TXE2- Telephone Exchange

Please excuse the layout and spelling of this page, more work needs to be done on it.

The POEEJ Vol 62 Part 1

April 1969

The TXE2 is explained in a 1969 edition of the post Office engineers journal

Electronic Telephone Exchanges: TXE 2-A Small Electronic Exchange System 

R. C. LONG, C.ENG., M.I.E.E., and G. E. GORRINGE 

The first type of electronic exchange to be used in large numbers by the British Post Office is a small exchange, covering the size range 200-2,000 lines. The space-division system is used, and switching is carried out with reed-relays. This article describes the trunking and operation of the exchange, the physical design and the features, both inherent and in the form of maintenance facilities, which enable a high quality service to be given to subscribers. 


INTRODUCTION The small electronic exchange system, TXE 2, has been designed for use as a local exchange in the approximate size range of 200-2,000 lines, and is now in full-scale production for the Post Office. It was designed under the auspices of the Joint Electronic Research Conunittee by Ericsson Telephones Ltd. in collaboration with the Post Office. The system is register controlled, and employs a multistage network of co-ordinate reed-relay switches. It gives a full range of subscriber facilities, and has simple growth capabilities within the designed size range. This range can be increased, should it be found to be necessary, after installation. The equipment is compact, has inherent reliability, is simple to install and maintain, and is compatible with the existing telephone network. 

SYSTEM OUTLINE 

The reed-relay with four dry-reed inserts is used in the switching network for the speech path cross-point and for logic purposes in the control circuits. It has a high reliability as the contacts are sealed against the atmosphere, a high operating speed, requires a small amount of space and lends itself to fully-automatic production techniques. Fig. 1 shows a typical reed-relay with four dry-reed inserts and twentyfive such relays mounted in a 5 x 5 matrix. When used as cross-point relays, two of the inserts are used for switching the speech path, the third for switching the control and testing wire (P-wire), and the fourth switches the operate and hold circuits. 

A basic switch used in the system consists of 25 reed-relays arranged to form a 5 x 5 cross-point matrix. Fig. 2 shows the exchange system in simple schematic form. The exchange can be divided into three main areas. (a) The switching area, with three cross-point-type switching stages, A, B and C, for originating and terminating traffic, and an additional switching stage, D, for terminating traffic. (b) The supervisory and register area with outgoing junction, incoming junction and own-exchange supervisory relay-sets, together with a common pool of registers serving alI three types. (c) The common control area with calling-line selection, speech-path selection and call-control equipment. Subscribers' lines are connected via their line circuits to A-switches. Outgoing supervisory relay-sets are connected to C-switches and incoming supervisory relay-sets to D-switches. 


Own-exchange supervisory relay-sets, which are used on local calls, combine the outgoing and incoming functions, and are connected between C-switches and D-switches. The speech paths between subscribers' lines and supervisory relaysets are connected by operating the cross-point switches in switching stages A, B, C and, when appropriate, in stage D. The A-switches, B-switch's and C-switches are assembled from basic 5 x 5 cross-point matrix and the D-switches from 4 x 5 matrices. Calls through the switching network are controlled by one centralized common control which, for security reasons, is duplicated and consists of a combination of reed-relays, semiconductors, and other solid-state components. Calls are processed on a one-at-a-time basis in a sequential manner. The relatively fast processing time of 50 ms ensures that there is very little delay in setting up calls which arrive almost simultaneously. When a subscriber calls, a path is set from the subscriber's line to either an own-exchange supervisory relay-set or a primary junction-route supervisory relay-set, from each of which access is given to the registers. The dialled information received by the register will indicate the route required. If this is different from the route initially selected at the time the call originated, the register discriminates during an interdigital pause, releasing the original connexion and establishing a new path to the required route. 


TRUNKING: General

In the A-switching stage, the basic 5 x 5 switch allows five subscribers' lines to be connected to five B-switches in the B-switching stage. Each B-switch has ten outlets giving access to ten C-switches in the C-switching stage. Thus, each subscriber has access to every C-switch and so to every supervisory relay-set. The trunking arrangements are shown in Fig. 3. 

TRUNKING: A-Switching Stage

The A-switching stage concentrates subscribers' traffic by commoning the A-switch outlets on to the A-switch to B-switch (A-B) links. The outlets of the A-switches are interconnected so that no two switches share more than one link. The interconnecting arrangement results in a good traffic handling capability and permits a mixture of high and low calling rate subscribers to be allocated at random to a group of A-switches. A typical A-stage group consists of 25 basic 5 X 5 switches connected on to 25 A-B links. The total traffic which can be carried on the 25 links in a group is 7·5 erlangs. 

TRUNKING: B-Switching Stage

Up to three A-stage groups may be connected to a given set of five B-switches, resulting in 5 x 10, 10 x 10 or 15 x 10 B-switches. The rack wiring is provided to cater for the 15 x 10 B-switch, but the switches are mounted as separate 5 x 10 plug-in units. Thus, increased traffic in 7 · 5 erlang steps can be catered for by additional plug-in switches. When the traffic is greater than 22 · 5 erlangs, the five 15 X 10 B-switches are fully loaded, and more than three A-stage groups are required, hence another set of five B-switches are provided and will grow as described for the first group of B-switches. When the traffic is greater than 45 erlangs, a third group of B-switches is required, and so on, to a total capacity of 240 erlangs 

TRUNKING: C-Switching Stage

Each B-switch has 10 outlets serving 10 C-switches. The number of inlets to the 10 C-switches depends on the number of 22·5 erlang B-switch groups. The number of C-switch outlets depends on the number of outgoing junctions, own exchange supervisory circuits and D-switch inlets required. Provision has therefore been made for the C-stage switches to grow in the two co-ordinate dimensions. This has been made easier by mounting the B-switches and the related co-ordinates of the C-switch on the same rack. 

TRUNKING: D-Switching Stage

D-switches have been introduced so that a terminating call can be given a choice of C-switches from which to find a path to the wanted subscriber, and thereby reduce traffic blocking. The D-switch gives own-exchange and incoming supervisory relay-sets access to five of the 10 C-switches 

Plessey Assembly 71610

SYSTEM OPERATION: ORIGINATING CALL

SYSTEM OPERATION:ORIGINATING CALL

ORIGINATING CALL The exchange equipment has two main functions: to provide a speech path between a subscriber's line and the supervisory equipment, and to select and set-up this path. The selection and setting-up procedure requires that the identities of both the calling and called subscribers are stored in the register. The calling condition must be detected and the line identified so that its connection into the switching network can be controlled. 

5 Line and A switch units able to support 15 Subs

Cards at Avoncroft Museum of Historic Buildings, Bromsgrove

The Line Circuit

Each exchange line is jumpered to a line circuit designed to operate to a loop calling condition from the subscriber's instrument. The operation of relay LR to this condition starts the calling-line selection and identification process (Fig. 4). The line circuit is identified by an equipment number, and any subscriber's line or directory number may be connected to any equipment nurnber. The line circuit handles both originating and terminating calls. Because the complete circuit requires change over contact actions on both the LR and K relays, the present equipment uses Type 12 relays(3) for these functions. Fig. 4 shows how equipment-engaged tone can be inductively coupled into the line feed, to ensure that the subscriber is not left without tone if network blocking prevents a connection being completed. 

Line and A switch unit

1A1/71200

Front of Unit

Location of A switch and LR and K Relays

The Calling Number Generator 

As a result of relay LR operating in the line circuit, the identity of the calling line in directory number form is generated and stored, so that, subsequently, the line circuit can be identified, or marked. On operation of relay LR in the line circuit, a current pulse passes in a jumper wire threading an array of pulse transformers (Fig. 5). This array, known as the calling number generator, consists essentially of four rows of pulse transformers, known as cores, each row representing, respectively, the thousands, hundreds, tens and units parts of the directory number. There are 10 transformers in each row, each of these representing a digit 0-9. The jumper is connected to the equipment number (EN) tag from the line circuit and passes through a start core, the appropriate thousands, hundreds, tens and units cores and is terminated on a similar EN tag at the bottom of the field. Where the local exchange uses 5-digit or 6-digit directory numbers, the last four digits only are generated, as these are sufficient to identify the calling subscriber. When the circuit is idle the capacitor C1 is charged to 50 volts (Fig. 4). 

Operation of relay LR in the line circuit causes the capacitor to discharge around the loop of jumper wire between the EN tags. The jumper forms a single-turn winding on each core through which it passes, and the output signals from the secondary windings of the cores are amplified and stored. To reduce the storage required for the calling-line identity, each decimal digit is coded into a 2-out-of-5 form. The K-relay contact performs two functions; on originating calis it ensures that when the LR relay is disconnected from the line, the capacitor cannot be charged, so that on clear-down any bounce on the switch-hook contacts of the telephone instrument will not cause spurious calling conditions into the exchange. On incoming calls, contact K3 slowly discharges the capacitor so that no pulse is generated. Thus, when the calling party clears, relay K releases and connects relay LR to the called subscriber's loop. Contact LRl is now ineffective. It is important for the exchange to be buffered against spurious calling signals, hence two start signals are generated in response to the pulse from the line circuit. These are S and SS as shown in Fig. 6. The start signal, S, is fed to an insensitive amplifier so that, unless a pulse of adequate amplitude is received, the call caru10t proceed. The second start signal, SS, is taken to a sensitive amplifier so that, in the event of signal SS being produced in the absence of signal S, a spurious noise condition is assumed with the chance that a partial code may have been generated. The store is then cleared as a precautionary measure. The outputs from the pulse transformers, through which the calling subscriber's calling number generator jumper connexion is wired, pass into a temporary high-speed store. From there the signals pass rapidly into a second high-speed store, and the process of selecting a free register is started. When selection is complete the calling line's directory number passes into the register store. The use of two high-speed buffer stores has two advantages; calls originated in quick succession can be distinguished, and time is allowed to select a free register, thus permitting the use of inexpensive reed relay access and storage in the register (Fig. 7). This arrangement can handle calls arriving at not less than 100 us apart.

The Calling Number Generator cont.

When the circuit is idle the capacitor C1 is charged to 50 volts (Fig. 4). Operation of relay LR in the line circuit causes the capacitor to discharge around the loop of jumper wire between the EN tags. The jumper forms a single-turn winding on each core through which it passes, and the output signals from the secondary windings of the cores are amplified and stored. To reduce the storage required for the calling-line identity, each decimal digit is coded into a 2-out-of-5 form. The K-relay contact performs two functions; on originating calis it ensures that when the LR relay is disconnected from the line, the capacitor cannot be charged, so that on clear-down any bounce on the switch-hook contacts of the telephone instrument will not cause spurious calling conditions into the exchange. On incoming calls, contact K3 slowly discharges the capacitor so that no pulse is generated. Thus, when the calling party clears, relay K releases and connects relay LR to the called subscriber's loop. Contact LRl is now ineffective. It is important for the exchange to be buffered against spurious calling signals, hence two start signals are generated in response to the pulse from the line circuit. These are S and SS as shown in Fig. 6. The start signal, S, is fed to an insensitive amplifier so that, unless a pulse of adequate amplitude is received, the call caru10t proceed. The second start signal, SS, is taken to a sensitive amplifier so that, in the event of signal SS being produced in the absence of signal S, a spurious noise condition is assumed with the chance that a partial code may have been generated. The store is then cleared as a precautionary measure. The outputs from the pulse transformers, through which the calling subscriber's calling number generator jumper connexion is wired, pass into a temporary high-speed store. From there the signals pass rapidly into a second high-speed store, and the process of selecting a free register is started. When selection is con1plete the calling line's directory number passes into the register store. The use of two high-speed buffer stores has two advantages; calls originated in quick succession can be distinguished, and time is allowed to select a free register, thus pern1itting the use of inexpensive reed relay access and storage in the register (Fig. 7). This arrangement can handle calls arriving at not less than 100 us apart. 

SYSTEM OPERATION: ORIGINATING CALL: Speech-Path Selection

Speech-Path Selection The subsequent testing, selection, marking and switching processes are performed on a one-at-a-time basis, and hence the register signals a de1nand to the call control for the exclusive use of the exchange common-control equipn1ent. When the register receives a signal fron1 the call control to proceed, it passes the calling subscriber's directory number to the decoder where it is changed from 2-out-of-5 code to a 1-out-of-N signal, which appears on a specific output lead at the top of a translation field. A jun1per wire from this lead to the line-circuit marker relay is threaded through an appropriate pulse-transforn1er core. The core generates the class of service of the subscriber, e.g. ordinary, shared-service, coin box, and passes this inforn1ation to the register. The class-ofservice signal enables a norn1al setting-up program to be modified when required. Operation of the marker relay in the line circuit operates the access relays which extend the busy/free condition of the five A-B links available to the subscriber into the B-switch selector (Fig. 8). The marker relay also defines the section or plane in the C-switches to which the calling line has access. A C-switch, with at least one free supervisory relay-set of the required type on its output, is chosen by the C-switch selector, and the five B-C links of the chosen C-switch in the plane defined by the calling subscriber are extended into the B-switch selector. The type of supervisory relay-set chosen for the initial connexion will depend upon the progran1 used in the particular exchange. It will generally be a circuit in the major traffic route. The B-switch selector will now attempt to find a free path bet\veen the A-switch and the selected C-switch. If a free path cannot be found, an alternative C-switch is chosen and the above process repeated until all possible paths have been tested. When a free path is found, a selection is made to determine which supervisory relay-set \Vill be used on the chosen C-switch. The selectors used in choosing a speech path are all similar in principle, having five inputs and five outputs. They operate in a sequential, non-ho1ning ma1U1er; the potentials on the inputs are exa1nined and one output, corresponding to a free input condition, is defined. Hence the circuit is referred to as a 1-out-of-5 selector. As a fail-safe feature, battery-testing principles are en1ployed to indicate free or busy conditions. 

The elernents of the B-switch selector are shown in Fig. 9. Since the selector is testing for link pairs, it is preceded by five, 2-input transistor AND gates. The 1-out-of-5 selection is performed by interconnected diode AND gates. 

The marking of the path to be S\vitched is shown in Fig. 10. The selected supervisory relay-set is niarked by the commoncontrol equipn1ent and as a result, the supervisory relay-set applies positive battery to the H-wire. The line circuit being marked has earth applied to the M-wire of its A-switch cross-point relays. Positive potentials are applied to the M-wires of the B-switch and C-switch by the selector and access circuits. Hence the C-switch cross-point operates in the C-switch, followed by the B-switch cross-point and the A-s\vitch cross-point. The operation of the A-switch cross point relay replaces the niarking earth with negative potential. The initial end-to-end holding potential is thus 100 volts, but the positive battery is subsequently removed, leaving the cross-point relays holding in series to an earth at the supervisory relay-set.4 The supervisory relay-set operates the K relay in the line circuit by applying an earth to the P-wire, thereby disconnecting the LR relay fron1 the line. The line is extended fron1 the supervisory relay-set into the register by nieans of a fully-available register switch. 

The register supplies linefeed current, returns dial tone and awaits dial pulses (Fig. 11). The time taken from the operation of the line relay to the return of dial tone is approximately 50 ms. 

Dialling

Dialling When relay A operates (Fig. 11), it operates the relief relay AA, a contact of which energizes the 690-ohm bias winding of relay A. The effective ampere-turns in relay A are reduced, since the flux produced by the bias winding is in opposition to the other windings. Hence, when the dial contacts open, relay A will release quickly. This reduces positive-pulse distortion on highly capacitive lines. The capacitor-resistor circuit connected to the bias winding of relay A eliminates contact bounce on relay A, and ensures that relay A does not flick-operate when relay AA releases. Diode D 1 ensures a short charge-time for the capacitor, while the 3 · 3 kohm resistor limits the discharge current, and hence protects contact AA 1. The register pulsing-relay feeds a 10-bit counter. An inter digit-pause detection circuit causes read-out of the counter via a 2-out-of-5 coder and distributor into a reed-relay store. Fig. 12 sho\vs the element in schen1atic fonn. Provision is inade for simultaneous pulse-repetition via the supervisory relay-set to a junction when required. 

Register Discrimination

The register, in conjunction with other co1nn1on-control equipment, can detern1ine the route required from the dialled digits. If this is different from the route chosen when the register connexion was niade, the supervisory relay-set releases the connexion during an inter-digital pause and the selection process is repeated. This ti1ne a path to a supervisory relay-set of the required type is established. The register can be reconnected before the arrival of the next digit. If the register finds that the outgoing route required was chosen on the original connexion, then it can transfer control to the supervisory relay-set and the register releases; any further digits are repeated directly by the supervisory relay-sets. The principle functions of the supervisory relay-set are to provide a transinission bridge, control the call, and, when appropriate, provide junction signalling. 

SYSTEM OPERATION: TERMINATING CALLS

SYSTEM OPERATION: TERMINATING CALLS Own-Exchange Call The system organization for connecting a calling subscriber to an own-exchange supervisory relay-set has been described above. When the complete nun1ber of the wanted line has been dialled and stored in the register, a new demand for call control is made and the sequence used for the initial connexion is repeated. The directory nu1nber of the wanted subscriber is passed to the decoder and translated to niark the line circuit of the \Vanted line. The class of service of the wanted line is determined, and also the state of the line circuit, i.e. free, busy, or parked. A line circuit is parked as the result of a pern1anent calling condition. If the wanted line is engaged or parked, the appropriate tone is sent to the calling line frorn the register. If free, the own-exchange supervisory relay-set handling the call is 1narked so that access relays in the associated D-switch operate and extend the five C-switch to D-switch links into the C-switch selector. The path-selection process proceeds as for the originating call, the wanted line being switched to the calling side of the supervisory relay-set. The register and common-control equipment are then released for further calls 

Incoming-Junction Call Seizure of an incoming supervisory relay-set causes an identity number to be generated in the calling-number generator in the same way in which, on an originating call, the calling subscriber's directory number is generated when a line circuit is operated. The subsequent program is very similar to that for an originating call in that the identity of the calling junction is transferred into a register. A priority signal is generated so that a register handling an incoming call will have preference in obtaining the use of the commoncontrol equipment. The identifying number is passed via the decoder to mark the inco1ning supervisory relay-set, and the register switch operates to extend the junction into the register. The number of the wanted line now passes into the register, and the procedure is then identical to that for an own-exchange call. Standardization of the switching sequences for originating, incoming and own-exchange calls has enabled the commoncontrol logic to be kept to a minimum, which is an essential requirement in an exchange system designed to be economic with only 200 lines. 

SECURITY AND MAINTENANCE 


Despite the need for economy, especially at the smaller sizes, exchange service has to be safeguarded under fault conditions. Security and maintenance features include the use of inherently reliable components with adequate working tolerances in the circuit design.5 In the setting-up and control of each call, use is made of equipment common to the whole exchange, so for security this equipment is duplicated. Under normal conditions the two sets of common equipment are brought into service alternately every eight minutes. If a fault occurs in one set of con1mon equipment, then this set is locked out and the other maintains service continuously. An alarm is given to the Maintenance Control Centre requesting the attendance of a maintenance engineer. To eliminate any possibility of a fault re1naining undetected for long periods, as could occur during a low calling-rate period, test calls simulating originating calls are automatically passed through the exchange after each periodic change-over. A feature of the system, which allows a repeat attempt to be made at setting up a path through the switching network if a fault is detected on the initial attempt, has a profound effect on the quality of service given to the subscriber. The primary object is to give service whenever possible. regardless of faults. Therefore, when a faulty speech path cross-point is detected, or a call fails to mature owing to a transient fault, the common equipment rnakes a second attempt to establish the call using a different network path wherever possible. The system incorporates many automatic fault-detection circuits. When a fault is detected, the particular equipment section affected is usually indicated. Accessible test points are provided to check the circuit condition before removal of the equipment unit. Faults that are detected in the system are automatically printed on a 5-inch wide paper tape. Ninety bits of information can be recorded on the tape to describe the state of the equipment when a fault is detected. The information includes the directory number of the calling and called subscribers, the speech path chosen, and the nature of the fault. Other maintenance aids designed into the system include the provision for setting up calls to any particular supervisory relay-set, displaying the directory number of the calling and called lines at the register, and call and path tracing. 

FACILITIES The facilities provided by a TXE2 exchange include all the facilities offered by the equivalent Strowger exchange, and in addition a number of new features. 

(a) A radical improvement in p.b.x. working has been achieved. Because any equipment number may be associated with any directory number, thecustomarypractice of reserving groups of numbers for p. b.x. lines is not required. Any size of p.b.x. group may be formed without the need for consecutive directory or equipment numbering. 

(b) Barring of outgoing or incoming calls is easily achieved. A particular exchange line may be barred by threading the jumper wire on the translation field to include the outgoingcall-barred core or the incoming-call-barred core. 

(c) Complete flexibility is provided between subscribers' equipment positions and the directory numbers of the subscribers. Therefore, if due to changing traffic conditions it is required to reconnect subscribers to a different section of the exchange, there is no necessity to change their numbers. 

(d) Subscriber-controlled transfer. When connected for this service a subscriber can, by dialling an appropriate code, arrange for his incoming calls to be transferred to any of a group of specified subscribers' lines. The controlling subscriber's telephone line is atlocated an ex-directory number as well as the directory number. Close associates of the subscriber can thus communicate with the subscriber's line even when the directory number is on transfer, as in the case of an off-duty doctor. 


In addition, anticipatory provision has been made for the following further features:

 (a) Keyphones. The calling number generator and register association arrangements have been organized so that keyphone registers may be readily added. 

(b) Freefone service. The arrangements for giving freefone service, which enables the called subscriber to pay for incoming calls, are being developed. 

(c) Calling-line identification. The availability of the subscriber's directory number in the register will assist in the later provision of calling-line identification to a centralized point, if required in the future. 

(d) Short-code dialling. Short-code-dialling equipment could be made available to the registers, which would then pass to the equipment's the identity of the calling line to enable the appropriate routing translation to be given when a preallocated abbreviated code is dialled into the register. 

SIZE RANGE

The system has been designed for use as a local exchange in the size range of 200-2,000 lines, or 240 erlangs of traffic. Developrnent work has now been completed on increasing the capacity of the exchange, to approximately 4,000 directory numbers, by physical redesign of the calling-number generator and class of service fields. However, the total traffic must not exceed the 240 erlangs Iimit. A method of connecting two exchange units to effectively double the capacity of the electronic exchange up to 8,000 directory numbers, or 480 erlangs of traffic, has been developed. The principles adopted en1ploy close interworking between the two exchange units so that the subscriber services are identical to those provided by a single exchange. Terminating supervisory relay-sets are given access to both exchange units, the unit required on tenninating calls being determined by the initial digit(s) received. Further developn1ent is planned to increase the capacity still further, and hence the field of application of the syste1n is continually expanding. 

CONCLUSIONS TXE2 exchanges are now in quantity production, and the exchanges that are in public service are justifying the claims advanced for electronic syste1ns. The n1aintenance commit1nent is much less than that required for a Strowger exchange of equivalent size. The service given to the subscriber is much improved due to the comprehensive fault-checking arrange1nents and repeat-atte1npt facility. The flexibility inherent in the systenm organisation gives the ability for future subscriber and service facilities to be provided when required. 

References 

I MARTIN, J. Electronic Telephone Exchanges: An Introduction to Switching Network Design. P.O.E.E.J, Vol. 60, p. 124, July 1967. 

2 ELEY, A. C., and LowE, W. T. Electronic Telephone Exchanges: Reed Relays for Exchange Systems. P.O.E.E.J., Vol. 60, p. 140, July 1967. 

3 ROGERS, B. H. E., and KNAGGS, A. The Post Office Type 12 Relay. P.O.E.E.J., Vol. 57, p. 27, April 1964. 

4 TIPPLER, J., LONG, R. C., and RIGBY, D. F. Electronic Telephone Exchanges: Speech-Path Control in Reed-Electronic Exchanges. P.0.E.E.J., Vol. 61, p. 103, July 1968. 

5 FRENCH, J. A. T., and LO\VE, W. T. Electronic Telephone Exchanges: Component Selection and Mode of Operation. P.O.E.E.J., Vol. 59, p. 233, Jan. 1967.