Telephone Switching History · Volume 5
Inside the Circuits

5.1 Why the Circuits Matter
Volume 4 followed a call through the switch train. This volume opens the switches and looks at the relays, because that is where faults live. A two-motion switch that steps wrongly is usually not mechanically broken; it is a relay that releases too soon, a contact that does not quite make, or a power supply a few volts out. Knowing what each relay is for turns a mysterious fault into a short list of contacts to look at.
The circuits described here follow the Automatic Electric-pattern switches walked through relay by relay in the US Air Force course Telephone Switching Equipment Repairman (Electromechanical), Volume 3 (CDC 36251, ERIC ED194781), cross-checked against Doug Kerr’s Bell-practice papers (Kerr, The selector switch; The connector switch). Kerr notes that the Bell System’s relay lettering (A, B, C …) was “inherited directly from the practice in systems made by Automatic Electric,” though he adds that the letters were assigned by the relays’ physical positions, so the same letter need not mean the same job in every switch. The circuits are representative, not any one drawing. Every switch on a real shelf has its own circuit drawing, and the drawing is the authority. The Reference Library holds Automatic Electric’s own selector and connector drawings, H-580156 and H-580260, with its circuit explanation E-580260.
5.2 What the Dial Sends
A rotary dial is a cam and a pair of contacts. Turning the finger wheel winds a spring; letting go lets the spring drive the wheel back at a speed set by a flyball governor, and on the way back the cam opens the line once for each digit step. Dial 3 and the loop opens three times; dial 0 and it opens ten times. Off-normal contacts in the dial short the receiver while it runs, so the caller does not hear the clicks (Kerr, Overview; Kerr, Rotary telephone dials).
Table 1 — What the Dial Sends
| Parameter | Value | Source |
|---|---|---|
| Speed | ”A dial speed of 10 pulses per second is considered ideal, while extreme test limits for dial speeds are 8 and 14.” | US Army TM 11-2111 (1950), ¶9 |
| Ratio | ”Normal percent make … is approximately 38½ per cent”, about 61½ per cent break | TM 11-2111 ¶9 |
| US break range | 58–64 per cent break, 61 per cent nominal | US 3,997,740; secondary sources |
| Bell station dials | adjusted to 9.5–10.5 pps | AT&T specification 4566 (1926), via Wikipedia |
| Coding | n breaks for digit n; 0 = ten breaks | Kerr |
Touch-Tone arrived in 1963. A step-by-step office could accept it only by giving a tone caller a pooled converter that answered with dial tone, received the tones, and sent ordinary dial pulses on to the selectors, as in Western Electric’s US 4,186,278 (Google Patents). The switches themselves never learned anything but pulses.
5.3 Seizure: A and B
When a switch is seized, the calling loop (the telephone’s own circuit, through the line finder and any switches already in the train) is extended to the switch’s A relay, the pulsing relay, whose two windings also form the battery feed to the caller. A operates on the loop current, and A’s contact operates B, the holding relay. B does two things at once: it puts ground on the C lead (the sleeve) back toward the previous switch, which holds that switch and everything behind it, and it prepares the stepping circuits.
B is slow to release. A copper slug on its core keeps it operated for a fraction of a second after its current stops, and that is the whole trick of step-by-step. Every dial pulse opens the loop, so A releases on every pulse. If B released with it, the call would collapse. Instead B rides through every break and drops only if the loop stays open for longer than a dial break: that is, when the caller hangs up. Smith and Campbell stated the principle in 1921: “the introduction of these slow-acting relays makes possible the operation of the switches by simply making and breaking the calling party’s loop” (Automatic Telephony, p. 56).
5.4 One Digit: A, B, C and the Vertical Magnet
From the USAF course (Vol. 3 §2-2), one digit on a selector goes like this:
- Break of the first pulse. A releases. B holds, being slow. A’s back contact closes a circuit through the C relay in series with the vertical magnet. Both operate, and the shaft rises one step.
- First step. The vertical off-normal springs operate, preparing the release-magnet path and the auto-rotary relay E.
- Make. A reoperates. The vertical magnet releases. C, also slow to release, stays up through the short make.
- Repeat for every pulse: “If the digit 5 is dialed, relay A restores and reoperates five times.”
- End of digit. A stays operated. With no more breaks to feed it, C finally releases, “perhaps 200 ms” after the last pulse in Kerr’s estimate, and its release switches the stepping circuit from the vertical side to the rotary side.
So B and C are both slow, for different reasons. B must hold through the longest break, about 60–70 ms at 10 pulses a second, so that the switch is not released by dialling. C must hold through the make between breaks, so that one digit is not mistaken for several, but must also drop out well inside the pause between digits, so that the switch is ready for the next one. Most dialling faults on a step-by-step switch come down to one of those two timings, or to A failing to follow the pulses at all.
5.5 Hunting: E, the Rotary Magnet and F
A selector’s second motion is not dialled. When C releases, the selector hunts along its level by itself. The auto-rotary relay E and the rotary magnet “interrupt each other’s circuits”: the magnet’s interrupter springs open E’s circuit, and E’s release opens the magnet’s (USAF course). The shaft steps round, one contact per cycle, testing each trunk’s sleeve as it goes. A grounded sleeve, a busy trunk, keeps the cycle going.
The cut-through relay F is held short-circuited the whole time by three overlapping shunts: through C during vertical stepping, through E, and through the ground on the sleeve wiper while it sits on busy trunks. Only when the wipers land on an ungrounded (idle) trunk do all three shunts disappear. Then F operates through its high-resistance (1800 Ω) winding in series with E’s low-resistance (210 Ω) one, which is enough for F but not for E, so hunting stops exactly where it should. F’s contacts drop A off the loop and extend the tip, ring and sleeve to the next switch. The B relay of the next switch operates and takes over the holding ground, and the selector’s own A and B release: the hand-over Volume 4 calls “the basic principle of step-by-step.”
Smith and Campbell’s 1921 Automatic Electric circuit names the same partnership a rotary-interrupter relay: “The rotary trunk-hunting action consists of mutual vibration between the rotary magnet RM and the rotary-interrupter relay RIR.”

5.6 The Connector, Relay by Relay
The connector takes both of its digits from the dial and then does everything an operator’s cord circuit used to do. The USAF course lists eight relays in a regular Automatic Electric-pattern connector:
Table 2 — The connector takes both of its digits from the dial and then does everything an operator's cord circuit used to do. The USAF course lists eight relays in a regular Automatic Electric-pattern connector
| Relay | Job |
|---|---|
| A | Pulsing, and the calling party’s battery-feed coils |
| B | Holding; slow release; grounds the C lead |
| C | Transfer from vertical to rotary stepping; slow release |
| D | The called party’s battery-feed coils; reverses battery on answer |
| E | Busy-test timing; slow release, held during rotary pulsing |
| F | Secondary cut-through and ring cut-off (ring trip) |
| G | Busy |
| H | Primary cut-through and ring start; a two-step relay |
Tens digit. A pulses; C and the vertical magnet step the shaft up. After the digit, C releases and transfers pulsing to the rotary magnet in parallel with E.
Units digit. The rotary magnet steps the wipers round one contact per break. E, slow to release, stays up for the whole digit and connects the busy relay G to the sleeve wiper. G flickers as the wipers pass busy lines on their way, but the caller hears nothing, because the dial’s off-normal contacts mute the receiver while dialling (Kerr).
Busy test. A fraction of a second after the last pulse, E releases, and the connector looks at the called line’s sleeve.
- Ground on the sleeve: the line is busy. G operates, locks, returns busy tone, and blocks H. Nothing more happens until the caller hangs up and tries again.
- Battery through the cut-off winding: the line is idle. E’s release puts H’s low-resistance (125 Ω) winding on the sleeve wiper. H’s first step closes its X contacts; its 1300 Ω winding then pulls it fully home and locks it. H opens the rotary magnet’s path, so further dialling has no effect; grounds the sleeve, which makes the called line busy to everyone and operates its cut-off relay; connects ringing to the line through F; and connects ringback to the caller.
Ring trip. Ringing is “ringing current superimposed on negative battery,” fed through F’s 200 Ω winding. While the telephone is on-hook, the capacitor in its ringer blocks the direct current, so F cannot operate on ringing alone. When the handset is lifted, direct current flows through the telephone’s loop and F’s winding; F operates, locks through its 1800 Ω winding, cuts off ringing and ringback, extends the called line to D, and completes the talking path. Because the battery is present in the silent intervals too, ringing is “tripped promptly” even when someone answers between rings (Kerr). Smith and Campbell’s 1921 ring cut-off relay was built the same way, with a copper collar and sleeve so that the alternating ringing current could not operate it: “Dependence is placed upon the battery alone to operate the relay.”
Answer supervision. D operates on the called party’s loop and reverses the battery on the calling side through make-before-break springs, so there is no click. That reversal is a signal: it could operate a message register, a coin collector, or an operator’s cord supervision.
Transmission. A (two 200 Ω windings) feeds the caller and D feeds the called party, and the two halves are joined by two 2 µF capacitors, one in each side of the line. The relay windings are low-resistance to the direct current that powers each telephone’s transmitter and high-impedance to speech, and the capacitors pass speech between the two halves while keeping their direct currents separate.
5.7 Release: Who Controls the Call
When the caller hangs up, A releases, then B. While B is releasing, E briefly opens the sleeve, and every switch behind the connector (selectors and line finder) releases. What happens to the connector depends on the design.
- In the Automatic Electric circuit of the USAF course, D, held by the still-off-hook called party, keeps F and H operated and the sleeve grounded: “The connector cannot be released by the release magnet until both parties hang up.” When the called party hangs up too, D releases, the release magnet operates, and the switch drops.
- In Kerr’s Bell circuit, the connector releases the called line at once. If the called party is still off-hook, that line’s own line relay then calls a line finder, and the called party hears dial tone.
When the called party hangs up first, D releases and the caller’s battery polarity returns to normal, but the connection stays up until the caller hangs up. The two descriptions are of different designs, and both were built. The drawing for the connector on the bench is the only way to know which one it is.
A release is mechanical once it starts: the release magnet kicks out the detents and the switch rotates home and drops (Volume 3). A selector that has run to the 11th step releases the same way, “the wipers … moved mechanically by the switch shaft spring” off the end of the bank.
5.8 Battery, Ringing and Tones
Talking and switching battery. The nominal office battery is 48 V, with its positive side grounded (Kerr, Overview); the USAF course gives the working range as 46–52 V DC. Smith and Campbell’s 1921 connector rings over “the main battery, 48 volts”. Kerr calls the reason for grounding the positive side “esoteric but very important” and does not give it. The usual explanation, found only in secondary sources, is that wires held negative to earth corrode far less than wires held positive, which mattered for miles of lead-sheathed outside cable.
Tones. The USAF course’s office makes its tones from “an Alnico permanent magnet-type generator … a tone of 600 cycles per second modulated at 120 CPS,” fed to a dial-tone transformer on each selector shelf. Busy tone is “dial tone interrupted 60 impulses per minute,” and all-trunks-busy is the same tone interrupted 120 times a minute. Ringback (“audible ringing”) is not a separate tone at all: it is part of the called line’s ringing voltage, coupled to the caller through capacitors. Pure 20 Hz is nearly inaudible on a telephone receiver, so ringing generators were deliberately made non-sinusoidal for the purpose (Kerr). A Holtzer-Cabot cam-driven interrupter unit supplied the slow ringing cadences, the busy and all-trunks-busy interruptions, and timing pulses for the alarm circuits.
Ringing. Kerr gives the ringing signal as “an AC signal at perhaps 75–90 V RMS at 20 Hz, to which is added a DC voltage of perhaps −48 volts,” with a typical Bell cadence of rings “2 seconds long separated by 4 seconds.” A Lenkurt Demodulator of 1968 describes a 6-second “standard interrupter” cycle of 1.2 s ringing and 4.8 s silence, with five groups of lines taking turns on the generator to even out its load (Demodulator, June 1968). No Automatic Electric document giving AE’s own cadence was found. The USAF course’s office ran its ringing from duplicated dynamotors (a motor on office battery driving an alternator) with automatic transfer from machine 1 to machine 2. AE’s 1929 P-A-X bulletin used vibrating-reed ringing converters in its smallest systems.

5.8.1 Party Lines, Briefly
Several subscribers on one line needed some way to ring only one of them. Bridged ringers sat across tip and ring; divided ringers sat from one side to ground. The Bell System used superimposed ringing, adding a DC polarity to the 20 Hz and fitting each ringer with a gas tube so that it answered only to the right sign. Independent companies, Automatic Electric’s customers among them, more often used frequency-selective ringing with tuned ringers: decimonic (20, 30, 40, 50, 60 Hz), harmonic (16⅔, 25, 33⅓, 50, 66⅔ Hz) or synchromonic sets (Lenkurt Demodulator, 1968). Hill records that Automatic Electric offered four-party harmonic selective ringing around 1905. A small private system has no need of any of this, but a connector from a party-line office may carry the relays for it.
5.9 In the Reference Library
- USAF CDC 36251, Telephone Switching Equipment Repairman (Electromechanical), 1978
- AE H-580156, selector circuit, local and incoming (1952)
- AE H-580260, connector circuit drawing
- AE E-580260, connector circuit explanation
- AE Training Bulletin 800, Electrical Principles of Telephony (1956)
- AE Training Bulletin 812, Power and Supervisory Equipment (1960)
- AE Training Bulletin 810, Pulse Repeaters (1955)
- D. A. Kerr, Rotary telephone dials: two families
- US 1,151,541, calling device (1915)
- D. A. Kerr, The connector switch
Sources
- USAF Extension Course Institute, CDC 36251, Telephone Switching Equipment Repairman (Electromechanical), Vol. 3, ERIC ED194781. https://archive.org/details/ERIC_ED194781
- D. A. Kerr, The step-by-step telephone switching system: Overview, The selector switch, The connector switch; Rotary telephone dials: two families. http://dougkerr.net/Pumpkin/
- A. B. Smith and W. L. Campbell, Automatic Telephony, 2nd ed., 1921. https://archive.org/details/automatictelepho00smit
- US Army TM 11-2111, Tools, Testing Equipment and Common Supplies, Step-by-Step Dial Central Office Equipment, May 1950. https://www.radionerds.com/images/9/97/TM_11-2111.pdf
- Lenkurt Demodulator, June 1968, “Multiparty service through ringing.”
- Automatic Electric circuit drawings H-580156 (selector) and H-580260 with E-580260 (connector), in the Reference Library.
- US 4,186,278 (Western Electric, Touch-Tone to dial-pulse converter, 1980).