Jump to content

IBM PC AT Troubleshooting Guide: Difference between revisions

From RetroTechCollection
← Older edit

Josh (talk | contribs)

8 user groups
18,330 edits
undo
VisualWikitext
Expand troubleshooting: RIFA/tantalum + Dallas/planar battery (161/163) + POST codes; cited
Rewrite from IBM HMS MAP 0000/0020 and minuszerodegrees: fix 109 (low meg chip select, not DMA), 201 format (address + bits), CMI drive (6426-S was the 20 MB; 30 MB was IBM-made); remove HelpPC generic codes, invented beep rows, drive-type counts, stiction advice, slot keying claim; add POST card codes, 17xx order, power procedure
 
(2 intermediate revisions by the same user not shown)
Line 1: Line 1:
This guide covers diagnostic procedures for the IBM PC AT (5170), including POST audio and numeric error codes, hard drive faults, the well-documented 601 / CMOS battery / CMI HDD issues, and other known AT-specific problems.
<templatestyles src="Template:StyledTable/styles.css" />
[[File:IBM PC AT (photo).jpg|thumb|right|300px|IBM PC AT (5170)]]


== Power-On Audio Beep Codes ==
Fault finding on the '''IBM PC AT (5170)''' follows IBM's own service sequence: check that the supply fan runs, listen for the beep at the end of the power-on self test (POST), read any error on the screen, and run the Advanced Diagnostics. IBM's 1986 service manual turns each POST indication into a procedure; minuszerodegrees.net documents the causes it has found for each. Supply limits, the battery check and the load resistor are covered in the [[IBM PC AT Maintenance Guide]].


Beep codes on the 5170 are produced by the BIOS through the PC speaker before video is initialised. Same coding scheme as on the 5150 and 5160.
== Before you start ==


{| class="wikitable styled-table" style="width:100%; text-align:center;"
* IBM's start procedure begins with the power off, all connectors and jumpers checked, the 115/230 V selector set for the local supply, and the configuration confirmed by running SETUP.<ref name="map0000">IBM, ''Hardware Maintenance Service'', 6280087, Revised Edition (March 1986), MAP 0000 "Start (AT)": checks before starting p. 0000-1; power cord and Power Good p. 0000-11; POST error table p. 0000-13; failure symptom table p. 0000-14. Hosted as [[:File:IBM Hardware Maintenance Service 6280087 MAP 0000 Start (AT).pdf]], described at [[IBM Hardware Maintenance Service (March 1986)]].</ref> The AT has no SETUP in ROM; see [[IBM PC AT Maintenance Guide#SETUP]].
|+'''IBM 5170 power-on audio codes'''
* Many POST errors are followed by a prompt to press F1 to continue.<ref name="map0000" />
! Pattern !! Meaning !! Stage
* Unlike the 5150 and 5160, the 5170 POST writes checkpoint codes to I/O port 80h, which a POST card can display. The code is written just before each test, so a card that stops at a code points at the test that failed.<ref name="mzcodes">[https://www.minuszerodegrees.net/5170/post_errors/5170_post_codes.htm IBM 5170 – POST – Codes], minuszerodegrees.net.</ref>
* All of this assumes IBM BIOS ROMs. Third-party BIOS replacements behave differently.<ref name="mzpost">[https://www.minuszerodegrees.net/5170/post_errors/5170_post_errors.htm IBM 5170 – POST – Some errors (on-screen, and otherwise)], minuszerodegrees.net.</ref>
 
== IBM's POST error table ==
 
IBM's start procedure sends each POST indication to a maintenance analysis procedure (MAP):<ref name="map0000" />
 
{| class="wikitable styled-table" style="width:100%;"
|+'''IBM 5170 POST indications (HMS MAP 0000, p. 0000-13)'''
! Indication !! IBM's action
|-
| No beep, with a blank display, a blinking cursor, an unreadable display, or even with the machine working || MAP 0020: Power
|-
|-
| (no beep) || Power supply or motherboard not running &mdash; check 5 V/12 V, PSU minimum-load resistor, bank-0 RAM, CPU socket, system clock || Before POST
| No beep and a 1xx error || MAP 0100: System board
|-
|-
| 1 short || POST complete, system OK || End of POST
| 1 long, 1 short beep || Replace the system board
|-
|-
| 2 short || Numeric error code displayed; read the screen for the 1xx&ndash;1Fxx code || Any
| 1 long, 2 short or 1 long, 3 short beeps || Display adapter MAP (EGA: MAP 2400)
|-
|-
| 1 long, 1 short || Motherboard speed test failed (Type 3 BIOS, motherboard speed outside expected window) || POST checkpoint
| 2 short beeps and a blank, unreadable or distorted display || Display adapter MAP
|-
|-
| 1 long, 2 short || Display adapter failure (MDA/CGA/EGA) || POST
| 2 short beeps and an error code || 1xx: MAP 0100; ''xxxxxx xxxx'' 201: MAP 0200; 30x or ''xx''30x: MAP 0300; 601: MAP 0600; 17xx: MAP 1700; 30xx and 31xx: PC Network MAPs
|-
|-
| 1 long, 3 short || Enhanced graphics adapter (EGA) memory failure || POST
| C8000 ROM error || Replace the fixed disk adapter
|-
|-
| 3 long || 3270 keyboard test failure (variant boards only) || POST
| ROM error || Replace the system board
|-
|-
| Continuous beep || Power supply problem, motherboard problem, or stuck speaker || Any
| Continuous beep || MAP 0020: Power
|-
|-
| Repeated short beeps || Power supply problem, often a shorted tantalum on +12 V or +5 V || Any
| Repeating short beeps || MAP 0020: Power
|}
|}


== Numeric POST Error Codes ==
IBM's table of other symptoms sends an incorrect memory size and PARITY CHECK to the memory MAP; display problems to the display adapter MAPs; a flashing cursor only to the coprocessor check and then the diskette MAP; a BASIC screen, a boot from fixed disk or a diskette boot failure to the diskette MAP; and an inability to finish the diagnostic tests to the power MAP.<ref name="map0000" />


If the BIOS detects a fault during POST after video has initialised, it displays a numeric code on the screen and (usually) halts or prompts F1 to continue. The codes below are the IBM-specified set for the 5170 with original IBM BIOS; third-party BIOS replacements may use different codes for some classes.
=== Beep causes ===


=== 1xx &mdash; System board ===
{| class="wikitable styled-table" style="width:100%;"
 
|+'''IBM 5170 beep indications: causes found by minuszerodegrees'''
{| class="wikitable styled-table" style="width:100%; text-align:center;"
! Indication !! Causes
|+'''5170 1xx system board codes'''
! Code !! Meaning
|-
| 101 || Interrupt failure (8259A PIC)
|-
| 102 || Timer failure (8254 PIT channel 0)
|-
| 103 || Timer interrupt failure
|-
| 104 || Protected mode failure (80286 internal)
|-
| 105 || Last 8042 (keyboard controller) command not accepted
|-
| 106 || Converting logic test (post-protected-mode return)
|-
| 107 || Hot NMI test
|-
| 108 || Timer bus test
|-
| 109 || DMA test error (8237A)
|-
| 110 || System board parity error
|-
|-
| 111 / 112 / 113 || Adapter card parity / I/O channel check
| 1 long, 1 short || 06/10/85 and 11/15/85 BIOS only. System board speed outside certain bounds, typically after a faster crystal has been fitted.
|-
|-
| 121 || Unexpected hardware interrupt occurred
| 1 long, 2 short || A faulty video card, a poorly seated card, dirty edge contacts, or interference from another card. An IBM EGA card set up for an MDA or CGA monitor can also produce it when its own ROM finds a fault.
|-
|-
| 129 || Cache error (later board steppings with cache support; unsupported CPU upgrade)
| 1 long, 3 short || The ROM on an IBM EGA card reporting a fault with itself.
|-
|-
| 131 || Cassette wrap test failed (legacy code, no cassette port on the AT)
| Memory count completes, then nothing || On a type 3 board, an MDA card with SW1 in the colour position. Also seen with a faulty Fixed Disk and Diskette Drive Adapter.
|-
| 161 || CMOS configuration empty &mdash; '''battery failure''' or freshly fitted board
|-
| 162 || CMOS checksum error (adapter ID mismatch) &mdash; run SETUP
|-
| 163 || Time and date not set &mdash; run SETUP and re-enter
|-
| 164 || Memory size error &mdash; CMOS does not match installed RAM
|-
| 166 || Adapter card system board error
|-
| 181 || Bad CMOS configuration
|-
| 199 || User-indicated configuration not correct
|}
|}


=== 2xx &mdash; Memory ===
Source: minuszerodegrees' 5170 POST errors page.<ref name="mzpost" />
{| class="wikitable styled-table" style="width:100%; text-align:center;"
 
|+'''5170 2xx memory codes'''
== On-screen errors ==
! Code !! Meaning
 
{| class="wikitable styled-table" style="width:100%;"
|+'''IBM 5170 POST error messages'''
! Code !! Message or IBM's description !! Notes (minuszerodegrees)
|-
| 101 || Interrupt failure ||
|-
|-
| 201 || Memory test failed &mdash; the four hex digits to the left of "201" identify the failing byte's segment, the digit pair to the right identifies the failing bit position in that byte
| 102 || Timer failure ||
|-
|-
| 202 / 203 || Memory address line failure
| 103 || Timer interrupt failure ||
|-
|-
| 215 || 64 KB memory module failure on a 16-bit memory expansion card
| 104 || Protected mode failure ||
|}
 
=== 3xx &mdash; Keyboard ===
{| class="wikitable styled-table" style="width:100%; text-align:center;"
|+'''5170 3xx keyboard codes'''
! Code !! Meaning
|-
|-
| 301 || Keyboard error (or stuck key) &mdash; the two hex digits before "301" are the scan code of the stuck key
| 105 || Last 8042 command not accepted ||
|-
|-
| 302 || Keyboard locked (5170 keylock is in the locked position) or keyboard test failure
| 106 || Converting logic test ||
|-
|-
| 303 || Keyboard or 8042 keyboard controller error
| 107 || Hot NMI test ||
|-
|-
| 304 || Keyboard or 8042 keyboard controller error &mdash; CMOS does not match keyboard
| 108 || Timer bus test ||
|}
 
=== 4xx / 5xx &mdash; Display adapters ===
{| class="wikitable styled-table" style="width:100%; text-align:center;"
|+'''5170 4xx / 5xx display codes'''
! Code !! Meaning
|-
|-
| 401 || Monochrome display adapter (MDA) failed
| 109 || Low meg chip select test ||
|-
|-
| 408 || MDA display attribute failure
| 161 || System Options Not Set (dead battery) || Replace the battery, then run SETUP; the error stays until SETUP is completed. Also a faulty system board.
|-
|-
| 432 || Parallel printer port test failed (on the MDA "monochrome and printer adapter")
| 162 || System Options Not Set (CMOS checksum error) || Corrupted SETUP, a flat battery on some boards, or a video card that does not match SETUP.
|-
|-
| 501 || Colour Graphics Adapter (CGA) failed
| 163 || Time & Date Not Set (clock not updating) || Invalid date or time, SETUP started and not finished, or a clock circuit fault.
|-
|-
| 508 || CGA display attribute failure
| 164 || Memory Size Error || RAM added without running SETUP, wrong amounts entered, or faulty RAM.
|}
 
=== 6xx &mdash; Floppy drive / controller ===
{| class="wikitable styled-table" style="width:100%; text-align:center;"
|+'''5170 6xx floppy codes'''
! Code !! Meaning
|-
|-
| 601 || Diskette or controller adapter error (see "601 quirk" below)
| 201 || Memory Error || Shown as ''AAAAAA BBBB'' 201: ''AAAAAA'' is the hexadecimal address of the failing word and ''BBBB'' the failing bits. The address is sometimes one word beyond the fault. A ''BBBB'' with many F digits, such as FFFE, usually means RAM that is not there is being tested, for example SETUP set for 640 KB with 512 KB fitted.
|-
|-
| 602 || Diskette boot record failure
| 202, 203 || Memory Address Error || Address bits 0 to 15 (202) or 16 to 23 (203).
|-
|-
| 606 || Diskette verify error
| 301 || Keyboard Error || A preceding byte identifies a stuck key. Otherwise no keyboard, a PC/XT keyboard, a faulty keyboard or faulty keyboard circuitry on the system board. The 5170 has no keyboard fuse.
|-
|-
| 607 || Diskette write-protected
| 302 || System Unit Keylock is Locked || The keylock is locked, the keylock switch is faulty, or the keyboard controller misreads it. Unplug J20: if the error goes, the keylock is the cause.
|-
|-
| 608 || Bad diskette command
| 303 || Keyboard Or System Unit Error || Remove the keyboard and power on. If 303 goes, the keyboard is faulty; if it stays, the system board.
|-
|-
| 611 || Time out (controller did not respond)
| 304 || Keyboard Or System Unit Error (keyboard clock line high) || As 303.
|-
|-
| 612 / 613 || Bad NEC controller chip / DMA error
| 401 || CRT Error || A CGA card with SW1 in the monochrome (rear) position.
|}
 
=== 17xx &mdash; Hard drive ===
{| class="wikitable styled-table" style="width:100%; text-align:center;"
|+'''5170 17xx hard drive codes'''
! Code !! Meaning
|-
|-
| 1701 || Fixed disk POST error &mdash; controller initialisation failed, or no drive ready
| 501 || CRT Error || An MDA card with SW1 in the colour (front) position.
|-
|-
| 1702 || Fixed disk adapter error
| 601 || Diskette Error || See [[#Diskette faults]].
|-
|-
| 1703 || Drive error
| 602 || Diskette Boot Record Error || The boot sector was read but its contents were not as expected.
|-
|-
| 1704 || Adapter or drive error
| 1780, 1781 || Disk 0 Failure, Disk 1 Failure || See [[#Fixed disk faults]].
|-
|-
| 1780 || Fixed disk 0 failure
| 1782 || Disk Controller Failure ||
|-
|-
| 1781 || Fixed disk 1 failure
| 1790, 1791 || Disk 0 Failure, Disk 1 Failure ||
|}
|}


=== Other ===
The messages and IBM's descriptions are as minuszerodegrees quotes them from the IBM BIOS.<ref name="mzpost" />
* '''7xx''' &mdash; Math coprocessor (80287) error
* '''9xx''' &mdash; Parallel printer adapter error
* '''10xx''' &mdash; Reserved (parallel adapter)
* '''11xx / 12xx''' &mdash; Asynchronous communications (16450 UART) adapter errors
* '''13xx''' &mdash; Game adapter error
* '''14xx''' &mdash; Printer error
* '''15xx''' &mdash; SDLC adapter error
* '''18xx''' &mdash; I/O expansion unit error


== Known Problems ==
=== Memory and parity errors ===


The following AT-specific symptoms are documented at minuszerodegrees.net and confirmed by other long-term restorers.
The BIOS always follows PARITY CHECK with 1 (system board RAM) or 2 (RAM on an expansion card). Below it is the failing address, or five question marks if it cannot be found. A PARITY CHECK with no number is not from the IBM BIOS; the Parity_Boot virus is known to display it.<ref name="mzpost" /> IBM's service manual decodes the parity and 10-character memory error codes to a bank and a module; see [[IBM PC AT (5170)#Memory errors]].<ref name="fsim">IBM, ''IBM Personal Computer Family Service Information Manual'', SA38-0037-00 (July 1989), Chapter 9: power supply failures p. 9-8, coprocessor socket p. 9-9, memory error isolation and parity checks pp. 9-10 and 9-11, fixed disk types p. 9-7. Hosted as [[:File:IBM Personal Computer Family Service Information Manual SA38-0037-00.pdf]], described at [[IBM Personal Computer Family Service Information Manual]].</ref>


=== 601 error with 06/10/85 or 11/15/85 BIOS ===
When a memory error cannot be placed, IBM's procedure is to remove all Memory Expansion Adapters, run SETUP, repeat the failing test, and then refit the adapters one at a time, running SETUP each time.<ref name="fsim" />


With either of the 1985-dated BIOS revisions, removing the IBM Fixed Disk and Diskette Drive Adapter and substituting a third-party replacement '''can''' (not always) result in a 601 error at POST. There is something in those revisions that expects the IBM "combo" card. The error does not stop the boot; an F1 prompt allows boot to continue, although floppy operation may be compromised.
A fault in the first 64 KB of bank 0 shows as "000000 ''xxxx'' 201" with an MDA or CGA card, but as a dead system board with EGA or VGA. A POST card then shows DD. Reseat the bank 0 chips first; the fault may be a poor contact. On a type 1 board, swapping the bank 0 and bank 1 chips turns the dead board into a 201 error if a bank 0 chip was at fault.<ref name="mzprobs">[https://www.minuszerodegrees.net/5170/problems/5170_known_problems_issues.htm IBM 5170 – Known Problems/Issues], minuszerodegrees.net.</ref>


Workarounds:
== No beep, fan not running ==
* Use a replacement controller that the IBM BIOS recognises as the "combo" card: confirmed working replacements include the Western Digital '''WD1003A-WA2''', '''WD1002-WA2''' and '''WD1003-WA2'''.
* Use a patched IBM 5170 BIOS that disables the combo-card check.
* Use a non-IBM BIOS (AMI, Award, Phoenix). This is the cleanest workaround if 1.44 MB diskette boot support is desired.


The 01/10/84 BIOS does '''not''' produce 601 errors when the combo card is missing; this check was added in the second BIOS revision.
IBM's start procedure:<ref name="map0000" />


=== 161 / 162 / 163 errors &mdash; CMOS battery failure ===
# Check that the power cord is in a working, properly earthed outlet.
# Switch off, unplug, and check the power cord for continuity. Replace it if it is open.
# Reconnect and power on. Check for 2.4 to 5.2 V between pins 1 and 5 (ground) of power supply connector P8. If Power Good is present, replace the supply; if not, go to the power procedure below.


These three errors usually appear together when the external 6 V CMOS battery has run down. The MC146818 loses the CMOS configuration and the date/time. Replace the battery (see [[IBM PC AT Maintenance Guide]]) and re-run SETUP.
== Power procedure ==


If 161/162/163 persist with a fresh battery, suspect a failed MC146818, a dirty or oxidised battery header, or a shorted CMOS line on the motherboard.
IBM's power MAP for the AT, in order:<ref name="map0020">IBM, ''Hardware Maintenance Service'', 6280087, Revised Edition (March 1986), MAP 0020 "Power (AT)", pp. 0020-1 to 0020-7. Hosted as [[:File:IBM Hardware Maintenance Service 6280087 MAP 0020 Power.pdf]].</ref>


=== CMI 5616 / CMI 6426 hard drive failure ===
# Check that the 115/230 V selector matches the supply. Disconnect all cables and external devices except the display, and power on. If the symptom goes, reconnect the devices one at a time, powering off before each, until it returns.
# Remove the cover, reseat the power connectors, and check Power Good (2.4 to 5.2 V between P8 pins 1 and 5).
# If Power Good is present, check the system board and drive connector voltages; see [[IBM PC AT Maintenance Guide#Voltage checks]]. A machine with no fixed disk needs the load resistor on P10 or P11. If any voltage is wrong, replace the supply.
# Check the speaker for continuity with its cable off the system board.
# Unplug the power connector from each drive in turn. If the symptom goes, that drive is at fault.
# Remove the option adapters one at a time, leaving the Fixed Disk and Diskette Drive Adapter and the display adapter.
# Remove the math coprocessor, if fitted.
# Remove the Fixed Disk and Diskette Drive Adapter (expect a 601), then the display adapter (expect one long and two short beeps).
# Power off, wait 10 seconds, power on and check Power Good again. If it is missing, replace the supply, then the system board; if it is present, replace the system board, then the supply.


The 20 MB CMI 5616 drive (early 5170 models) and 30 MB CMI 6426 drive (later models) suffered well-documented reliability problems, leading to a PC Magazine article opening with the line "If you own an IBM PC AT and your hard disk hasn't crashed yet, don't worry &mdash; it probably will." If the drive does not spin up, makes a continuous click, or returns 1701/1780 errors, the drive itself is most likely faulty. Many enthusiasts retire the CMI drives in favour of a Seagate ST-225 (20 MB) or ST-251 (40 MB), or an MFM emulator such as XT-IDE / GoTek.
Other causes of a machine that will not start:


=== PSU will not start with diskette-only configuration ===
* A diskette-only machine without its load resistor: the +12 V output is below its 2.5 A minimum load. The 5170 may or may not start with its hard drive removed.<ref name="mzprobs" /><ref name="trat">IBM, ''Technical Reference, Personal Computer AT'', First Edition (September 1985), Section 3 "Power Supply", pp. 3-4 and 3-5. Hosted as [[:File:IBM PC AT 5170 Technical Reference Sep85.pdf]].</ref>
* A coprocessor fitted the wrong way round. Some 5170 system boards were made with the coprocessor socket's notch to the rear; a coprocessor fitted to match it gives a blank display and no beep. Fit it with its notch to the front of the board.<ref name="fsim" />
* An XT-class supply: a type 1 board with the 01/10/84 BIOS may not start at all, and other boards may start intermittently or lose SETUP.<ref name="mzprobs" />
* Bank 0 RAM with an EGA or VGA card (see above).
* A supply that fails after two or more hours and works again once cool. IBM's instruction is to replace it.<ref name="fsim" />


The 192 W AT PSU requires a minimum load on +12 V to start. Diskette-only ATs shipped with a 5 Ω 50 W resistor on the spare hard-drive power connector. If the resistor has been removed or has failed open, the PSU may not start. Verify the resistor is present and resistive (about 5 Ω across the +12 V/GND pins, fan off).
== Diskette faults ==


=== Floppy drive incompatibility ===
A 601 has several causes: the POST cannot find the diskette part of the Fixed Disk and Diskette Drive Adapter; a hard disk-style twisted cable used in place of the diskette cable; drive A faulty, without +5 V or +12 V, wrongly jumpered, or a model that does not work in the 5170.<ref name="mzpost" />


A 360 KB diskette written in the 1.2 MB drive cannot be reliably re-read in a 360 KB drive. The head track width of the 1.2 MB drive is half that of the 360 KB drive; the wider 360 KB read head picks up the half-track plus the remnants of any earlier track, producing garbled data. Use HD media in HD drives only.
With the 06/10/85 and 11/15/85 BIOS, replacing IBM's adapter with a third-party controller can produce a 601 at POST, because those revisions look for the IBM adapter. The boot continues after an F1 prompt, although diskette operation may be affected. The 01/10/84 BIOS does not do this. Fixes are a controller the POST takes for IBM's, such as the Western Digital WD1003A-WA2 (the WD1002-WA2 and WD1003-WA2 may also work), a patched IBM BIOS, or a third-party BIOS.<ref name="mzbios">[https://www.minuszerodegrees.net/5170/bios/5170_bios_revisions.htm IBM 5170 – BIOS Revisions], minuszerodegrees.net.</ref>


=== Memory size error (164) ===
* Corrupted diskettes: a 1.2 MB drive must generate DISK CHANGED on pin 34. A 360 KB drive that puts READY on pin 34 also causes trouble; disconnect pin 34 when fitting a 360 KB drive.<ref name="mzprobs" />
* 360 KB diskettes that other machines cannot read: a 160/180 KB or 320/360 KB diskette written in the high-capacity drive can afterwards be read only in a high-capacity drive.<ref name="fsim9">IBM, ''IBM Personal Computer Family Service Information Manual'', SA38-0037-00, Chapter 9, "High-Capacity Diskette Drive", p. 9-2.</ref> IBM offered the optional 360 KB drive for writing disks for 360 KB machines.<ref name="mz360">[https://www.minuszerodegrees.net/diskette/360K_floppy_in_1.2M_drive.htm 360K Floppy in 1.2M Drive], minuszerodegrees.net.</ref>
* No boot from drive B: the POST does not boot from B.<ref name="mzbasics">[https://www.minuszerodegrees.net/5170/misc/5170_basics.htm IBM 5170 – Basics, for 5170 beginners], minuszerodegrees.net.</ref>
* 1.44 MB drives are fully supported only by the 11/15/85 BIOS.<ref name="mzbios" />


A 164 error means the BIOS POST detected an installed memory size that does not match the CMOS configuration. Run SETUP and re-enter the actual installed RAM size. If the size cannot be set correctly, suspect a failed RAM chip on the motherboard or memory expansion card &mdash; the 201 code (if present) will identify the failing address.
== Fixed disk faults ==


=== Slot keying and AT bus problems ===
The POST tests a fixed disk configured in SETUP in the order 1782, 1780, 1781, 1790, 1791.<ref name="mzpost" />


Some XT-era 8-bit cards have a long tab that physically blocks the 16-bit AT extension. Cards designed for the AT are dual-tab: an 8-bit portion plus a shorter 16-bit extension. The slot itself does not damage incompatible cards mechanically, but the card will not seat fully and the system will refuse to boot.
* 1782: the controller failed its internal diagnostic or did not answer within a minute. The controller is missing, unseated, dirty at the edge connector, faulty, wrongly jumpered, or an XT-class controller.<ref name="mzpost" />
* 1780 (1781 for drive D): the controller passed but the recalibrate command failed. Check the 34-pin control cable (missing, unseated, faulty, on J1 instead of J5, or reversed on J5), the drive select jumper, the cable type, the drive's power, the drive, the controller, and that the controller is in a 16-bit slot.<ref name="mzpost" />
* 1790 (1791 for drive D): the recalibrate passed but a read failed. Check the 20-pin data cable (on J4 for drive C, not J3), the low-level format after fitting a new drive, and that the drive has at least as many cylinders and heads as the SETUP type says. Minuszerodegrees also lists +12 V missing at the controller and a controller without −12 V.<ref name="mzpost" />


== No-Beep / No-Video Diagnostics ==
The first hard drive IBM supplied in the 5170 was the 20 MB CMI 6426-S (also seen as CM-6426). Its very high failure rate led IBM to move to the Seagate ST-4026.<ref name="mzcmi">[https://www.minuszerodegrees.net/5170/cmi_6426_problems/IBM%205170%20-%206426%20problems.htm IBM 5170 – Failures of the CMI 6426 Hard Drive], minuszerodegrees.net.</ref> In SETUP, IBM's 20 MB drives are type 2 and its 30 MB drives type 20; type 20 needs diagnostics at level 2.0 or later and a BIOS that supports it.<ref name="fsim" /> Controllers with their own BIOS ROM, such as SCSI and XT-IDE cards, usually ignore the SETUP type.<ref name="mzbasics" />


If the AT shows no signs of life:
== Cards ==
# Verify the PSU fan is turning. If not, check the wall power, the PSU mains lead, and the front-panel power switch.
# With the PSU running, check +5 V at a free drive power connector. If absent, the PSU is faulty or one of its rails is shorted (likely a tantalum capacitor on the motherboard or an ISA card).
# Pull all ISA cards except the video card. Re-test.
# If still dead, swap in a known-good video card.
# If still dead, reseat the CPU, the math coprocessor (if fitted), and the BIOS ROMs.
# If still dead, suspect bank-0 RAM on the motherboard. The AT motherboard requires bank-0 RAM populated to POST.


== Hard Drive Diagnostics ==
* Some cards designed for the 5150 and 5160 do not work in a 5170, partly because the slot timing differs, and some cards designed for the 6 MHz 5170 do not work in the 8 MHz model.<ref name="mzprobs" />
* IBM notes that interrupt conflicts do not always make the diagnostics fail, and that a card isolated as the cause of a fault is not necessarily defective: it may be incompatible with the other cards in the machine.<ref name="fsim" />


* 1701: Run SETUP and confirm the drive type matches the drive's label. The 5170 SETUP table has 15 drive types in the 01/10/84 BIOS, 22 in the 06/10/85 BIOS, and over 40 in the 11/15/85 BIOS.
== Power supply ==
* 1780/1781: Drive 0 or Drive 1 failure. Verify cable seating (pin 1 marked with a red stripe), drive power, drive ID jumpers (drive 0 vs drive 1), and that the drive is correctly terminated.
* Drive does not spin up: head stiction. Carefully twist the drive (with power off) to free the heads &mdash; this is a stop-gap, replace the drive afterwards.


== Math Coprocessor (80287) ==
The supply's mains suppression capacitors tend to rupture with age. Old RIFA parts do so with smoke and a strong smell, and may leak fluid that should not be touched. A replacement must match the capacitance and class (X2 for X2, Y2 for Y2) with an AC voltage rating at least as high.<ref name="mzfail">[https://www.minuszerodegrees.net/failure/failure.htm Commonly Failing Electronic Components (in vintage computers)], minuszerodegrees.net, section "Line Suppression Capacitor".</ref> Opening the supply can expose hazardous voltages, and minuszerodegrees warns not to assume they disappear soon after it is unplugged.<ref name="mzpsu">[https://www.minuszerodegrees.net/5170/psu/5170_psu.htm IBM 5170 – Power Supply Unit (PSU)], minuszerodegrees.net.</ref> See also [[IBM PC AT Capacitor Replacement Guide]].
 
The 5170 socket accepts the '''Intel 80287''' running at a divided clock. POST does not test the 80287 unless software explicitly invokes it; a missing or failed 80287 is reported as a 7xx error only by application software. Use IBM AT Advanced Diagnostics to test the 80287.
 
== ⚠️ Power-supply RIFA capacitor and tantalum shorts ==
 
Two age-related failures are near-universal on this era of IBM hardware:
 
* '''RIFA mains-filter capacitors''' in the power supply are metallised-paper parts that crack and fail '''short''' with age, producing acrid smoke shortly after power-on. Replace them pre-emptively with modern X2-class parts.<ref name="ibm_rifa">[https://www.minuszerodegrees.net/failure.htm minuszerodegrees.net — IBM failure symptoms]; [https://www.classic-computers.org.nz/blog/2010-11-04-restoring-an-IBM-xt.htm Repairing and Restoring an IBM XT]; and [https://retrorepairsandrefurbs.com/2025/05/15/1983-ibm-pc-5160-xt-power-supply-rebuild-modifications/ Adam's Vintage Computer Restorations]. Source for the RIFA mains-filter capacitor failing short (smoke) and the tantalum capacitors failing short and preventing the PSU from firing.</ref>
* '''Tantalum capacitors''' on the planar (system board) and on ISA cards fail short with age. A shorted tantalum will '''prevent the power supply from starting''' (dead machine, PSU protection latched) &mdash; look for a cracked or discoloured tantalum and lift suspect ones to find the short.<ref name="ibm_rifa">[https://www.minuszerodegrees.net/failure.htm minuszerodegrees.net — IBM failure symptoms]; [https://www.classic-computers.org.nz/blog/2010-11-04-restoring-an-IBM-xt.htm Repairing and Restoring an IBM XT]; and [https://retrorepairsandrefurbs.com/2025/05/15/1983-ibm-pc-5160-xt-power-supply-rebuild-modifications/ Adam's Vintage Computer Restorations]. Source for the RIFA mains-filter capacitor failing short (smoke) and the tantalum capacitors failing short and preventing the PSU from firing.</ref>
 
IBM PC/XT switching supplies also need a '''minimum load''' to start, so a bare supply on the bench may not run without a dummy load.<ref name="ibm_rifa">[https://www.minuszerodegrees.net/failure.htm minuszerodegrees.net — IBM failure symptoms]; [https://www.classic-computers.org.nz/blog/2010-11-04-restoring-an-IBM-xt.htm Repairing and Restoring an IBM XT]; and [https://retrorepairsandrefurbs.com/2025/05/15/1983-ibm-pc-5160-xt-power-supply-rebuild-modifications/ Adam's Vintage Computer Restorations]. Source for the RIFA mains-filter capacitor failing short (smoke) and the tantalum capacitors failing short and preventing the PSU from firing.</ref>
 
== ⚠️ CMOS / RTC battery ==
 
This machine keeps its configuration in battery-backed CMOS, and the battery is a common failure. On AT-class boards the clock/CMOS is often a '''Dallas DS1287/DS12887''' module with the cell sealed inside; it lasts about ten years and then dies, giving '''161 / 163''' CMOS and clock errors at POST (and sometimes spurious floppy-drive errors). PS/2 planars use a rechargeable barrel or pack battery that '''leaks''' and corrodes the board. Replace a dead Dallas module (or rework it with an external coin cell), and on a leaking planar battery remove it and clean the corrosion before it eats the traces.<ref name="ibm_batt">[https://www.classic-computers.org.nz/blog/2009-10-10-renovating-a-dallas-battery-chip.htm Fixing a Flat Dallas DS1287 RTC], Classic Computers; and [https://www.ardent-tool.com/misc/Dallas_Rework.html Reworking Dallas RTC Modules], Ardent Tool. Source for the Dallas DS1287/DS12887 internal-battery death (161/163 CMOS errors) and the leaking planar battery.</ref>


== References ==
== References ==
Line 249: Line 192:
<references />
<references />


== Related Pages ==
== Related pages ==


* [[IBM PC AT (5170)]]
* [[IBM PC AT (5170)]]
* [[IBM PC AT Maintenance Guide]]
* [[IBM PC AT Maintenance Guide]]
* [[IBM PC AT Capacitor Replacement Guide]]
* [[IBM PC AT Capacitor Replacement Guide]]
* [[IBM PC XT Troubleshooting Guide]] &mdash; preceding model
* [[IBM PC XT Troubleshooting Guide]]
* [[IBM PC (5150) Troubleshooting Guide]] &mdash; preceding model
* [[IBM PC (5150) Troubleshooting Guide]]
 
== References ==
 
* [https://minuszerodegrees.net/5170/post_errors/5170_post_codes.htm IBM 5170 &mdash; POST checkpoint codes], minuszerodegrees.net.
* [https://minuszerodegrees.net/5170/bios/5170_bios_revisions.htm IBM 5170 &mdash; BIOS Revisions], minuszerodegrees.net (601 error notes for the 06/10/85 and 11/15/85 BIOS).
* [https://helppc.netcore2k.net/hardware/diagnostic-codes HelpPC &mdash; Diagnostic Codes]. Reference for the 1xx&ndash;18xx numeric error code meanings.
* IBM, ''Personal Computer AT &mdash; Hardware Maintenance and Service'' (IBM service manual; multiple revisions 1984&ndash;1987).
* IBM, ''Personal Computer AT &mdash; Technical Reference'' (1502243, March 1984; 6280070, September 1985).


{{Navbox-IBMComputers|state=collapsed}}
{{Navbox-IBMComputers|state=collapsed}}

Latest revision as of 20:53, 30 September 2026

IBM PC AT (5170)

Fault finding on the IBM PC AT (5170) follows IBM's own service sequence: check that the supply fan runs, listen for the beep at the end of the power-on self test (POST), read any error on the screen, and run the Advanced Diagnostics. IBM's 1986 service manual turns each POST indication into a procedure; minuszerodegrees.net documents the causes it has found for each. Supply limits, the battery check and the load resistor are covered in the IBM PC AT Maintenance Guide.

Before you start

[edit | edit source]
  • IBM's start procedure begins with the power off, all connectors and jumpers checked, the 115/230 V selector set for the local supply, and the configuration confirmed by running SETUP.[1] The AT has no SETUP in ROM; see IBM PC AT Maintenance Guide#SETUP.
  • Many POST errors are followed by a prompt to press F1 to continue.[1]
  • Unlike the 5150 and 5160, the 5170 POST writes checkpoint codes to I/O port 80h, which a POST card can display. The code is written just before each test, so a card that stops at a code points at the test that failed.[2]
  • All of this assumes IBM BIOS ROMs. Third-party BIOS replacements behave differently.[3]

IBM's POST error table

[edit | edit source]

IBM's start procedure sends each POST indication to a maintenance analysis procedure (MAP):[1]

IBM 5170 POST indications (HMS MAP 0000, p. 0000-13)
Indication IBM's action
No beep, with a blank display, a blinking cursor, an unreadable display, or even with the machine working MAP 0020: Power
No beep and a 1xx error MAP 0100: System board
1 long, 1 short beep Replace the system board
1 long, 2 short or 1 long, 3 short beeps Display adapter MAP (EGA: MAP 2400)
2 short beeps and a blank, unreadable or distorted display Display adapter MAP
2 short beeps and an error code 1xx: MAP 0100; xxxxxx xxxx 201: MAP 0200; 30x or xx30x: MAP 0300; 601: MAP 0600; 17xx: MAP 1700; 30xx and 31xx: PC Network MAPs
C8000 ROM error Replace the fixed disk adapter
ROM error Replace the system board
Continuous beep MAP 0020: Power
Repeating short beeps MAP 0020: Power

IBM's table of other symptoms sends an incorrect memory size and PARITY CHECK to the memory MAP; display problems to the display adapter MAPs; a flashing cursor only to the coprocessor check and then the diskette MAP; a BASIC screen, a boot from fixed disk or a diskette boot failure to the diskette MAP; and an inability to finish the diagnostic tests to the power MAP.[1]

Beep causes

[edit | edit source]
IBM 5170 beep indications: causes found by minuszerodegrees
Indication Causes
1 long, 1 short 06/10/85 and 11/15/85 BIOS only. System board speed outside certain bounds, typically after a faster crystal has been fitted.
1 long, 2 short A faulty video card, a poorly seated card, dirty edge contacts, or interference from another card. An IBM EGA card set up for an MDA or CGA monitor can also produce it when its own ROM finds a fault.
1 long, 3 short The ROM on an IBM EGA card reporting a fault with itself.
Memory count completes, then nothing On a type 3 board, an MDA card with SW1 in the colour position. Also seen with a faulty Fixed Disk and Diskette Drive Adapter.

Source: minuszerodegrees' 5170 POST errors page.[3]

On-screen errors

[edit | edit source]
IBM 5170 POST error messages
Code Message or IBM's description Notes (minuszerodegrees)
101 Interrupt failure
102 Timer failure
103 Timer interrupt failure
104 Protected mode failure
105 Last 8042 command not accepted
106 Converting logic test
107 Hot NMI test
108 Timer bus test
109 Low meg chip select test
161 System Options Not Set (dead battery) Replace the battery, then run SETUP; the error stays until SETUP is completed. Also a faulty system board.
162 System Options Not Set (CMOS checksum error) Corrupted SETUP, a flat battery on some boards, or a video card that does not match SETUP.
163 Time & Date Not Set (clock not updating) Invalid date or time, SETUP started and not finished, or a clock circuit fault.
164 Memory Size Error RAM added without running SETUP, wrong amounts entered, or faulty RAM.
201 Memory Error Shown as AAAAAA BBBB 201: AAAAAA is the hexadecimal address of the failing word and BBBB the failing bits. The address is sometimes one word beyond the fault. A BBBB with many F digits, such as FFFE, usually means RAM that is not there is being tested, for example SETUP set for 640 KB with 512 KB fitted.
202, 203 Memory Address Error Address bits 0 to 15 (202) or 16 to 23 (203).
301 Keyboard Error A preceding byte identifies a stuck key. Otherwise no keyboard, a PC/XT keyboard, a faulty keyboard or faulty keyboard circuitry on the system board. The 5170 has no keyboard fuse.
302 System Unit Keylock is Locked The keylock is locked, the keylock switch is faulty, or the keyboard controller misreads it. Unplug J20: if the error goes, the keylock is the cause.
303 Keyboard Or System Unit Error Remove the keyboard and power on. If 303 goes, the keyboard is faulty; if it stays, the system board.
304 Keyboard Or System Unit Error (keyboard clock line high) As 303.
401 CRT Error A CGA card with SW1 in the monochrome (rear) position.
501 CRT Error An MDA card with SW1 in the colour (front) position.
601 Diskette Error See #Diskette faults.
602 Diskette Boot Record Error The boot sector was read but its contents were not as expected.
1780, 1781 Disk 0 Failure, Disk 1 Failure See #Fixed disk faults.
1782 Disk Controller Failure
1790, 1791 Disk 0 Failure, Disk 1 Failure

The messages and IBM's descriptions are as minuszerodegrees quotes them from the IBM BIOS.[3]

Memory and parity errors

[edit | edit source]

The BIOS always follows PARITY CHECK with 1 (system board RAM) or 2 (RAM on an expansion card). Below it is the failing address, or five question marks if it cannot be found. A PARITY CHECK with no number is not from the IBM BIOS; the Parity_Boot virus is known to display it.[3] IBM's service manual decodes the parity and 10-character memory error codes to a bank and a module; see IBM PC AT (5170)#Memory errors.[4]

When a memory error cannot be placed, IBM's procedure is to remove all Memory Expansion Adapters, run SETUP, repeat the failing test, and then refit the adapters one at a time, running SETUP each time.[4]

A fault in the first 64 KB of bank 0 shows as "000000 xxxx 201" with an MDA or CGA card, but as a dead system board with EGA or VGA. A POST card then shows DD. Reseat the bank 0 chips first; the fault may be a poor contact. On a type 1 board, swapping the bank 0 and bank 1 chips turns the dead board into a 201 error if a bank 0 chip was at fault.[5]

No beep, fan not running

[edit | edit source]

IBM's start procedure:[1]

  1. Check that the power cord is in a working, properly earthed outlet.
  2. Switch off, unplug, and check the power cord for continuity. Replace it if it is open.
  3. Reconnect and power on. Check for 2.4 to 5.2 V between pins 1 and 5 (ground) of power supply connector P8. If Power Good is present, replace the supply; if not, go to the power procedure below.

Power procedure

[edit | edit source]

IBM's power MAP for the AT, in order:[6]

  1. Check that the 115/230 V selector matches the supply. Disconnect all cables and external devices except the display, and power on. If the symptom goes, reconnect the devices one at a time, powering off before each, until it returns.
  2. Remove the cover, reseat the power connectors, and check Power Good (2.4 to 5.2 V between P8 pins 1 and 5).
  3. If Power Good is present, check the system board and drive connector voltages; see IBM PC AT Maintenance Guide#Voltage checks. A machine with no fixed disk needs the load resistor on P10 or P11. If any voltage is wrong, replace the supply.
  4. Check the speaker for continuity with its cable off the system board.
  5. Unplug the power connector from each drive in turn. If the symptom goes, that drive is at fault.
  6. Remove the option adapters one at a time, leaving the Fixed Disk and Diskette Drive Adapter and the display adapter.
  7. Remove the math coprocessor, if fitted.
  8. Remove the Fixed Disk and Diskette Drive Adapter (expect a 601), then the display adapter (expect one long and two short beeps).
  9. Power off, wait 10 seconds, power on and check Power Good again. If it is missing, replace the supply, then the system board; if it is present, replace the system board, then the supply.

Other causes of a machine that will not start:

  • A diskette-only machine without its load resistor: the +12 V output is below its 2.5 A minimum load. The 5170 may or may not start with its hard drive removed.[5][7]
  • A coprocessor fitted the wrong way round. Some 5170 system boards were made with the coprocessor socket's notch to the rear; a coprocessor fitted to match it gives a blank display and no beep. Fit it with its notch to the front of the board.[4]
  • An XT-class supply: a type 1 board with the 01/10/84 BIOS may not start at all, and other boards may start intermittently or lose SETUP.[5]
  • Bank 0 RAM with an EGA or VGA card (see above).
  • A supply that fails after two or more hours and works again once cool. IBM's instruction is to replace it.[4]

Diskette faults

[edit | edit source]

A 601 has several causes: the POST cannot find the diskette part of the Fixed Disk and Diskette Drive Adapter; a hard disk-style twisted cable used in place of the diskette cable; drive A faulty, without +5 V or +12 V, wrongly jumpered, or a model that does not work in the 5170.[3]

With the 06/10/85 and 11/15/85 BIOS, replacing IBM's adapter with a third-party controller can produce a 601 at POST, because those revisions look for the IBM adapter. The boot continues after an F1 prompt, although diskette operation may be affected. The 01/10/84 BIOS does not do this. Fixes are a controller the POST takes for IBM's, such as the Western Digital WD1003A-WA2 (the WD1002-WA2 and WD1003-WA2 may also work), a patched IBM BIOS, or a third-party BIOS.[8]

  • Corrupted diskettes: a 1.2 MB drive must generate DISK CHANGED on pin 34. A 360 KB drive that puts READY on pin 34 also causes trouble; disconnect pin 34 when fitting a 360 KB drive.[5]
  • 360 KB diskettes that other machines cannot read: a 160/180 KB or 320/360 KB diskette written in the high-capacity drive can afterwards be read only in a high-capacity drive.[9] IBM offered the optional 360 KB drive for writing disks for 360 KB machines.[10]
  • No boot from drive B: the POST does not boot from B.[11]
  • 1.44 MB drives are fully supported only by the 11/15/85 BIOS.[8]

Fixed disk faults

[edit | edit source]

The POST tests a fixed disk configured in SETUP in the order 1782, 1780, 1781, 1790, 1791.[3]

  • 1782: the controller failed its internal diagnostic or did not answer within a minute. The controller is missing, unseated, dirty at the edge connector, faulty, wrongly jumpered, or an XT-class controller.[3]
  • 1780 (1781 for drive D): the controller passed but the recalibrate command failed. Check the 34-pin control cable (missing, unseated, faulty, on J1 instead of J5, or reversed on J5), the drive select jumper, the cable type, the drive's power, the drive, the controller, and that the controller is in a 16-bit slot.[3]
  • 1790 (1791 for drive D): the recalibrate passed but a read failed. Check the 20-pin data cable (on J4 for drive C, not J3), the low-level format after fitting a new drive, and that the drive has at least as many cylinders and heads as the SETUP type says. Minuszerodegrees also lists +12 V missing at the controller and a controller without −12 V.[3]

The first hard drive IBM supplied in the 5170 was the 20 MB CMI 6426-S (also seen as CM-6426). Its very high failure rate led IBM to move to the Seagate ST-4026.[12] In SETUP, IBM's 20 MB drives are type 2 and its 30 MB drives type 20; type 20 needs diagnostics at level 2.0 or later and a BIOS that supports it.[4] Controllers with their own BIOS ROM, such as SCSI and XT-IDE cards, usually ignore the SETUP type.[11]

Cards

[edit | edit source]
  • Some cards designed for the 5150 and 5160 do not work in a 5170, partly because the slot timing differs, and some cards designed for the 6 MHz 5170 do not work in the 8 MHz model.[5]
  • IBM notes that interrupt conflicts do not always make the diagnostics fail, and that a card isolated as the cause of a fault is not necessarily defective: it may be incompatible with the other cards in the machine.[4]

Power supply

[edit | edit source]

The supply's mains suppression capacitors tend to rupture with age. Old RIFA parts do so with smoke and a strong smell, and may leak fluid that should not be touched. A replacement must match the capacitance and class (X2 for X2, Y2 for Y2) with an AC voltage rating at least as high.[13] Opening the supply can expose hazardous voltages, and minuszerodegrees warns not to assume they disappear soon after it is unplugged.[14] See also IBM PC AT Capacitor Replacement Guide.

References

[edit | edit source]
  1. ↑ 1.0 1.1 1.2 1.3 1.4 IBM, Hardware Maintenance Service, 6280087, Revised Edition (March 1986), MAP 0000 "Start (AT)": checks before starting p. 0000-1; power cord and Power Good p. 0000-11; POST error table p. 0000-13; failure symptom table p. 0000-14. Hosted as File:IBM Hardware Maintenance Service 6280087 MAP 0000 Start (AT).pdf, described at IBM Hardware Maintenance Service (March 1986).
  2. ↑ IBM 5170 – POST – Codes, minuszerodegrees.net.
  3. ↑ 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 IBM 5170 – POST – Some errors (on-screen, and otherwise), minuszerodegrees.net.
  4. ↑ 4.0 4.1 4.2 4.3 4.4 4.5 IBM, IBM Personal Computer Family Service Information Manual, SA38-0037-00 (July 1989), Chapter 9: power supply failures p. 9-8, coprocessor socket p. 9-9, memory error isolation and parity checks pp. 9-10 and 9-11, fixed disk types p. 9-7. Hosted as File:IBM Personal Computer Family Service Information Manual SA38-0037-00.pdf, described at IBM Personal Computer Family Service Information Manual.
  5. ↑ 5.0 5.1 5.2 5.3 5.4 IBM 5170 – Known Problems/Issues, minuszerodegrees.net.
  6. ↑ IBM, Hardware Maintenance Service, 6280087, Revised Edition (March 1986), MAP 0020 "Power (AT)", pp. 0020-1 to 0020-7. Hosted as File:IBM Hardware Maintenance Service 6280087 MAP 0020 Power.pdf.
  7. ↑ IBM, Technical Reference, Personal Computer AT, First Edition (September 1985), Section 3 "Power Supply", pp. 3-4 and 3-5. Hosted as File:IBM PC AT 5170 Technical Reference Sep85.pdf.
  8. ↑ 8.0 8.1 IBM 5170 – BIOS Revisions, minuszerodegrees.net.
  9. ↑ IBM, IBM Personal Computer Family Service Information Manual, SA38-0037-00, Chapter 9, "High-Capacity Diskette Drive", p. 9-2.
  10. ↑ 360K Floppy in 1.2M Drive, minuszerodegrees.net.
  11. ↑ 11.0 11.1 IBM 5170 – Basics, for 5170 beginners, minuszerodegrees.net.
  12. ↑ IBM 5170 – Failures of the CMI 6426 Hard Drive, minuszerodegrees.net.
  13. ↑ Commonly Failing Electronic Components (in vintage computers), minuszerodegrees.net, section "Line Suppression Capacitor".
  14. ↑ IBM 5170 – Power Supply Unit (PSU), minuszerodegrees.net.
[edit | edit source]


Retrieved from "https://wiki.retrotechcollection.com/index.php?title=IBM_PC_AT_Troubleshooting_Guide&oldid=16504"