In 1963, I started work with the GPO as British telecom was then known. After a few years with the exchange construction division, I moved over to Circuit Provision Group.
The work involved providing and setting up long distance telephone circuits from Portsmouth to other major cities throughout the United Kingdom. My work took me into various repeater stations in the Portsmouth Telephone area,
including the one situated within the Fort
Southwick tunnels and the GPO PRI
repeater station. The former was use to route telephone traffic generated from within the Fort complex to various destinations.
I will endeavour to explain the purpose of repeater stations in as non-technical language as I can.
In early days of telephony, all telephone traffic was carried over copper conductors in telephone cables. One pair of wires usually carrying one telephone conversation.
Copper wire shows resistance to electrical signals, and of course, over a long length, the power of the transmitted signal diminishes so that ambient noise on the line can swamp the signal altogether.
Certain criteria was laid down such that the loss of signal between a subscriber and his telephone exchange must not fall below -3dB, i.e. lose not more than half power. Loss between adjacent telephone exchanges was kept to 0 dB. (Zero loss).
When a telephone call has to be transmitted to another distant exchange, it is attenuated and for all circuits longer than about 30 km it will need amplification. Circuits connecting district and/or master switching centres were operated with zero overall loss. Group switching centre-to-district switching centres had a loss of 3.5dB, and group switching centre-to-local exchange junctions had a loss, which must not be greater than 4.5dB.
The required gain or amplification required cannot be provided simply at either the sending end or receiving end of the circuit, and so to achieve these figures, it is necessary to provide amplification at a number of points along the line. This is where the repeater station comes in. The circuits are fed into repeater stations where high-grade amplification is applied.
Another factor now has to be considered. I would be most uneconomical to provide one pair of wires in a cable per telephone call, either cable would have to be huge, or the number of circuits severely limited, thus proving uneconomical.
Consider for a moment the principle of Amplitude Modulation (AM) in radio signals. Here, a voice signal is superimposed, (modulated) onto a carrier wave of higher frequency, which is then transmitted. At the receiving end, the modulated signal is removed from the carrier wave and amplified if necessary before being feed to the recipient.
We can apply this principle to telephone traffic. The human ear is capable of responding between
300 Hz and 3400Hz, but if we then modulate a group of telephone signals on a range of higher frequencies, we can transmit several signals along the same pair of wires at the same time. These are known as 'Group Carriers' and a single group carrier would consist of twelve channels shared between frequencies ranging between 64 kHz and 108 kHz, each carrier frequency working within its own bandwidth so as not to over lap with its neighbour.
Let us take it a stage further, suppose we take 5 twelve-carrier groups and combine them to form a 60-channel 'Super Group'. Taking it one stage further, if 15 super groups are modulated over even higher frequencies, we can have 900 individual transmissions over our single pair of wires. These were called 'Hyper Groups'. Cable made of as many as 1080 pairs were manufactured and here we can see that such a cable would be capable of carrying no less than 972,000 conversations, quite staggering.
Repeater stations would have the necessary circuitry to monitor and maintain several racks of equipment, each rack comprising shelves of Group, Super Group, and Hyper Group circuitry.
Modern technology has seen the development of optic fibre cables, replacing the old and bulkier copper ones. Nevertheless, amplification of signals is still necessary, but the technology and quality of equipment has vastly improved from that of fifty or so years ago.
As with all forms of technology, advancements are made and systems improved upon, and the original carrier system was no exception. Engineers developed a system called Pulse-amplitude Modulation. (PAM). Again, multiple signals are transmitted over a single line, each signal is 'sampled' for a fraction of a second and that sample is transmitted to line. Each successive sample from following signals is then transmitted in turn, and the process is repeated over the whole range of signals. The sample rate is so fast, (in milliseconds), that the human ear cannot detect any break between each sampled piece of a signal. This system worked quite well, but then came the age of digital technology, and Pulse-code Modulation was born.
The signals are each 'sampled' as in PAM, but each sample is assigned a value in an 8-digit code, each 8-digit code describing the amplitude level of the sample. The sampling rate is 8000 times per second, and since each sample has 8 digits, there are 64,000 bits per second for each PCM channel.
A single speech channel made of an 8-bit PCM 'word' is generated every 125 microseconds (because the sampling rate is 8000 times per second). Again, the speed of 'sampling' is such that the human ear is incapable of detecting the breaks between samples, and the signal is received as one continuous stream of information.
There are many books containing pages of information describing these systems in detail, but I hope this brief explanation has not been too technical and has served to explain the reason for repeater stations.
Brian Wells (BT engineer retired) - 14 February 2004