What a switch does
A telephone switch connects one line to another for the length of a call. Early on, a person did it by plugging a cord into a jack. A machine has to do the same three things: notice that a caller wants service, learn which line is wanted, and join the two lines, then undo the connection when the call ends. The history of switching is the history of finding better ways to do that.
Step-by-step switching
The earliest widely used automatic switch was the step-by-step, or Strowger, system, named after Almon Strowger, who patented the basic device in the early 1890s. Its key part is a selector with a bank of contacts arranged in rows and levels, a bit like a curved panel of many small contacts. When the caller dials a digit, the dial sends a series of electrical pulses, one pulse for each number on the dial (ten for zero). The pulses make the selector’s arm step up, level by level, to the row matching the digit. It then rotates horizontally to find a free contact in that row.
With several stages in a row, each controlled directly by one digit, a call reaches the wanted line one step after another. The first digit selects a group of lines, the next digit chooses a subgroup, and so on, until the last digits choose the individual line. This is where the phrase “step by step” comes from.
The system was simple and durable, and each call was set up by its own dialed digits with no central brain. It was also noisy and mechanical, and each call tied up a chain of selectors. Parts wore out and required continual adjustment. Even so, step-by-step exchanges were built in large numbers across many countries and some remained in service for decades.
Panel and rotary systems
Larger cities needed switching that could handle huge numbers of lines. The Bell System developed the panel switch in the early twentieth century, which used motor-driven rods that moved up and down through banks of contacts. Unlike step-by-step, a panel system stored the dialed digits in a register, then used the stored number to direct the switching. This idea, separating “remembering the number” from “making the connection,” was important. Other systems, including rotary designs used in Europe, followed similar ideas.
Crossbar
The crossbar switch, which entered service in the late 1930s, used a different approach. Its main component is a grid of horizontal and vertical bars. When a control circuit energized a particular horizontal and a particular vertical magnet, the crossing point closed, and the line was connected through that crosspoint. A crossbar switch is much like a grid of intersecting wires with a switch at every junction.
The main advantage was in control. A shared control circuit, often called a marker, could look at the number dialed, check which paths were free, and choose the best route, then set up the connection quickly and release itself for the next call. Crossbar switches were more reliable and quieter than earlier designs, and the control logic could be shared among many calls. Crossbar exchanges became a standard in many countries after the Second World War.
Stored-program control
The next change was to replace the wired logic of the control circuits with a computer. Instead of relays that could only do one job, a processor running a stored program could decide how to route a call, and the program could be changed without rebuilding the hardware. The Bell System’s No. 1 Electronic Switching System, introduced in 1965 in Succasunna, New Jersey, is the famous early example. It still used metal contacts to connect calls, in the form of tiny sealed switches, but the control was electronic.
The change made new services easy to add. Features such as call forwarding, call waiting, speed dialing and conference calling were largely software and data. It was a case of the same logic that drove early electronic computers being applied to the telephone network.
Digital switching
The following step was to switch the voice itself in digital form. Digital switches convert each call into a stream of numbers, as pulse-code modulation does, and move them through memory and time slots rather than through physical contacts. Instead of dedicating a distinct set of wires to a call, a digital switch can share a path among many calls by dividing time into small slots, a method called time-division switching.
Digital switches appeared in the 1970s and became widespread in the 1980s. They were made with integrated circuits and silicon chips, were smaller and used less power than electromechanical switches, and could be diagnosed and updated by software. A modern central office may hold what once would have filled a large building.
Circuit switching and what came after
All of the systems above use circuit switching: for the length of the call, a path is reserved between caller and callee, whether physical or in time slots. Computer data networks use a different approach, packet switching, in which messages are broken into pieces that share links and are reassembled at the destination. Packet switching is what lets wide area networks share links among many users. As internet-based voice services became common, many telephone calls began to travel as packets for at least part of the route, with software “softswitches” running on ordinary servers taking over functions once done by dedicated switches.
A view of the whole story
Each generation of switching reduced the human and mechanical effort per call. The progress can be described as a shift in where the intelligence sits: in the operator, in the dial and its selectors, in a shared control circuit, in a computer program, and now in software that runs on general-purpose hardware. The basic job of connecting two people stayed the same.