This page is where I am writing up my initial notes and experiences in playing with “Storage Module Device” (or SMD) drives.
It is presently just a somewhat random collection of notes. I may divide it up into separate pages later on – so please don’t create a permanent link to this page.
The SMD interface
The interface consists of:
- The “A” cable: A 60-way control cable that is daisy-chained between the drives, with a terminator plugged in to the last drive
- The “B” cable: A 26-way data cable that runs radially from the controller direct to the drive
- A ground cable
The connectors appear to be standard IDC headers, though the pin numbering is unconventional. One cable I have seen has (on the 60-pin connector) labelled the pin numbers as 31 to 60 on the upper side, with Pin 60 being where Pin 1 would normally be.
Generally speaking, it seems that if you line up the tiny “pin 1” triangle on the male and female connectors, you should be fine. But I still don’t have a working SMD system, so I may be wrong about this.
Webster WQSMD
This is a two-drive quad-height QBUS SMD controller.
It looks to be a good product as it has the disk configuration and formatting utilities onboard, as well as a massive 1MB of cache RAM, so that the controller can do “look ahead” reads. It is seen by the operating system (RT-11 etc) as an MSCP controller.
The WQSMD supports only flat ribbon cable, not twisted pair cable (see section 8.1 of the Webster manual).
To the right is a picture of the jumpers on the WQSMD.
The controller can be programmed to appear at different addresses. From the manual, it appears that the alternative addresses may depend upon the PROMs that are fitted to the board. But all boards that I’ve tried seem to respond as follows:
- Jumper A1 ON : 172150 (in 16-bit systems) or 772150 (in an 18-bit system) or 17772150 (in a 22-bit system)
- Jumper A2 ON : 160334 (in 16-bit systems) or 760334 (in an 18-bit system) or 17760334 (in a 22-bit system)
Here’s how to start WOMBAT on an 18-bit system with the controller at the first address:
[Need to take a log of the dialog and update this] @772150/000000 250 @R7/xxxxxx 2000 @RS/xxxxxx 340 @P
Sabre 9720-1230 SMD drive
This is the first SMD drive that I am attempting to get running.
I’ve no idea what the correct manufacturer name is for this drive. It has “Magnetic Peripherals, Inc”, “Control Data Corporation”, “Seagate” and “Sabre” branding on it.
I’m just going to call it the “Sabre drive” for now.
The “-1230” is a reference to the drive’s unformatted capacity: 1230MB.
I first powered this drive up with no interface cables connected to it. The drive seemed to spin up OK when I pressed the “Start” button (the green light flashed for about 60 seconds as the RPM increased, then changed to steady green), and spun down OK when I pressed Start again. But after sitting idle for about 10 minutes (with the drive spun down), the power supply’s cooling fan noise stopped. It seems the power supply has failed. I opened the power supply up and there were no obvious indications of what failed. The internal input fuse is still OK. The power supply is not designed to be serviced – there are PCBS and components on both halves of the case, and many interconnecting wires. Repairing it isn’t going to be an easy exercise, so I’ve put it aside for another day.
Here are some pictures of the original power supply, after disconnecting it from the Sabre drive and removing it from the sled that the drive and power supply shared:
I found an SDI drive that had a similar looking power supply attached to it. I removed that power supply, and fitted it to the Sabre drive. Here is a photo gallery showing my Sabre SMD drive with the replacement power supply fitted to it.
There are 38 dipswitches on this beast.
Switches on the rear panel of the drive
|
The 18 switches on the rear panel of the drive are called the “I/O Board Switches” and a picture showing their current setting is shown above.
Also shown above is Figure 3-21 from the Sabre drive’s manual, showing the function of these switches. The configuration of the drive according to these switches is as follows (from top to bottom):
Note: Move A213/1+2 both to OPEN, to send I & S on A and B cables. Note: Move A213/7 to OPEN, to change to SMD-E mode. |
Switches on the top of the drive
|
Current status (11-AUG-2016)
I’ve not yet been able to get this drive working with the Webster WQSMD. Changing the rear panel dipswitches (so the drive is in SMD-E mode, and putting the I&S signals on the A Cable) seemed to help. It goes through the motions of writing the formatting data (first pass of the drive-format process) without any reported error, but then fails with errors on every block (or on every second block, depending on the WQSMD board that I used) during the second pass of the formatting process.
There are many possible things to still investigate (dipswitches, cables, faulty drive, terminator). But it could be that the Webster board isn’t fast enough for this drive. The specifications for the Webster board say that it’s maximum data transfer rate is “3.0 Mbyte/sec maximum” (page 12 of Webster PDF), and the specifications for the Sabre drive say that it has a transfer rate of 3.024 MB/s (page 24 of Sabre PDF). So maybe I should try a less-advanced SMD drive instead?
Drives known to be supported by the Webster WQSMD
In the back of the Webster manual, it provides settings details for the following Fujitsu drives, so these appear to be supported by the WQSMD:
- M2322 8 inch 160 Megabyte drive
- M2333 8 inch 337 Megabyte drive
- M2351A Eagle 490 Megabyte drive
- M2361A Super Eagle 689 Megabyte drive
2 thoughts on - SMD Drives
Your problem is likely the sector size (set by the dip switches under the removable cover on the top of the drive module). The “raw” sector size (not 512 bytes) varies depending on the controller type. You’ll probably need to look at the switch settings for one of the supported drive types in the controller manual, then look at that drive’s manual to see what those settings do, and work the sector size back from there. IIRC, on this drive you can either ask for a specific number of sectors per track (implying the size) or number of bytes per sector (implying number of sectors). There is also an option to suppress “runts” (partial sectors). I’d guess that suppressing runts and increasing the sector size (bigger doesn’t hurt, up to some limit) will get things going.
I doubt the data rate is an issue. I think you’re being led astray by a “kilobytes vs. kibibytes” (I hate that stupid term) mistake. No manufacturer would create a drive that only runs a few kilobits faster than any other drive, only to be incompatible with most controllers.
These drives are batteships and it probably isn’t broken. Years ago I bought one of these from someone who couldn’t get it to work, also couldn’t get it to work, sent it back to MPI / whoever (see below) and they said the problem was in the sealed HDA and it wasn’t worth fixing. I asked them to send it back to me instead of scrapping it and they sent it back with cardboard wrapped over the bare drive (they thought it was scrap)! I decided “what do I have to lose?” and opened the sealed HDA (in my parents’ spider- and dust-infested basement), found a cable wasn’t plugged in, plugged it in and closed it back up. It worked after that, and continued to work as long as I used it (over 5 years) without ever throwing an error or growing a defect. I even bought the fancy front panel (optional) and 19″ tray / faceplate for it, since I originally only got the drive, power supply and power cable with mine.
The confusing branding is due to the original manufacturer (CDC) deciding to sell all of their profitable product lines to “concentrate on [going bust in] our ‘core competency'”. Various branding is CDC / MPI (Magnetic Peripherals, Inc.) / Imprimis / Seagate. Seagate buying this group is why they (Seagate) stopped making junk drives and started making good ones. All of the “bird or fish name” drives from Seagate came out of this group. The official model is Sabre 9720 and your particular drive is a PA8N2A. Bitsavers have a bunch of manuals under /pdf/cdc/discs/sabre which may be helpful.
I must say you don’t really know what you are talking about. The bird and fish drIves you are talking about didn’t come from Seagate. The sabre, the wren, the wren runner etc… were in production way before Seagate took over the Control Data Corporation. Sabre was a drive line geared tward super computers at their time. Seagate just pulled the plug on their manufacturing of these drives and sent all to singapore. Taking away American jobs. The drive you have is cool though. Those drives were put in the Cray super computers at the time.