DEC VAxstation 3540
2026 Restoration

4/18/26

Dan-II Found a 3.5" 4.3GB ST15150N disk in the Museum's collction. He used simh on bismuth (FreeBSD PC that has a SCSI adapter) to install NetBSD/vax 10.1 to an RAUSER image of the same size as the disk. It was a full install, and he copied the GENERIC.MP kernel to /netbsd. I wrote this image to the SCSI disk after verifying it boots under simh using the MicroVAX 3900  personality. First, he connected it to the VAX as SCSI ID 1, keeping the OpenVMS disk at 0. The console command BOOT DKA100 would just hang after VMB printed the "-DKA0" line.

He changed the disk to ID 0 and disconnected the OpenVMS disk. The console command BOOT DKA0 also hangs.

He connected the disk back to bismuth and ATTACHed /dev/da0 directly to simh and moved the UP GENERIC kernel back into place and reconnected it to the VAX. Same hang as before.

He put a 10BASE-T MAU on the AUI port of the VAX, installed mopd on bismuth and placed the NetBSD stage 1 loader (/boot) in the /tftpboot/mopd directory with the filename <VAX MAC address>.DAT

The console command  BOOT ESA0 hangs even earlier than booting from SCSI attempts did.

He found from mopd log output that the /boot ELF image needs to be converted to MOP format mopa.out. But the version of mopa.out in the mopd distribution does not support ELF32/ELF64 (it can also be used for Alpha binaries).

He discovered that the version of mopd that OpenBSD used to ship before it was removed from the tree was modified to work on ELF32 binaries. The ELF #ifdefs will not compile on FreeBSD. He fired up an old NetBSD 9 VM and fiddled with the source to get it to build. Eventually got a binary that worked on the /boot binary.

He placed the binary in the directory with the proper filename and noticed mopd was getting further, but then the VAX dumped back to firmware console halting on an exception.

He connected a NeXT SCSI CD-ROM unit to the VAX. The device would not enumerate in firmware SHOW SCSI.

4/22/26

We replaced the batteries behind the front panel with new ones from eBay. Hopefully the boot device configuration will stay now.

We connected a DEC RRD42-AA SCSI CDROM to the system, installed an Ultrix V4.5 CDROM, and tried to boot from the CDROM. After a pause of about a minute and some blinking lights on the CDROM, the VAX displayed an error message on the console. It is hard to say if the CD is bad or the CDROM drive is broken. More testing is needed.

CPU08 >>>show scsi

 ADDR   VMB     DEVTYP  DEVNAM   NUMBYTS  REV   CHAR

 5.0.0  DKA0    DISK    1588-15  667  MB  AS0C

 5.1.0  DKA100  RODISK  CD-ROM   .......  4.3d  RM

 5.7            HOST    SII-A

CPU08 >>>boot dka100

VMB V1.0

%VMB-F-ERR, PC = 000008EF

%VMB-I-STS, R0 = 000001A4

CPU08 ?06 HLT INST

CPU08     PC = 000010DE

CPU08 Failure.

CPU08 >>>

We tried with a Ultrix 4.3 CDROM and got a similar failure.

CPU08 >>>boot dka100

VMB V1.0

%VMB-F-ERR, PC = 000008EF

%VMB-I-STS, R0 = 000001A4

CPU08 ?06 HLT INST

CPU08     PC = 000010DE

CPU08 Failure.

CPU08 >>>

We tried a NetBSD 10.1 CDROM and got the same result.

CPU08 >>>boot dka100

VMB V1.0

%VMB-F-ERR, PC = 000008EF

%VMB-I-STS, R0 = 000001A4

CPU08 ?06 HLT INST

CPU08     PC = 000010DE

CPU08 Failure.

Using the command boot /R5:0 dka100 didn't change anything.

The nice people at the LSSM said that this would decode the CDROM status:

3_SERVER::$ exit %x1a4
%SYSTEM-F-MEDOFL, medium is offline

The Simh instructions for installing Ultrix say to copy the install image to a disk drive, and boot from the disk. Even though the installation media is referred to as a CDROM, it is not in ISO-9660 format and is really a UFS filesystem. So maybe copying the installation to a disk makes sense. 

4/25/26

Dan-II copied the Ultrix CDROM install image onto SCSI disk, installed the disk into the 3540, When we booted from the SCSI disk it printed the Ultrix bootloader information, and the address in the Ultrix kernel that it jumped to, and then hung. Simulating the boot process with Simh showed exactly the same addresses and the Ultrix kernel banner, so we know that the bootloader is working correctly. Mike installed Ultrix on a Simh VAX 3900. That is the closest model to the 3540 because Simh doesn't emulate the 3540. We copied the installed disk image onto a SCSI disk and installed in the 3540. The behavior was the same as with the installation CDROM image.

During the week Mike will try booting a VAX 3100 and VAX 3500 from the installed SCSI disk. Maybe it will boot? If not, Mike will try booting from the Ultrix CDROM and running an installation.

DEC also sold a lobotomized VAX 3100 desktop system with an expansion chassis called an Infoserver. It booted a minimal operating system that allowed it to be a MOP boot server and file server. It could also share it's tape and CDROM drives over the network. You could boot a VAX with an empty disk over the network from the Infoserver using the MOP protocol, and then run an operating system installation. We could try MOP booting 3540 from the Ultrix installation CDROM from the Infoserver and then running the installation.

4/29/26

We disassembled the TK70 tape drive and manually unloaded the TK50 tape cartridge. The tape in the TK50 was bound up and the leader was disconnected from the tape. We used packing tape to tape the leader to the magnetic tape and reassembled the TK50 tape cartridge. This is really tricky to do because it is full of springs and spring loaded levers. We found that the tape drive guide rollers were not moving easily. We sprayed some WD-40 into the bearings and used an electric drill to spin the rollers. After a few minutes the rollers moved much more freely. The tape drive will now load and unload a tape cartridge.

5/2/26

A kind person donated caddys for our RRD40 CDROM drive. This drive was made before the standard caddy was invented so they are a little unusual looking.

We tried booting Ultrix-32 from the RRD40 and just get errors

CPU0B >>>boot dka200

VMB V1.0

%VMB-F-ERR, PC = 000008EF

%VMB-I-STS, R0 = 000001A4

CPU0B ?06 HLT INST

CPU0B     PC = 000010DE

CPU0B Failure.

CPU0B >>>

This is the same SYSTEM-F-MEDOFL, medium is offline error that we saw with the other CDROM drives.

After some fiddling we got past the medium is offline error and now the write protect light goes on when it is writing the label on the tape,.

At this point we don't know if VMB  (Virtual Memory Boot) is actually loading the NetBSD bootloader, or the NetBSD kernel. Dan-II looked through the NetBSD bootloader code and found that support for the VAXstation 3540 was commented out. Maybe this code was untested or just didn't work? We convinced Dan-II to spend a little time on this project. He modified the bootloader and kernel to add support for the VAXstation 3540. He copied the updated bootloader and kernel onto the SCSI disk and we tried to boot it again. The results were the same.

It is possible that there is a problem with the serial console code and we are not seeing any of the boot messages.

5/13/26

We tried booting OpenBSD 5.8, the last version that supported the 3520 & 3540 from three different CDROMs. No luck on any of them even after cleaning the lenses.

Saturday we will get MOP booting working and see if we can determine why NetBSD doesn't display any messages when booting.

5/16/26

Dan-II spent more time on getting the VAX to boot NetBSD. He replaced the standard stage-1 bootloader with his modified version that contains HALT instructions for debugging. We tried to boot NetBSD from this disk drive and the behavior was the same as before. Dan-II also booted the VAXstation 4000 VLC from the same disk drive. It booted NetBSD without problems, and also didn't halt during boot.

We spent quite a while speculating about how VMB loads the stage-1 bootloader from the first block on the disk, and then the stage-1 bootloader loads the stage-2 bootloader from the filesystem. From our experiments there is no evidence that the stage-1 bootloader is being executed.

We found source code for VMB and the stage-1 bootloader. It is very difficult to understand how VMB loads the stage-1 bootloader, and if the stage-1 bootloader is actually executed or just has pointers to the stage-2 bootloader.

Mike has a SCSI bus analyzer, but it is broken. We talked about using a SCSI disk emulator so we can see if VMB actually reads the stage-1 bootloader from the disk, and if it ever reads the stage-2 bootloader from disk.

We will go back to MOP booting the VAX because it is very easy to see what files the VAX is requesting. With the Museum's current network configuration the FIOS router is the DHCP server. It will handle IP address requests, but it will not supply information to the VAX about where it can TFTP the stage-2 bootloader and the rest of the NetBSD files. The Museum's Linux server can handle all of these DHCP and TFTP requests, but needs to be on an isolated subnet. The VAX and Linux server are connected to a Cisco switch. Dan-II will reconfigure the Cisco switch next Saturday so we can go back to MOP booting.

5/20/26

Mike brought a Sun badged Toshiba XM-4101B CDROM and got the VAXstation 4000 VLC to boot from the NetBSD 10.1 CDROM and install NetBSD on a RZ29 disk.

Saturday we can use the same CDROM to install Ultrix on the VAXstation 3540.

5/23/26

We connected Mike's Sun badged Toshiba XM-4101B CDROM jumpered for SCSI ID #6 to the 3540, installed the Ultrix V4.5 CDROM and connected Mike's RZ25 disk drive. We booted from theCDROM and were pleased to see:

CPU0B >>>boot dka600

VMB V1.0

-DKA600

..

...

Ultrixboot - V4.5  Sun Sep 17 13:03:13 EDT 1995

Loading (a)vmunix ...

Sizes:

text = 945424

data = 1290240

bss  = 757940

Starting at 0x5219

Unfortunately after blinking the LEDs on the disk and CDROM it sat at that point forever. We removed the base graphics board in case it was sending the installation messages to the graphics display. That didn't help.

We tried to boot NetBSD 10.1 from CDROM and it didn't even load the kernel.

CPU0B >>>boot dka600

VMB V1.0

-DKA600

..

...

We tried OpenVMS V7.3 from CDROM and got to to the 

CPU0B >>>boot dka600

VMB V1.0

-DKA600

..

...

%SYSBOOT-I-SYSBOOT Mapping the SYSDUMP.DMP on the System Disk

%SYSBOOT-W-SYSBOOT Can not map SYSDUMP.DMP on the System Disk

%SYSBOOT-W-SYSBOOT Can not map PAGEFILE.SYS on the System Disk

   OpenVMS (TM) VAX Version X7G7 Major version id = 1 Minor version id = 0

%WBM-I-WBMINFO Write Bitmap has successfully completed initialization.

PLEASE ENTER DATE AND TIME (DD-MMM-YYYY  HH:MM)  23-MAY-2026 10:46

Configuring devices . . .

Now configuring HSC, RF, and MSCP-served devices . . .

Please check the names of the devices which have been configured,

to make sure that ALL remote devices which you intend to use have

been configured.

If any device does not show up, please take action now to make it

available.

Available device  DKA400:                          device type RZ25

Available device  DKA600:                          device type TOSHIBA XM-4101TA

Available device  MUA0:                            device type TK70

Enter "YES" when all needed devices are available: yes

%BACKUP-I-IDENT, Stand-alone BACKUP T7.2; the date is 23-MAY-2026 10:49:22.58

$

So it looks like VMS installation will work OK.

We made an OpenBSD V5.8 CDROM and booted from that. It ran the kernel but paniced after probing the SCSI bus.

CPU0B >>>boot dka600

VMB V1.0

-DKA600

..

...

>> OpenBSD/vax boot [1.18] <<

>> Press enter to autoboot now, or any other key to abort: 3

> boot bsd

changing bootrpb.unit from 600 to 6

cannot open /etc/random.seed: No such file or directory

2503596+374060=0x2bea5c

Copyright (c) 1982, 1986, 1989, 1991, 1993

        The Regents of the University of California.  All rights reserved.

Copyright (c) 1995-2015 OpenBSD. All rights reserved.  http://www.OpenBSD.org

OpenBSD 5.8 (RAMDISK) #88: Tue Aug 11 08:15:35 MDT 2015

    deraadt@vax.openbsd.org:/usr/src/sys/arch/vax/compile/RAMDISK

VAXstation 3540 [0A000004 03010002]

cpu0: KA60

real mem = 50331648 (48MB)

avail mem = 45527040 (43MB)

mainbus0 at root

mbus0 at mainbus0

uba0 at mbus0 mid 0: Q22

mtc0 at uba0 csr 174500 vec 508 ipl 17

mscpbus0 at mtc0: version 3 model 14

mscpbus0: DMA burst size set to 4

mt0 at mscpbus0 drive 0: TK70

ECC memory at mbus0 mid 1 not configured

cpu at mbus0 mid 2 not configured

cpu at mbus0 mid 3 not configured

ECC memory at mbus0 mid 4 not configured

fwio0 at mbus0 mid 5

dz0 at fwio0 vec 424: console, 4 lines

lkkbd0 at dz0 line 0

wskbd0 at lkkbd0

le0 at fwio0 vec 420: address 08:00:2b:0f:7d:06

le0: 64 receive buffers, 16 transmit buffers

sii0 at fwio0 vec 416

scsibus0 at sii0: 8 targets, initiator 7

sd0 at scsibus0 targ 4 lun 0: <DEC, RZ25 (C) DEC, 0700> SCSI2 0/direct fixed

sd0: 406MB, 512 bytes/sector, 832527 sectors

cd0 at scsibus0 targ 6 lun 0: <TOSHIBA, XM-4101TASUNSLCD, 1084> SCSI2 5/cdrom removable

panic: Segv in kernel mode: pc 80020e52 addr 6

lkkbd0: no keyboard

syncing disks... done

resetting system...

5/27/26

We determined that the RRD40 CDROM drive that we tried to boot from earlier was broken. We are now using a Sun badged Toshiba XM-4101B CDROM. We don't have any documentation for Ultrix 4.4 or 4.5, but the 4.3 docs show that the VAX 3520/3540 is supported. We booted from an Ultrix 4.3 CDROM. VMB loaded Ultrixboot, Ultrixboot loaded the Ultrix kernel, and when Ultrixboot jumped to the Ultrix kernel nothing happened. This was the same behavior that we saw with Ultrix 4.5.

We speculated that Ultrix may be using the graphics diskplay and keyboard, and not using the serial port. We connected a ViewSonic VGA monitor to the RGB graphics connector, We were very surprised to see the Ultrix kernel messages and the installation menu displayed on the VGA monitor. We tried to enter characters on the terminal, but nothing showed up on the VGA monitor. It looks like we will have to make a DB15 female to two RJ11 cable to connect a keyboard and mouse. 

We tried NetBSD 10.1 with the VGA monitor attached, but never saw any messages after VMB.

5/30/26

Mike brought an AUI cable and an RJ11 socket. We can cut up the cable and turn it into a keyboard cable.

Hardware Notes

CPU0B >>>t 50

KA60 V1.1

MID  MODTYPE     ID    SLOT ERR

 0   01010001  L2008    9

 1   01020310  L2007CA  8

 2   01010108  L2001    7

 3   01010108  L2001    6

 4   01020110  L2007BA  5

 5   01010004  L2003    4  ?

 7   01010002  L2004    3  ??

CPU0B >>>t 5 4 100 0 1

 ADDR   VMB     DEVTYP  DEVNAM   NUMBYTS  REV   CHAR

 -----  ------  ------  -------  -------  ----  ----

 5.0.0  DKA0    DISK    ST43400  2.91 GB  0103

        VID: SEAGATE   PID: ST43400N

        ANSI = 2, ECMA = 0, ISO = 0, RSPDATFMT = 2

CPU0B >>>t 5 4 100 400 1

 ADDR   VMB     DEVTYP  DEVNAM   NUMBYTS  REV   CHAR

 -----  ------  ------  -------  -------  ----  ----

 5.4.0  DKA400  DISK    1588-15  667  MB  AS0C

        VID: MICROP    PID: 1588-15MB1036511

        ANSI = 1, ECMA = 0, ISO = 0, RSPDATFMT = 1

CPU0B >>>t 5 4 103 400

%SCSI-E-STS = 01510392

%SCSI-I-EXTSTS, USRMOD

CPU0B >>>t 5 4 103 400

%SCSI-E-STS = 01510392

%SCSI-I-EXTSTS, USRMOD

CPU0B >>>t 5 4 104 400 0

%SCSI-E-STS = 01510492

%SCSI-I-EXTSTS, USRMOD

CPU0B >>>t 5 4 106 400

%SCSI-S-STS = 00000601

CPU0B >>>t 5 4 108 400

BLKSIZ = 00000200

MAXBLK = 0013E5F3

%SCSI-S-STS = 00000801

CPU0B >>>t 5 4 10B 400

%SCSI-E-STS = 01510B92

%SCSI-I-EXTSTS, USRMOD

VR295 Monitor

It has a resolution of 1280 x 1024 pixels and a 66-Hz noninterlaced refresh rate.

DB15 to RJ11 & PS2 for LK201 Keyboard & VSXXX-AA Mouse or VSXXX-AB tablet

The table above shows the keyboard. It is reversed from the pinout below because of the cable.

Keyboard and Mouse/Tablet Connector (I/O Cover)

Pin Source Signal Description

1 GND Keyboard ground           to RJ9 Pin 2 Yellow

2 L2003 kTX0 Keyboard transmitted data to RJ9 Pin 1 Orange

3 KEYBRD kRX0 Keyboard received data    to RJ9 Pin 4 Green

4 L2003 +12V Keyboard power            to RJ9 Pin 3 Brown

5 MONORTN Monochrome video return

6 MOUSE nRX1 Mouse received data

7 L2003 nTX1 Mouse transmitted data

8 GND Mouse ground

9 GND Keyboard power return

10 MFMD1 Manufacturing mode input

11 L2003 MONO Monochrome video data

12 MFMD0 Manufacturing mode input

13 L2003 +5.3 V Mouse power

14 L2003 –10 V Mouse power

15 GND Mouse power return

  1    2    3    4

 -------------------

| "    "    "    "  |

| D    G    +    L  |

| E    N    1    K  |

| C    D    2    -> |

| ->        V    D  |

| L              E  |

| K              C  |

 --               --

   |             |

    --         --

      |       |

       -------

Looking into the DECstation Keyboard Connector

(Socket on DECstation)

VT220 Keyboard Connector

1 KB-R From Keyboard, pulled to -6V

2 GND

3 +12V

4 KB-T To Keyboard

      Cable Connectors

      ================

   DB15S, computer.  (Female, as seen from the male point of view)

      08  07  06  05  04  03  02  01

        15  14  13  12  11  10  09

   DIN-7, mouse.  (Female, as seen      4-pin modular phone, keyboard.

   from the male point of view)         (Female, as seen from the male

                                        point of view)

         _   _

          |_|                                 1   2   3   4

       7   6   5                                  _   _

      4  [   ]  3                                  |_|

         2   1

      (The mouse and keyboard connector numbers may be non-standard.)