Malcolm Bigg, Romilly Bowden, Ted Braunholtz, Geoff Crome, Peter Harrild, Gordon Harvey, Peter Holland.
Frank Taylor, the 8th member of the group, was unable to attend.
Just before setting off for the meeting, Gordon had learned that Jack Templeton had died in April 2004 in hospital following a heart attack.
At the Computer Conservation Society seminar on 29 October 2003, presentations on the Ferranti Orion 2 computer were given by Frank Taylor, Ted Braunholtz and Malcolm Bigg who had all been actively involved in the development of the computer in the1960's. Peter Hall, who had been Manager of the Ferranti Computer Department at that time, was prevented by illness from contributing to the CCS meeting, but subsequently wrote some comments which were published in 'Resurrection' No.33 (Spring 2004).
Some other members of the development team who were at the CCS seminar were concerned that some errors and potentially misleading statements arising from that meeting might be perpetuated in the CCS archives. Subsequent correspondence resulted in the setting up of an 'Orion 2 Contact Group' with the object of trying to get the record straight. This report attempts to summarise the outcome of the group's first meeting. However, for completeness, some details have been included which were not specifically discussed or verified by those present.
Those present had received re-typed copies of Peter Harrild's project diaries for the crucial years, 1961 to 1966, in 6 volumes totalling nearly 40 pages, which proved to be invaluable although, inevitably, many of the entries were cryptic and naturally lacked 'context’ which would have been common knowledge at the time.
It quickly became dear that the term 'prototype' in the project documentation is misleading in that it did not relate to a specific, well-defined, hardware/software configuration. The original hardware assembled at Lily Hill House, Bracknell, was often referred to as 'the prototype' or sometimes as 'DP102'. The equipment delivered to the first customer, the Prudential Assurance Company (‘the Pru'), in September 1964 was also thought of as ‘the prototype'. In fact it was evidently an extended and upgraded version of the original configuration and it was subsequently further enhanced on site. Some of the original hardware was left at Lily Hill and, after upgrading, it was used there for development of the Type C magnetic tape system and the Nebula Compiler. This set-up was also known as ‘the prototype' and sometimes as ‘DP102’.
The original magnetic tape system at Lily Hill used Ampex TM2 decks (90 kchar/s) and a tape controller operating in hesitation mode. Before delivery, this system, known as Type A', was upgraded to Type B' with Potter 120 decks (90 kchar/s) and operating in direct access mode via a 'Store Access Control’ (SAC) unit. It was further upgraded on site to ‘Type C' (200 kchar/s) in September 1966.
After completion of acceptance tests at Lily Hill, the complete system was delivered to the Pru on 19/20 September 1964 and installed as their Ordinary Branch (OB) system, required to handle 1.2 million policies.
Site acceptance tests were completed on 18 November 1964. This was one week ahead of an agreed (though re-negotiated) deadline beyond which a penalty of £4,000 per week would have been payable.
An outline specification of the Pru OB system is shown in Table 1 together with similar information on three other delivered Orion 2 systems, and on the Nebula development set-up left at Lily Hill.
The main store was a magnetic core store of 16K words with a cycle time of 2 microseconds. Plessey Ltd supplied the original core store in the OB system, but this was replaced on site by a unit supplied by Ampex, and Ampex became the preferred supplier.
Initially Ferranti MD5 magnetic drums were used. Each drum had a capacity of 16K words, i.e. 128k characters. The Pru OB system had four such drums in two pairs, each pair being mounted one above the other in a single cabinet. The two drum cabinets were positioned on either side of a third cabinet containing the drum controller. Before delivery to the Pru, the Ferranti drums were replaced by compatible drums obtained from Sperry Ltd.
An essential requirement for both the Pru OB and their IB (Industrial Branch) operations was the ability to read and punch 80-column cards, in the OB set-up, the data on over a million cards had to be read and transcribed to magnetic tape, and in the IB case the figure was more than 30 million. Controllers had to be designed from scratch to meet this requirement.
The primary reason for embarking on a replacement for the original Orion hardware developed in Ferranti Manchester was the unreliability of the 'neuron' majority-logic elements in a large system. Discrete transistor logic circuits had been developed by Maurice Gribble in Ferranti Wythenshawe (where they were dubbed 'griblons') and used successfully in the Ferranti-Packard FP6000 computer. A decision was made in November 1961 to adopt that basic circuit design for the cards or 'packages' for Orion 2.
In order to isolate the computer circuits from mains transients, a motor-alternator set was interposed between the raw ac supply and the internal (ac-dc) power units. One such set (supplied by Mawdsleys of Durstey, Glos.) was installed at Lily Hill. The Pru insisted on having two, one being a stand-by for the other. The alternator output rating was 42 kVA.
A firm figure for the price to the Pru of their OB system has not been established from the available records, but in March 1963, the price quoted to the Air Ministry for a comparable but slightly larger system was reduced from £900k to £750k!
(a) Lily Hill House was abandoned several years after the last of the project team members left - mostly on redundancy - in 1967. The house was more or less derelict until acquired and refurbished for commercial offices. The (croquet) lawn is now a car park.
(b) It was hoped that John Travis would be able to get his 8mm film of Lily Hill activities transcribed to DVD and copied to members of this group.
A corrected draft of this report would be sent to those present plus Frank Taylor and then submitted to the editor of 'Resurrection', Nicholas Enticknap, with a suitably worded comment on Peter Hall's remark in Issue No. 33 that "The change in circuits [ from Orion 1 to Orion 2 ] meant a significant logical redesign, but essentially the job was only to re-engineer Orion with 'Griblons' instead of 'Neurons' - nothing else"! A copy would be passed to Simon Lavington as an input to the 'Our Computer Heritage' Pilot Study.
TABLE 1 - Summary of Configurations
|
Customer |
Pru OB |
Pru IB |
SAMLAS |
Beecham |
Nebula dev't |
|
Equipment |
|
|
|
|
|
|
Core Store size (K words) |
32 |
32 |
16 |
16 |
16 |
|
Drum Store (No. of 16K drums) |
4 |
4 |
4 |
3 |
2 |
|
Mag Tape Decks - Ampex TM2 |
- |
- |
- |
10 |
7 |
|
Mag Tape Decks - Potter 120 |
10 |
10 |
8 + 3* |
- |
- |
|
Line Printer - Anelex |
2 |
2 |
2 |
- |
1 |
|
Line Printer -ICT665 |
- |
- |
- |
2 |
- |
|
Card Reader - ICT 593 |
2** |
2** |
1 |
2 |
- |
|
Card Punch - ICT 582 |
1 |
1 |
- |
- |
- |
|
Paper Tape Reader - TR5 |
2 |
2 |
2 |
3 |
2 |
|
Teletype Punch - 7-track |
1 |
1 |
1 |
2 |
1 |
Note: All above systems included a Flexowriter 'Z’. "16K words" = 128K characters.
* 8 decks with 200 kchar/s tape plus 3 with 90 kchar/s tape.
** Modified for 65-column (interstage-punched) cards as well as 80-column cards.
(The above configuration information, dated April to June 1967, was obtained from Malcolm Bigg)
GW Harvey
1961
|
Sep. |
21 |
TGHB started to describe his ideas for Orion 2. |
|
Oct. |
20 |
Meeting to discuss packages for Orion 2. |
| Oct. | 30/31 | Ferranti-Packard engineers arrived from Canada. |
|
Nov. |
2 |
Decision to use F-P 'Gemini' packages. |
|
Nov. |
14 |
Plessey 4K 2 microsec. core store "looks promising". |
1962
|
Jan. |
9 |
Peter Dorey announced setting up of Data Systems Division. |
| Jan. - Feb. | Frequent meetings to discuss basic engineering details, e.g. back wiring, waveform names, wire lists, cooling method, &c. | |
|
Mar. |
9 |
First batch of 20 boxes (package shelves) arrived. |
| Apr. | 25 | First batch of F-P packages arrived (170 total of two types). |
|
May |
25 |
Drum (Ferranti MD5?) powered up. |
|
Sep. |
10 |
Motor-alternator set installed in conservatory. |
|
Sep. |
18 |
Start of assembly of racks. |
|
Oct. |
21 |
Discussion on faster tape transfer system (became ‘Type B’). |
|
Nov. |
14 |
Merger with ICT believed confirmed. |
|
Nov. |
16 |
Core store (Plessey) in position on plinth. |
|
Nov. |
20 |
ICT Directors et al visited Lily Hill to see 102 set-up - "impressive". |
|
Nov. |
30 |
Visit from West Gorton production staff. Hope to be ready to start production of Orion 2 by May 1963. |
|
Dec. |
3 |
Core store basically working. |
1963
|
Jan. |
7 |
'Prototype 102’ officially switched on by Peter Hunt. |
|
Jan. |
16 |
Commissioning of logic started. |
|
Jan. |
23 |
Technical Clerks started processing mods. |
|
Feb. |
15 |
Plessey core store working with >5% margins. |
|
Feb. |
18 |
Peter Hall gave go ahead for Type B tape system. |
|
Mar. |
6 |
Visit of Prudential staff + Basil de Ferranti et al. |
|
Mar. |
19 |
Peter Hall proposed Pru should have (extended) prototype for their OB application. BdeF promised September 1964 handover. |
|
Apr. |
2 |
Ferranti MD5 drum "graunched again". ?Use Sperry instead? |
|
May |
1 |
Peter Hunt confirmed prototype sold to Pru with heavy penalty if late. |
|
May |
30 |
Peter Hall agreed Racks 2 to 4 to be replaced by production modules. |
|
Jun. |
5 |
First Ampex TM2 tape deck arrived. |
|
Jun. |
14 |
Anelex printer and 2 TM2’s arrived. MD5 drum returned. |
|
Jul. |
3 |
Type A tape system "almost working". |
|
Jul. |
27 |
Seven TM2's arrived. |
|
Aug. |
20 |
First Type B tape control arrived. |
|
Sep. |
23 |
Decision to put pin numbers on logic diagrams. |
|
Sep. |
26 |
ICT takeover. |
|
Oct. |
3 |
SAC (Store Access Control) arrived. |
|
Nov. |
6 |
Plessey core store failed. |
|
Nov. |
8 |
Demonstration of 102 to high-level ICT delegation (Humphries, Organ et al) |
|
Nov. |
15 |
Basil de Ferranti and Sir Edward Playfair to see 102. |
|
Nov. |
27 |
Second Sperry drum arrived. |
|
Dec. |
6 |
Second Type B Tape Control arrived. |
|
Dec. |
13 |
First Sperry drum working |
1964
|
Jan. |
4 |
Start of major re-arrangement of racks etc. with new and replacement hardware from West Gorton to implement Pru OB system requirements. |
|
Jan. |
10 |
New desk &c arrived from W.G. |
|
Feb. |
1 |
More racks &c arrived from W.G. |
|
Feb. |
5 |
Ampex core store arrived. |
|
Feb. |
24 |
MD Committee decision to sell Orion 2. |
|
Mar. |
10 |
4th Sperry drum arrived. |
|
Apr. |
2 |
Successful cooling tests overnight. |
|
Apr. |
7 |
2nd Type B Tape Control installed. |
|
May - Aug. |
|
Intensive testing of Pru system at Lily Hill. |
|
Aug. |
25 |
Successful demo of Nebula to Pru. |
|
Sep. |
7 |
Testing finished and packing for delivery started. |
|
Sep. |
19 |
First load delivered to Pru site (Holborn). |
|
Sep. |
20 |
Second load delivered. |
|
Sep. |
21 |
All hardware in position on site. |
| Oct - Nov. | Commissioning on site - 7 days a week | |
|
Nov. |
10 |
Acceptance tests started. |
|
Nov. |
18 |
Acceptance tests completed. |
|
Dec. |
1 |
Celebration party at Admiral Cunningham hotel. |
|
Dec. |
29 |
Problems at Pru - core store and mag. tape. |
1965
|
Jan - Feb |
|
Problems on site (Pru OB) being investigated and corrected. |
|
Apr. |
26 |
Beechams, using Pru OB system, found logical error in 102E instruction. |
|
Apr. |
27 |
Pru negotiated extra 16K core store at nominal price, claiming Nebula takes up twice the expected space. |
|
May |
10 |
First (of 10) Potter high density tape deck arrived. |
|
Jun. |
26 |
First Type C Tape Control installed at Lily Hill. |
|
Sep. |
15 |
SAMLAS Orion 2 factory acceptance tests completed at W.G. and .... |
|
Oct.-Nov. |
|
... hardware shipped from Southampton. |
1966
|
Jan. |
11 |
SAMLAS Orion 2 passed acceptance tests on site (Cape Town). |
|
Mar. |
18 |
Pru IB Orion 2 factory acceptance tests completed. |
|
Apr. |
23 |
Pru IB system delivered to site. |
|
Jul. |
5 |
Pru IB acceptance tests started, and ... |
|
Jul. |
13 |
... due to be completed, but ?delayed by problems with tape system. |
|
Aug. |
5-13 |
'DP 102' (at Lily Hill) re-located in Room 1 (old Library). |
|
Sep. |
2 |
Type C Tape Control, 10 HD decks + extra core store delivered to Pru OB site. |
|
Oct. |
31 |
Site acceptance tests on enhanced Pru OB system due to start. |