Most recent revision of this page Feb 11 2012. (C) Herb JOhnson 2012
This is a description of my work on three Intel-designed 80386 terminal servers, Multibus based. My thanks to David Comley who provided these systems to me in Sept 2008; and to Northwest Technical for providing a Xenix on a hard drive. - Herb JOhnson
Here's the Intel 320 front and Here's the rear.
Here's a lable showing the System 320 board manifest (list). It lists:
SBC 386/22 2MB CPU, 16MHz (Multibus card) with two iSBC MM03 memory modules
SBC 214, Peripheral Controller
SBC 547, 8 channel serial comm card
HH 5.25" Floppy 320K
Archive Scorpion cartridge drive 5945C, 60MB Streaming tape drive
Archive Scorpion SAC (stand alone controller) SASI board (for tape drive)
Quantum 540 40MB Full-height MFM hard drive
As of mid-2011 I had THREE 320 chassis, THREE CPU boards, THREE controllers, and three comm cards. I got one running. In Oct 2011 I sold one 320 chassis and one CPU board. Below are the magic numbers for the current active and working system and its cards.
CPU board: 386/12 PBA 149129-007, AA 149644-016, PB 149693-001
ROM set 452112-001 452112-002
two RAM boards: PBA 454073-002, SCB MM02; PBA 148332-010 SCB MM02
The jumper areas for this board are here and also here.
controller board SBC 214: PBA 135837-008 SNP 356325
PB 148298-001 N 885414-C-1B
ROM set 455467-001, 455468-001; 8742 455466-001
SBC 547: 8-port serial card with local processor. Two boards in current active system.
Archive Corp SASI board: 20574-0141A, ROM 80182-018
Archive cartridge drive 5945C
There are two 128Kbyte ROMS with the 386/20 CPU board, Intel numbers 452112-001 and 452112-002. These are "high-byte" and "low-byte" ROMS, as the processor reads both ROMS at once as a 16-bit word. here's the hex -001 and the hex -002 , and the binary -001 and binary -002 .
I looked at the ROM contents in November 2008. When I dumped the ROMS on a reader, I was left with these two high and low files. I wrote a C program to merge the two; a copy is at this link. The program reads two Intel hex format files, and produces a merged hex file result.
Having the merged ROM image, a disassembler and a hex dump/inspection tool can be used to examine the binary. (There are many hex/binary programs around.) Inspection shows that the ROMS occupy the upper-most part of memory space, where the 80386 looks for reset and other vectors. HEre's a disasembly file in progress, showing portions of ROM memory. An ASCII dump of ROM also shows some useful commands and descriptions buried in the ROM. Here's some exerpts of text from the ROMs.
Mentioned in the ROMS are "V6.2", "iRMX86 Bootstrap loader", file dates for 19 Jan 1987, and various test commands. From three CPU cards, I found one of them would try to boot and then come up to a command prompt. In this mode, I could run "wh" for help, "wt" for tests, "wu" for utilities. These tests allowed me to verify the other cards. I could not "verify" the two Quantum MFM hard drives which came with the systems (see below); a MFM drive I had on hand seemed to respond.
The two Quantum 540 (?) drives with these systems look physically the same, and they turn out to have the same problem. Neither drive would enable, one would spin up but not the other. In handling one of the drives I heard a wrattle in the HDA (head-disk assembly, where the platters and head are encased). That suggested to me a broken head - the drive would be useless, so I decided to open it up.
This photo shows the drive's HDA open to the world. See any problems? The scale that is used to mark the location of each track is fallen off the head assembly. HEre's a closeup of where the scale mounts.. It's shocking to see that it is glued on, with no mechanical registration marks or holes! A closer look at the mounting area confirms this. And, here's a closeup of the scale, with the Quantum copyright. Not seen in the photo is the "smudge" where some residue of glue remains. I'll try to photograph that another day.
the Intel OEM systems like the System 320, ran Intel's iRMX operating system. iRMX system was sold with 8086, 80186, and 80286 systems. About the time of the 80386, Intel offered iRMX and Xenix, a Microsoft Unix licenced from AT&T and only sold to OEM'S (Original Equipment Manufacturers) such as Radio Shack and IBM and other system producers. Xenix was derived from Unix System 6 and had some BSD-related features. The following is from discussions I've had with Patrick Wong of Northwest Technical about the System 320, in Feb 2009. - Herb Johnson
As far as the documentation we have shows, the version of RMX-86 that we are familiar with, only supports builds for the 86, 186 and 286 series boards--not the 386 boards. Are you certain that the 386-based SYP320 systems you acquired actually successfully ran a standard iRMX86 version on the Q540 drives? So far we have not been able to get the 386/2x to boot any RMX-86 or RMX286 disks. RMX-III, that is supposed to run on the 386/2x platform, requires a minimum of 75MB disk space and at least 4MB of RAM memory. I think you have 2MB on you CPU board.
(I have two 2MB memory boards on the running 386/22 CPU board. My guess about running iRMX was based on a text string in the ROM. My 1987 Intel OEM Systems Handbook discusses System 310 with the 286, the 386/20 board, and barely mentions 386 products which "address up to 16MB of memory". iRMX286 apparently will run in as low as 700KB of memory. - Herb)
Intel's Xenix 3.5 for the SYP320 Systems (latest version) would probably be a better choice, depending on your need. It is FAR MORE hardware tolerant and supports multi-users. There are several good downloadable Xenix manuals at tenox.tc. Using a Xenix 3.5 Q540 disk, we have successfully booted 286/14, 386/21 & 386/24 boards....[Otherwise] you will need to find RMX-III (for your 386/2x boards), a larger HD and more RAM memory.
The same SBC386/2x firmware EPROMs run both the RMX and Xenix OS, so you shouldn't automatically assume that because you saw "iRMX86" in the firmware listing you created, that the Q540 HD's had RMX on them.
In fact, by far, most of the hundreds of SYP310 and dozens of SYP320 sytems that have come across our test benches, have had various releases of Intel's Xenix on them. Some had Unix and relatively few had RMX-I and RMX-II. In fact, many of the Xenix-based systems were identical to your configuration.
Intel's software and hardware evolved separately. Roughly ...
RMX-I (RMX86) supported 86, 186 & 286 - Multibus I
RMX-II (RMX286) supported 286 & 386 - Multibus I & II
RMX-III (RMX386) supported 386 & 486 - Multibus I & II
The comments in the RMX forum seem to support this. Considering the fact that your systems are 386-based, and that your HDs are only 40MB, chances are, your Q540 HDs had Xenix on them. - Northwest Technical
In 2012 I got some OS comments from Bill Beech: > Under OS's, Intel sold Xenix on the government contracts handled by SMS. We could NOT buy iRMX. It was NOT just their OEMs they sold to. I don't believe SMS was an OEM in this case - Intel was an active partner in this contract.Also, Xenix for the 310 with the 286/10 or 12 was full multiuser.
In late March, Northwest Technical provided a formatted Quantum hard drive with Xenix for a similar System 320. On April 1st, we installed the drive, and reinstalled a tape drive, onto the system. Here's an inside view of the chassis with tape drive, floppy drive and hard drive in place. Then we attached our Wyse 30 terminal and powered up.
With not much fuss, we booted into the diagnostic monitor (that's on ROM). It ran the usual diagnostics for the drives and boards. With the hard drive in place, it passed all of them. For the first time, the ROM monitor asked me about iMON-386 or iSDM, two ROM monitors for further tests and utilities. Saying no to both, it entered the "bootloader".
Then we loaded Xenix R3.5!. It found the 214 card and its UART. There was a choice to run single-user or multiuser mode. Simply waiting without entering "return", Xenix came up in multiuser mode. I entered the date, logged in as root, did a few simple commmands, looked around the file system. Then I entered "shutdown 0", to sync the hard drive and shut down the OS.
I've been advised to immediately backup the hard drive to tape; I was given a boot floppy with a backup program for that purpose. Meanwhile, I've obtained a few QIC 1/4-inch DC300 and DC600 cartridges, hoping they are in good shape and of the right capacity and tape "density" to match the tape drive.
On April 6th (2009), I set up the System 320, booted from the provided floppy with backup program. After several tapes, I was able to back up the drive and verify the tape. The successful tape was an unused DC6150 tape; two unused DC600A tapes failed with verify errors. Four used DC300 tapes failed, because they were so old the internal rubber "band" broke! The DC6150's seem to be available from media dealers for about $25-$30 each.
In July 2009 I was contacted by Bill Beech. We traded some Intel manuals relevant to the System 310 and 320. I now have a manual for the SBC 547 serial comm card. We corresponded at length over the next two years: Here's a Web page that links to our discussions and his work on 8080, ISIS, some Multibus systems he has, his prior System 310 and 320 work, and various software tools he's come up with. As of 2011, he may have some disassembled ROM code for the 310 or 320.
Over the summer of 2009, I acquired another Archive tape drive (with a mushy capstan) and an Archive Scorpion SASI board to match mine. Manuals for these are available on Web archives of Archive brand tape drives.
Al Koskow of bitsavers.org says he has a System 310 he needs to power up, sometime. "The DSD 5215 controller is program compatible with the Intel 214/15, FWIW."
My Web page for Multibus stuff in general is at this Web link.
My Web page for other's Multibus stuff in general is at this Web link.
this SLAC Web site for some Intel iRMX manuals; they mention one or two of the cards of this system. To search the SLAC system for old documents, , follow this link.
tenox.tc seems to be a private site of Antoni Sawicki. He has a variety of vintage computing interests.
I have not found many Web links about Xenix. In fact, Wikipedia points to THIS Web page. But this Web site by Dr. Nikolai Bezroukov has a curious but detailed document about Xenix history. You can check "Wikipedia" also. The short story is that Microsoft got "in bed" with Santa Cruz Operation (SCO) and licensed a System 7 Unix and they turned it into Xenix. Microsoft sold to manufacturers only - Intel, Tandy, etc. - and was serious about Xenix for some years, through their development of MS-DOS. By the late 1980's, the time of the 80286 and just before the 80386, Microsoft got OUT, sold their interests to SCO. Today, we know about SCO as a company who went after license fees to anyone who once owned Unix.
Since 2002, US Technologies is the current owner of Multibus intellectual property. Probably owns Intel's last stock of cards, plus their development stuff. They bought it from Radisys, who bought it and the iRMX rights from Intel back in Feb 1996. Inbus opened a Multibus-specific Web site in 2009. But Radisys still has some Multibus info on their Web site.
There is a Yahoo! discussion group on iRMX, at this Web link. It has posts about the Intel "System 320", and mentions it runs or ran iRMX Release IV (release 4); and there may be some current resources (2008) to help with that OS.
Some of the Web archives of manuals include iRMX or Multibus manuals include:
Here is the bitsavers.org iRMX archive. Check for Intel manuals as well.
Here is the Harte Technologies Intel archive.
an iRMX archive as part of Yahoo group irmsxstuff
INtel manuals accumulated by Bill Beech, 2012
Other iRMX resources on the Web may be found by specific Web searches.
Copyright © 2012 Herb Johnson