Edmund Berkeley's

Relay Processor

The cover of the October 1950 issue of Radio-Electronics magazine featured a picture of a desktop-sized relay-based "electric brain" called Simon. The issue contained the first of a series of 13 articles presented over the next year about the principles of digital automated computing. Edmund Berkeley, the principle author of the articles, had also just written a book "Giant Brains and Machines that Think" (Wiley & Sons, 1949).

If you write a program for Simon to run in the simulation, please submit it for inclusion on these pages.

See the Programs for Simon page.

The design of Simon would be discussed in the articles and the book, however the Radio-Electronics articles do not present a complete technical description of the machine. There is some suggestion that additional plans filling in missing details of the design were made available, but I have not seen them. These pages attempt to rectify the shortcomings by presenting:

Most of the the material here has been deduced from the articles, some of it is new, and the occasional presumption is made about how Simon may have been implemented.

Contents (this page): Sub-pages: Related pages:

 Historic Significance

Simon was likely the first attempt at producing a minimal or affordable automatic digital computer for educational purposes. Several large computer construction projects had been completed or were underway at the time but only the cognoscenti had access to them. Simon would introduce the principles of binary arithmetic, logic and automatic computation to a wider public.

 Brief Technical Overview

Simon is a Harvard architecture machine constructed from relays. The program is executed directly from paper tape. In these regards it was behind the state-of-the-art as several electronic stored-program machines were under design or construction by the late 1940s. The main data registers and ALU are 2 bits wide, data input is via either the paper tape or 5 toggle switches, output is via 5 lamps.

As an educational instrument, Simon was directed more towards the electrical implementation of logic than towards programming. As such, and as a minimal machine, programmibility is rather limited. The one saving grace may be that the program can be quite long (limited only by paper tape handling), a feature that certainly would not have been feasible in an attempt to make an inexpensive stored-program machine at the time.

The name "Simon" came from the Mother Goose rhyme "Simple Simon", the designers freely referring to the machine as "the little moron".

 Simon and the Radio-Electronics Articles

The Radio-Electronics "Simon" Articles
Month Page Part Title & Description
1950 Oct 29 I
World's Smallest Electric Brain
Introduction to Simon and relay logic.
1950 Nov 44 II
How an Electric Brain Works
Binary addition using relays.
1950 Dec 35 III
Relays Do Simple Arithmetic
Binary subtraction and multiplication using relays.
1951 Jan 70 IV
How an Electric Brain Works
Binary division using relays, decimal-binary and binary-decimal conversion.
1951 Feb 70 V
How an Electric Brain Works
Limited discussion of programs, control and timing/sequencing.
1951 Mar 56 VI
How an Electric Brain Works
Some description of the physical layout of Simon.
1951 Apr 54 VII
How an Electric Brain Works
Machine cycle timing and instructions of Simon.
1951 May 64 VIII
How an Electronic Brain Works
Electronic flip-flops, delays lines, drums, electrostatic storage.
1951 Jun 38 IX
How an Electronic Brain Works
Electronic (vacuum tube) logic circuits.
1951 Jul 56 X
How an Electronic Brain Works
Bit-serial multiplication and division.
1951 Aug 48 XI
How an Electronic Brain Works
Some timing and control issues.
1951 Sep 45 XII
How an Electronic Brain Works
Minor discussion of code conversion logic, computer overview and instructions.
1951 Oct 53 XIII
How an Electronic Brain Works
Overview of SEAC construction.
(Sorry, but scans of the articles are not presented here.)

The 13 Radio-Electronics articles are broadly broken into two sets, the first 7 articles covering "electric brains" and the final 6 covering "electronic brains". The final 6 articles are not relevant to Simon.

Even within the first 7 articles, much of what is presented is not applicable to Simon. Most of articles III and IV are devoted to binary multiplication and division with relays, operations which it later turns out are not implemented in Simon.

Inconsistencies, ambiguous references and missing details can leave one perplexed as to Simon's design and capabilities. Some diagrams in the articles show registers and data paths 4-bits wide, while the text states that they are 2-bits wide in Simon. Large portions of the circuitry are never shown in detail. There is no overall functional or block diagram of the machine and no complete schematic. Even the instruction encoding is not fully described.

Fortunately, the timing of the machine cycle of Simon is covered in some detail in the articles, being very helpful to deducing aspects of the machine.

I initially ran across the Simon articles a few years ago when I was sorting through some boxes of early R-E magazines. While they certainly caught my attention I confess I was disappointed. The authors do admit to the incompleteness of the Simon description but such mention is left for the final paragraphs of the 7th article.

After studying the articles more closely however, teasing out as many details of Simon as possible, and going through the exercise of creating the machine definition, the ingenuity of the design becomes apparent.

For amusement, I just have to mention the tortured attempt by the authors to bring Cold War politics and patriotism into a technical article. Witness this extract from the first article (note: the reference to a telegraph system is from an analogy being made between a computing system and a telegraph system earlier in the article):

[ Extract from Article I (Radio-Electronics 1950 Oct) ]

There are two ways to get the central traffic control to function properly. One is to have all the orders ready ahead of time, and tell it to do just as it is told. This is dictatorship. The second way is to have some special wires of the telegraph system run into the central control, and let information from time to time (though not all the time) come from the system into the central control - feedback. The central control then knows what is going on and can direct the steps. This is democracy. This second technique is of course a honey, even with electric brains, and a good electric brain does compute some - or even most - of its own instructions.

As a programmer of Simon one could feel much more as though one were working under the formerly mentioned system than the latter.

(Another digression: If one happens to have the magazine issue containing the first article, take note on page 33 of the photo of Jay Forrester with a holding-beam storage tube for Whirlwind.)

 References, External Links

For a stored-program relay computer see:

  Machine Definition | Relay Implementation | Programs | Simulation | Alterations
2006 Sep