Number One Electronic Switching System
The Number One Electronic Switching System (No. 1 ESS), also known as 1ESS, was the first large-scale, stored-program-controlled electronic telephone switching system designed and developed by Bell Telephone Laboratories for the Bell System, capable of handling local, toll, and tandem switching applications with capacities ranging from 4,000 to 65,000 lines and up to 100,000 calls per busy hour.[1] Introduced as a general-purpose switching machine, it marked a pivotal shift from electromechanical systems to electronic control, emphasizing flexibility for existing and future telephone services while ensuring high reliability through duplicated central processors and automated fault detection mechanisms.[1] The system was first placed into commercial service on May 30, 1965, in Succasunna, New Jersey, following laboratory testing in Holmdel, New Jersey, earlier that year.[2] Development of the No. 1 ESS began with foundational research at Bell Labs in 1945 and built upon the experimental electronic switching trial installed in Morris, Illinois, in 1959–1960, which validated key concepts like stored-program operation and maintenance diagnostics.[1] As the largest engineering project in Bell Labs' history at the time, it involved interdisciplinary teams from systems engineering, device development, and Western Electric, culminating in a design finalized by early 1964 that incorporated novel hardware such as ferreed cross-point switches in an eight-stage space-division network and low-level logic circuits using silicon transistors and diodes.[1] The system's central processor featured a 24-bit word length, a 5.5-microsecond cycle time, and memory technologies including twistor and ferrite-sheet stores with error-correcting Hamming codes, enabling efficient call processing and scalability for growing telephone networks.[1] Key objectives of the No. 1 ESS included economic competitiveness with prior electromechanical switches, minimal service interruptions (targeting no more than two hours of downtime over 40 years), and adaptability to new services like call forwarding and multi-line hunting through programmable software rather than hardwired logic.[1] Innovations such as dual duplicated controls, an emergency-action sequencer for rapid fault recovery, and the PROCESS III compiler-assembler for managing over 100,000 instructions facilitated its role as a foundational technology in telecommunications.[1] By the early 1970s, over 250 No. 1 ESS offices were in operation across the United States, serving as the dominant voice switching platform until the widespread adoption of digital systems in the 1990s.[3]History and Development
Origins and Design Objectives
The development of the Number One Electronic Switching System (1ESS) was initiated in the late 1950s by Bell Telephone Laboratories as part of a broader effort to modernize the Bell System's telephone network. Development of the No. 1 ESS built upon foundational research in electronic switching that began at Bell Labs in 1945.[1] This project built upon exploratory research in electronic switching that had begun after World War II, including a pivotal experimental trial in Morris, Illinois, starting in 1960, which demonstrated the feasibility of stored-program control using cold-cathode gas tubes for serving approximately 400 subscribers.[1][4] The primary motivation was to overcome the limitations of existing electromechanical systems, such as Step-by-Step and Crossbar switches, which were inefficient in space, power consumption, and adaptability to the rapidly growing demand for telephone services and new features.[1] Key design objectives centered on achieving exceptional reliability and scalability to support the Bell System's nationwide operations. The system targeted 99.999% uptime, limiting downtime to no more than two hours over 40 years through extensive redundancy and automated diagnostics.[1][5] It was engineered to handle up to 65,000 lines and approximately 100,000 calls per busy hour, with configurations supporting 4:1 concentration ratios for up to 16,384 trunks.[1] Additionally, 1ESS was designed for versatile two-wire or four-wire switching to accommodate local, toll, and tandem applications, while incorporating robust error detection and correction mechanisms—such as parity checks, Hamming codes, and continuous self-diagnostics—to ensure uninterrupted operation.[1] The design emphasized innovative shifts from traditional electromechanical relays to electronic control, utilizing reed relays for switching and ferrite-core memory for program and call data storage, which enabled stored-program control for greater flexibility.[1] Modularity was a core principle, allowing scalable growth through standardized frames and circuit packs without major rewiring, facilitating future upgrades and adaptations to evolving service needs.[1] The project, recognized as Bell Labs' largest development effort to date, was led by key engineers including J. R. Harris, who played a central role in the central processor design and overall system planning.[1]Initial Deployment and Milestones
The first commercial deployment of the Number One Electronic Switching System (1ESS) took place on May 30, 1965, in Succasunna, New Jersey, where Western Electric installed the system to serve an initial 4,000 customers with plans for expansion to full capacity.[6] This installation marked the transition of the Bell System to large-scale electronic stored-program control switching, fulfilling design objectives for high-capacity handling of up to 80,000 calls per hour in configured systems.[6] The system's organization and objectives had been outlined earlier in a series of articles published in the Bell System Technical Journal in September 1964, providing the foundational documentation for its architecture and goals.[1] Following the Succasunna cutover, 1ESS saw rapid adoption within AT&T's network, with subsequent installations enabling nationwide service integration and replacing electromechanical systems in key locations.[7] By the mid-1970s, thousands of 1ESS and its variants were deployed across the Bell System, handling a significant share of telephone traffic and demonstrating the system's scalability for urban and suburban offices.[6] Early reliability testing highlighted its robustness, with duplicated controls and diagnostics contributing to extended operational uptime during initial field trials.[1] A major milestone came in 1976 with the introduction of the 1AESS upgrade, which offered a plug-compatible enhancement to the original 1ESS through a faster central processor and reduced equipment volume, facilitating smoother evolution in existing offices.[6] The system also supported adaptations for specialized applications, including the No. 101 ESS variant for remote Centrex service using time-division switching units connected to central office controls.[8] In the 1980s, many 1ESS installations underwent retrofits, such as the addition of digital interface frames and time slot assigners, to accommodate emerging digital signaling requirements in the evolving network.[9]System Architecture
Switching Fabric
The switching fabric of the Number One Electronic Switching System (1ESS) consists of an eight-stage space-division network employing ferreed crosspoints, which combine reed relays with ferrite cores for magnetic latching and reliable metallic transmission paths.[10] These crosspoints are organized into basic building blocks such as 8x8, 8x4, 4x4, and 16x8 matrices, enabling non-blocking connectivity across line link and trunk link networks.[11] The structure supports path hunting by the central control, which selects idle paths through sequential stage activation using wire-spring relays for selection logic, ensuring efficient call routing without mechanical crossbars.[1] Each stage incorporates ferreed crosspoints in frame-based assemblies, with approximately 1,000 reed relays per stage to handle the matrix configurations and provide scalability for office sizes up to 65,000 lines.[11] Error detection in the switching fabric is integrated through specialized circuits that monitor relay states and path integrity, including group check circuits that verify single paths and test verticals equipped with current-sensing ferrod detectors to identify false crosses or grounds.[11] These mechanisms perform parity-like checks on crosspoint activations during call setup and supervision, triggering false cross/ground (FCG) tests if anomalies are sensed.[1] Upon detection, the system initiates automatic rerouting via alternate paths in the multi-stage design, while diagnostic buses and maintenance programs isolate faulty crosspoints by sequentially testing stages, quarantining defective elements, and switching to duplicated controllers for continued operation without service interruption.[11] The fabric's capacity supports 4:1 concentration in trunk networks, allowing efficient resource allocation for varying traffic loads, with a total switching capacity of up to 100,000 calls per busy hour in a typical configuration.[1] This design achieves minimal blocking probability, less than 0.5% under normal operating conditions (e.g., approximately 0.04% at 0.15 erlang per line), through multiple access paths and flexible junctor redistribution.[11] The switching fabric integrates briefly with line interfaces to accept incoming signals at the initial stages, facilitating seamless entry into the network for call processing.[10]Line and Trunk Interfaces
The line networks (LN) in the Number One Electronic Switching System (1ESS) serve as the primary interfaces for connecting subscriber lines to the switching fabric, supporting up to 65,000 lines organized into 512-line groups for efficient management and scalability.[1] These networks employ line link frames equipped with ferreed switches in 4x4 or 8x4 configurations and utilize a 2:1 concentration ratio to optimize resource allocation, where concentrators—such as eight units each handling 32 lines in 2:1 setups—route calls from lines to the core network.[1] Key functions include supervision via ferrod sensors that detect off-hook conditions through line current flow, with scanners cyclically monitoring 1,024-point modules every 100 milliseconds to identify service requests; ringing is provided by dedicated circuits delivering 20-cycle voltage from 110A generators, adjustable by subscriber class and ceasing within 0.25 seconds upon answer; and dial pulse detection occurs through scanner interrupts or customer dial pulse receivers interrogating pulsing relays up to 100 times per second to accurately count and correct distortions.[1] Trunk networks (TN) interface with external toll and tandem trunks, accommodating up to 10,000 trunks per office through trunk link networks that connect to the switching fabric for path completion.[1] These networks feature a four-stage 8x8 switch design with 4:1 concentration in certain configurations, using trunk switching frames that support 256 trunks each and junctor switching frames providing up to 4,096 access paths per trunk group, while enabling four-wire transmission to minimize signal loss.[1] Signaling protocols such as single frequency (SF), multifrequency, and dial pulse are handled by trunk circuits that detect incoming signals and generate outgoing ones under central control, with supervision via Type 1C and 1D ferrods monitoring trunk states.[1] Both line and trunk interfaces incorporate robust protection mechanisms, including gas tube arrestors in protector frames to safeguard against lightning surges and overvoltages, with each module handling up to 6,000 outside plant pairs.[1] Scanner interfaces, distributed across network frames, use 64-row matrices of 16 points each to monitor line and trunk status, feeding data to the central processor via peripheral buses with 17-lead answer and 48-pair monitor lines at rates up to every 5.5 microseconds.[1] Distributor outputs, employing signal distributors with 768 points per frame and duplicated controllers, actuate relays for ringing, cutoff, and trunk operations using bipolar pulses from central pulse distributors.[1] These components ensure reliable signal conditioning and endpoint connectivity while integrating with the overall switching network for call handling.[1]Control and Processing
Central Processor
The central processor of the Number One Electronic Switching System (1ESS), known as the central control, features a dual redundant CPU architecture in a Harvard configuration, with separate program and call stores to enable parallel instruction fetching and data access. Each CPU utilizes high-speed semiconductor logic and operates with a cycle time of 5.5 microseconds (approximately 182 kHz), executing 44-bit instruction words that include 37 data bits and 7 check bits.[1] This design supports efficient processing of switching tasks, with the two CPUs running in synchrony and continuously cross-comparing outputs for fault detection.[1] The program store employs core rope memory, a read-only medium providing 131,072 words of 44 bits each, for storing the system's microcode and fixed programs.[1] To ensure reliability, it incorporates Hamming code for single-error correction, using 7 parity bits calculated as approximately \lceil \log_2(n) \rceil + 1, where n is the number of data bits (applied here to the 37 data bits per 44-bit word, enabling correction within the overall structure).[1] In contrast, the call store consists of ferrite sheet random-access memory (RAM) with capacities ranging from 8,192 words (basic unit) up to approximately 300,000 words of 24 bits each, depending on configuration, dedicated to temporary data for active calls and system state.[1] Processing occurs via microcode execution, where the central control handles call setup and teardown by performing logical operations, such as path hunting and connection establishment, in response to inputs from peripheral scanners.[1] Redundancy is achieved through full duplication of the processors, program stores, and call stores, with automatic switchover to the standby unit in less than 10 milliseconds upon fault detection, minimizing service disruption.[1] The processors and associated stores consume several kilowatts of power, drawn from dedicated battery plants to maintain operation during outages.[1]Scan and Distribute Mechanisms
The scan and distribute mechanisms in the Number One Electronic Switching System (1ESS) serve as essential peripheral units that interface between the central control and the external telephony environment, enabling efficient monitoring of subscriber lines and trunks while executing control commands with high reliability. These subsystems employ ferromagnetic (ferrod) technology to handle the demands of large-scale switching, supporting up to 65,000 lines and trunks in a typical installation. By periodically polling for events such as off-hook conditions or dialed digits and distributing orders to relay networks, they ensure real-time responsiveness without overburdening the central processor.[1] The scan subsystem utilizes ferrod magnetic scanners to detect service requests and supervise ongoing calls by sampling the states of lines, trunks, and diagnostic points at discrete intervals. These scanners, configured in 1024-point matrices (such as 64 by 16 arrays), employ specialized ferrod sensors—types 1B, 1C, 1D for line and trunk monitoring, and type 1E for master control—to sense current changes indicative of events like off-hook detection or digit reception. Scanning operates on a 10-millisecond cycle for directed tasks, interrogating all office lines approximately 10 times per second, while cyclic scans for administrative purposes occur every 100 milliseconds; dial pulse detection may use faster 5- to 10-millisecond intervals triggered by interrupts or a 5-millisecond clock. The resulting data is serialized and recorded in temporary memory locations or call store tables, such as the scanner appearance (SCA) or line link (LL) areas, for subsequent processing by the central control. To distribute the workload evenly, scans are balanced across 5-millisecond intervals, with half the signal receivers polled per period.[1] The distribute subsystem, comprising signal distributors and the central pulse distributor (CPD), activates relay drivers in response to central control commands broadcast over communication buses, thereby controlling peripheral equipment like trunk relays and solenoids. Signal distributors manage low-speed operations through a relay-tree structure, supporting up to 1 million relay points via magnetic latching relays, mercury relays, or wirespring types, with drivers capable of 1.35-amp pulses over twisted-pair cables. The CPD, an all-electronic unit, handles high-speed actions by connecting specific peripheral units to the buses and generating bipolar or unipolar pulses—such as 0.5-microsecond high-speed pulses or 300-microsecond nominal pulses at 2.5 to 9 amperes—using pulse-width modulation for precise timing in operations like network enabling. Commands are buffered in peripheral order buffers (POBs) and executed at a rate of up to 100,000 per hour, with verification through scan points or current-sensing circuits to confirm relay states.[1] Reliability in both subsystems is achieved through redundancy and self-checking features, including duplicated controllers for scan matrices—since a controller fault could affect an entire group, while individual ferrod failures impact only one line or trunk—and fault-detection programs that monitor error counters and parity checks. The ferrod matrices themselves are non-duplicated for cost efficiency, but the overall design targets failure rates below 10 per 10^9 device-hours, with bit error rates exceeding 1 in 10^7 triggering module rewrites and single-bit errors corrected via program store checks; this contributes to a predicted one component failure per month per central control and less than 0.02% incorrect calls. Hardware safeguards, such as group check circuits and emergency-action sequencers, further ensure fault isolation and system continuity.[1]| Parameter | Value | Description |
|---|---|---|
| Maximum Lines/Trunks | 65,000 | Total supported by scan matrices |
| Relay Points | 1,000,000 | Controlled by distribute relay trees |
| Scan Cycle Time | 10 ms per frame | For directed line and digit scans |
| Commands per Hour | 100,000 | Distributed orders to peripherals |
| Device Failure Rate | <10 per 10^9 hours | Overall reliability metric |