top of page

Anatomy of an Automatic Telephone Exchange

Two years after A.G. Bell patented the telephone, the first manual switchboard went live in New Haven, CT, 1878. It was spearheaded by George Coy, a telegraph office manager.  Manual boards proliferated and by 1900 there were ~20,000 operators in the domestic US market plus more worldwide.  

 

Creative engineers began looking for ways to automate calling and reduce the need for operators 24x7. Over the period from 1878 until about 1960 they developed a wide variety of automated exchange architectures and methods using electromechanical relays and inventive switching gear.

 

All exchanges strive to do the same thing; connect any two phones with the minimum of cost and equipment while providing near 100% uptime with low maintenance. No single exchange can serve the world so they must be interconnected using trunk lines (talking paths) to create a holistic network. 

 

The first ever automatic exchange was very modest and became the forerunner of the worldwide telephone network. Almon Strowger, an icon, installed the world's first automatic system. Here is an overview of his La Porte system 

Truly, the automatic telephone system seems the very high-water mark of human creative power.''

                -- Dr. J. A. Fleming, eminent British electrical engineer, circa 1920.

 

The coverage to follow provides a “Gray’s Anatomy” overview of automatic exchanges (also called central offices).  With the development of many different exchange types and associated switching devices over an ~80-year period, a learner can quickly become overwhelmed by the trees and miss the forest.  The approach taken here is to explore, at a high level, the major system parts and their relationships (the forest) then focus on the implementation details (the trees).

 

Here is a simple view of an exchange system with lower-level functions not shown. The switch example is generic and not a specific type. 

Simple telephone exchange diagram showing call control and the main switch

Fig 1

Abstract exchange diagram

 

Assume subscriber #288 goes off hook. They receive a dial tone, and the “Call Control” registers the digits dialed (calling #315). For this example, after digit 3, Call Control creates a connection (red dot) so that phone #288 connects to phone #315. See Endnote 4.

 

The engineer's challenge is to design the most cost effective and reliable, call control and switch components.  Let's first examine the switch component. If this is old news for you, skip ahead

 

 

The Switch Part

 

Assume a single exchange with 10,000 subscribers. Assume too that half the subs desire to speak to the other half at the same time. This would require a huge 100 million (10K x 10K) point switch. This is not a realistic scenario for many reasons.

 

Knowing the subscriber calling rate and conversation hold time allows designers to make wise switching tradeoffs.  For example, if the expected traffic in a small exchange was 30 callers per hour each holding for 5 minutes, this provides guidance for the exchange switch designer. 

With clever use of multiple smaller 100-point switches (for example), engineers can build a switching network to meet all the realistic traffic needs for an exchange. Using interconnected smaller switches (not one huge switch), reduces cost, saves space and meets the needs of real-world telephone usage.

 

The diagram below breaks down the single (could be massive) switch in Fig 1 into three stages. This is a general pattern for most exchange types. Authors writing about this switching plan describe the central stage as a "switching network" or "distribution stage". For this coverage Central Switching Fabric (CSF) will be used. This division of labor allows for more flexibility in traffic routing as we will see below. 

 

The composition and size of each stage vary considerably and depend on the exchange technology used. For example, a 2-digit PBX  (Private Branch Exchange) for a business may not have the center CSF stage. Examples to follow will add clarity. 

 

The Call Control part is not shown here but is vital for switch control. Its operation is covered later in this section.

Anatomy Fig 2

Fig 2

Switching plan for a telephone exchange

After [Chapuis] and [ATT1]

 

Upon examining the components, we observe:

Subscriber lines/phones, up to 10,000 per exchange. See Endnote 1.
 

The Concentration stage, comprising switches, links a new caller to the Central Switching Fabric (CSF) even before the first number is dialed.

The Expansion stage, comprising switches, works like the inverse of the concentrator. The outputs of the CSF are inputs to the Expansion stage. For some designs this stage may route only the last 2 digits of a 7-digit number. 
 

The Central Switching Fabric (CSF) stage is a distributed mesh of many smaller switches and routes between its input and output ports. Importantly, "trunks" (conversation paths to/from other offices) are also switched here.  It is at the core of the network. See more on trunks here. 

The CSF normally routes/switches 3-7 digits of a 10-digit phone number. This stage can become complex since it routes calls intra office, to/from offices in the same city and distant offices. See the popup for more Fig 2 insights. 

 Exchange Anatomy

 

The diagram below (not displayed on mobile devicesis segmented into three sections: top, middle, and bottom.  The top section outlines the fundamental components for any exchange. These elements combined become the framework for an automatic exchange.

  • Switching devices: This section delves into various inventions over time. While some did not gain traction, others achieved global acceptance. Fifteen types are explained. 

  • Fabrics: Using many smaller switches to create an equivalent large switch (the CSF and other types).

  • Relays and Logic: With ~50-60K relays in a large exchange, they are key to its operation.

  • Call Control:  The brains of the outfit. This orchestrates a call's progress including instructing the switches to route calls based on the dialed number. Two methods are used: progressive direct dial control and common control

  • Supporting Actors: structural framing, wiring, ringing/tone generators, power systems, maintenance/testing and more.

The middle section shows specific exchange categories: Panel, Step-by-Step, Rotary, and Crossbar offices. These are the Big Four. The "Exchange Floor Plans" box provides  a few simple exchange floorplans for 1, 2 and 3 digit systems. Not every conceivable exchange type is shown. Importantly, there are many varieties of each genre. See Endnote 2.

 

The bottom section illustrates a hypothetical representation of a domestic telephone network. The boxes labeled TE are “in series” Tandem Exchanges and route callers between end offices. They use similar technology and techniques as an end point exchange but do not connect directly to subscribers. 

 

A full-featured central office may have up to 10K lines. With local trunks and toll interconnecting trunks to Tandem Offices, it’s easy to appreciate how one phone can potentially reach any other phone on earth. See the exchange problem and Endnote 3. 

Click/select any box in the diagram below for a perspective on how the trees align to create a forest. Generally going from top to bottom is best but starting with Exchange Floor Plans then Switching Devices (15 different switches explored) will give you a solid foundation. 

Anatomy master diagram

     Traffic
Engineering

Switching
Fabrics

Exchange types

Switching
Fabrics

Exchange Anatomy anchor
Master hierarchical telephone exchange diagram

San Francisco

New York City

Public Switched Telephone Network

These are the menu choices when using a mobile device.
Explore any topic with images and videos.

Switching Devices

Call Control
Switching Fabrics      
Relays & Logic
Supporting Actors

Exchange plans     
Traffic Engineering       

Unified Telephone Network

Exchanges: 
Step-by-Step       
Rotary
Panel
Crossbar
 
Rotary Phones 

Endnotes


Endnote 1

For a 7-digit phone number, the first 3 digits are the “Office Code” and this is the exchange ID. So, if the called number is 552-1677, the 552 digits identifies a specific exchange (like the KLondike exchange in San Francisco in 1970’s).  The 1677 subscriber is supported by that exchange. A large city exchange handles the last 4 digits of the called number. This is up to 10K subscribers. 

 

Adding a 3-digit area code (415 for SF) enables 999 (many fewer in practice) exchanges for that area. Adding a “Country Code”, like 44 for the UK, adds yet more reach to cover the world.

 

Endnote 2

The Big Four exchanges are all named after the core switch for that exchange. That is, a Crossbar exchange is based on crossbar switches. Exchange technology varied widely over an 80-year period and between countries too. The Big Four took the majority of installations worldwide. 

 

Endnote 3

Both the Bell System (and its manufacturing arm, Western Electric) and Independent Telephone Operators co-existed for many years. By 1912 there were 5.1 million Bell System phones and 3.6 million independent company telephones. The proportions of independents decreased progressively until about 1940 due to Bell acquiring the independents [Gabel].

 

Endnote 4

A.G. Bell had a vision for a grand system, a worldwide telephone network. From a company prospectus he is quoted in 1877;

 

I believe, in the future, wires will unite the head offices of the Telephone Company in different cities, and a man in one part of the country may communicate by word of mouth with another in a distant place… that all present arrangements of the telephone may be eventually realized in this grand system.” [Craft]


References

 

ATT1: Switching Systems, Chapter 4, 1961, AT&T

Chapuis 100 Years of Telephone Switching (1878-1978) Part 1, Robert Chapuis1982

Gabel: The Early Competitive Era in Telephone Communication, 1893-1920, Richard Gabel, Law and Contemporary Problems, Vol. 34, No. 2, Communications: Part 1 (Spring, 1969).

 

Craft: E.B. CRAFT, and others, Machine Switching Telephone System for Large Metropolitan Areas, Bell System Technical Journal, April 1923.

bottom of page