After Step-by-Step — Telephone Switching History

Figure 1 — A Western Electric crossbar switch from a No. 5 Crossbar office. Horizontal select bars and vertical hold bars close spring contacts at whichever crosspoint they share. There are no wipers, no bank… — Figure 1 — A Western Electric crossbar switch from a No. 5 Crossbar office. Horizontal select bars and vertical hold bars close spring contacts at whichever crosspoint they share. There are no wipers, no banks and no sliding contacts, and nothing moves more than a fraction of an inch. — Unknown photographer, via Wikimedia Commons (public domain)

7.1 The Limits of Direct Control

Step-by-step’s great virtue is that it needs no brain. Each switch obeys the dial pulses as they arrive and then hands the call on. Its great weakness is the same fact. Because the switches act on the digits as dialled, the telephone number is the route: the first digit picks a level on a first selector because that level is physically wired to the trunks for that group of lines. That has three consequences.

Numbering is frozen into the wiring. Renumbering a district, or reaching another office by a code that does not match its trunk wiring, means rewiring banks.
Large cities don’t fit. A city of many offices, each reachable from every other, needs office codes that bear no relation to which selector level leads where.
Every switch must be fast enough to follow the dial, and each stage is held for the whole call, whether it is doing anything useful or not.

The Bell Laboratories switching history summarises the century in one sentence: “The step-by-step system was directly controlled from the customer’s dial; the panel system introduced indirect control enabling the selectors to be independent of the customer’s dial; and the crossbar system enabled common control by high-speed markers” (BTL Switching Technology, ch. 5). This volume follows that progression to the point, in 1976–1982, where digital switching took over. It is told more briefly than the step-by-step story, because the rest of this dive is about step-by-step.

Figure 2 — Five generations of switching control. Only step-by-step lets the dial drive the switches directly. Every later system puts common equipment in between that stores the number, works out a route and… — Figure 2 — Five generations of switching control. Only step-by-step lets the dial drive the switches directly. Every later system puts common equipment in between that stores the number, works out a route and then sets up the path, first in relays, then in a stored program, and finally in a digital computer switching speech as numbers. — Original diagram drawn for this dive

7.2 Panel: Bell’s Big-City Machine

Around 1906 AT&T set up two Western Electric development groups, one working on a rotary system and one on a linear panel system. In 1912 the panel was chosen for the large American cities. Its essential idea came from E. C. Molina’s translator (US 1,083,456, 1914; the idea dates from 1905): the caller’s digits are stored in a sender, translated, and then used to drive the switches at the switches’ own pace (Google Patents). The switches themselves were tall frames of flat contact banks swept by brushes on vertical rods, driven up by a continuously running motor through electromagnetic clutches. They were power-driven rather than stepped by the caller’s pulses.

The first trials were semi-mechanical, with operators keying the numbers: Newark’s Mulberry office (3,640 lines) on 16 January 1915 and Waverly (6,480 lines) in June 1915 (BTL Early Years). The first fully automatic panel offices opened at Omaha, Nebraska, in December 1921, and New York’s Pennsylvania office followed in October 1922. The lettered dial (“PEN 5000”) belongs to this era. Panel was never adopted outside the United States. The last Bell System panel office was retired in 1982 or 1983: the Bell Laboratories history, written in 1982, says 1982, while Western Electric Magazine reported Newark’s Bigelow office as “The Last Panel Office” in its first issue of 1983.

Figure 3 — A panel line-finder frame at the Connections Museum, Seattle. Selector rods carry brushes up the face of flat multiple banks, driven by power through clutches rather than stepped by the caller's dial. — "Panel Switchman", via Wikimedia Commons (public domain)

Figure 4 — A panel office outgoing-trunk test desk. Commons gives no location or date. Common-control offices were maintained from test desks like this one, which could reach into any trunk and observe it. — "Panel Switchman", via Wikimedia Commons (public domain)

7.2.1 Rotary, for Europe

The other Western Electric system, Rotary, went to the company’s international arm in Antwerp and became the 7A family, adopted in France (from October 1912), the Netherlands and New Zealand (1913–14). In Britain a Western Electric rotary exchange opened at Darlington in October 1914, before the Post Office standardised on Strowger (Wikipedia, Rotary system). ITT bought International Western Electric in 1925, and Rotary went with it.

7.3 Crossbar

The crossbar switch has two inventors on two continents. J. N. Reynolds of Western Electric invented a crossbar selector in 1913 (US 1,139,722, 1915) (Google Patents). In Sweden, the engineers Betulander and Nils Palmgren patented the links-and-markers principle in 1912 and applied for a crossbar switch proper in 1919. The Swedish Telecommunications Administration put its first large crossbar station into service at Sundsvall in 1926 (Ericsson history). The word “marker” was Betulander’s before the Bell System used it.

A crossbar switch is a grid. A set of horizontal select bars, each able to rock up or down to pick one of two rows, crosses a set of vertical hold bars, one per path through the switch. To close the crosspoint at row 4 of vertical 6, the control equipment operates the select magnet that rocks bar 2 toward row 4, then operates hold magnet 6. The hold bar catches the fingers at the crosspoint and closes its spring contacts. The select bar is then released for the next call, while the hold magnet keeps the path closed until hang-up. Each motion is short, and the contacts are precious-metal springs pressed together rather than wipers sliding over a bank. That is why crossbar wore so much better than step-by-step.

Figure 5 — How a crossbar switch closes one crosspoint. A select bar rocks to choose a row, a hold magnet operates a vertical, and the contact where they cross closes and stays held by the hold magnet alone. … — Figure 5 — How a crossbar switch closes one crosspoint. A select bar rocks to choose a row, a hold magnet operates a vertical, and the contact where they cross closes and stays held by the hold magnet alone. The switch does no thinking: a marker decides which row and vertical to use. — Original diagram drawn for this dive

What made crossbar powerful was not the switch itself but the common control that operated it. A register collected the caller’s digits; a marker, very fast relay logic shared by the whole office, examined which paths through the switching network were free, chose one, and operated the right select and hold magnets; then it dropped the call and moved on to the next. Hundreds of calls shared a handful of markers, each in use for well under a second per call. The dialled number was now just information, and the route could be anything.

The Bell System’s work on coordinate switches began in 1921 and was shelved in 1926. W. R. Mathies saw crossbar in Sweden in 1930, and urban crossbar development began in 1934. The landmarks (BTL Switching Technology, chs. 4, 6, 7):

Table 1 — The Bell System's work on coordinate switches began in 1921 and was shelved in 1926. W. R. Mathies saw crossbar in Sweden in 1930, and urban crossbar development began in 1934. The landmarks ([BTL Switching Technology, chs. 4, 6, 7](https://archive.org/details/bell-system-science-history))

System	First office	Date	Notes
No. 1 crossbar	Troy Avenue, Brooklyn	February 1938	big-city local office, initially 1,400 telephones
755 PBX	—	1938	the first crossbar PBX, 20 lines and 4 trunks
No. 4 crossbar	Philadelphia	October 1943	toll switching
No. 5 crossbar	Media, Pennsylvania	11 July 1948	the general-purpose local office; first commercial system with automatic message accounting
No. 5 crossbar	Englewood, N.J.	10 November 1951	first customer-dialled long distance

No. 5 crossbar, designed to be economical in towns as well as cities, was the system that finally displaced step-by-step as the Bell System’s most common: it overtook step-by-step’s line count in 1973, the year step-by-step peaked. In the independent industry, the North Electric NX-1 brought crossbar to US independents from the mid-1950s, and Automatic Electric is reported to have drawn on North Electric engineers for crossbar and electronic know-how. Neither point was confirmed from a primary source.

Figure 6 — No. 5 Crossbar equipment at the Telephone Museum in Ellsworth, Maine. The frames of crossbar switches are the switching network; the markers that decide which crosspoints to close are separate, sha… — Figure 6 — No. 5 Crossbar equipment at the Telephone Museum in Ellsworth, Maine. The frames of crossbar switches are the switching network; the markers that decide which crosspoints to close are separate, shared relay equipment. — Ddickenson, via Wikimedia Commons (CC BY-SA 4.0)

7.4 Step-by-Step Lives On: Directors, Tandems and CDOs

Step-by-step did not stand still while common control developed. Britain’s director system (Volume 2) grafted a register-translator onto Strowger switches for London and the large provincial cities. In the United States the Bell System adapted the two-motion switch as a line finder in 1926, replacing the line switch. It built step-by-step tandem networks (Los Angeles, 1926) and step-by-step intertoll dialling from 1936, and it put thousands of small unattended community dial offices into rural and suburban service: the 350A at San Clemente, California, in July 1927, and the mass-produced 355A from 1939 (BTL Switching Technology, ch. 3). Automatic Electric built its own community dial offices for the independents, such as the 35E97.

Step-by-step’s weakness was the age of nationwide direct dialling. A step-by-step office could not easily identify the calling number for billing, or translate the long codes that customer-dialled long distance required. Equipping a small step-by-step office with automatic number identification often cost more than it was worth. Those offices became the last to receive modern services, and the first candidates for replacement.

7.5 Stored Program: Electronic Switching

The next step removed the relays from control altogether. After a field trial of an electronic central office at Morris, Illinois, from November 1960, the Bell System’s No. 1 ESS (Electronic Switching System) went into commercial service at Succasunna, New Jersey, on 30 May 1965, a 4,000-line office (BTL Switching Technology, ch. 9). Its control was a stored program running on a duplicated special-purpose computer, so new features meant new software rather than new wiring. Its voice path was still a space-division analog network, a grid of sealed reed switches called ferreeds. Each conversation still had its own metallic path through the office, as in crossbar or step-by-step.

The Bell System’s first time-division switch came earlier, in the No. 101 ESS PBX, which went into service at Cocoa Beach, Florida, on 30 November 1963. Its time-division network used pulse-amplitude modulation: samples of speech, but analog samples, not numbers. The same is true of the DIMENSION PBX of 1975, which is sometimes called the first digital PBX but was in fact a PAM time-division system (BTL Switching Technology, chs. 9 and 13). Automatic Electric’s own space-division electronic exchange, the No. 1 EAX, followed in the early 1970s.

Figure 7 — A Type 2A ferrite-sheet memory card from a No. 1 ESS. The office's program and its record of every line's state lived in memory like this. The switches no longer decided anything, and the stored pr… — Figure 7 — A Type 2A ferrite-sheet memory card from a No. 1 ESS. The office's program and its record of every line's state lived in memory like this. The switches no longer decided anything, and the stored program decided everything. — Jim.henderson, via Wikimedia Commons (public domain)

7.6 Where Digital Took Over

Digital switching converts speech to numbers (pulse-code modulation, PCM) and switches the numbers. Calls share the switching network in time slots, with no metallic path through the office at all. It arrived in three steps:

Toll and tandem: No. 4 ESS, Chicago, January 1976, the first digital time-division toll switch in the Bell System (BTL Switching Technology, ch. 9).
Local offices: Northern Telecom’s DMS-10, whose first production installation was at Fort White, Florida, on 21 October 1977. Then the DMS-100 (1979), Automatic Electric’s GTD-5 EAX (first deployed 26 June 1982) and the Bell System’s own No. 5 ESS at Seneca, Illinois, on 25 March 1982 (Wikipedia, Digital Multiplex System; Automatic Electric; 5ESS).
PBXs: the computer-controlled, PCM digital ROLM CBX of 1975 or 1976 (the sources differ).

That is where this history stops. The analog No. 1 ESS and its successor the 1A ESS stayed in service for decades more, and step-by-step itself was not gone until the last offices were cut over to digital switches. The date usually given for the United States is 1999, though no source consulted names the office. Nantes, Québec, is reported still to have been step-by-step in 2000 and to have been replaced by a DMS-10 in 2002. Neither claim could be confirmed from a primary source. The last step-by-step office of the Bell System has not been identified either.

Figure 8 — A typical Northern Telecom DMS-100 central-office installation. Rows of identical cabinets hold digital switching that has no moving parts and no metallic speech path, the end of the line of develo… — Figure 8 — A typical Northern Telecom DMS-100 central-office installation. Rows of identical cabinets hold digital switching that has no moving parts and no metallic speech path, the end of the line of development that began with Strowger's collar box. — Pdesousa359, via Wikimedia Commons (CC BY-SA 3.0)

7.7 What Survived

Almost every idea in the step-by-step system outlived the machinery. The loop that signals off-hook by drawing current from a 48-volt office battery is still how an analog telephone asks for service. Dial pulses are still accepted by many digital offices. The busy test, ringing at about 20 Hz with ring-trip on answer, dial tone, the line circuit that hands a line over once something has answered it, and the idea of hunting for an idle trunk all came through step-by-step and are still in use. A small step-by-step system on a bench, such as the one this project is building, is therefore a working model of the ideas underneath every telephone call made since.

7.8 In the Reference Library

Sources

A. E. Joel Jr. et al., A History of Engineering and Science in the Bell System: Switching Technology (1925–1975), Bell Telephone Laboratories, 1982, chs. 1, 3–7, 9, 11, 13. https://archive.org/details/bell-system-science-history
M. D. Fagen (ed.), The Early Years (1875–1925), same series. Same archive.org item.
S. Hochheiser, “Electromechanical Telephone-Switching,” IEEE ETHW. https://ethw.org/Electromechanical_Telephone-Switching
Ericsson history, “The crossbar switch: from concept to success.” https://www.ericsson.com/en/about-us/history/products/the-switches/the-crossbar-switch—from-concept-to-success
Western Electric Magazine, 1983 Q1 (“The Last Panel Office”). https://archive.org/details/westernelectricmagazine%201983%20q1
Western Electric, The Panel Dial Telephone System (1953). https://archive.org/details/panel-dial-system-1953
Patents: US 1,083,456 (Molina, 1914); US 1,139,722 (Reynolds, 1915), via Google Patents.
Wikipedia articles Rotary system, Digital Multiplex System, 5ESS Switching System, Automatic Electric, used for dates not covered by the Bell Laboratories histories, as cited inline.