The scrapbook restoration team visited the National Museum of Consulting at Bletchley Park on the weekend of the 8th and 9th of February 2019. Weather conditions were pretty good – we only mention it as the last time we attended in February last year it was baltic!
Attending for this session was once again Sean Allison of Askaris, his brother Ray Allison; and ex Systime colleague Malcolm Gill. Scrapbook is a hypertext system written by the National Physical Laboratory and preceded the invention of the World Wide Web by Tim Berners-Lee by several years. The system being restored operates on a DEC PDP 11/70 CPU with two DEC RM03 (67Mb) disk drives. The system ran an operating system called IAS which is a derivative of RSX.
The intention was to run two work paths during our two days. The 11/70 CPU fault finding and diagnostics was run by Ray and Malcolm, whilst Sean concentrated on building a disk sub-system to enable low risk disk copies using one of the museum’s spare PDP 11/34 machines.
On our previous visit we had managed to get the 11/70 CPU to boot XXDP which is DEC’s diagnostics system. Even if the system it is being run on has severe problems, there is always a chance to run XXDP. Our 11/70 has problems! The last stage we got to was to establish the 11/70 had cache memory issues and also memory management issues. We had several basic areas of verification before we could get into circuit level fault finding. We confirmed all voltages were in tolerance.
We also had to test the memory cables – the memory is a separate box weighing 50Kg and contains 128Kw (256Kb) of memory. This is connected to the CPU via four 20ft ribbon cables. We had always suspected a potential cable issue so Malcolm had spent time since the last visit building a test rig for ribbon cables. All cables were tested and two were found to be faulty, so we re-crimped some new connectors on them to resolve that issue.
We are now at a stage where we are ready to use circuit diagrams and an oscilloscope to find out exactly where the cause of the remaining cache and memory issues are. more on this during the next visit.
The ultimate aim of this project is to recover the data stored on the only set of Scrapbook disks anywhere in the world. At this time we are nowhere near this stage, but the ultimate plan is to provide a foundation to allow us to back up this critical media to a robust destination. We are using a modern disk drive to achieve this. When we say modern, the drive is 8 inches wide and stores 134Mb! However it is a Winchester disk which is the same technology as modern magnetic drives, where the entire drive is enclosed and sealed, so is not susceptible to physical or environmental damage.
The drive will be attached to an Emulex SC21 disk controller and is configured to be seen as 2 x 67Mb drives (823 cylinders, 5 heads, 32 sectors) which is exactly the same geometry as the source Scrapbook (RM03) disk packs. The Emulex controller will present this single drive as unit 0 and 4, which allows one partition to be installed with an OS capable of reading any disk block by block. The second partition will be used for the destination of any disk copy – ultimately the Scrapbook disks. The SC21 also allows up to four physical drives to be attached which will enable the simultaneous operation of the Fujitsu M2322K and also a DEC RM03 (with some modifications).
We then needed a system to attach the controller to. The obvious candidate was the Scrapbook PDP 11/70, however as this is still being investigated for cache and memory management problems, Sean sourced a PDP 11/34 from museum stock. The PDP 11/34 is a 16-bit Unibus DEC machine (the same as the 11/70) with an 18 bit address bus. Our machine (right) has 32Kw (64Kb) of memory installed. It also has the optional Programmer’s Console, a memory parity module and a Floating Point Processor installed.
With a small amount of work this machine is now operational. It has two disk controllers – an SC21 (as above) and a RL11 controller to run the DEC RL02 drive – a 10.4Mb drive used for media distribution, of which the museum has lots. We were unable to get the SC21 controller to see the M2322K drive as we found its terminator resistors were missing. New ones have been ordered.
The next visit to the museum should be in the next 2-3 months where we intend to spend an entire week working on the system – subject to spouse approval! A final note is to acknowledge the contributions from Geert Rolf – Geert is a self-confessed PDP nut based in The Netherlands and very kindly donated the Emulex and Fujitsu equipment to assist with this project. Geert has the most amazing collection of PDP computers – details can be found at http://www.bejaardecomputers.nl/index-en.html