This paper has been written so that consideration can be given to the growth of the telephone network of the DFR. Growth is best catered for by the installation of more strowger equipment as the present staff have little interest in introducing more modern exchanges.

Growth in the forseeable future is likely at Lydney Signal Box, particularly for S&T in their new compound and C&W when their proposed carriage shed is built. Such growth may exceed the capacity of the cables feeding the Signal Box and an additional small exchange will be required there. It can be accommodated on the line concentrator rack which will provide the power and common services a small exchange will require.

Should the railway extend to Speech House Road, a new cable network will need to be provided along the railway extension. Should this cable extension be "bought in" then the conductor size may well be smaller than currently in use, putting Speech House Road out of limits for the Parkend PABX4 exchange equipment. Line limits are around 200 ohms. This may require a small exchange to be provided there, or if there are few circuits, long line relays could be fitted at Parkend.

When several exchanges serve the same general area it is usually most convenient to the user if all the exchanges share a numbering scheme that allows users simply to dial the number required.The alternative is to give each exchange its own numbering scheme and then allow interconnection by the dialling of code digits where the code digits will necessarily be different at different exchanges.

This paper uses the DFR layout to illustrate the arrangements and so is not very typical of GPO practice.

Direct Interconnections The simplest arrangement provides each exchange in the scheme with its own initial digit. Dialling the first digit of the wanted number then connects you to that exchange via a direct junction circuit. Such a scheme removes the need for any complex equipment and became the favoured scheme for some of the smaller non director telephone areas.

However the scheme requires each exchange to have its own outgoing and incoming junction routes to all other exchanges and the number of routes rises rapidly as the number of exchanges within the scheme increases. This produces many small junction routes between distant exchanges within the scheme, which individually are rather inefficient and make poor use of the junction cable pairs.

At present we only have three exchanges which would require six groups of junctions using this method of interconnection. If the DFR had a further exchange at Lydney Signal Box and all four exchanges were directly interconnected by junctions, we would require twelve separate groups of junctions. If yet another exchange were added, say at Speech House Road, we would require twenty groups of junctions. Such a scheme would be getting out of hand and require major cable laying works.

In practice there is another problem with the exchanges. All junctions would terminate on an incoming final selector, yet in Norchard there is only room for two such selectors. Parkend is a little better in that it can accommodate four incoming finals.

In practice therefore this simple scheme is not really suitable for the Dean Forest Railway should we desire to increase the number of exchanges in use.

We need to look back at the satellite schemes used by the GPO when they too had cable pair shortages but still required linked numbering schemes.

In general one exchange was designated as the "Central" Exchange and all the others in the scheme were known as satellite exchanges. Each satellite had its own junction route to the central exchange and calls to others satellites were made by using the central exchange as a tandem.

Such an arrangement appears to be much simpler and to make savings on line plant. However the line plant savings are offset by the need for complex discriminating equipment in the satelite exchanges. It can be seen that dialling 3, 4 or 5 at Lydney Signal Box requires a call to be made via a single junction group to Norchard. Norchard needs to know whether the caller dialled 3, 4 or 5 when the call arrives at Norchard and yet the number has already been dialled to get the call to Norchard. This implies that the Lydney Junction exchange would need to store the initial digit dialled and repeat it to Norchard. Then there would be a need to store the final two digits dialled as they would be received whilst the initial digit was being pulsed out to Norchard. Lastly the two stored final digits would need to be pulsed out to Norchard for the call to be completed. This was the sequence of events used in the final versions of the satellite exchanges.

Three main schemes were used.

Trombone Working This was the simplest arrangement and it did not last long with the GPO as it used junction pairs even when the call was between two customers on the same satellite exchange.

When the customer lifts his handset he receives dial tone from a first selector at the central exchange. It is then simple to route the call to any incoming final at any of the exchanges within the linked numbering scheme. The big disadvantage is that all calls that start and end at Lydney Junction would be routed via Norchard, thus using two junctions (tromboning) for the whole duration of the call. However for such a location as Lydney Signal Box, where most calls are to Norchard, the scheme would probably be quite effective.

Discriminating Repeater Working

This is a variant on trombone working which was fairly widely used with Siemens 16 equipment. When a caller lifts his handset, he is connected to a discriminating repeater relay set. This has two uniselector hunters. Both operate on seizure. The selector hunter rotates to find a local discriminating selector and the junction hunter looks for a junction to the central exchange. Dial tone is therefore obtained from the central exchange, unless all junctions are busy when the discriminating selector can handle a local call.

When the customer dials, his pulses are repeated over the junction to the central exchange first selector and also to the local discriminating selector. The discriminating selector has a vertical marking bank which allows it to know whether the call is local or junction.
If the call is for a local number, the selector tells the repeater to release the junction for other calls. The selector also restores and now acts as a local selector; in the case of Lydney Signal Box it would be a final selector. The junction will have been held in use only for a very short time.
If the call is for a distant exchange, the selector tells the repeater to release it and continue the call via the junction.

The repeater has the auto-auto facilities built into it and separate relay sets for each junction are not necessary. This is not quite the case, as each junction does need a P wire guard facilty to keep a junction busied for some time after it has been in use. This ensures that the equipment in the central exchange has completely restored before the junction is freed for a follow on call.

This is probably an over elaborate scheme for Lydney Signal Box, although it would work satisfactorily if implemented.

Discriminator Working

The most used and final scheme for satellite working in a non director area employed discriminators. A caller obtained access to a special first selector which in turn had a discrimator hunter which found a free discriminator. No junction was taken into use at this stage and dial tone was obtained locally.

The number was dialled via the first selector into the discriminator which stored it on a mechanical impulse regenerator. Once the discriminator had sufficient digits dialled in to be able to determine where the call was going, It could start to set up the call via the first selector.

If the call was local, the discriminator could absorb the local routing digit(s) and then repeat the rest of the stored number to the first selector. In this case the first selector will act as a second (or third) selector.
If the call was for a distant exchange, the call would be set up via an otherwise spare level on the first selector and the whole number would be repeated to the central exchange.
In either case, when the regenerator has finished pulsing out the stored digits, the discriminator is released for use on other calls.

This was an elaborate system requiring quite complex exchange equipment using mechanical regenerators, but it did make the very best use of the junction group to the central exchange. It also saved complete ranks of group selectors as one or possibly two digits could be simply absorbed by the discriminator.

In practice, the Asterisk exchange can be set up to mimic such a system.

Parkend Exchange

Parkend exchange is of the simple group selector type and does not have any discriminating features built in. The linefinder is also part of the first selector circuit and therefore even simple tromboning cannot be considered.

At present dialling 5 gives a caller access to junctions to Norchard final selectors. There is no way of altering this so that calls from Parkend can come into Norchard on a first selector without considerable changes at Parkend. At present we will need to add a level 4 junction to the Asterisk and a level 6 junction to Lydney Signal Box for outgoing calls. We do not have the cable pairs at present.

The options are to add cable pairs if possible, perhaps by conversion of the Norchard junctions to both way working. This really adds complication and the possibilty of junction circuits being held in a locked situation and is not a desirable way forward.
It is just possible to find two pairs for these two junctions by reorganising the line test jacks in the cabinets between Whitecroft and Norchard DP5C and providing a relief cable between DP5A and DP5C, though it might be preferable to take the cable all the way back to Norchard exchange.

Another option is to design and build new outgoing relay sets for Parkend which include discrimination and storage facilities. If level 5 were also to be included in the relay set we would not require any additional junctions and could reterminate the Parkend junctions on first selectors at Norchard and free the incoming finals for use as normal finals.

Such relay sets would need five inputs, from levels 4, 5, 6, 7 and 0 and storage of the next two digits. The set would generate the 4, 5, 6, 7 or 0 for transmission to Norchard and would then repeat the two stored digits. One minor effect of this system is that it will introduce a short delay between completion of the dialling and the reception of ring tone etc. The final result though would probably be the most satisfactory solution for Parkend.

Possible Linked Numbering Scheme for the DFR

Norchard will be the central or tandem exchange for the system.
Incoming junctions from Lydney Signal Box will terminate on incoming first selectors which will return the dial tone to the caller at the Junction.
Incoming Junctions from Parkend and the Asterisk will terminate on incoming first selectors which do not return dial tone to the callers.

Parkend will have the storage outgoing relay sets as described above for use on calls to Norchard, Lydney Signal Box and the Asterisk. Incoming junctions from Norchard will terminate on incoming final selectors, but incoming junctions from Speech House Road will terminate on the incoming first selectors already available..

The Asterisk will remain much as it is with calls to it coming from level 4 of the Norchard selectors. Outgoing calls to all exchanges will be made via Norchard first selectors.

Lydney Signal Box will be very simple and will trombone any calls via Norchard using first selectors which return dial tone for outgoing calls. The only selectors will be 2 digit finals attached to incoming junctions and will be of a simple uniselector type providing only a 22 line multiple.

Lydney Junction wil be designed when and if major development work occurs on the site. There is a five bank plus five bank selector shelf available which could form the basis of an exchange using two motion selectors.

Speech House Road will be a simple exchange tromboning calls via Parkend and will be similar to the Lydney Signal Box exchange currently under development.

Notes

1) Relay set performance in repeating digits may become more onerous as we will be using junctions in tandem and therefore taking two relay sets into use on some calls. It will be important not to design any arrangement that requires three auto-auto relay sets on the one call. Such a call is destined to fail unless regeneration is provided..
2) Transmission will probably not be a problem due to the short length of our railway line and the generous conductor size over most of the network.

the trunking diagram below shows how the DFR telecom network is proposed to expand. It involves the provision of a trombone satellite at Lydney Junction and the conversion of the Parkend exchange to a Regenerator satellite.

Display diagram as PDF (80Kb)

The provision of this network requires the following equipment to be built and installed :

1) A 22 line exchange at Lydney Signal Box consisting of
(a) 2 outgoing auto-auto relay sets.
(b) 2 incoming 22 line finals based on uniselectors.
(c) All common services are already available at Lydney Signal Box.

2) 2 auto-auto relay sets with regenerators at Parkend. These sets will have to be designed as modifications of the level 1 and 0 U13 relay sets. They will need five inputs (from levels 4, 5, 6, 7and 0) and storage of the two subsequent digits. The sets must then generate the 4, 5, 6, 7 or 0 as required followed by the two stored digits. ie All three digits of the called number will be received on a first selector at Norchard. Note that it is possible to have one junction from Parkend level 3 terminating on an incoming final at Norchard and only the second junction converted to a regenerator auto-auto relay set. This does mean that overflow level 5 calls, Asterisk level 4 calls, level 6 calls to Lydney Signal Box, level 7 calls to Lydney Junction and Level 0 calls to an assistance point would all be routed via one junction only. This would probably be quite acceptable and would certainly be an option whilst moving to a complete regenerator satellite scheme.

Other points to note are :

1) Long lines at Parkend to serve any possible telephones at Speech House Road could be served by long line extension relays at Parkend. These relays merely assist the L relay to work correctly to the high resistance line. Once the line is switched to a first selector, the line resistance will not be a problem. However, technically it may be more interesting to install a small Lydney Signal Box exchange at Speech House road, in conjunction with any line concentrator that may be required.
2) Incoming junctions at Norchard from Parkend will be switched from incoming finals to incoming first selectors as they are converted to regenerator working, making the incoming finals available for reconversion to ordinary finals if thought necessary.
3) Growth beyond the four incoming junctions is likely to happen, and Norchard will not cope as it stands, as there are only four incoming first selectors available. The situation could perhaps be covered by providing incoming junction hunters on all incoming junctions which will in turn switch to any of the four selectors available. Such hunters are described in Atkinson and are relatively simple. Possibly a better solution is to convert some of the local first selectors to incoming selectors as this gives the opportunity to strap permitted levels differently and as required on each incoming junction.

Possible Future Equipment Designs

Norchard Equipment

This circuit is based on that in Atkinson Vol 2 Page 521.

There can only be four working outlets on the banks of these hunters as there are only four available incoming first selectors. As most uniselectors available to the DFR have bridging wipers, it will be necessary to use every other outlet to avoid clicks from passing wipers bridging adjacent circuits. The hunters are non homing and will distribute traffic evenly over the incoming first selectors. The outlets could be wired to appear twice around the bank to reduce hunting time.

Display diagram as PDF (63Kb)

Further diagrams will be added as designed.

Page provided by John Bathgate This page was last updated on 24th March 2012