Parallel Port Trivia (DMA Arbitration)

How should I fill the memory? One 8 and two 4MB or 2 8MB SIMMs?
8MB, so you can use interleaved memory access. Use MEM1 and MEM2
  (Ed. Aron Eisenpress reported this really applies to the 8556/57 models, he tried the 9556/57, and saw no improvement. )

Bordshtles asks:
>does anyone know of a card to increase the memory on a 9556, I know its a 386 based machine with a 486 upgrade, Will the 386 memory boards (from kingston or ibm) work in these machines

Peter fields it:
Yes. Highly recommended is the Kingston KTM-609 II - since it supports XMS memory - what the IBM 1-2 and the 0-8 XMA don't.

NEC CDR-222 and IBM model 9556
Andrew Acton wrote:
   Installing CD-ROM Drive Support under DOS running on a PS/2 MicroChannel Machine with NEC SCSI CD-ROM Drives.

I recently installed CD-ROM support with the following configuration:
Machine : IBM PS/2 Model 9556 (MicroChannel/SCSI)
CD-ROM  : NEC MultiSpin 6X (External Drive)
O/S     : IBM PC-DOS 7.0

1) Obtain a copy of the following file from NEC tech support.  I obtained the following file from my local NEC BBS service in Sydney Australia:
   PS2.ZIP   101,052   22/09/93 | DOS Drivers for PS/2 SCSI machines

2) Unzip the file, run the install program (creates a directory on the dos boot drive called scsi), reboot and you now have CD-ROM access!

3) The installation program updates (in my case):
  CONFIG.SYS    -> Device=c:\scsi\neccdr.sys /d:NECCD  -> Lastdrive=E
  AUTOEXEC.BAT  -> C:\scsi\mscdex /d:NECCD /m:10

The PS2.ZIP package is great because it supplies the the device driver (neccdr.sys), the Microsoft CD rom extensions (mscdex) and installs the lot without and problems.

Workaround: enter machines configuration and set the Parallel port DMA to "Disabled". This causes the parallel port to work in "compatible bidirectional" mode - and the ZIP works fine. A little slower maybe, but works. I run a ZIP on all of my PS/2 with the PP-DMA disabled.

Ian Brown chimes in with:
   Good tip that one Peter. It is also relevant to Ditto drives, certain versions of Lap Link, and just about anything that is connected to the parallel port for bi-directional data transfer.

George Jefferson
   Funny thing, i use a parallel zip on my ps2 77 with no problem at all. Matter of fact it is far and away faster than any other machine, including ones that clam to be EPP.

Peter cuts in with:
   Depends on. IBM changed the specification a little -or the BIOS-support on that respectively- on the 95xx-machines as it seems. The 9556 / 9557 still suffer some problems with the DMA enabled, the later 9576 / 9577 (all planars) seem to be a lot better. The *85*90 / 95 are known for having problems with various parallel devices (even printers) when leaving the PP-DMA enabled. My 8595-AKD refuses to handle the Iomega ZIP-drive properly until I switched the parallel port DMA to "disabled".
   Basically it is a good idea to disable the DMA if one might experience problems with parallel CD-drives or scanners on all PS/2 (which use DMA printer port) to test whether the device is working at all or if there is a cabling problem at all. 

CHRISTIAN A. ROBERTS
  *SHRUG* My Iomega Ditto2Gb didn't get along with my 9556 slc3-75's parallel port, so I found a Boca Research MCA Parallel adapter...it worked for awhile, then went belly-up, too.

Peter finishes up with:
A: try the Ditto directly on the parallel-port with the DMA disabled B: there is a lot trouble reported with BOCA cards (Hi Allen !)  which I can sort only under "strange incidents". I had some Boca cards installed in various servers and *none* of them caused any trouble - unlike to similar AMS-cards, which scrambled the arrangement of LPT-ports (LPT1 becomes 2, LPT3 becomes 1 and LPT2 becomes inactive ... or such). This "sudden death" of Boca PP-cards is more than myterious. I probably could understand if that happened on very fast PS/2 (like "Lacuna"-77i, Server 95A, 8595 with Type 3 DX-50 or all Type 4 platforms) ... but not with a Mod. 80 or a 9556 ... !

Peter has a flashback and sez:
   You had just experienced the "standby power warming effect" :-)  The power supplies on various PS/2 (33, 35/40, all 56/57, all 76/77, 85, all 90/95) do not really "switch off" - a part of the PSU is always active and the frontside power button in fact switches only a "sense voltage" from the standby part of the PSU against GND ... which starts up the main power supply.

   Especially the PS/2 Mod. 9556 PSUs are known for a high failure rate. Once having opened the PSU you will find parts of the printboard having gotten dark-brown from the heat emmitted by components. Particularly know for "sudden death by aging" are the Italian Magnetek PSUs for the 56. They cannot be repaired ! I have tried that various times. They use a sort of "hybrid circuit" for the PSU-internal failure detection, which seemed to be fried after some components died by thermal overstress. I have replaced various diodes, resistors and the main switching hi-voltage VMOS transistor (all were defective) - and the PSU did not work but fried some larger resistor I'd replaced some minutes before.
   Recommendation: try to get some spare PSUs. The IBM models 35, 56 and 76 small desktops use the same PSU (and Mod. 40, 57 and 77 large desktops use the same larger PSU as well) And: use a common line breaker to fully separate the machine from the AC-power when not in use for a longer period of time.
   The machine itself is a nice little thing with reasonable performance. The case might be a little tight and does not offer much room for expansions, but you could replace the harddisk against a more modern and much faster 2.16GB DCAS-32160 from IBM (or anything up to 3.94GB), install a CD-ROM drive and expand the memory up to 16MB ... XGA-2 video comes standard and gives acceptable results even with Win95.

BTW: the machine might not run with the cover removed. There's a little blue security switch at the frontside, which shuts down the PSU when released (= when the cover is removed). You need to push the security switch inside/up to start the machine with the cover removed. Just for completeness.

Peter has another flashback:
  It is no PSU failure. It is simply a serial failure that occured during the assembly of the speaker/frontpanel cable. The "cold" speaker wire is tied to +5V (of the HD-LED AFAIK) instead of being tied to GND. There was (once) an "adapter cable kit" available that was plugged between the board and the cable plug and fixes this misbehaviour. There was an ECA on IBM about that .. must have it somewhere ... Ah ... here it is:

Symptom:
   Some 8556/57 and 9556/57 systems may exhibit low volume background speaker noise that alters during screen refresh or mouse movement.

Problem Isolation Aids:
   Problem is specific to 85xx and 95xx 56/57 models. Symptoms will probably be most noticable when switching from one application session to another.

Fix:
Two cable jumpers have been released to modify the speaker cable wiring. The jumper should be plugged between the speaker cable socket on the planar and the speaker cable connector. 

Details are as follows:
FRU P/N 8130978 (8 pin jumper for i386 8556/8557 systems)
FRU P/N 8130979 (12 pin jumper for i486 9556/9557 systems)

  To fix it by yourself: You could unsolder the speaker wires and measure the voltages on the wires against GND (power supply case) with the machine running. Use the one which has *not* +5V for the speaker and another wire directly attached to GND.
  I fixed my 9556 with that trick. If I find the time I try checking out the wiring on my machine and make a diagram of the proper wiring then ... but that will take time.

56 Model Summary Matrix
Model CPU   Memory   Disk  Floppy  Software
 DBA  SLC2  8-16MB   208   2.88MB  OS/2 or  DOS/Win
 DB6  SLC2  8-16MB   104   2.88MB  OS/2 or  DOS/Win
 DE9  SLC3  8-16MB   170   2.88MB  Choice
 DEB  SLC3  8-16MB   245   2.88MB  Choice
 DED  SLC3  8-16MB   340   2.88MB  Choice
 KBA  SLC2  8-16MB   208   2.88MB  OS/2 2.1
 KB6  SLC2  8-16MB   104   2.88MB  OS/2 2.1
 QBA  SLC2  8-16MB   208   2.88MB  DOS/Windows
 QB6  SLC2  8-16MB   104   2.88MB  DOS/Windows
 0BA  SLC2  8-16MB   208   2.88MB  OS/2 2.0
 0B6  SLC2  8-16MB   104   2.88MB  OS/2 2.0
 1EX  SLC3  4-16MB   None    None  N/A(Ethernet)
 2EX  SLC3  4-16MB   None    None  N/A(TokenRng)