The Panel Switch
Highlights
Switch category: Linear motion
​
Inventors: F. Jewett and other Bell System engineers. Mr. Jewett oversaw the panel exchange project.
Important dates: Circa 1906, switching investigations began into the “large city problem”; first panel-type exchange installed Newark, New Jersey, 1915; first full 7-digit panel in NYC 1922.
Legacy: More than 500 panel exchanges were working in NA in 1950; all in large cities. The switch became famous as the backbone of the Bell System’s first metro system.
Backgrounder
​
While A.E. Keith’s Strowger (Step-by-Step exchange) development was continuing at Automatic Electric, Bell System engineers were pioneering a new switch and exchange type known as the "panel system". In parallel, Bell was also developing a "rotary system” with very similar control logic to panel but using a completely different master switch type.
​
Both developments were influenced by the Lorimer system. It was based on motor driven rotary switches and had some modest success during the years 1905-1915. Early on, Bell engineers saw promise in Lorimer and licensed their patents in 1903.
​
The panel switch found a home mostly in large metro exchanges in North America. Bell’s preference was to use panel where, in their view, Step-by-Step (Strowger) was inferior due to a myriad of scaling reasons.
The Rotary system, it was decided, should be used initially for European customers to meet the growing demand there then expand elsewhere. It was built and sold by the International Western Electric Company (AT&T) in Belgium.
​
Panel Switch concepts
​
Bell’s panel switches were motor driven and used clutches to control the up/down movement of vertical shafts with attached connecting brushes (wipers). The brushes made contact with a linear panel of terminals, hence the name panel switch.
​
Control of the panel switch action departed from the Strowger in several important ways. One was the method of switch control. Strowger uses a “dial direct progressive” style and panel uses “common control”. Check out the differences here.
​
The switch, and some derivatives, was designed to work within a 10,000-subscriber exchange. More importantly, each exchange needs to connect to many other local and remote exchanges (over trunk lines). NYC metro area had over 100 panel offices in the 1940’s. Hence, a switch of this size made interconnecting easier.
​
Fig 1 below, from a panel exchange, was first published by the engineer and educator Elsworth Goldsmith. See the switch, center, among other devices necessary to complete a call. This section only focuses on the switch part. See panel exchange for more info on the exchange's workings.
Fig 1, Panel selector switch [Goldsmith]
Compared to Keith’s Strowger, for instance, this switch is significantly larger.
Let’s break down some of the salient aspects for a selector-style switch:
​
-
Five terminal banks each with 100 rows of terminals, arranged in groups of 3 leads: tip, ring and sleeve. So, there are 500, 3-lead, connections available per selecting rod.
-
Each panel is double-sided, with the rear identical to the front.
-
Each rod has 5 brushes (one per bank) attached. A brush is responsible for making the physical contact with bank terminals. So, any of the 500, 3-lead, terminals is accessible by at least one brush. There are 30 selecting rods per side, 60 total. Up to 60 independent, 3-lead (talking path), connections can be made per switch.
-
The selector rods are powered by a motor and driven up/down by engaging clutches.
Various panel switch types existed, all based on the same core mechanism. In a typical panel office there were four main types of selector; District, Office, Incoming, and Final. These designs had only slight differences to adapt to their needed function.
​
The Line Finder switch was the most different of the panel types. It had 15 small banks (not 5 big) with with 20 sets of terminals vertically (not 100) accommodating 300 subscribers (not 500). Providing dial tone as quickly as possible was a goal and the LF design reduced brush travel time to meet this goal.
​
The Selector Switch
The image below [Goldsmith] spotlights a single bank of terminals for a selector switch. It’s easy to see the 30 columns of T/R/S pins (tip/ring/sleeve leads). Each set of 3 pins on a row repeat across the 30 columns. So, for say row 50, the same T/R/S leads are accessible by any of the 30 rods on either side (30 + 30 total).
Fig 2, Double sided panel selector bank with 100 lines (T/R/S) and 30 columns
To get a feel for both the line finder and selector switches in action, view the making a 7-digit panel call. The purpose here is switch movement, not switch function. There is no dialog. The line finder switch is activated before any digits are dialed; it is the first switch shown in the video. Then three selector switches follow and move based on the dialed digits, 722-7234. The functions of each selector switch is detailed in the panel exchange section and not in this video.
Lets break down the operations needed to make a connection using Fig 3. This is a model for what a selector switch does and not a diagram for exactly how it does it.
​
By way of explanation, let’s assume we want the switch to connect to point #498.
Fig 3, Concept model for a typical panel selector switch
To begin, a switch has 5 banks that contain 500 total available connection points. A single connection point always connects 3 leads (T/R/S). Each bank has 100 rows and 30 columns of 3 collocated terminals. So, each row has 90 pin (3x30) terminals total.
Importantly, for a given row there are only 3 different terminals. For that row, the same T, R and S triplet of pins is repeated for each respective column on a given bank. See Fig 2. This way, any free vertical selector rod can access the same T, R and S leads. This double-sided design was ingenious.
The model above illustrates a single connection for column 15 (vertical selecting rod 15), bank 4 and row 98. #498 is the desired connection point. Control logic knows that brush #4 needs to be selected (the other brushes do not make physical contact with their bank) and that the brush must rise to level 98 of the bank and then stop. A relay-based switch controller engages the clutches to raise the brush to level 98 and to engage brush 4 at the right time.
Given that the rods are propelled upwards by a motor rather than through a stepping action, how does the controller determine when to halt the brushes and establish a connection? See the Appendix below for the answer.
Fig 4, Portions of a panel selector switch, selecting rods and terminal banks
​
There are a few excellent references that dig much deeper than the coverage here. Learn about brush trip rods, the commutator, the motor drive, bank construction (w/ 90,000 terminal points), switch control and much more from [Craft], [Goldsmith], [Miller] and [Hersey].
​
The three figures below show some of the components discussed above for switch operations.
Fig 5
Contact brush, 3 terminals, top is closed, bottom is open (no contact) [Miller]
Brushes are made to open and close for making contact with a bank's 3 terminals.
Looking at Fig 4, it's not obvious how the Fig 5 brushes make contact with the terminal bank. Fig 6 provides a color coded explanation. The terminal bank has insulated layers of Tip, Ring and Sleeve terminals (see Fig 2). Each of these terminal sets are offset vertically. Notice the green Tip layer, blue Ring layer and pink Sleeve layer.
The lower part of Fig 6 illustrates how the brushes make contact with the terminals of the bank. Notice the vertical stagger of the 3 contact points. Why 3 when there are 4 contacts per brush? The center 2 contacts touch the same sleeve terminal to keep the brush assembly balanced and thereby provide even wear patterns.
The panel switch is double sided so the same T/R/S terminals are available on both sides of the switch. No other exchange type switch was built with two sides. Fig 6 was derived from a Western Electric document and color coded by Sarah Autumn, associated with [CMoS].
Fig 6, Panel terminal bank with brush contact points
In Fig 7 the black fabric-like layers of material are sandwiched between the T/R/S terminals and provide insulation. The vertical offsets between each T/R/S terminal are clearly seen in this photo. Note too the horizontal spacing between triplets to allow clearance for each rod/brush assembly.
Fig 7, Panel bank showing offsets between T/R/S terminals [CmoS]
Fig 8, Switch closeup; selector rods, trip rods, rack, brushes, bank terminals [Craft]
The brush attached to the rod (Fig 8) in the center is making contact with 3 bank terminals at once.
Fig 9 clearly shows a brush in its home position and the T/R/S triplet bank terminals. Per bank there are 100 rows of 30 triplets per row (typical selector switch) and 5 banks per double-sided switch.
Fig 9, Panel brush at rest and bank terminals [Williams]
Fig 10, Three clutches per vertical selector rod [Miller]
Fig 11, Partial view of base infrastructure of a panel selector switch [MS]
Below is a brief video showing Line Finder and Final Selector switches in operation at [CMoS].
In essence, the design and construction of this switch transformed AT&T (along with Western Electric) into a manufacturing powerhouse. Panel was a masterpiece in electromechanical design and found its place in metro exchanges in North America.
​
Unlike the Strowger type switch, this switch required a substantial control infrastructure. Bell engineers ended up designing dial pulse registers, sequence switches, many new relays (E and R types), motor and gearing systems, power systems and more.
​
​
References and Acknowledgements
​
CMoS; Connections Museum of Seattle, several working exchanges including panel, step and crossbar. Special thanks to Sarah Autumn for consulting advice, panel video clips and images of the panel brushes and banks.
​
Craft, E.B & Morehouse: Machine Switching Telephone System for Large Metropolitan Areas, Bell System Technical Journal, 1923b.o
​
Goldsmith, E. H., The Panel Type Dial Telephone System. New York Telephone Company,1926 (author’s original copy)
​
Hersey, R.E., Panel Dial Systems, Bell Telephone Laboratories, 1928
​
MS: Machine Switching Equipment Notes, Vol 2, 1920-21, Illustrations and Drawings. Western Electric Company publication for the Panel exchange
​
Miller, Kempster, Telephone Theory and Practice V3, 1933, New York, McGraw Hill.
​
Williams, David, image of Panel bank and brush from [CMoS].
​
Appendix
Analog and Digital Switching Mechanisms
Electromechanical switching devices can be grouped into digital or analog types. For sure, this is not a conventional way to classify an exchange switch. Nonetheless, let’s stick with this naming as an aid to understanding the need for feedback position pulsing for the analog switch types.
Digital: Using discrete units
Analog: Using continuously variable quantities
Using these definitions, exchange switches typically move either in discrete steps (digital) or continuous movements (analog). Let’s look at some examples of "digital" and "analog" exchange switches starting with digital.
The Strowger-type switch is categorized, for our purposes, as digital because it responds directly to a subscriber’s dial pulses (current on/off), advancing the mechanism's contact wipers accordingly. For example, dialing a digit 3 yields exactly three mechanical switch steps either vertically or horizontally. Each step is like clockwork, with unambiguous position control. Almon Strowger, Alexander Keith, Ericsson Company, and others invented digital switches in this context.
On the other hand, some switches are analog in nature and powered by continuously rotating motors and a drivetrain moves contact wipers/brushes. The gearing, shafting, and clutching mechanisms are non-trivial and reliably positioning and stopping a wiper/brush precisely on a desired terminal is not possible without some real-time position feedback. A "digital switch" does not generally need auxiliary position information, but an "analog switch" does.
The panel, 7A rotary, and Lorimar switches are examples of motor-based, analog switching in the context of this discussion. The discussion to follow uses the rotary switch to exemplify the position feedback concept. The panel switch real-time position feedback means and rotary means are implemented differently for sure but the control principles are the same.
Revertive feedback needed with analog switches
One definition of the word revertive is motion in a certain direction (OED). This term is used often in exchange literature when describing the real-time position feedback signal for (analog) switch control. Another commonly used phrase is revertive pulsing.
Fig A, Example of revertive feedback from a rotary switch
The image above shows a partial 7009 Rotary Selector made by Western Electric. Notice the brush trip spindle, toothed cam, and interrupter switch at the bottom of the image. As the spindle turns, under motor power, the interrupter switch opens/closes as it follows the cam. This spindle position signal (on/off pulses) is sent to the controller that keeps account of the exact position of the spindle. The spindle’s rotation is stopped by releasing a holding clutch (not shown) when the desired position is reached.
As an aside, the rotary switch rotates at ~15 positions per second. Compare this to a Strowger switch operating at 10 steps per second.
Below is a short video (rotary switch at [Ferrymead]) showing an interrupter switch riding a cam and providing real-time revertive pulsing as the spindle rotates. The important action occurs in the first 10 seconds, so a replay may help. If you look closely, you can see the spindle's electromagnetic clutch engaging and disengaging.
The panel switch also uses revertive feedback for contact positioning. The motor driven system employs a “commutator strip and brush” (refer to Fig 1, top of frame) to transmit revertive pulses to the controller. The pulses track a moving vertical selector rod’s position during travel time. The controller keeps track of the pulses, enabling precise start/stop accuracy of a brush for making terminal connections.
References for this Appendix
​
Ferrymead Post and Telegraph Historical Society, Switch Room, Christchurch, New Zealand. Thanks to Brian Cameron for the raw video.