My first Data General restoration!
Having looked on various auction sites for many months it became clear that very few systems become available, even more so for those systems that are of a younger generation. In October 2018 I came across a MV/9800 upon examination and correspondence with the owner it transpires that it was pulled from a failed business, it had no software or manuals and I could see from the pictures sent to me it also had no interconnecting cables between each component in the rack and the console. Well I left it a few days to mull it over and then decided to see if I could negotiate a deal whereby the seller would remove each chassis from the rack and pack them well in boxes for shipment. The seller agreed and a week later I have 4 large boxes arrive from the USA with all of this – minus the 19″ rack:
The Arrival
The first task was to unpack everything and check for physical damage (both from its removal from the rack (by non-technical people) and the shipping – thankfully nothing was broken.
The tape drive turns out to be a 6125 unit – a very nice compact unit compared to the vacuum drives I used to repair many years ago. In the next box is the console a DG D464, in the next the Clariion chassis which is a CSS3 (SCSI II Combined Storage Subsystem 3) with dual PSU, 643Mb HD and DDS2 Tape and finally the MV/9000 chassis which contained a surprise!
In the chassis slots I found:
005-020141 : DC Power Supply
005-040190 : MV/9800 dual CPU PCB with 2 x 32Mb
005-038704 : MV/9600 CPU PCB with 1 x 32Mb
005-038601 : SCSI II controller (3)
005-015289 : Mag Tape Controller
005-039526 : Intelligent Lan Controller (ILC) 4217
005-019071 : ISC
005-022462 : IAC/16 Serial Line Controller
005-016764 : IAC/8 Serial Line Controller
Zetaco Line Printer Controller
I was quite surprised to find a second CPU (MV/9600) and three SCSI II controllers – this more than made up for loosing the rack (due to shipping constraints)!
Conversion from 110v to 240v
So the first task is to convert all of the kit to run on 240v. I’ll start with the easiest, but before I do – please do not attempt to make changes to a Power Supply unless you are aware of the risks and you are confident of what you are doing. If you do not understand what you are doing do not undertake any changes yourself and refer the task to a qualified technician:
D464 Console: The D464 (DG 6865) terminal benefits from a switched mode PSU it is capable of running from 110v thru 240v without modification.
Data General Clariion CSS3: This chassis has two Power supplies located centrally. Each PSU has a voltage selection switch on the top. All that is required is to move the switch to the 240v position on both PSU’s.
Tape Streamer 6125: The PSU for this device is mounted on the rear panel. Remove the panel and the cage covering the PSU. At the side you will find a voltage selection jumper move this from the 110v to the 240v position and reassemble. Be careful exposing the PSU as voltages can remain stored long after the power has been removed.
MV 9000 Chassis: This has two power supplies (a VNR at the rear and a separate DC PSU in the chassis). The input voltage is converted by the VNR to 300v DC which is fed into the chassis PSU which provides that various low voltage supplies. The VNR has two voltage selection (loopback) plugs one external and one internal these both select the input voltage and are labelled 110v on my VNR. The A/C input filter is also rated at 110v so that will need changing as well. I am still trying to obtain schematics for the VNR (or source/borrow a 240v one) so that I can make a new internal and external plug. In the interim I have purchased a 240v to 110v step down transformer.
Visual Inspection, PSU Checks and Initial Power On
I tend to take the same approach (whilst taking notes/pictures before I remove or disassemble anything) with most of the items I acquire and that is a complete strip down, clean, inspection and base PSU test and this was carried out on the MV Chassis, Tape Streamer, CSS and Display. My procedure for this is documented here as its the same for each restoration.
Once I was happy with the health of the PSU’s each one was powered on and basic functions were tested.
D464 System Console:
- Power Light is showing Green.
- Display Beeps initially and again a few seconds late – good code seems to be running in the display.
- Keyboard CAPS lock works.
- DISPLAY IS BLANK – brightness controls ineffective.
Tape Streamer 6125:
- Power Light On
- Fails to load a tape, starts load procedure and then stops.
- Second load attempt works and tape loads to BOT.
Unit will switch to On-Line. - Unload Button causes tape to rewind and unload.
- Very High Frequency sound being emitted by PSU.
Data General Clariion CSS3:
- Power switch works and both PSU indicators are lit.
- All fans are operational.
MV 9000 Chassis:
- Power Applied and VNR displaying Green LED.
- Power Rocker ON selected and PWR LED lights on Chassis and fans run.
- Fans are very noisy!
- All voltages present and correct on the back-plane.
- Power off and insert MV/9800 CPU in slot 5.
- Power on – green power light on on MV/9800 CPU, self test LED is off.
- Tape Streamer does not always load the tape.
System Console has no video.- MV/9800 CPU Fails Self Test
First Signs of Life
On inspection of the D464 console I could find no obvious signs of failed components or damage to the PCB. However on closer inspection on the underside of the PCB I noticed some suspect solder joints on the Flyback Transformer. I desoldered each of the tags, cleaned and resoldered and this resulted in a working display.
So on to a RS232 cable, the MV Chassis presents the console on a DB25 connector, so I set about constructing a DB25 to DB25 cable as per the example shown in one of the Dasher Terminal Manual’s (Thanks Tommie) DG part no: 005-094614.
With my homemade console cable plugged into the D464 and the console switched on I applied power to the MV, within a few moments the CPU PCB started to communicate!
So we have life and with it brings some good news and some bad. It seems that the MV/9800 CPU PCB has a problem as it stops the POST at character U, however the MV/9600 CPU passes POST and is ready to rock!
At the moment the system has nothing but the CPU in the chassis. The next step is to introduce a peripheral to the system and see if we can get it to boot! I have a choice of two; a 6125 streamer (with a single tape unmarked so I don’t know if its bootable or not) or a SCSI HD that was in the CSS3 chassis.
Purely because I love the visual aspect of reel to reel magnetic tape I opted to try the tape first.
1st Boot
I removed the Mag Tape Controller (005-015289) PCB from storage and confirmed that the device select jumpers were set to 22 and placed the PCB into slot 13 of the MV chassis. I then wire-wrapped the Interrupt and Data Channel priority pins A93 and A95 from the CPU to A94 and A96 on slot 13. Lastly I installed the back-plain to bulkhead cable onto side A and B of the back-plane for slot 13.
Next comes the guessing part – this system came without any interconnecting cables so I opted to use a DB50 to DB50 straight through cable in the hope that DG opted for a pin-to-pin solution.
I loaded the tape that was on the drive when it arrived (un-labelled) and with the MV/9600 sitting at the SCP prompt I placed the tape on-line and issued a boot 22. The tape moved for approx 2 seconds and stopped but nothing was returned on the console. Um, before I can draw any conclusions I needed to know that I was booting from an actual bootable tape, a quick search on e-bay found someone selling a DG SCP tape for an MV/7800. I duly purchased this as even though it was the wrong tape for the MV/9600 or MV/9800 it would at least be bootable.
A few days later the DG tape arrived and I repeated my test but alas I got the same results with this tape as well. So I decided to put the 6125 to the side for now and take a look at the SCSI side of things whilst I try and obtain manuals and schematics for it.
2nd Boot
So with the 6125 out of the equation I turn to the SCSI-II HBA, VIP Mario very kindly sent me some pictures from a DG fiche containing the field engineering manual for the SCSI-II controller. From these pictures I was able to configure the controllers Device ID and placed the PCB in Slot 14. I then attached the DG back-plane to DB50 bulkhead cable, a DB50 to CN50 cable connected the MV to the CSS3 chassis. With the MV powered-up I issued a boot 13 (SCSI-II HBA Hard Device Code) and I have my first (kind of) boot event!
The console screen cleared and I was presented with the SCSI-II controllers firmware menu 🙂
With a big grin 🙂 (having got more than just a SCP-CLI prompt finally) I set about defining a disk and giving it a soft device code of 24.
Option 5 returned me to the SCP-CLI and I issued a boot 24 – nothing 🙁 my grin was gone! I hooked up a SCSI monitor to the bus and tried again and I could see no activity on the SCSI bus. Deflated I contacted VIP Mario to see if the manual he had sent me pictures of previously had any referenced in to to the back-plane signal or and reference to cables.
Whilst I wait for Mario to take a look I turned my attention back to the 6125.
Magnetic Tape Investigations
I contacted VIP Bruce and explained what my 6125 was doing, he provided me with advice and a short piece of code to enable me to test if the tape was being read:
If the tape moves when executing a IORST followed by an NIOS MTA instruction (i.e. auto boot load requests), that generally indicates that at least the PIO (Programmed I/O) instruction signals are getting to the tape drive. A short program to test if a tape record is read in (without assuming the data is an executable program) could be placed in high memory (for example starting at location 070000):
062677 IORST 060122 NIOS MTA 063622 SKPDN MTA 000777 JMP .-1 063077 HALT
Start the program at the IORST instruction; if online and at BOT, the tape should move, data should be read into location 0, and the program should halt.If it does not halt, then the DCH priority chain should be [very] suspect.
This code was duly ran and the tape did move and then stopped which reassured me that PIO and DCH were functioning, so I started looking at other possibilities.
VIP Lothar described to me how to use code to read the status register of the device, he explained that using the following code you could read into AC0 the status of the device:
60422 DIA 0,22
60377 DOC 0
Using this code I conducted the following tests:
6125 Off – AC0 contains: 0000000000000000
Tape Loaded / On-Line – AC0 contains: 0000000000104304
Tape Loaded / Off-Line – AC0 contains: 0000000000104304
So we can see that Data-Bit 15 is staying low as there is no change between the off-line and on-line status returned by the controller. VIP Lothar offered to test my controller and cabling in his MV so I sent these off to him for testing.
Whilst I await the results of the controller and cabling tests I decided to return to the SCSI-II controller as VIP Mario has been able to find signalling information for the HBA.
SCSI-II Controller Investigations
So the lack of SCSI data seen on the bus becomes clear on reading the fiche pictures VIP Mario was able to send me. The presentation Data General opted to use on the back-place to DB50 connector is non-standard!
So I set about building a spreadsheet with all of the SCSI-II SE and differential signals and their standard DB50 presentation and cross referenced these with the back-plane presentation from the SCSI-II HBA. I then set about building a new back-plane to DB50 cable to SCSI standard presentation, this task is documented HERE in a separate post.
So with my new back-plane to DB50 cable made I install it in the MV/9000 chassis and connect up the CSS3, power-up the MV/9600 and boot to the SCSI-II HBA firmware. I then go and assign the Hard Disk a soft ID of 24 but this time around the controller returns a PIO error and exits to the SCP-CLI:
I happened to mention this to VIP Lothar and he had experienced a similar error with a Zetaco SCSI controller where the HD had not been initialised with a Zetaco software utility. This reminded me that I have plugged in a SCSI2SD device which I was using to monitor the SCSI bus and I wondered if this was causing the problem. With the suspect device removed from the SCSI bus I tried again and I was able to assign my soft ID of 24. I was now ready to try a boot the 643Mb HD that came with the MV/9800 system (if there is anything on the disk).
643Mb SCSI Disk
So with the soft device code set I returned to the SCP-CLI prompt and issued a ‘boot 24’ and the SCSI HD activity light flashed, the console screen cleared and I was presented with the following on the console:
I now had a big smile on my face as clearly my home made SCSI back-plane cable worked and more importantly I had a SCSI HD that had something bootable on it!
This SCSI HD contains ADEX (Advanced Diagnostic EXecutive) which is Data General’s Diagnostic Operating System.
Magnetic Tape Controller Returns From Testing
VIP Lothar has tested my controller, back-plane cable and chassis interconnect and all is well with these components which have arrived back from Germany. So my problem lies somewhere within the tape unit. Another visual inspection of the cable does not reveal anything untoward so I am going to strip the unit down and setup on the bench so I can take a look at the interface design and draw my own logic diagram of the initial input/output side of the Tape Drive.
To avoid cluttering up this system restoration page I have decided to create a new page for this work HERE.
Lots more to follow…………………
2 thoughts on “The MV/9800 Restoration”