9585 "K"/"N" Planar

rf9585a.exe Reference v1.32, 23 Feb 94
rd9585a.exe Diagnostics v1.30, 23 Feb 94
Rev71upd.exe F/W SCSI Adapter Firmware Upgrade 7.1
193-295 SERVER 85 433 AND 466 - New Models

"F5JW15UK" BIOS Upgrade - If you know where it is, tell ME!

9585 Power (same form-factor as 95, but 288W)

Operator Panel Upgrade for 9585 K/N

K/N Planar
Blown Nichicon Capacitors!
85 K/N Ports
Problems Installing OS/2 on 9585 (2.x and Warp)
L2 Cache
Odd bracket in Basil's ONT (why is a fan power bracket there, see the plug holder?)
ADF sections


K/N Planar 64G2405 (DX), 64G2401 / 82G2481 (DX2), 61G2401 / 82G2480 (DX2)

BH1 Battery (CR2032)
CR1 SCSI LED
CR3, CR4 MBRD630CT (Autoterm.)
F4 Keyboard Fuse
F12, F13, F14, F15 PTC Resistors
F16, F17 SCSI Autoterm. (?)
J1, 3, 4, 7, 10, 12 32-bit MCA slots
J2 Indicator Panel Connector
J8 Floppy Connector (44-pin, 2 keys)
J9 32-bit MCA slot with BVE
J11 32-bit MCA slot with AVE
J16 Power Supply Connector
J27 Keyboard Port
J28 Mouse Port
J201 Parallel Port
J203 Serial Port
J452 External C68 FW SCSI Port
J460 Internal C68 FW SCSI Port
JMP1 Password Override (J5)
JMP2 Privileged-access (J6)
JMP4 Remote Power On (J469)
JMP3 LogicLock
JMP5 Remote Flash (J18)
JMP6 ServerGuard
JMP7 SCSI Boot 0- Allow (J463)
R462 Temperature Sensor?
U2 Dallas DS1585S RTC + NVRAM
U4 LM386 Audio Op Amp
U9 14.3181 MHz Osc (adapters)
U16 82077SL Floppy Controller
U22 24.0000 MHz Osc (FDC)
U49 10G7808 DMA Controller
U65-G 10G4672
U77 22.1184 MHz Osc (UART)
U78 ST93C46A 1kbit EEPROM (PAP etc.)
U84 Flash ROM BIOS 71G2540
U97 Flash ROM BIOS 71G2539
U134 61G2444 Int SCSI Controller
U135 61G2444 Ext SCSI Controller
U136, U138 Sony CXK581001M-70L
U139 SCSI BIOS 61G3930 (Even)
U140 SCSI BIOS 61G3929 (Odd)
U142 AMD N80C186-20
U146 40.0000 MHz Osc (planar I/O bus?)
U151 61G2323 MCA interface (?)
U179 L2 cache connector (see Lacuna)
U188 Socket 3 LIF (5 V only)
U198-PAJ 69G1212
U199-PG 69G1204
U205 66.0000 MHz Osc (CPU)
Y1 32.768 KHz Crystal (RTC)

U78 Pin 6 of the ST93C46A EEPROM chip is not grounded, setting it to the 64 word x 16 bits mode. This setting is reflected by the BIOS code.

JMP6 Remote Maintenance Service Connector can be used to add a hardware reset button.


Blown Nichicon Capacitors!

My planar F/W adapter refused to respond to a Controller Reset in Advanced Diags. It would not flash to the C9 level of BIOS. What really plugged things up was a Corvette controller in a slot could not format an HD, because the planar SCSI had thrown an error and the diags had flagged it.

   I stripped the system down and started to blow out the slots and around the chip leads with computer duster (cheap brand from Staples). It was then that I noticed the green on some chip leads and on top of the planar. From experience, it was necessary to pull the planar in order to wash the corrosion off effectively.

   First wash was with white vinegar and that really cleaned things up. The next wash is with baking soda so we neutralize any acidic compounds still on the board.

   Further investigation showed that ALL three Nichicon 10 uF, 25 V, 105 C electrolytic capacitors (N251) had blown.

The three caps were:
   C293 under U84
   C503 by upper right corner of internal F/W connector.
   C504 by U131

C293 on left lower corner

C503 and C504 by External SCSI port

   Of the three, C503 was the most significant because it is to the upper left of U135. It managed to corrode some pins at the upper left corner, and the PTC F17 made it a bit cramped to apply the vinegar.

Tom says:
   My board had the very same problem. C293, C503, and C504 leaked and corroded some nearby vias and parts. Thankfully it didn't case any permanent damage it seems. Recapped the board, and gave it a good cleanup...


9585-K/N Ports and Indicator Panel

9585-K/N Ports

COM supports up to 345K
LPT is bi-directional parallel
SCSI is a Fast/Wide

9585-K/N Indicator Panel


Reference v 1.31 on the -xXx?

   Another word for the BIOS: Once you'd put the inappropriate version of the Reference / Diags on an 85 and start for example OS/2 ... you could see really odd effects. Putting the 1.31 code from the -xNx on a -xXx result in a "speed cursor": a fast flashing icon on PM and you cannot click that fast to do anything with the machine anymore with using the mouse. You need to shutdown the beast over the keyboard. The xxx.BIO files differ - and these set the onboard timers to false base values. I had that in '92 or '93 when customers complained about inability to use their machines after "an update". They used the 1.31 "just because it was there and had a higher revision number". Without thinking and reading the DOCs as it seemed.


Parity / ECC in System Programs

   The 85 can emulate ECC using parity SIMMs. This is not worth it. Leave the Memory Detected setting in Change Configuration to "Parity". Go to "ECC-P" for more details.

   System memory can be expanded up to 256 MB using 32 MB Parity SIMMs. 2 MB to 32 MB Parity SIMMs are supported, INTERLEAVED ONLY. All 256 MB are addressable by Direct Memory Access (DMA). SIMMs must be installed in pairs of the same size. speed, and type. (Ed. I have used 32 MB EOS SIMMs successfully. If you do, use the "Parity" setting under System Programs. Remember, the ECC function is performed on-SIMM so using ECC-P is redundant at best).

Mixed Memory Sizes in K and N Models

   When a mixture of 4 MB, 8 MB and 16 MB (or larger) SIMMs are installed in the 9585-xKx and -xNx computers, install the larger pairs in the lower numbered connectors (A1/B1 lowest) and the smaller pairs must be installed in the higher-numbered connectors (A4/B4 highest).

Parallel Port

Bidirectional with DMA support. 100 KB/s max supported speed.


Fast/Wide SCSI

   The IBM SCSI-2 Fast/Wide busmaster controller is integrated on the system board. It uses the same chipset as the Fast/Wide SCSI. 40MB/s Data Streaming to MCA bus (32 bit), 20 MB/sec on SCSI bus (16 bit).

Server 85 - Sharing External SCSI DASD fails

   External SCSI DASD expansion shared between two system fails when one of the systems is powered down.

   On the 9585 0N* the trace on the solder (back) side of the planar running parallel below resistor R351 must be cut. On the SCSI-2 F/W controller the trace running next to C30 on the component side of the card must be cut. This trace runs from the fourth pin from the right on the bottom of the larger IC next to the external connector.

They did not EC the planar due to the expected low instances of this problem.

Ed. Unless you >actually< share SCSI devices externally, I wouldn't touch a thing.

50 to 68 pin adapter

   A 50 pin cardedge to a female 68 pin half pitch Centronics is FRU61G3594. The female 68 pin socket is an AMP part.

Can't Use >7 SCSI Devices

From John Poltorak:
   How come your machines can always use 15 devices and mine max out at seven?

From Peter:
   Q1: Which internal cable do you have? The odd-colored woven double wire cable with 50 pin connectors ? Or the "later" internal Wide-SCSI cable with 68-pin plugs at the device ends?

   Q2: (results from Q1:) which port on the internal adapter are you using ? The 68-pin Molex (black plastic) directly on the board or do you have the Molex-to-50 card-edge converter in between ? If so - the adapter is switched into 8-bit narrow mode and will take only 7 devices. You need to use the 68-pin Molex without the converter to have advantage of the 16-bit Wide mode. If you have the converter attached to the Molex the external port is also run in 8-bit mode as far as I know - and limits the adapter to... seven devices.

   The planar F/W SCSI on the Server 85 is *similar* to the "3 connector" F/W SCSI adapter /A (8EFC) - but it differs a lot. The F/W card has a separate 50-pin card-edge connector for the internal devices - and using this connector with only "narrow" devices does not afflict the ability to use "Wide" devices on the external port - given that you use a "wide" cable as well. You cannot use an internal narrow cable and external narrow cable as well and hope that you can address 15 devices with that. This will not work. As soon as you use an 8-bit cable you are cut to 7 devices on the planar F/W.


Enhanced security The Server 85 433 and 466 with LogicLock and new tamper-evident, locking covers, Privileged Access Password, selectable drive startup and support for optional features (such as, the new Cable Cover 4 and the Enhanced 2.88MB 3.5-inch Diskette Drive with electronic eject) provide system security exceeding C2 enabling requirements.

Remote power-on feature that allows the server to be powered-on remotely via an external modem, or from an internal timer. Remote power-off can be achieved through software control and is beneficial for running reports or printouts remotely, during off-shift periods.

Unattended start mode The Server 85 433 and 466 models can automatically restart the server after a power failure and resume normal operation, without operator intervention.

Vital Product Data (VPD) enabled for unique ID (model/submodel); Type/model/serial number; Planar serial number; Manufacturing ID; Planar FRU number; Planar part number. A system administrator can view from the LAN the PS/2 Server 85 models that reside on the network, including configuration. This function can also serve as a security measure by confirming if unauthorized servers are connected to the network.


Problems Installing OS/2 on 9585 Computer (HERE)

From Joltin' Joe Kovacs:
   One of the following error messages is displayed at Diskette 1 while installing OS/2 on an IBM PS/2 9585 computer:

  • OS/2 Warp: OS/2 is unable to operate your HD or diskette drive.
  • OS/2 2.x: The system cannot find the file 'A:\COUNTRY.SYS'.

The diskette-drive light stays on, and pressing Ctrl+Alt+Del has no effect. The system must be turned off and then turned back on.

Note: This problem is particularly troublesome because at the first sign of trouble during OS/2 installation, users remark-out (REM) all unneeded drivers in the CONFIG.SYS file; for example, BASEDEV=IBM1*.ADD. This should be a line of action because the PS/2 9585 is a Micro Channel system. However, if the statement BASEDEV=IBM1FLPY.ADD is remarked-out, OS/2 installation will fail with one of the above error messages.
   Make sure the REM in front of BASEDEV=IBM1FLPY.ADD is removed in CONFIG.SYS file. If the line was deleted, restore it. If there are no other problems, the installation will work.


L2 Cache

0Kx - 128 KB L2 WB cache. 0Nx - 256 KB L2 WB cache.
Both use the same cache modules as the 76/77 Lacuna planars.

Uh, Maybe Not
   I tried a variety of L2 modules in my N. They all failed (with POD and interposer) until I tried an IDT 256K WB module, 7MP6150. To my surprise, the Cypress OEM P/N 06H3306 and P/N 06H3307 failed, either jumpered for WB or WT. The failures were a continuous looping reboot, or a black screen (with interposer).


0Kx and 0Nx models are also:

  1. Intel Pentium Overdrive upgradable (socket is 19x19 LIF)
       NOTE: The CPU socket is right behind Bay #3. The original DX2-66 in my 85 had about 3/8ths of an inch of the heatsink fins milled off on the left side of the 486. If you use a POD (like me) be cautious of trying to ram a 5.25" drive or CD into that bay... "It won't fit! @%#&!!! Bam! Bam! Crunch!"
       Without the interposer, the POD does not tolerate any of the L2 cache modules I have, either IDT or Cypress (IBM). I had to take out the cache. The interposer looks like THIS. Details how to make your own are there too.
  2. SurePath BIOS, upgradeable via Flash EPROM. (IBM never released an upgraded BIOS. My ONG has BIOS 00.)
  3. SCSI-2 F/W adapter (Oh yeah baby! Oh, behave!) 
  4. Capable of 256 MB w FPM -OR- EOS [Ed. I have used both 256 MB FPM and EOS]

System Board 486DX-33 w/SCSI 61G2405
L2 Cache 128KB (for 61G2405) 61G4098
System Board 486DX-33/66 w/SCSI 61G2401
L2 Cache 256KB (for 61G2401) 61G4098


ADF Sections AdapterId FEDF

Num Lock
   How the Num Lock key is set when the operating system is started. Please note that your operating system environment might change the setting of the Num Lock key.
      <"Off">, On

Display F1 prompt to access System Programs
   During startup, your system normally displays a prompt that tells you to press F1 for access to the system programs. If you wish to suppress this prompt, change this to <No>"
      <"Yes">, No

Hands-off Configuration [this is commented out]
   Normally, when you add or remove adapters, devices, or memory, you provide input to reconfigure the system. If you change this setting to <Enable>, the system will attempt a hands-off configuration when hardware is added or removed. No user input will be required unless the default values cannot be used.
      < "Disable" >, Enable

Serial Port
   The serial port can be assigned as Serial 1 through 16, or disabled. Standard COM port interrupt levels are IRQ 4 for serial 1 and IRQ 3 for any other serial level.
      <"SERIAL 1, IRQ 4">, many choices, Disabled

Parallel Port
   The parallel port connector can be set as Parallel 1 through 4 or the port can be disabled.
      <"PARALLEL 1">, 2, 3, 4, Disabled

Parallel Port DMA Arbitration Level
   If the level selected is shared then other devices can be set at the same level. If the level selected is dedicated then only this device can be set to that level. Select <Disabled> to use the port in compatibility mode.
      <"Shared level 7">, 6, 5, 4, 3, 1, 0, Dedicated 7, 6, 5, 4, 3, 1, 0, Disabled

SCSI Address (ID)
   ID of the built-in SCSI controller. Under normal circumstances, select <7>.
      <"7">, 6, 5, 4, 3, 2, 1, 0

SCSI I/O Address
   I/O address of planar SCSI controller. Normally, use <3540h-3547h>.
      <"3540h-3547h">, 3548-354F, 3550-3557, 3558-355F, 3560-3567, 3568-356F, 3570-3577, 3578-357F, Disabled

SCSI DMA Arbitration Level
   Arbitration level planar SCSI controller uses to transfer data. Normally, select <Level C>
      <"Level C">, D, E, 8, 9, A, B, 1, 3, 5, 6, 7

Move Mode Support
   Enable or Disable Micro Channel Subsystem Control Block Move Mode for the built-in SCSI controller. Under normal circumstances, select <Enabled>
      <"Enabled">, Disabled

Wait State Support
   Enable or Disable Bus Master wait states for SCSI controller. Normally, select <Enabled>
      <"Enabled">, Disabled

Selected Feedback Return Exception
   If the SCSI controller uses the MCA Selected Feedback Return Exception feature.
      <"Ignored">, Enabled

100ns Streaming Data Transfer Support
   SCSI controller supports 100ns Streaming Data Micro Channel protocol, which provides better performance on the Micro Channel.
      < "Enabled">, Disabled

Target Mode
   Target mode should be disabled only if this system is sharing SCSI devices with another system and there are more than 15 devices to be shared. Only 15 devices can be configured on the built-in SCSI controller. When target mode is enabled, the built-in SCSI controller appears as a processor device on the other system and unless you have specialized software that can communicate between the two systems through these processor devices (peer-to-peer communication), there is no advantage in having target mode enabled. When target mode is disabled, the built-in SCSI controller does not appear as a device to the other system, and one more device can be shared by the two systems. If your system is not sharing any SCSI devices with another system on the built-in SCSI controller, it does not matter whether you enable or disable target mode. The normal default for this option is enabled.
      <"Enabled">, Disabled

SCSI Disconnect
   Disconnect is a SCSI-bus procedure in which a device logically stops communicating with the built-in SCSI controller during certain operations and then reestablishes communication with the built-in SCSI controller when the operation is complete. Disconnect should not be confused with the 'Presence Error Reporting' option for a device in 'Set and view SCSI device configuration.' If you are using an operating system that is single-threaded and issues commands to only one device at a time (such as DOS), disabling SCSI disconnect might result in a slight performance improvement. If your operating system is multi-threaded (such as OS/2), disabling SCSI disconnect will degrade the performance of the SCSI subsystem. The normal default for this option is enabled.
      <"Enabled">, Disabled

Fast SCSI - External
   Enabling Fast SCSI external improves performance if you have one or more Fast SCSI devices attached externally in one of the following configurations:
1) One external SCSI device enclosure model 3511.
2) Up to three external SCSI device enclosures model 3510.
3) Any external configuration in which the SCSI cable length does not exceed 3 meters. The normal default for this option is disabled.
      <"Disabled">, Enabled

Wide SCSI messages - External
   This should be 'Enabled' unless you have a Wide SCSI device attached to the built-in SCSI controller through a narrow (8 bits wide) external interface cable. Refer to the documentation that came with the device and cable you are using to determine whether it is necessary to disable Wide SCSI messages. The normal default for this option is enabled.
      <"Enabled>, Disabled
Ed. This changes if the adapter terminates the high byte (disable). If you have a 50 pin cable, disable wide messages.

Wide SCSI messages - Internal
   This should be 'Enabled' unless you have a Wide SCSI device attached to the built-in SCSI controller through a narrow (8 bits wide) internal interface cable. Refer to the documentation that came with the device and cable you are using to determine whether it is necessary to disable Wide SCSI messages. The normal default for this option is enabled.
      <"Enabled">, Disabled
Ed. this changes if the adapter terminates the high byte (disable). If you are using the 50 pin edgecard adapter, FRU61G3594, disable wide messages.

ADPItem 1 Processor
   Type of processor currently installed on the system board.

ADPItem 2 Bypass System Programs on Error
   When the power-on self-test (POST) detects an error, POST normally starts the system programs. If you want POST to start the operating system instead, choose <Enable>. Warning: Setting this to <Enable> could result in a partially configured system when an adapter or device is added. A partially configured system may cause some operating systems and applications to be inoperable.

ADPItem 3 Memory-Checking Method
   Method that the computer uses to check the system memory, either parity or ECC (error-correcting code). The ECC-checking method allows the computer to continue to operate in the presence of single-bit memory failures.

Note: If a bad-battery error (161) or a configuration-integrity error (173) occurs, the configuration will be reset to use the parity-checking method.


9595 Main Page