent row.


      Reseat Chips Before Switching

      Before switching chips in the "suspect" row, however, remove all chips
       and reseat them in their sockets.  Run the diagnostics again.  Because
       many parity problems are due to poor contacts between chips and
       sockets, this trick may eliminate the parity error.  RAM chip failures
       are rather rare.

      The most difficult parity error or memory error to locate is one which
       occurs "intermittently".  For example, heat generated by expansion
       boards, disk drives or other add-ons may cause a memory chip or some
       other memory-related component to shift sufficiently to break the
       electrical contact.  You may never find this particular problem since
       the same set of conditions may not be duplicated exactly during
       diagnostics.


      A Simple, Do-It-Yourself Remedy

      What has been discussed here is a simple method to troubleshoot your
       PC.  If in doubt, professional service technicians have the tools to
       diagnose memory problems instantly.  But performing these simple checks
       yourself can save you money, especially if the cause is a poorly
       socketed chip.





















                                         5