neil's webbly world
Academic Pages

Roland W-30 Information

This is my little trove of technical information on the Roland W-30 Music Workstation.



SCSI Chips

SCSI chips are available from:

Steven Bukin in UK

Jose Pereira in Portugal

Sylvain Labrecque in Canada

Derek Taubert in CA, USA

Vintage Planet has some new-old-stock SCSI chips for sale (do a search for "W30" to find it), for about €33. Also has replacement disk drives, LCD panels, and lots of other parts available.
NOTE: this is not a personal recommendation, I have no connection or other interest in Vintage Planet.

Yep, the W-30 has a SCSI port on the back. The standard model has the connector fitted, but not the SCSI interface chip needed to communicate with other SCSI devices (mmmm...some cunning marketing ploy?) An optional upgrade kit, KW-30, was available to activate the port. It included the SCSI controller chip, a revised version of the OS, and a manual.

The SCSI interface is the venerable old version 1 (SCSI-1) which came out in 1986, and in the W-30 it is hard-coded to SCSI ID 7. Apparently, all SCSI-2 devices should be compatible with SCSI-1. The engineer in me says "yeh, right" so here I'd like to have a catalogue of all known SCSI drives that work with the W-30. I kindly ask all W-30 owners with SCSI devices to let me know the details of their successes.

It would also, I think, be beneficial to know which devices definitely DO NOT work with the W-30, so that we know which devices to avoid in future exploratory tests.

The Good Guys -- These Work Perfectly

Manufacturer Type Capacity (MB) Notes
HD Conner (Seagate) CP3040 40 The first drive I've tried. No problems encountered, formatted at 40MB, with system 1.07
HD Quantum ProDrive LPS 42S 40 An ex-Apple drive.
HD LaCie Atlas 9GB Another ex-Apple drive, but HUGE (sadly most of it ignored). AND it is BOOTABLE!!!!
(Thanks to Richard Fouchaux for this valuable information)

The Sad Guys -- These Partly Work

Manufacturer Type Capacity (MB) Notes
None so far...

The Bad Guys -- These Do Not Work At All

Manufacturer Type Capacity (MB) Notes
None so far...

Drives waiting to be tested

I have rumours and hints that the following drives might work. As yet I have no information either way.

  • CD-ROMs
    • Sony CDU-541 (internal) CDU-6111 (external) caddy-loading 1x speed, apparently Roland CD-5 drives by another name.
    • Sony CDU-8001 in Apple CDSC drive (1x speed)
    • Sony CDU-8002 in Apple CDSC+ and CD150 drives (1x speed, but faster transfer rate)

MO Drives

Derek Taubert has successfully connected an Epson OMD-5010 (rev 3.09) Magneto-Optical drive to his W-30. After some kludging with jumpers, here is how he managed it:

I decided to give my 128MB optical drive a shot (EPSON OMD-5010 Rev 3.09). It has two jumpers to make it 1) appear as device type 00H instead of 07H, and 2) not issue "UNIT ATTENTION". The W-30 will then gladly accept it as a hard drive, but fails when trying to format a disk.

Not yet giving up, I decided to format an optical disk on another machine, and then dd'ed one of my above quantum drives to the formatted optical disk (dd if=/dev/sdXh of=/dev/sdXh obs=2048 count=163840 -- see below about my h partitions). Not only does it work (can read and write songs/sounds), the W-30 will even boot from the optical disk (say, that is fast)!

For anyone that has an OMD-5010:

  • Jumpers 1,2, and 3 select the SCSI ID.
  • Jumper 4 is reserved.
  • Jumper 5 changes the "INQUIRY COMMAND" result (07H off, 00H on).
  • Jumper 6 enables parity.
  • Jumper 7 turns off "UNIT ATTENTION".
  • Jumper 8 enables termination (backward from the FaxAdvice EPSON documentation).

    I have jumpers on 5,6, and 7 for the W-30 (1-3 and 8 depend on your particular SCSI chain).

So there you have it, a contender for removable media for the W-30 that is bootable! So far the only issue is formatting a drive (Derek got round this by copying the format structure from a hard drive).

And sadly we're still stuck at the 80MB (64 volumes) limit...


The W-30 is limited in the main software to a maximum of 80MB of disk space (well, 64 areas, which I guess comes to 80MB). Hacking of the boot ROMs indicates that this is a limit imposed by the main operating software -- the BIOS code would be quite happy accessing much larger drives (upto 2,000GB -- 32-bit block number, 512-byte blocks). This is a problem of not using a filing system, but rather specific disk vectors (either <side, track, sector> or <block>). Yet another feature to add to the list...


The W-30 boot ROMs include a lot of software functionality for talking to the SCSI bus. While investigating the boot ROM software I find a number of interesting pieces of information, which I present here.

Firstly, the only types of devices the W-30 recognizes are:

  • Type 00h -- Direct-access devices (e.g. magnetic disks)
  • Type 05h -- CDROM devices

The type numbers refer to official SCSI device types.

Thanks (again) to Dave Fenwick, we now have a clearer idea of why the W-30 cannot boot off ZIP drives. Over to Dave:

After some pain-staking reverse engineering of the OS, I believe I've finally figured out why we can't boot from Zip disks. It is a bug in the ROM BIOS that causes this (and probably a lot of other SCSI-related issues.)

Basically it comes down to a very odd issue with how the original developer read from the mb89352 scsi chip. There's a register called the SSTS register (the chip status register) that holds information about the state of the mb89352. When in initiator mode, the chip won't register properly until the REQ line on the PSNS line is high. Unfortunately, there is a BIOS routine (kindly labeled as BIOS_Synth_Function_37_Select_Device by Neil Johnson in his original attempt at reverse engineering the BIOS) that sits in a tight loop waiting for the initiator flag to go high on the SCSI chip. The unfortunate part is it sits in this tight loop waiting for it to go high, but every time it reads the SSTS register, it clears the REQ bit in PSNS and it never gets a chance to go high.

Inserting a very small delay during the reading of the register clears up the problem nicely. I don't have an EPROM emulator for the 27c64-20 chips, but I've mimicked the entire boot process with some code I wrote that boots from floppy. I hope that I'll be able to transfer control to the zip drive soon to boot.

The crux of the problem is really that the Zip drive doesn't identify itself fast enough in the scsi-select process to allow the REQ line to go high. Compound that with the bug I just talked about, the Zip drive will never be detected and the OS will drift off into la-la land waiting for the REQ line to go high.

So, there you have it: its a bug in the W-30 firmware.

The SCSI disk that the W-30 tries to boot from is the last SCSI disk (not CDROM) on the SCSI chain. This disk must have 512-byte sectors, and have the string

* ROLAND S-550 *

in the first sector. Are we to assume that the W-30 boots of S-550 SCSI disks? Seems bizarre....


The SCSI controller chosen was the Fujitsu MB89352AP SCSI Protocol Controller (SPC), also sourced by Sharp. Sadly, it is no longer in production, although there are still quantities of them left in specialised distributors.

I have recently found the datasheet for the MB89352 (1.4 MB).


The SCSI connector on the rear of the W-30 is a standard DB25 SCSI connector, as used on PCs etc. While not recommended for faster data rates, due to the lack of twisted-pair grounds, it is fine for the relatively slow 5MB/s (max) data rates from the W-30.

DB-25 Pin Signal Description Signal Function
8D0Data bit 0
10D3Data bit 3
11D5Data bit 5
12D6Data bit 6
13D7Data bit 7
15C/DCommand / -Data
20DPData Parity
21D1Data bit 1
22D2Data bit 2
23D4Data bit 4
25TermPwrTermination Power (+5V)

Copyright © 2001-2024 Neil Johnson