The ICL 7502 - The Cabinet_

The ICL 7502 was designed for unattended operation in an office environment, so it was small and quiet and could be housed in special, wood-veneered furniture with matching accessories for all system units e.g. VDU and 132-column matrix printer. The system housing had two levels, giving space for a second system or a floppy-disc unit, and was lined with dense acoustic foam to soak up the noise of cooling fans. All cabling was designed to enter the rear of the housing to maintain the smart and orderly facade. In the picture below, a telephone is included, but it is a con because there is no nearby modem! Note the 8-inch floppies for scale.

7502/5 Enclosure with dual 8-inch diskette drive
7502/5 Enclosure with dual 8-inch
diskette drive.

The processor unit was about 17 inches square and 8 inches high, with a detachable cosmetic front grille with an internal dust filter and a hinged door 4 inches square which carried the operator controls, such as they were. Early versions had two rows of chromed lever switches on the door, while the later build had one row of switches and a 16-position rotary switch for selecting the system options at load time. Please note that the Edinburgh picture is of an early 7502/3 system, while above is a later 7502/5. A row of 4 LEDs gave a minimal diagnostic display, with adjacent LED indicators for PSU voltages. Lever switches were used to select modem control functions, and the leftmost, slightly larger than the others and with a flattened dolly, was biased to its centre position and labelled GEN RES for a complete system reset when lifted, and ACTIVATE when depressed. GEN RES would do what it said on the packet, but as it was the only way of controlling the destiny of the system, this caused entry to the ROM, wher the code that was executed was just a tight loop apart from one switch

7502/5 Front panel drawing
7502/5 Operator's controls diagram.

The ROTARY SWITCH communicated the user's startup options to the bootstrap loader contained in the ROM, selecting local or teleloading or diagnostic operation as follows (positions 11-15 undefined):

  • 0 - Engineer Functions and 7502/3 emulation
  • 1 - Modem Loopback Test
  • 2 - Restart
  • 3 - Diskette Load
  • 4,5 - Teleload Request A or B (7181)
  • 6,7 - Teleload Request A or B (XBM)
  • 8 - Diskette Dump
  • 9 - Teledump Request (7181)
  • 10 - Teledump Request (XBM)

The FAULT lamp indicated that the processor had lost the plot completely, and a system reset had been forced by the WATCHDOG TIMER not being satisfied. Other controls and indicators were labelled as follows:

  • POLLING - Modem Tx Data
  • REC DATA - Modem Rx Data
  • DSR - Modem Connected to Line
  • DUPLEX - Full or Half
  • STANDBY - Put modem on standby
  • DATA RATE - High or Low
  • DATASET - Voice or Data
  • Lamp 0-3 Codes
  • 0 - OK
  • 1 - Option not available
  • 2 - I/O device inoperative
  • 3 - Modem loopback fail
  • 4 - Store clear error
  • 5 - Diskette data error
  • 6 - Diskette not dump enabled
  • 7 - Store error on load
  • 8 - ROM Sumcheck error
  • F - Operation complete
7502/3 with 7561/1 VDU
7502/3 System (left on table) with 7561/1 VDU
and diligent Edinburgh Student.
©Edinburgh Computer History Project

A grey-coloured cosmetic sleeve covered the top and sides of the processor unit; when it was removed the fold-down rear panel could be released, and two large screws undone to release the PSU which would hinge upwards to rest on the top of the cabinet. Access could then be gained to the rear of the backplane. The rear panel carried a large number of DIN-type sockets for 8 VDUs and 4 serial printers, a male and female 25-way D-type connector for modem input and daisy-chain output and a 34 way DIL socket for the diskette interface. There were also a set of 8 toggle switches for the engineer's functions, which were also interrogated under normal operation to select a TERMINAL EXECUTIVE to TELELOAD at certain positions of the rotary switch.

Logic boards were inserted from the front and lay horizontally in nine slots. The positions were not identical despite the bus structure of the backplane. The 7502 was based on an implementation of a 'slow' X2 HIGHWAY, which was also found in units of the 2900 mainframe systems. The 2900 DEVICE CONTROL UNIT (DCU) had a fast end and a slow end for different peripheral access characteristics, which meant that the 7502 was able to share COUPLERS for some big-system peripherals; cryptic versions of the 7502 were found in several areas of the 2900 such as the OPER terminal. In the 7502 backplane the X2 HIGHWAY was on the right-hand side connectors, while the function-specific signals were carried on the left-hand side. This avoided the need for any boards to have connections to their front edge, but meant that cabling to the peripheral connections was specific to particular board positions. In addition, the topmost position was only half-width and could only be used for memory boards which had no peripheral connectivity. The PROCESSOR board always occupied the topmost full-width slot followed by the first STORE board, with VIDEO STORE boards filling slots towards the bottom of the backplane. The lowest slots were the home of the principal I/O boards which connected to keyboards, serial printers and the modem. Where a diskette drive coupler was employed it was forced to use the slot above the first I/O board. It can be seen that there was competition for space and there were significant constraints on how many VDUs could be connected to a system with diskette dives, for example.

The VDUs were composite video monitors with no direct connection to the keyboards they accompanied. Flyleads from the monitor and keyboards were attached at GPO type junction boxes to wiring which could be up to tens of metres to the back of the 7502, so a cluster of terminals could be quite widely spread about a building. The junction boxes were a nightmare because the miniature coaxial cable for the VDU used a very delicate single centre-conductor which was virtually certain to snap as you tightened the screw terminal.

Reference: ICL Technical Publication TP4804 7502 Operating (1976)
Acknowledged Source: Brian Spoor's ICL 1900 website Link