The Computer Laboratory Relics Database

Uid:  Class:  calculating machine computer /part of documentation memory /part of network /part of other peripheral /part of
Name:  Machine:  ? CAP CDCS EDSAC I EDSAC II Elliot 405 Fairisle IBM 3420 IBM ? IBM? PDP11 Rainbow Ring Titan n/a other
15 matches found

Unique id/year of acquisition: CL-1/97
Name: EDSAC I log book

Other No marked on object: none
Inscription: Copy of EDSAC log 1949 (front cover)
Dimensions: 297x209x17
Description: Bound book, green front cover
Class: documentation
Machine name: EDSAC I
Condition: good
Notes: DW: This is the copy of the EDSAC 1 Log book. It was started I think on May 6 1949 and was almost a commencement of practical Computing. It did in fact print a table of squares and the programme was completely automatic. In other words the programme was read in from punched papertape, did a little computation of two minutes thirty seconds and then printed out the results on a teleprinter and the whole was automatic apart from pressing the start button. The machine had actually been working and been commissioned before then but no records were kept and so we can't give any authentic dates.
Q: Who would have written the log ? Would it have been the operators ?
DW: There were no operators to start with. It would usually be the engineers or the users who would fill it in. It started with a very informal system and the proper computing probably didn't start until about three months later, although it was used for quite a number of problems.
Q: Any people in particular who would have written the log ?
DW: Some important names. Some would be by Bill Renwick no doubt, some would be by me, some would be by Maurice Wilkes and various others who were around at the time. I don't think the records were all signed, though I believe some of them are signed.
See also: CL-149/00

Unique id/year of acquisition: CL-2/97
Name: EDSAC I chassis

Other No marked on object: none
Inscription: none
Dimensions: 820x60x60
Description: EDSAC I chassis with one remaining valve. Grey metal rack with 14 white valve sockets and 9 smaller black ones. One red valve (EF50, 7/43) is still in one of the larger sockets. A large number of components are found in the rear of the rack. The rack is badly scratched and rusty.
Class: computer
Machine name: EDSAC I
Condition: poor
Notes: DW:This is a chassis from the EDSAC 1 which had possibly fifteen racks, each of which held five to ten chassis. These were wired in permanently - the one shown here has had all its valves removed apart from one which shows the size and shape of typical valves. The wiring was underneath and the back wiring connected chassis across the full room which was about 15 ft by 15 ft for the racks. The purpose of this particular chassis is unknown but a chassis of this size would typically run one memory tank in the memory storing possibly 16 words or maybe one word of an accumulator or even half a word for the instruction register. Racks of similar size were used to make adders and other selection devices.In the very early days the memory chassis cleared the memory by the users actually using a finger to touch a diode at one end although later a press button to clear the entire store was fitted. But early on the individual 32 memory tanks were cleared by touching 32 individual diodes.
Q: How early would this chassis have been used ?
DW: This was a part of the machine so it came into use in May 1949. It may have been commissioned 6 months or a year before. It was made here in Cambridge. It was manufactured by - I think it's Bailey, there's an electrical wiring workshop here in Cambridge. I forget, it's something like Bailey - and after the prototypes had been made in the Computer Laboratory they did the manufacturing, e.g. for the memory chassis there were 32 copies to be made so this was a reasonable way of proceeding. The machine was turned off round about 1957. The machine was designed by Bill Renwick with the assistance of Maurice Wilkes. Maurice Wilkes went to the Moore school where he learnt about the EDVAC design and he came back and headed the project to build the EDSAC. He deliberately under-engineered it. In other words - an example of this - the pulses were 2 microseconds long rather than 1 microsecond to make it easy to make. He believed, correctly, that he would get great experience by using, rather than making, a highly refined computer.

Unique id/year of acquisition: CL-4/97
Name: EDSAC I delay line

Other No marked on object: none
Inscription: none
Dimensions: 274x50x50
Description: Brass tube in several parts mounted on a base, central section of different metal. 5 screws protrude from top and 3 from both sides. Screw fixings at ends.
Class: memory
Machine name: EDSAC I
Condition: good
Notes: DW: I don't quite know what to say about this. I did write some notes. I'll try and audio describe it but if you want to include the principles and how it worked which you probably do it may take a little time. This is a mercury delay line. It was used for memory in EDSAC. The main memory consisted of two batteries of 16 long tanks 6 feet long. This is a delay line which was used for holding one word of about 35 bits. The way that mercury delay lines worked is roughly as follows: There was an oscillator which drove a transducer which is a quartz crystal at one end. The sound waves travelled all the way along and were picked up by quartz transducer at the other end. The length is such that you can have 35 pulses possibly here between the beginning and the end. There was a circuit in the chassis the like of which we've just seen which took the signal coming out. It amplified it and converted this into a voltage pulse. The voltage pulse was then Anded with a clock and possibly sent back or possibly, according to control wave forms some new information was fed in, so it could hold the information as long as you pleased. There was some trouble in the design of these. The attenuation was so small on the ultrasonic pulses that it was possible for a pulse to go from one end to the other to be reflected and reflected again and come back and cause interference and there is a little screw at the top which screwed into the mercury to increase the attenuation to make this not happen. The longer six foot tanks did not need the increased attenuation. The tube was filled with mercury which happened to match the characteristics of quartz crystals,so there was a good impedance match which was why mercury was chosen and the possible pulses were sent one every two microseconds and there were-about 35 of them in this and then they were sent back. The pulses were about one microsecond long but they were held on a carrier of 13 and a half megacycles very much like information is carried on a radio signal and therefore the receive end included a detector to change the 13.5back into a voltage pulse. It was designed with the aid of - name forgotten. Bill Renwick had worked on ultrasonic delay lines during the war. These were used for echo cancellation and so the technique was fairly well known. Bill Renwick had been involved in that but we had assistance from an Austrian whose name I've forgotten, and those two collaborated in making the design and there it was. There were some side issues like should you control the temperature or should you keep the frequency fixed. It was originally intended to let the frequency wander and track the temperature that way, but finally a thermostatic box was used to hold all of these to keep the boxes in constant temperature.
Q: Was it made in Cambridge?
DW: Yes, I cannot remember if it was made in our own workshops or in the engineering lab workshops but it was made in Cambridge University.

Unique id/year of acquisition: CL-25/97
Name: EDSAC thermostat

Other No marked on object: 545967 (Patent number, on scale inside tube)
Inscription: "T.D.M." (on rear of scale inside tube)
Dimensions: 383x33x33
Description: Thermostat, mercury, used to control temperature of delay lines in EDSAC. Device is a mercury thermometer, in a clear glass tube with white plastic scale inside, to which has been added at the top an electrical thermostat in black plastic casing, from which there are two yellow wires. The thermostat is attached by tape and is not secure.
Class: other
Machine name: EDSAC I
Condition: fair
Notes: DW:This object is a thermometer - actually it's a thermostat. It was made from a thermometer of mercury type and at the appropriate height of mercury two circuits shorted and it was used to control the temperature of the mercury delay lines in EDSAC. The delay lines were put in a coffin-like box, the thermometer was inserted and the thermostat arranged to keep the temperature constant. So the number of pulses in each delay lines remained invariant.
Q: This looks sort of quite home-made really - this was something just built in the Lab?
DW: This was built in the Lab. Standard thermostats for house equipment didn't exist then. But this was a standard thermometer into which two contacts were put at a particular point. Of course it really didn't matter where as long as you controlled the temperature. You arranged the position of the wire so the temperature was right for what you wanted, but you are quite right - it was made in the Lab by hand.
Q: Is this then typical of EDSAC that you would sort of cobble together these things in the Lab because nothing existed ?
DW: The modern things didn't exist - almost everything was done by hand like there were no automatic wiring things - all wiring was done by soldering, the valves were replaced when they were broken, by hand. The mechanical tape readers, punches and so on were constructed in the Lab or in fact the punches were converted from a Creed punch to electrically driven, rather faster affairs. The tape reader for the EDSAC 1 was a photo electric one unlike the later one which goes back 30 or 40 characters a second and was mechanically driven through its sprocket hole.

Unique id/year of acquisition: CL-63/99
Name: EDSAC I Valve lifter

Other No marked on object:
Inscription:
Dimensions:
Description:
Class: other
Machine name: EDSAC I
Condition:
Notes: A device, made in the Mathematical Laboratory, for lifting hot valves during operation.

Unique id/year of acquisition: CL-64/99
Name: EDSAC I paper tape (various)

Other No marked on object:
Inscription:
Dimensions:
Description:
Class: other
Machine name: EDSAC I
Condition:
Notes: A program written by L.Slater. It calculated the mathematical constant $e$ as 2.718281, dated March 1951.
See also: CL-81/99
See also: CL-96/99
See also: CL-100/99
See also: CL-126/99

Unique id/year of acquisition: CL-65/99
Name: EDSAC I tape splicer

Other No marked on object:
Inscription:
Dimensions:
Description:
Class: other
Machine name: EDSAC I
Condition:
Notes:

Unique id/year of acquisition: CL-66/99
Name: EDSAC I horn

Other No marked on object:
Inscription:
Dimensions:
Description:
Class: other
Machine name: EDSAC I
Condition: fair
Notes: Don Hunter (Research Assistant 1949-51) states
"I still have a ribbon (tweeter) loudspeaker horn calculated on EDSAC because I did not know the closed form of the solution giving the shape of the sides. These were bent on the metal rolling machine by Bill Renwick, presumably during the lunch hour."

Unique id/year of acquisition: CL-82/99
Name: Assorted components

Other No marked on object: none
Inscription: Various labels and leaflets on boxes
Dimensions: various
Description: 3 large valves, 6 smaller, misc. unidentified components, some large ferrite cores.
Class: other
Machine name: EDSAC I
Condition: good
Notes: 3 large valves, 6 smaller, misc. unidentified components, some large ferrite cores.

Unique id/year of acquisition: CL-86/99
Name: EDSAC Z-function results
Photograph not applicable

Other No marked on object:
Inscription:
Dimensions:
Description:
Class: documentation
Machine name: EDSAC I
Condition:
Notes:

Unique id/year of acquisition: CL-96/99
Name: EDSAC I tape spools

Other No marked on object:
Inscription:
Dimensions:
Description:
Class: other
Machine name: EDSAC I
Condition:
Notes:
See also: CL-64/99
See also: CL-81/99
See also: CL-100/99
See also: CL-126/99

Unique id/year of acquisition: CL-126/99
Name: paper tapes (2 boxes)

Other No marked on object: none
Inscription: none
Dimensions: n/a
Description: Two boxes of many 5-hole paper tapes. Tapes are mostly white but some blue, green or red. Some are in boxes with labels, eg H5, F2 MASTER etc. Tapes also appear all to have written labels
Class: other
Machine name: EDSAC I
Condition: good
Notes: Codes refer to library subroutines. coloured ones later, some may be EDSAC II
See also: CL-64/97
See also: CL-81/99
See also: CL-96/99
See also: CL-100/99

Unique id/year of acquisition: CL-141/00
Name: Core tester

Other No marked on object: none
Inscription: none
Dimensions: 292x292x35
Description: Case consists of grey metal back and sides with perspex top. Open on one side. Inside case is a square sheet of brown plastic. There are a number of holes of varying sizes in the middle and a double set of solder connections around all 4 sides. Thin wires are fixed to some of the solder connectors and form a mesh going through some of the holes. There are 2mm diameter ferrite cores at the intersection of some of the wires on one side of the sheet.
Class: memory
Machine name: EDSAC I
Condition: fair
Notes:

Unique id/year of acquisition: CL-142/00
Name: Engineer

Other No marked on object: none
Inscription: none visible
Dimensions: 1800x450x300
Description: Tall humanoid
Class: other
Machine name: EDSAC I
Condition: fair
Notes: Answers to Richard, Dick or Dickie

Unique id/year of acquisition: CL-149/00
Name: EDSAC coding pad
Photograph not applicable

Other No marked on object: none
Inscription: UNIVERSITY MATHEMATICAL LABORATORY, CAMBRIDGE EDSAC PROGRAMME SHEET
Dimensions: 345x214x8
Description: Cardboard backed pad of paper forms, each of which has a printed grid with numbered rows and columns headed Order and Notes
Class: documentation
Machine name: EDSAC I
Condition: good
Notes: DW: These programming pads were used for EDSAC 1. Maurice [Wilkes] had a large number printed and they were in common use. Programs were made of a main program and subroutines, each starting with a fresh sheet. The numbering started with zero on each set of sheets,the directive GK resetting the references. Every tenth location was usually filled in. This meant that most references were local and simple faults did not lead to renumbering the entire program. A typical way of programming was to first write the function letters known addresses. A second scan would enable the programmer to fill in all the local references.
After a program was complete it would be taken to a keyboard punch to make paper tape (5 hole). The separate parts of the program would be copied by a tape reader tape punch combination onto a single tape including the needed library routines. The results were often checked using a tape comparater which had two tape readers and stopped when the tapes were different. As the tapes were translucent, some people compared the tapes by superposing in alignment, holding up to the light and looking.
This would then be taken to the computer to be run. The results were punched on a paper tape and printed by an offline reader-teleprinter combination.
See also: CL-1/97

Notes

• Identifiers are written on objects in the format CL-uuu/yy/gg (u = unique identifier, y = year of acquisition or entry into database, g = group id. The group id is only used if there are more than one of a given object.) eg 80/99/3. This format was adopted to make the identifiers from this exercise distinct from any used in previous cataloguing attempts.
• Photographs. Where relevant the database contains a comma-separated list of Unix path names of pictures of the object, in jpeg format. If photos are available a thumbnail image will be shown as a search result. Click on this to see the full size image.
• Date format: dd.mm.yy
• Name format: Full first name (where known), else initial and surname
• Dimensions: max x next x min (all in mm)
• Lists in all fields are comma separated.
• n/a = not applicable
  blank = no information entered
  ? = information uncertain or unknown
  The database search allowed by this page is selective and will only display information that is present, i.e. it will not display labels for empty fields.
• Not all information in the database is displayed on the web page produced by this search facility, some of it is private (eg location of object, for security reasons).
• The Notes section of the database may contain html formatting commands to allow it to be displayed to best advantage on the web pages output by this search facility.