ERICSSON REVIEW 2 1986 Cabinet Construction Practice for Electronic Systems ERICOOL for Cooling Telecommunications Equipment Operational Experience of ERICOOL for Active Cooling ERICOOL Systems for Passive Cooling Automatic Teller Machine E281 Frequency Planning of Digital Radio-Relay Networks Modulation and Switching Using Optical Components in Lithium Niobate New Hardware in AXE 10 ERICSSON REVIEW Number 2 1986 Volume 63 Responsible publisher Gosta Editor Gosta Llndberg Neovius Editorial staff Martti Viitaniemi Address S-126 25 S t o c k h o l m , Sweden S u b s c r i p t i o n one year $ 16 Published in Swedish, English, French and Spanish with four issues per year Copyright Telefonaktiebolaget LM Ericsson Contents 42 49 52 58 63 71 80 • • • • • • • Cabinet Construction Practice for Electronic Systems ERICOOL for Cooling Telecommunications Equipment Operational Experience of ERICOOL for Active Cooling ERICOOL Systems for Passive Cooling Automatic Teller Machine E281 Frequency Planning of Digital Radio-Relay Networks Modulation and Switching Using Optical Components in Lithium Niobate 86 • New Hardware in AXE 10 Cover One of the first installations of AXE 10 in cabinet construction practice BYB202, in Sevenoaks just south of London, England. Fully built out the exchange will have 16000 subscribers, handle 97000 Busy Hour Call Attempts and have approximately 500 systems for 2 Mbit/s Cabinet Construction Practice for Electronic Systems Bengt Hellstrom and Dick Ernmark Digital electronic equipment is subjected to increasingly stringent environmental endurance requirements, particularly as regards electromagnetic compatibility (EMC) and the ability to survive earthquakes and high temperatures. New system components mean demands for greater heat dissipation ability. Cabinet construction practice BYB202 has been developed in order to satisfy new system applications as well as these new environmental requirements. The authors describe the criteria of the design work, the construction of BYB202 and how it is adapted to the functional units and handling units called magazine modules. packaging environmental engineering Digital technology is a comparatively new phenomenon in the telecommunications field. It was introduced into a branch which already had its own established operating conditions. The experience obtained from the use of digital technology has led to more stringent requirements as regards the environmental endurance of the systems. This applies above all to the electrical environment, i.e. the electrical and magnetic effects of the environment on the system and the corresponding effects of the system on the environment. However, the requirements for the equipment to be able to withstand high temperatures and mechanical stresses, such as earthquakes, have also increased. The requirements vary slightly in different countries but international standardization work is well on its way to achieving a uniform standard. Magazine The mechanical building blocks in a cabinet consist of magazines. Their height can vary between single and triple height and their length from 3 to 24 building modules, BM (one BM = 40.64mm). A magazine consists of a printed board frame with a rear plate that constitutes the wiring unit. The printed board assemblies are inserted into slots in the magazine, plugged into the connectors (usually two for each printed board) in the wiring unit and locked into place by a front rail in the frame. The connectors and frame slots are placed at fixed distances of six or eight modules, M, from each other (M = 2.54mm). All cabling is connected to the front of the printed board assemblies or to cross-connection fields which are situated in the wiring unit but accessible from the front of the cabinet. Internal logic voltages are distributed by the wiring unit, usually from a plug-in power unit. The need for high transmitted power and low tolerances as regards voltage drop has led to the use of busbars made of 0.25mm connecting wire or printed conductors for the voltage distribution in the wiring units. Greater need for heat dissipation Fig. 1 The modular structure of system AXE 10 has made it possible to exploit new technology and new system design for various subsystems. The figure shows how this has affected the space and power requirements ARF (crossbar switches) AXE10 The development in the component field is rapid and the new components have made it possible to reduce the volume of the hardware drastically. However, it has not been possible to reduce the amount of power per component in step with the volume. For the new components to be used with reasonable packing density the heat dissipation must be made more efficient. Fig. 1 shows how the volume and power have gradually been reduced for an AXE 10 exchange with 10000 lines. Uniform heat transfer is a main objective in the thermal dimensioning of a construction practice. The operating conditions for a magazine must be the same 43 BENGT HELLSTROM DICK ERNMARK Public Telecommunications Division Telefonaktiebolaget LM Ericsson regardless of its position in the cabinet. This has been achieved in BYB 202, and the limit for cooling by means of selfconvection has been extended by using a combination of series and parallel cooling. The maximum permissible power dissipation is 1200W per cabinet and the average power per printed board assembly 5.5 W with a board spacing of 8 M. Fig. 2 shows the temperature differences recorded in the cabinet. Environmental endurance requirements and environmental stresses Environmental endurance requirements specify the environmental stresses a product must be able to withstand without the function deteriorating. The requirements span all stages in the creation and life of a product. The product is exposed to a number of environmental factors which interact in complicated ways. For example, during handling, storage, transport and operation the product suffers vibrations and shocks as well as variations in temperature and humidity. The effects of the environmental factors on the product are dependent on the product design, en- Spread within the magazine 5 W/board with a spacing of 8 M Fig. 2 Excess temperature at printed board assemblies with power evenly distributed in the magazines. The excess temperature is the increase in temperature above the ambient temperature. The magazines must be given equivalent conditions regardless of their position in the cabinet vironmental control arrangements and the contributions generated by the product itself during operation. To be able to design a product with good environmental endurance and economy it is necessary to know both the type and the magnitude of the environmental stresses it can be expected to encounter Once the actual environment and the requirements for operational reliability are known the environmental endurance requirements can be drawn up. At the same time the environmental endurance of the product makes demands on the environment. It is therefore necessary to specify the environment, i.e. set environmental requirements. Various standards are used for this purpose, and the relations between environmental endurance and environmental stresses have to be treated systematically. Standards of importance in this context are various rules, regulations and instructions that specify parameters, values, materials, methods etc. Fig. 3 gives a summary of environmental concepts that concern the environmental endurance of the products. 44 The actual product environment Environmental factors and characteristic environments during the life of the products Fig. 3 Summary of the different concepts that affect the environment and environmental endurance of the products Climatic environment Handling environment Chemical environment Storage environment Mechanical environment Transport environment Biological environment Operating environment Electrical environment Temperature and humidity requirements The upper temperature limit for safe function is determined by the equipment installed and is normally 45°C. This upper limit is primarily intended to safeguard operation in the case of a breakdown of the air c o n d i t i o n i n g equipment. Administrations exploit this feature for various purposes, for example to repair c o o l i n g equipment w i t h o u t standby. The room temperature and relative humidity are measured at a point 0.4m from the equipment and 1.5m above the floor. Electrical environment Electronic equipment works with the aid of electromagnetic energy, which is transported over signal conductors between the circuits in a specified way. In addition to the desired electromagnetic energy unwanted energy can appear, so-called electromagnetic interference, EMI. The interference can be divided into two basic categories: - Interference caused by the equipment: emission - Interference imposed on the equipment: reception. Some countries with requirements fixed by law are: - West Germany (VDE, Verband Deutscher Elektrotechniker) - the United States (FCC, Federal Communications Commission). Fig. 4 An example of electrical environmental requirements (from FCC). uV refers to a terminal voltage (conducted) and uV/m to field strength (radiated) Class A (industrial equipment), measured at a distance of 30 m Class B (consumer equipment), measured at a distance of 3 m The requirements of different countries are often based on recommendations by CISPR (Comite International Special des Perturbations Radioelectrique). This is the c o m m i t t e e that, under the International Electrotechnical Commission, w o r k s on the standardization of matters c o n c e r n i n g interference of broadcasting and television. Fig.4 shows an example of such requirements. When designing the B Y B 2 0 2 cabinets an electrically screened version was also developed. Measurements show an attenuation factor of 50 dB in the frequency range 3 0 - 5 0 0 MHz, fig. 5. The screened version consists of a basic cabinet equipped with screening strips, rear doors and screens at the top and bottom. Fig. 5 The screening properties of cabinet construction practice BYB. The diagram shows that the screening factor is 50 dB in the frequency range 30—500 MHz. The interference source used is a 10 MHz oscillator and a number of loaded current loops Fig. 6 Typical time sequence for an earthquake in the most severe of the four zones specified for the US. The earthquake strength specification has been given as a frequency domain with its response spectrum. In this illustration the earthquake tremors are shown in the time domain, which gives a clearer picture The screened cabinet can be supplemented by special cable runs with similar attenuation properties. stand discharges of 15kV against the cabinet casing. All electronic equipment is to a greater or lesser degree sensitive to electrostatic discharge, ESD. Low relative humidity, combined with insulated flooring, contributes to an impaired operating environment for the equipment. A general recommendation for flooring is that it should have good antistatic properties and a volume resistivity of 10 7 -10 9 ohms in order to prevent ESD. When preparing the requirements specification for cabinet construction practice BYB 202 consideration was paid to its ability to withstand ESD. Verification tests have shown that the equipment can with- Mechanical environment BYB 202 has been designed to meet new requirements for mechanical environmental endurance during transport and operation. Transport can entail great stress for cabinets. Special packing and transport securing requirements had to be met to make it possible to transport fully equipped cabinets. The mounting of magazines and printed board assemblies has also been adapted to meet these requirements. This has resulted in a robust construction practice which can also withstand earthquakes. Fig. 7 The ceiling height required is basically determined by two factors. It must not be so low that it is a risk to the heat dissipation and there must be a reasonable working space between overhead cable ducts and the ceiling. The figure shows the recommended ceiling height and the cable running, using overhead ducts as well as floor ducts together with Ericsson's cable floor Technical data BYB202 is a modular cabinet construction practice for systems with magazines as the basic units. The external dimensions of the cabinet are height 2120 mm depth 400 mm width 1 720 mm width 2 1 200 mm Its main features are - high ability to withstand ESD - good electrical attenuation properties - efficient heat dissipation through self-convection - robust mechanical construction which makes it possible to deliver cabinets fully equipped and enables them to be installed in earthquake zones. Fig. 8 Cabinet BYB202 Certain countries make special demands as regards mechanical stress, for example the US where four earthquake zones have been specified. Fully equipped cabinets have undergone a number of tests based on the American requirements. The results show that the requirements were met, fig. 6. During the tests the equipment was fixed to a vibrating table by means of angle brackets. The same type of bracket is used for installing the equipment. Installation and layout criteria A construction practice must be flexible enough to be adapted to different customer requirements within certain limits. For example, the cable running differs, with cables laid in ducts on top of the equipment or in ducts on the floor. In the latter case the cabinets are mounted on raised or cable floor. BYB202 cabinets can be installed singly and the layout can be adapted to suit local conditions, such as previously installed equipment and pillar spacing. Other basic principles of the row construction practice have been maintained, such as full frontal accessibility for magazines, printed board assemblies, connection blocks and cables. This means that very compact layouts can be obtained by placing cabinets back to back. Two steel profiles have been chosen for the cabinet base in order to compensate for any unevenness in the floor of the exchange room. They have a bearing surface of 400x40 mm and are adjustable vertically by 10mm. This construction spreads the load over a large surface and avoids the pressure normally exerted by supports. Ericsson has designed a cable floor as an alternative to the conventional raised floor. This floor is recommended for new exchange buildings. The equipment is mounted on a floor framework having the same dimensions as the base of the cabinet. The cable floor covers only the space between cabinets and the immediate surroundings. This means that the equipment is earthquake-resistant, since the strength of the cable floor is no longer a crucial factor. The floor is 180 mm high, sufficient for screened cable ducts. The cable ducts, with their close-fitting lids, also provide good fire protection. Overhead cabling can also be used with BYB 202, and optional assembly kits can be used to adapt the height of the cabinets to existing cable routes and Ericsson's water-based cooling system ERICOOL.1 Fig. 7 shows the recommended ceiling height. Fig. 9 The space required for functional unit ETC has been reduced from two six-shelf magazine groups to a single-shelf magazine module Characteristics of a magazine module - is a handling unit for the optimum arrangement of magazines in a cabinet - is a functionally delimited unit - can occupy one or more shelves or comprise more than one cabinet - does not include cables or cabinets - is documented in the form of rules. Mechanical construction Functional units The cabinet, which mechanically is a self-supporting unit, is built up of a sturdy top and bottom frame. Brackets for the sides are welded to the frames, and the sides are screwed to the brackets. One or more magazines equipped with printed board assemblies form a functional unit, which constitutes the building block in the cabinet in AXE 10. Previously the functional units were assembled in standardized magazine groups which included the associated row mechanical structure and internal cabling. The great reduction in the volume of AXE 10 has facilitated new dispositions for optimum utilization of the construction practice. For example, the functional unit ETC (Exchange Terminal Circuit) for 512 channels, with the associated control equipment and power unit, previously comprised two row sections having a height of six shelves. Today the corresponding functional unit occupies just one shelf, fig.9. The fittings, which consist of a cable chute, earth bars, shelves and ducts for internal cabling, are screwed to the framework. The cabinet front consists of two doors, which can be opened 180° and lifted off the hinges if necessary. The cabinet can also be fitted with supplementary equipment, for example rails for 19" equipment. The earthing, which is an important function in the cabinet, is supplied by two vertical aluminium rails, which via contact plates are connected to the horizontal shelf earth rails. Earth wires are used for the earthing between cabinets and between cabinet rows. The earth system meets the requirements for 1000 A short-circuit current, which gives a maximum voltage drop of 0.5V. Fig. 10 The number of cabinets required for a local exchange with 11 000 subscribers, of which 6000 are connected to the parent exchange. The remaining 5000 are connected to six remote subscriber stages. The figure also shows the number of magazine modules per subsystem. The traffic intensity is 0.12Erlang per subscriber IOS Input/Output Subsystem CPS Central Processor Subsystem GSS Group Switching Subsystem TSS Trunk and Switching Subsysten. CCS • I Common Channel Signalling Subsystem Subscriber Switching Subsystem Magazine modules By treating the functional units separately, not associated with the cabling and mechanics, a handling unit is obtained which gives more versatile use of 48 Magazine module document (positioning rules) Fig. 11 Flow from the choice of product to the installation. The magazine module documentation is an aid for planning the positioning of magazines in the cabinets. Standard arrangements of magazines in cabinets can be prepared if customers so desire and be productized with all internal cabling the mechanics while retaining standardization and rational handling. This handling unit is called a magazine module. The magazines in a magazine module can fill a shelf, several shelves or several cabinets. The documentation for magazine modules can be considered as menus, describing which magazines are included, and giving disposition examples and any layout restrictions. The documents are used in the project planning and design of exchanges. When an exchange has had its functional content defined and has been converted to hardware with the aid of a computer program it is time to position the magazines in cabinets. Technical data Cabinet Height Width Depth Required ceiling height Aisle width Load on the floor 2133 mm 1 200and 720 mm 400 mm 2650mm 800 mm 4 kN/m2 Magazine 1M 1 BM Height Width N Depth Printed board spacing 0.1" (2.54mm) 1.6" (40.64 mm) 6 BM (234.8 mm) Nx3BM 1-8 220 mm 6 M or 8 M Printed board Height Depth 2 or 4 layers Magazine modules are the basic units used in the positioning, which can be manual or computer aided. Fig. shows the arrangement of magazine modules for a whole exchange. The next stage is to plan the layout and prepare a floor plan. When this has been completed, cable types are chosen and cable lengths are calculated forthe cabling within and between cabinets. At this stage basic data for the production can be prepared, so that the cabling can either be manufactured in the factory and delivered ready-made to the exchange or prepared on site by the installation staff. Computer aids are available for the whole process, from the choice of product to the production documentation, see the flow chart in fig. 11. Summary BYB202 is a versatile, modular cabinet construction practice for systems whose construction is based on magazines. Its first application is for AXE 10. The construction practice is characterized by good heat dissipation, high ESD endurance and ability to withstand great mechanical stresses. Supplemented with EMI protection and earthquake security devices it will meet the most stringent customer demands. 222 mm 178 mm Wiring unit Wrapped or printed board 2 or 6 layers Pin spacing 1M Operating conditions Temperature Normal range Safe function Relative humidity Normal range Safe function 5-40°C 0-45°C 20-80% 5-90% References 1. Almquist, R.: A Cooling System for Electronic Telephone Exchanges. Ericsson Rev. 58 (1981):4, pp. 1 8 8 195. ERICOOL for Cooling Telecommunications Equipment Erik Albertsson Ericsson has developed an extensive product range for cooling electronic equipment in both small and large premises. The ERICOOL cooling systems have been designed and dimensioned to maintain the stipulated component and plant temperatures in all existing types of climate. The author emphasizes the importance of uninterrupted cooling function and describes the two basic cooling principles, active and passive cooling, and the conditions that determine which principle should be chosen for a certain plant. cooling telecommunication equipment The requirements for operational reliability of telecommunications equipment are very exacting. A reliable system requires reliable subsystems and components. These requirements have led to the development of uniquely reliable computers for controlling sophisticated telecommunications systems. However, modern electronic circuits, with their high component density, have much higher heat dissipation. Efficient cooling has therefore become a prerequisite for satisfactory operation of the systems. Ericsson's product range for uniquely reliable cooling is called ERICOOL. Uninterrupted cooling function matching the high exchange reliability - is ensured by means of energy storage, a system structure with parallel circuits and duplication of vital components. The use of reliable components and dependable system designs also en- Active Low ERIK ALBERTSSON Ericsson Power Systems RIFA AB sures low maintenance costs for the ERICOOL cooling systems. Periodic preventive maintenance is usually sufficient. Low maintenance costs, low energy comsumption and a long life are three important factors that contribute to the good overall economy of the ERICOOL cooling systems. ERICOOL cooling systems work according to two main principles: - Active cooling, which means that the system is equipped with a special cooling unit and a pump or fan that ensures circulation in the system. - Passive cooling, which means that natural convection is used for the transfer of heat and that the cooling system uses and stimulates spontaneous thermal processes. Several factors influence the choice of the main cooling principle. The advantages of the purely passive systems are most apparent in the case of remote and unattended units. The deci- ERICOOL Local energy costs Passive High Standard of comfort Temperature requirements Component standard Fig. 1 Choice of cooling principle, general guidelines. The size of the exchange, environmental requirements and the local power costs are some of the factors that determine the choice of basic cooling principle. A remote link unit which is driven by relatively expensive power, for example from solar cells, is best cooled passively, whereas a large exchange with a good mains supply is normally cooled actively Large Exchange size Small Fig. 2 The ERICOOL passive cooling systems are extremely reliable. In this container for small telecommunications units the heat dissipated by the electronic circuits is used as the driving power. The cooling system has no moving parts, which makes it maintenance-free. The standby cooling capacity is unlimited since the system is not dependent on any external power sources Fig. 3 ERICOOL offers several cooling alternatives for equipment installed in containers. This active system can be extended in stages to a total cooling power of 5-10kW Fig. 4 ERICOOL Comfort offers silent and draught-free cooling of modern work premises - laboratories, conference rooms, banks and offices. The pleasant climate is obtained by natural convection whereby the air circulates through cooling coils and by maintaining a fairly high water temperature in the collectors (13-15'C). ERICOOL Comfort is available as a complete system including the cooling unit or as a supplement to an existing cooling system Fig. 5 ERIC00L cooling systems for large exchanges work with self-convection at the heat source. Heat Is transferred from the exchange room by means of the active method. These systems offer uniquely reliable cooling of exchanges having a cooling requirement of up to several hundred kilowatts. There is no capacity limit for the ERICOOL cooling systems sive factors are then the extremely high operational reliability and independence of external power sources. Large attended exchanges with a good power supply are usually cooled actively. In the power range up to 10kWthe factors that decide the choice of cooling method are the climate, comfort requirements and access to power. A mainly passive system supplemented by a simple circulation device may be a suitable solution in such cases. Some ERICOOL systems of different size and design are shown here. The fol- lowing two articles describe the latest development of each main principle and recent operational experience from some plants. Summary ERICOOL cooling systems are characterized by extremely high reliability with uninterrupted operation also in case of mains failure. Low maintenance cost, low or no power consumption and long life are other characteristics of the ERICOOL cooling systems which ensure a low life cost and good overall economy. Fig. 6 Modular structure with a capacity range of 2 to 10 kW (illustrated here) makes it possible to design reliable and economical cooling systems for small and medium-sized exchanges References 1. Almquist, R.: A Cooling System for Electronic Telephone Exchanges. Ericsson Rev. 58 (1981):4, pp. 188195. 2. Alexandersson, R., Junborg, A. and Vesterberg, H.-J.: Passive Cooling of Premises for Electronic Equipment. Ericsson Rev. 61 (1984):3, pp. 128131. Operational Experience of ERICOOL for Active Cooling Ragnar Almquist Since its introduction in 1980, ERICOOL has become a familiar concept in telecommunications administrations around the world. Its main features are uninterrupted cooling and good environment for both eguipment and staff. The author describes the experience gained from ERICOOL and the further development of the systems. Finally, new applications are discussed. Brief system description ERICOOL has been described in a previous issue of Ericsson Review1, and hence only a brief description will be given here. The system is based on self-convection. The heat from the telephone system is dissipated into the air which rises, transfers the heat to the cooling coils and then flows down into the aisles, fig. 1. cooling telecommunication equipment Since the introduction of electronictelephone systems, the thermal density has increased from 100 to 500 W/m2 floor area. There are no signs today that this trend will be broken; the density instead seems likely to increase. When AXE 10 was introduced it was realized within Ericsson that the amount of heat dissipated from the system would create problems. The company faced up to the fact and decided to develop a cooling system that met all the requirements telecommunications make on climate regulation. So far Ericsson is the only manufacturer of telecommunications equipment to develop its own cooling system. Fig. 1 The telephony equipment is cooled by means of self-convection. This method gives efficient cooling and a quiet and draught-free environment Fig. 2 shows the full system function. The temperature in the exchange room is kept constant by means of the mixer valve (MV), which controls the ratio of returning water to cold water. During a power failure the cold water comes from tanks having a volume that ensures a coolant backup which corresponds to the power backup period provided by the exchange battery. Experience from five years of operation ERICOOL was designed to meet the following requirements: 53 RAGNAR A L M Q U I S T Ericsson Power Systems RIFA AB High reliability The availability of a telephone system is very much dependent on the reliability of its auxiliary systems for power supply and cooling. 2 3 The auxiliary systems are integrated subsystems and their reliability should be on par with that of the exchange system. Simple installation The telecommunication projects of today are usually turn-key projects. It is important that all subsystems should be designed for rapid installation. As far as possible the installation should not require specially trained staff. Fig. 2 The modular structure makes for simple installation and enables the equipment to be extended in step with the telephone system W Circulation during normal operation v' Circulation during a mains failure CC PU MV WT CU Cooling coil Pump unit Mixer valve Water tank Cooling unit Simple extension It may be difficult to forecast the final capacity and extension rate of a telephone system. It should therefore be possible to build cooling systems with the required initial capacity and then extend them to suit future demands. A minimum of maintenance Modern telecommunications networks consist to an increasing extent of a large number of small, unattended exchanges and only a few large ones, from which all network supervision is managed. The maintenance of a cooling system must be kept to a minimum and performed periodically, for example once a year. Staff should not need special training in cooling engineering to carry out maintenance. These requirements are related to the experience obtained during five years of operation. ERICOOL now in more than 20 countries Since its introduction in 1980, ERICOOL has been installed and is in operation in more than 20 countries in all parts of the world and in most climate types (Nordic climate, desert climate and warm and humid tropical climate). There are now approximately 80 systems in operation (October, 1985). Most installations are complete systems according to fig.2, or the compact version tailor made primarily to meet the space requirements of telephone systems in containers. In some cases only parts of ERICOOL have been installed and combined with existing cooling systems. For example, in one case ERICOOL was to be installed in a building which already had a cold water system for the air conditioning of offices. The cooling coils and pump rack from the ERICOOL system were used, the cold water being supplied from the existing system. High reliability, interruptionfree One of the unique properties of ERICOOL is that the cooling function can be maintained during a power failure. Experience obtained on several markets - Saudi Arabia, Egypt and Pakistan - confirms the importance of this function. Fig. 3 shows the temperature curve for an exchange in Cairo exposed to a prolonged mains failure. During the years ERICOOL has been in operation temperature curves obtained during mains failures have also been reported for exchanges without a standby, i.e. where conventional air conditioning systems have been used. These reports verify that the temperature exceeds the permissible limit shortly after the loss of the cooling function. Fig.4 shows the temperature increase with different thermal densities.4 With only five years of operational experience it is difficult to obtain a complete evaluation of the characteristics and performance of the system. However, it is clear that the objectives set out during the development of ERICOOL - high reliability, simple installation, and low maintenance costs - have been reached. The concern expressed regarding the risk of water leakage into the electronic system has not been justified. No leakage has been reported from any system in operation. The features of ERICOOL that are most highly regarded by the users are, in addition to the high reliability, the low energy costs and the quiet environment, free from draughts. Fig. 3 Correctly dimensioned ERICOOL provides good protection against overheating during long mains failures. The temperature falls to the normal value when the mains power is restored Temperature during a mains failure Temperature after the mains power has been restored Comments from customers ERICOOL has met with interest from telecommunications administrations around the w o r l d . Some of the administrations that have installed the system have made their o w n evaluations which have formed the basis of published articles. Some excerpts are given here: Comments from Finland "ERICOOL was installed in Jyvaskyla in 1984. 3 The system was easy to install. The telephone exchange (AXE 10) is placed in a600 m 2 room in rock, of which the telephone system occupies 80m 2 . The space requirements would have been reduced by 200 m 2 if ERICOOL had been chosen at the start. The building costs w o u l d then have been reduced from 4.5 to 3 million Finnish marks. The modular structure of ERICOOL permits extensions. As a consequence the initial investment was low, which has kept the capital cost low. A minimum of maintenance, together with low energy c o n s u m p t i o n and heat recycling, gives low operating costs." Fig. 4 With no cooling the temperature in the exchange room rises. The rise in temperature is mainly determined by the thermal density of the telephone system. The curves refer to an outdoor temperature of 25°C Temperature, °C A Comments from Singapore6 "In 1984 ERICOOL was installed in the Tampineo exchange building to cool an SPC exchange of the FETEX type. Telecom's annual electricity bill for the cooling of t e l e c o m m u n i c a t i o n s equipment is 10 million Singapore dollars. Telecom is therefore continually seeking new, less energy-demanding cooling systems. The new system, ERICOOL, is more efficient than the systems used previously. An energy saving of 20% is expected. Other advantages of ERICOOL are uninterrupted operation and the absence of air ducts and raisedfloor systems." New system design gives even lower energy consumption The design work on the ERICOOL system started in 1979. The experience gained since then has been collated and discussed. Only a few modifications have been necessary since the first generation of the system was launched. Other possible improvements sugaested bv the onerational pxnprience 55 kept at a suitable temperature by cooling unit CU1. The standby circuit consists of cooling unit CU2, tank WT2 and mixer valve MV2. CU2 keeps the water in the tank at 8°C. In case of a mains failure or CU1 being out of operation, standby mixer valve MV2 is activated and regulates the ratio of water from WT1 and WT2 in such a way that the water flow to pump unit PU is kept at a temperature of 17°C. Apart from the features described above, the new system works in the same way as the earlier ERICOOL system. Cooling units CU1 and CU2 are identical except that they work with different water temperatures. Having CU1 working at a higher temperature increases its capacity by 30%. Fig. 5 The second generation of ERICOOL. The system design facilitates low energy consumption l» Circulation during normal operation Hi Circulation during a mains failure CC PU MV1 MV2 WT1 WT2 CU1 CU2 DC Cooling coil Pump Mixer valve 1 Mixer valve 2 Operating tank Standby tank Operating cooling unit Standby cooling unit Cooler (heat exchanger) are now forming the basis of the second generation of ERICOOL. The system is dimensioned to give the desired capacity at a water temperature of 17°C. The previous system was based on 8°C from both units. The curve of fig.6 shows that the relationship between cooling capacity, supplied power and water temperature is not linear, which means that the energy consumption has been reduced by approximately 15%. The most important features, the high reliability and energy saving, have been extended further through different system designs, fig. 5. Alternative system designs The temperature of the water to the cooling coils must not fall below 17°C. This limit gives a safe margin between the ambient temperature of the room and the dew point. Outdoor air can be used to cool the water in areas where the mean daily temperature falls below 15°C for eight months of the year. This method gives a considerable energy saving and higher system reliability. The way the system functions in the exchange room has not been altered. However, the cooling of the water in the system has been divided into two separate circuits: one for water at a temperature of 17°C (operating circuit) and one for water at 8°C (standby circuit). In normal operation the pump unit (PU) is fed with water at 17°C from the operating water tank (WT1) via the mixer valve (MV1). The water in the operating tank is Cooling by outdoor air requires a cooler (DC) that supplements or replaces CU1 in the system, fig. 5. DC consists of a cooling coil and a fan, is placed outdoors, and works as a heat exchanger between the water and the outdoor air. During the part of the year when the temperature is lower than 15°C,all cooling is done by DC. During periods with high outdoor temperatures, CU1 or the standby circuit are Fig. 6 ERICOOL operates at high water temperatures. This ensures better utilization of the cooling units and leads to energy savings Coefficient of performance (cooling capacity/supplied power),% of nominal operating data Fig. 8 ERICOOL for the modern computerized office. Here features such as high capacity and quiet and draught-free environment are very important used. A suitable dimensioning of the combination of DC, CU1, CU2 and WT2 gives the optimum design for low energy consumption, fig. 7. New applications The unique ERICOOL properties of silent and draught-free operation have now become even more important with the increased use of computers and terminals in offices. A normal office in Sweden has a thermal load of 20-30W/m 2 . This value increases to 50-200W/m 2 when computer and terminal equipment is installed. By comparison it should be mentioned that computer centres have thermal loads of over 300 W/m2. To ensure the comfort of the office staff, the air conditioning system has been divided so that cooling is arranged by means of cooling coils for self-convection (often placed on a suspended ceiling), whereas the fresh-air ventilation is arranged in the traditional way, fig.8. The use of conventional air-conditioning systems to handle the new cooling problems would produce an environment that would be considered Fig. 9 ERICOOL for telephony equipment. A first step towards an integrated system Fig. 7 In a temperate climate a cooler, DC In fig. 5, can be used during the greater part of the year. The cooling unit is only used for peak loads during the warmest period of the year and as a standby for the cooler. This system version has an extremely low power consumption, t = °C outdoors Top: The capacity of DC Is high when the outdoor temperature Is low but falls to zero at 17 C. CU2 provides the extra cooling that DC cannot manage. Centre: Up to 15 C only DC consumes energy. At higher temperatures the consumption Increases as CU2 starts operating. Bottom: During the greater pari of the year only DC Is In operation. The figure shows the number of hours during which the temperature exceeds a certain value. The climate Is Central European. The curve refers to Zurich. -Capacity of the cooler, DC Energy consumption Capacity of the cooling unit, CU2 Only DC consumes energy Both DC and CU2 consume energy Only CU2 consumes energy Conclusion The telephone systems of today have a heat dissipation factor corresponding to approximately 500 W/m 2 . The trend is rising, and in a not too distant future it will reach 1000 W/m 2 . Tests in ERICOOL laboratories show that self-convection can be used in both telephone packaging structures and cooling systems to dissipate these thermal loads. Fig. 9 shows a joint construction of telephone and cooling systems. The next stage in the development is most likely that the cooling system will be integrated in the telephone packaging structure, so that the heat will be taken directly from the source instead of being transferred first to the air and then to the cooling coils for removal. References 1. Almquist, R.: A Cooling System for Electronic Telephone Exchanges. Ericsson Rev. 58 (1981):4, pp. 188195. 2. Wolpert, T.: The Reliability of Power and Cooling Systems. Intelec 1982 Proceedings, pp. 181-186. 3. Wolpert, T.: The Reliability of Auxiliary Systems - Power and Cooling; Further Insights. Intelec 1983 Proceedings, pp. 109-116. 4. livarinen, R.: The Effect on Telephone Modernization on Buildings. CEPT AP/ GT6 1984, pp. 76-83. 5. Hakkinen, M.: Water Cooled Telephone Exchange, a New Concept. Puhelin 6/84 (Finland), pp. 26-27. 6. Boon, O. E.: Towards Cooling Exchanges at Less Cost. Hello - Newsletter for Employees of Telecoms. July 1984 (Singapore), pp. 5. ERICOOL Systems for Passive Cooling Rune Alexandersson and Anders Junborg Ericsson has developed methods for dimensioning systems for passive cooling of electronic equipment. Detailed studies of flows in and around heat sources and exchange rooms have provided the experience that forms the basis of optimum dimensioning of components and complete systems for passive cooling. The authors describe three ERICOOL systems for interruption-free cooling of small telephone exchanges and give a summary of the calculation methods. cooling calculation telecommunication equipment Fig. 1, left Container 500, basic diagram. The heat in the primary cooling circuit is transferred out to the secondary cooling circuit through the walls of the container Fig. 2, centre Cooling module 700, basic diagram. The liquid thermosyphon has two heat exchangers, an internal heat collector and an external cooler Fig. 3, right Container 3500, basic diagram. The liquid thermosyphon has been equipped with a thermal mass that provides cooling reserves and reduces the daily temperature variations The following three examples have been chosen to illustrate the application of passive cooling: - Air-cooled container for small exchange units that dissipate up to 500 W. - Liquid thermosyphon for cooling an existing exchange building. - 20-foot container with liquid cooler for a temperate climate and 3500W dissipated power. The common denominator of these ERICOOL systems is that they work without any external power supply and therefore offer extremely high operational reliability - in practice interruption-free cooling. The development of systems and components in the telecommunications field tends towards structures with high- er functional density and high heat dissipation. A holistic approach to the phenomenon of heat being conveyed from a component, through the room and out to the environment, provides new possibilities of efficient cooling of the electronic equipment. Cooling is being introduced closer and closer to the heat source. Modern telecommunications networks, with an increasing number of small remote system units, require very high operational reliability and maintenancefree equipment. Special consideration should be paid to the requirement for balance between battery reserves and cooling reserves. Passive cooling systems, which in most cases work without any moving parts and any energy consumption, are an attractive choice for such applications. The systems described in this article are based on natural convection in air and liquids. The heat dissipated by the electronic equipment provides the driving force in the cooling systems. When the hot air leaves the electronic equipment it passes an adjacent heat exchanger, which absorbs the heat and lowers the air temperature, whereby the airflow isturned down, recirculated into the room and onwards through the equipment. This air flow is the primary circuit in the passive cooling system. 59 RUNE ALEXANDERSSON ANDERS JUNBORG Ericsson Power Systems RIFA AB The heat exchanger that lowers the temperature also forms part of the subsequent system circuit, which in the simplest case consists of the air flows around the exchange room, for example a cabinet or a container. Container 500 is one example of this type of system, fig. i • The heat exchanger often consists of a collector that forms part of a liquid thermosyphon, which transfers the heat to the cooling coils placed on top of the container. The heat dissipated by the electronics is used as the driving force also for the second circuit. The thermosyphon has closed circulation. The liquid takes up heat in the collector, and the heat rises through pipes up to the cooling coils, where it is transferred to the surrounding air through the flanged coils. The cooled liquid is piped down to the collector to be heated again and keeping the circulation going. Cooling module 700 consists of such a simple thermosyphon, fig.2. Systems for exchanges that require redundant capacity or which experience large daily variations in the heat dissipation and outdoor temperature are equipped with a thermal mass in the form of a water tank. The thermal mass evens out Fig. 4 Container 500 Fig. 5 The diagram illustrates the thermal properties of Container 500. The blue curve shows the outdoor temperature, the other curves the indoor temperature. Passive cooling systems are often dimensioned in accordance with component requirements. Higher temperatures and larger temperature variations are permitted than for attended premises with comfort requirements. For example, Ericsson's environmental specification allows the room temperature to vary between 5 and 40°C the daily variations and acts as a buffer in case of unexpected events. The cooling system in Container 3500 is of this type, fig. 3. Container 500 for tropical climate The cooling system for Container 500 is built up around a sturdy metal minicontainer. The internal collector circuit consists of air conduits specially designed for the packaging structure, which guide the heat from the electronic circuits out to the container walls. The external cooling circuit consists of the air conduits between the sun screens that surround the container and the outer walls of the container. Container 500 is a purely passive cooling system without any power consumption and with extremely high reliability. Passive cooling system for small buildings ERICOOL also provides facilities for passive cooling in connection with the extension and modernizing of existing exchanges. A flexible component program makes it possible to adapt the cooling systems to local requirements and conditions. Fig. 7 The two diagrams show the daily temperature variations of the thermosyphon system in Brobyvaerk during one week in the summer (top) and winter (lower diagram). During summer all four thermosyphons are in operation, the outdoor temperature varies between 7 and 22 C, the indoor temperature is slightly above 20 C. Sunshine is indicated on the diagram by a separate curve. One or more thermosyphons can be shut off during the cold season. The second diagram shows the outdoor and indoor temperatures during a week in December. Only one thermosyphon is in operation and gives sufficient cooling Fig. 6 Chimneys and air conduits at collectors and coolers improve the performance considerably. This system with four thermosyphons has been equipped with chimneys around the cooling coils on the roof In Brobyvaerk, Denmark, an exchange has been modernized and supplemented by electronic equipment which dissipates approximately 3kW. The building is an insulated brick house. The cooling system consists of four simple thermosyphons. The collectors in the house are connected to the external cooling coils by means of pipes straight through the roof. The cooling coils are supplemented by chimneys, fig. 6, which improve the efficiency considerably. The basic cooling system is not equipped with any special regulators for continuous control of the circulation. However, the thermosyphons can be regulated individually by means of manual valves for seasonal adjustment of the system. Container 3500 for general use Container 3500 is built up around a 20foot container and is cooled by a thermosyphon system with several circuits. The collectors are placed on the ceiling and the cooling coils are situated outside, on top of the container. The cooling system also contains a thermal mass. The container can be equipped with external sun screens, which in certain types of climate prevent heating by radiation and stimulate the external cooling air flows. 61 Fig. 8 Container 3500 can be equipped with a maximum of four cooling modules comprising heat collectors, coolers and tanks for thermal mass In the basic version the thermal mass is placed in complete cooling modules on the roof. Each module then comprises two thermosyphons with collectors, tanks and cooling coils. Container 3500 has the tanks placed below the container, which has some advantages as regards installation. For example, it gives great flexibility in the positioning of the collectors. However, this design must be supplemented by small circulation pumps, powered by the exchange battery. The temperature reduction process is still entirely passive. Calculation method Fig. 9 Cooling module for Container 3500 Fig. 10 Container 3500 B gives flexibility in the location of the heat collectors in the container room. The cooling coils are placed on the roof and the tanks underneath the container One characteristic of all calculations for systems working with natural convection is that temperatures and flow rates are coupled and interact in a complex manner. This applies to everything from individual components to the entire cooling system, including the room and peripheral equipment. It istherefore necessary to start with detailed studies of, for example, the process between two flanges in a cooling coil and successively work towards a whole, balanced unit. This is done gradually by switching from the narrow to the wide perspective making detailed calculations that are then matched to the environment. There are no shortcuts or simple solutions. Numerical methods must be used. The basic data that govern the calculations for a passive cooling system are the heat dissipation of the electronic equipment, environmental requirements, the thermal properties of the building and the local climate. For example, the dimensioning of a system with a cooling circuit, collector circuit and thermal mass for a container involves: - Flow calculations for the cooling circuit in order to obtain a balance between two coupled equations: the momentum equation and the equation of energy. Expressed in a simplified way it is a question of achieving a balance between the heat dissipation ability of the cooling coils, 62 Fig. 11 Ericsson's dimensioning program for passive ERICOOL systems gives good precision in the design work. The diagram shows the relationship between the outdoor temperature and the calculated and measured temperature in the container "the motor", and the internal friction in the circulation, "the brake". Some factors that affect the efficiency of the system are the size and shape of the cooling coil walls, the running of the pipes and the use of chimneys. The calculations are iterative. Qualified guesses give suitable starting values, and repeated application of the calculation programs provides the data for a preliminary assessment of the cooling circuit. • The collector circuit is then dimensioned. This stage includes the important preliminary work of calculating how much of the heat dissipated by the electronics will immediately disappear through the floor, walls and ceiling. It is then important to have access to reliable climate data. Apart from this the collector circuit is dimensioned in the same way as the cooling circuit. - The thermal mass is calculated with the aid of special subprograms. The input data comprises critical compo- nent temperatures, daily variations in the outdoor temperature and in the heat dissipated by the electronics and any special requirements on the duration of cooling reserves, for example in the case of mains failures. Development trends Important subprograms in the calculation system concern individual components in the cooling equipment. The interface between the heat exchanger and the electronic equipment is now a subject of great interest. Chimneys of different types and special collectors for small temperature differences and low air velocities offer new solutions for efficient cooling systems. Modern electronic cooling is on the threshold of a new and interesting stage. The next generation of cooling systems will constitute an integral part of the electronic packaging structure. References 1. A l e x a n d e r s s o n , R., J u n b o r g , A. and Vesterberg, H.-J.: Passive Cooling of Premises for Electronic Equipment Ericsson Rev. 61 (1984):3, pp. 1 2 8 131. 2. Wolpert, T.: The Reliability of Auxiliary Systems - Power and Cooling; Further Insights. Intelec Proceedings (1983), pp. 1 0 9 - 1 1 6 . Automatic Teller Machine E281 Weine Bernfordt and Bertil Olsson Automatic teller machine E281 has been developed jointly by Ericsson Information Systems AB and the Japanese company Omron Tateisi Electronics in close collaboration. The main objective was to create a product that met exacting requirements as regards availability, security and ergonomics. These requirements tend to grow even more exacting, particularly in the Nordic market. Ericsson E281 has already been chosen by the savings banks as well as several commercial banks in Sweden. The authors give a summary of the development of automatic teller machines and describe the system components in E281. history point of sale systems etts The work on automatic teller machine E281, fig. 1, started early in 1983 as a part of the continual product development work. At the same time Ericsson received an enquiry from the Swedish savings banks regarding the development of a new generation of cash dispensers. Ericsson already had an established project organization which was responsible for the company's previous generation of cash dispensers. The project organization was given the task of studying the requirements specification from the savings banks and recommending a suitable action plan. Fig. 1 Automatic teller machine E281 as seen by the customer The requirements were found to conform in all essentials to the general market requirements and the trends that were apparent for the next generation of cash dispensers. In certain aspects the requirements were even more far-reaching and would require a large investment in development if they were to be met. When contemplating such a large development project the possibility of establishing close collaboration with another manufacturer of cash dispensers had to be considered. This manufacturer should above all have experience of the dispensing of notes and automatic service. Such experience, combined with Ericsson's knowhow as regards system structure, communications, security and ergonomics, was considered a prerequisite for the success of the project. During the spring of 1983 possible partners around the world were studied and evaluated. In the autumn of 1983 the decision was made to initiate negotiations with the Japanese company Omron Tateisi Electronics for a cooperation agreement concerning the development of an automatic teller machine. Omron is one of the leading manufacturers of cash dispensers in the Japanese market and since the end of 1984 the company has also established itself in the US market. A cooperation agreement was concluded with Omron and work started on adapting Omron's units to European requirements, primarily as regards function and security. Ericsson had to adapt the basic system of the machine to facilitate a wider use of public communication networks. Data communication in connection with financial transactions is a basic requirement for modern bank terminal systems, of which automatic teller machines are an integral part. History and trends Automatic teller machines, ATM, have increased rapidly in number during the last ten years. Approximately 150000 ATMs have been installed, with about 1/3 in the US, 1/3 in Japan and 1/3 in the rest of the world. WEINE BERNFORDT BERTIL OLSSON Ericsson Information Systems AB In the US the growth started in earnest towards the end of the 1970s and has continued along the same trend. In Europe the growth has taken place in waves dependent on local restrictions and the times when collaboration has been established between different ATM networks, fig. 2. However, at present the curve shows a smooth development trend for Europe. In Japan the growth has also been rapid, and the number of installed units iseven slightly larger than the number in the US. In Europe there are considerable differences between different countries. Sweden has the highest ATM density with approximately 150 machines per million inhabitants, a number exceeded only by the US, Japan and Hongkong, all with about 200. Sweden is followed by France, England, Finland and Norway with 110-130, whereas for example Germany and Italy only have approximately 40 machines per million inhabitants. In countries with a high ATM density the number of transactions, mainly cash withdrawals, is also high. An average of 5000-6000 transactions per month and ATM is common. Some machines can have over 20000 transactions per month. Fig. 2 The growth rate of automatic teller machines, ATM, in the US and Europe the US Europe Cash dispensers have been accepted by the customers in most countries, and the transaction frequency increases in line with the installation density. Thefollowing trends are apparent: - A changeover to pure on-line communication with the host computerof the bank in question. - Increasing number of functions for customer self-service. - Greater collaboration between banks, otherfinancial institutionsand card companies. - More and more machines are being installed in public places, e.g. in shopping centres, stations and on company premises. - Greater demands on the availability of ATMs and associated communications networks. - Demands for larger note capacity. - More stringent requirements as regards both logical and physical security. - Greater demands for good ergonomic properties, both for the customers and the operators. There are two main reasons why banks and financial institutions invest large amounts in cash dispensers and associated communication networks: - Competition, which is met by expanded service in the form of new facilities and extended opening hours, sometimes a 24-hour service. - Rationalization, with a desire to take the load off the cashiers. In both cases it is important that the machine has high availability and also that the customers' attitude to it is positive, since it is often their main direct contact with the bank. At present there are approximately ten major ATM manufacturers, the largest being American and Japanese. With E281 Ericsson will be one of the three largest in Europe. Automatic teller machine E281 An automatic teller machine is seen by most customers as a machine from which one can conveniently obtain cash by using one's cash card, state the desired amount and key one's personal code The fact that the machines also work outside ordinary banking hours, and are sometimes installed conveniently close, increase their popularity even further. Since the banks will in future also increase the facilities offered by the machines by, for example, balance enquiries, statement requests and transfer of money between accounts, the availability of the machines must be very high. The rationalization work of the banks includes efforts to enable the customers to carry out an increasing number of simple and uncomplicated transactions themselves, using automatic teller machines. Not many people are likely to realize that the above-mentioned customer service without any manual supervision requires a very sophisticated system design. The "customers" who are not satisfied with the amount of money they can legally obtain from the machine Outdoor climate - Temperature - Humidity Air pollution Sunlight Rain Snow Ergonomics and design Security Protection against • burglary • fraud Availability Indoor climate - Temperature - Humidity - Pressure differences between the indoor and outdoor climate User The public, 24 hours per day Protection against vandalism Maintenance - Operator System and function Fig. 3 An ATM is subjected to many requirements; the main one being that it must provide the public with a 24-hour facility for obtaining money Capacity Installation through a wall in an unsupervised environment might have wondered how this money is kept. The notes, new and used, often of several denominations, are kept securely in the machine safe, which is approved for the storage of very large amounts. It is not possible to manipulate the machine by means of forged cash cards or to trick the system in any other way. The built-in security feature, encryption of essential data in combination with verification of the personal code in the central bank computer, makes any such attempt impossible. However, the protection against different types of attack must be improved Fig. 4 The division of the E281 development work between Ericsson and Omron. The safe is supplied by a local manufacturer - Serviceman continually, and the manufacturers of both hardware and software must always be at least one step ahead of the cleverest criminals. In addition to unique security requirements there are very stringent environmental requirements made on automatic teller machines. They can be subjected to rather tough treatment from some customers and they must also be able to withstand large variations in the weather. E281 is designed for installation through a wall, outdoors or in an entrance hall. The machine is equipped with Ericsson's terminal computer and 66 Fig. 5 Network configuration E281 can either be connected to the local computer in System E2100 via a two-wire cable or be remotely connected via the telecommunications or data network to a host computer. Alternatively E281 can be used completely off-line thereby forms an integral part of terminal system E2100. E281 has a modular structure w h i c h simplifies operation and maintenance. The design of the units operated by customers and their positions on the front of E281 meet exacting e r g o n o m i c requirements. Fig. 6 Automatic teller machine E281 is built up of four main parts: a front, electronics cabinet, printer cabinet and safe The division of the development work between Ericsson and Omron is outlined in f i g . 4 . Certain stages of the development work required very close collaboration, such as the actual construction of the machine so that it satisfies the stringent environmental and ergonomic requirements, and the adaptation of Omron's units to European security requirements. System connection E281 can either be connected to the local c o m p u t e r in system E2100 via a twowire line or be remotely connected via the t e l e c o m m u n i c a t i o n s or data network to a host computer. Alternatively E281 can be used entirely off-line, fig. 5. Structure E281 comprises four basic parts: a front, electronics cabinet, printer cabinet and safe, fig. 6. When positioning the four parts in relation to each other particular attention was paid to: - the need for safe and reliable transport of notes from the note dispenser, through the wall of the safe and out through the slot in the front - the need for a simple means of transporting withdrawal statements from the printer and out t h r o u g h the slot in the front - security and g o o d e r g o n o m i c characteristics in the positioning of units operated by the customers - easy handling when replenishing the note store and c h a n g i n g paper and ribbon in the printer - simplicity in the replacement or repair of faulty parts. Front Fig. 7 shows the positions of the various units on the front. The o p e n i n g of the magnetic card reader is placed bottom right and the slots for notes and withdrawal statements are at bottom left. The visual display unit is placed above the keyboard on the right-hand side, and the two units are so close together that they can easily be seen in one glance. The keyboard is placed at a convenient height for the hand and in such a way that the customer hides the keyboard and VDU from persons q u e u i n g behind. Fig. 7 The front of E281 The front is equipped with a combined handbag rest and writing shelf at a convenient writing height. The front also has a handle, bottom left, as an aid for elderly and handicapped customers. The front is made of thick steel plate in order to prevent vandalism and entry into the bank premises through the machine. It is equipped with a flange for efficient sealing between the machine and the wall. A lamp illuminates the units operated by the customers. Electronics cabinet The electronics cabinet holds most of the units in the machine. A fan that removes surplus heat is placed at the top of the cabinet. The fan also creates overpressure in the machine, which prevents dust from being sucked in through the slots in the front. The units are mounted on guide rails in the cabinet in order to simplify maintenance. One pair of rails holds the following units: - Magnetic card reader and writer, together with return slot - Visual display unit for the customer - Control units - Voice guidance unit - Operator panel - Visual display unit for the operator - Battery Physical environment The parts of E281 that are placed indoors, i.e. the units in the electronics cabinet and safe, satisfy the requirements of Ericsson's STD EIS/ T1025-315. The parts placed on the front, e.g. the customer keyboard, meet very stringent demands as regards outdoor environment. Ambient temperature indoors outdoors Humidity indoors outdoors Voltage Frequency Electrical safety in accordance with and + 10 to +40°C - 4 0 to +50°C 20 to 85% 20 to 100% 220V + 10% 5 0 H z ± 2% IEC 380/435 UL478 A mains failure will not damage the machine. Battery backup is provided for memories and automatic program loading. The other pair of rails holds the journal printer and its control unit. Other units, such as the computer, floppy disc unit and power supply unit, are placed on shelves in the cabinet. Printer cabinet The receipt printer is mounted on guide rails and placed in a separate cabinet on top of the safe. Safe E281 can be equipped with different types of safes. The actual design and the choice of material for the walls are governed by local theft-protection regulations. The note dispenser and its control unit are mounted on guide rails in the safe. Burglar alarm equipment can also be installed in the safe. Communications E281 can be used on its own, off-line, or on-line, either connected to a local computer in a bank off ice or as a terminal in a network. The following types of connection are possible: - Permanent or switched circuits in the telephone network with signalling in accordance with CCITT Recommendation V.24/V.28. - Public or private data networks with signalling in accordance with CCITT Recommendations X.21 and X.25. - Standardized local two-wire bus (SS3) having a maximum length of 1500 m, a tranfer rate of 300 kbit/s and with up to 16 drop points per bus and local computer. Synchronous as well as asynchronous and bit orientated (HDLC/SDLC) protocols can be used. External communications can also be arranged by means of System Network Architecture (SNA) communication, which can be used towards permanent lines or X.25 networks. E281 has sufficient computer capacity to accommodate the software for SNA. Operating system The E281 operating system offers the usersawell-defined and secureenvironment in which several different application programs can be executed. The application programs and the core of the operating system are protected from each other by a function which ensures that the applications programs do not carry out unauthorized operations, e.g. overwrite other programs, and provides a user-friendly interface for the basic functions of the operating system. The operating system has the following subfunctions: - A real-time executive system, RTX, which handles the communications between processors, allocation of processor time, allocation of memory, real-time clock, calendar etc. The 68 RTX process communication is used throughout the system, including communication with a superior local computer. RTX also manages restarts after power failure without the user having to be aware of it. - Loaderfor loading E281, eitherfrom a floppy disc unit connected to E281 or - Technical data for units Computer - Central unit with connection of terminal bus (SS3) - Memory module for 256, 384 or 512 kbytes in RAM. Battery backup during mains failures - Input and output controller for one RS422 and three RS232 interfaces and a parallel interface, e.g. for video camera and alarms - Control unit for two floppy disc units - Transmission line interface for V.24/V.28 and X.24/X.27. Note dispenser - Two, three or four denominations - Up to 25 notes per transaction - A total capacity of more than 10000 notes. Visual display unit - 9" with green text on a dark background - 31 characters per line and 14.5 lines with normal character size - The size of the characters can be varied and combined in the same picture - The characters can be displayed flashing or reversed - A part of the screen can show a moving picture, which is achieved by shifting between four patterns - RAM of 40 kbytes for approximately 150 set pictures - Character generator for standard characters in PROM of 16 kbytes, for customized text in RAM of 8 kbytes and for moving pictures in RAM of 8 kbytes Customer keyboard unit - For personal identification and selection of function and amount - Equipped with a security module for identification of the customer's personal code. Sensitive data in the security module can be protected against access or manipulation by means of built-in security devices. The security module is programmable for easy adaptation to verification systems that are unique to the bank. - The security module can be used for encryption of sensitive information between the machine and the host computer - The keyboard has ten numerical keys, eight function keys and keys for READY and FAULT - The keyboard is designed to withstand vandalism and outdoor environment - In the numerical block the key for the figure 5 also has a dotted relief to aid people with impaired vision. Magnetic card reader and printer - For standard magnetic cards in accordance with ISO 2894 and ISO 3554 - Motor driven, with battery backup for the return of cards in the case of a power failure - Combined read and write head - Magnetic tracks in accordance with ISO standards o Reading and writing, track2, standard o Reading, track 1, option o Reading and writing, track 3, option - Watermark test in accordance with the specification of the manufacturer, EMI/MALCO, option - Receipt printer - Character generator for 256 characters in PROM - Matrix printing in both directions - Printing rate 100 characters per second, corresponding to 2.1 lines per second including line feed - Character size normally 7 x 9 or 8 x 9 dots and for enlarged print 7x 18 or 8x18 dots - Up to 40 characters per line with normal characters and 20 characters per line with enlarged characters - Up to 16 lines per transaction - Paper roll for 3000 transactions - Indications for paper almost finished and paper run out - Equipped with inked ribbon cassette - Semi-automatic loading of paper - Function for retrieval of forgotten withdrawal statements. Journal printer The technical data for the journal printer are identical with the data for the receipt printer with the following exceptions: - Only normal character size - Up to 36 characters per line - Up to 13 lines per transaction - Monitoring of the paper feed. - Floppy disc unit - One or two units, controlled by the control unit in the computer - 51/4" disc - A capacity of 0.65 Mbyte per unit. Voice guidance - Speech recorded on a magnetic drum - Maximum message length 3.5 seconds per track - Up to 32 messages, which can be linked together. Customer detector - Senses when a customer is within approximately 1 m of the machine. Can be used to start up the VDU or voice guidance. from a rigid disc unit connected to a superior local computer. During operation there is only one loader in E281. The loader carries out the initial loading of the computer as well as the reloading of applications programs, and hence applications programs can be changed during operation. A transport system, which handles internal data transfers between the computers in a system. Outside the transport system the system is entirely independent (except as regards performance) of the type of link used to interconnect the computers, e.g. fast wire bus, permanent circuit or switched circuit in a public data network Remote terminals are thus fully integrated in the system. Management functions for different interfaces towards external units. During operation each management function is stored in the computer. The following interfaces are available: o RS422 towards Omron units o Interface for RS232 units o Floppy disc interface. This function also starts file processing systems at the lowest level o Transmission adapter interface, which is used by the communications software o Parallel interface for control of external units such as video camera, lights and alarms. The parallel interface consists of two gates, each for eight bits. Each eight-bit gate can be defined as an input or output gate. A maintenance module, MM, which on request from an operator diagnoses the units in E281. The module is controlled from an operator panel at the rear of E281. A system diagnostics module, SDM, which handles the reporting of faults to the operator, gathers fault and operating statistics and on request presents the gathered information. Applications system All functions offered by E281 are controlled by the application program, written in the program language DIL. DIL has been specially developed for the management of units connected via a serial interface (RS) and of transactions. These properties make the language particularly suited for use in E281. 69 The application program is divided into a basic part, which is unique to E281, and a customizing part. In the basic part the properties of the operating system and the various units are optimized. The basic application has well defined interfaces for the customizing of texts and layouts for the VDU, the keyboard, printer, floppy disc unit, magnetic card reader and printer, parallel interface and transaction format. Customizing is done by encoding unique program sections and is controlled by means of parameters. The basic application also contains functions for the recording of alarms and forwarding them to a central system in the case of, for example: - Hardware fault in a unit - All notes gone - Form for withdrawal statements finished. The basic application comprises three types of functions: - System functions - Customer functions - Service functions. Access to the different functions is controlled via the operator panel at the rear of the machine and can be tied to an authorization check using magnetic cards or codes. System functions E281 normally contains the following system functions: Fig. 8 E281 is the result of collaboration between Ericsson Information Systems and Omron Tateisi Electronics Initiated from E281: - Insertion and removal of note cassette - Reading of cassette status - Loading of encryption keys in the security module - Loading of date and time for off-line operation. Initiated from a central system: - Opening and closing of E281 - Status enquiry - Reading of cassette status - Message to the VDU or statement printer. Automatic: - Service supervision and status reports. Customer functions The customer functions include: - Withdrawal of notes - Account balance enquiry or enquiry concerning recent transactions. The answer is shown on the VDU or received as a printout - Blocking of accounts, for example after loss of the cash card - Choice of personal code - Transfer to another account. Service functions Initiated from E281: - Testing of the units, using a standard module in the application program. The unit status is displayed on the operator panel or output on one of the printers - Unauthorized withdrawal from the note dispenser is prevented by means of a lock controlled by the basic application program. Initiated from a central system: - Statusenquiry to the units forthe purpose of checking their function. Security A cash dispenser must meet two types of security requirements, physical and logical. A combination of these is necessary in order to reach a level of security that is acceptable to banks. In general it is also necessary to establish administrative routines for the handling of sensitive modules, program 70 Fig. 9 Automatic teller machine 281 media etc. These routines must then be followed during the whole life of the design, i.e. during development, production, delivery, operation and maintenance. Physical security Physical security comprises protection against burglary and vandalism. It is essential to protect the large sums of money kept in the safe. The safe has been tested by the Swedish Association for Protection Against Theft and was given very high marks for the protection of valuables against different types of attack. The safe can also be equipped with an alarm which can be connected to the general alarm system of the bank. The units placed in the front are protected against vandalism in various ways such as by unbreakable glass in front of the VDU and shutters for the slots for the magnetic card and notes. Logic security The logic security is built into E281 and can be used for the following functions: - Verification of the personal code - Orders to the note dispenser - Authorization check for operators. Code verification The identification of a customer is based on the information stored on the magnetic card and on a personal identification number, PIN, chosen by the bank or the customer. The identification can be reinforced by a water mark number that is permanently stored on the card. Identification can be performed using a common algorithm, such as Data Encryption Standard, DES, which is key controlled, oran algorithm uniquetothe bank. The input parameters for the verification process are any algorithm key, card data, for example the card number, and the PIN. If the watermark method is used the WM number read from card track 0 can also be used as an input parameter. The result of the algorithm calculation is compared with a check sum computed when the card is activated The electronic circuits for the PIN verification, i.e. algorithm circuits, key register and comparator circuits, belong to the E281 customer keyboard unit. The encapsulated circuits are effectively protected against unauthorized monitoring or tapping of confidential information. Orders to the note dispenser A PIN verification carried out by the keyboard unit gives a positive or negative result. The result is interpreted and checked by the protected electronic circuits in the note dispenser. A correct positive result from the PIN verification is a prerequisite for an order to dispense notes. Authorization check for operators The authorization of an operator is checked via the operator panel at the rear of E281 and is used in connection with changeover between the system, customer and service modes. The identification is similar to the customer identification but a specially allocated personal code is used. Frequency Planning of Digital Radio-Relay Networks Heinz Karl The performance characteristics of digital radio-relay systems differ significantly from those of analog systems. This also includes the influence of interfering signals from other radio-relay transmitters. The planning of radio frequencies for a digital radio-relay network must therefore be based on other criteria than for analog systems. The author discusses new criteria of interference predictions for a digital radiorelay network. The paper has been presented as Ericsson's contribution at Bell National Radio Engineers' Conference in Denver. USA, September 1985. digital systems radio links interference network topology Fig. 1 BER versus receiver input level, LR,, for various values of CIR and co-channel interference (4FSK - 6/8 Mbit/S) CIR = 12dB CIR = 15dB i • CIR = 20 dB CIR = * BER The mechanism of interference In radio-relay networks the various radio paths are isolated from each other either by different radio frequencies or, when using the same radio frequency, by appropriate antenna discrimination and/or topographical isolation. Despite that, a certain amount of undesirable energy from surrounding transmitters will always interfere with the wanted signal at a radio-receiver input. The task of the network-planning engineer is to select radio frequencies and antenna types in such a way that the influence of interfering signals is within HEINZ KARL Ericsson Telecom Telefonaktiebolaget LM Ericsson the margins of the planning objectives for the overall performance of the radio circuit. The performance is expressed in terms of bit-error ratio, BER. The influence of interfering radio signals depends mainly on the f o l l o w i n g parameters: - The radio signal-(=carrier)-to-interference ratio, CIR, that is the level of the wanted signal with respect to the combined levels of disturbing signals. - The radio-channel spacing between disturbing and disturbed signals. The larger the separation between the two signals, the higher the attenuation of the disturbing signal in the radio frequency (r.f.) and intermediate frequency (i.f.) bandpass filters in the receiver assembly. - The type of modulation applied to the disturbed and disturbing signals. In order to economize on the frequencies allocated to a particular service, it is desirable to use the same radio frequency as often as possible. This will also be of benefit for the management of spare parts. This paper will therefore mainly be devoted to interference caused by transmitters operating at the same radio frequency as the disturbed path: cochannel interference, particularly at high radio frequencies. The influence of interfering signals When conveying TDM (Time Division Multiplex) signals, the influence of interfering radio signals on the performance is not detectable d u r i n g fading-free time, as long as the carrier-to-interference ratio, CIR, is at least 2 0 d B . It is only during f a d i n g , w h e n the receiver input level, LRx, comes close to its threshold level, LTe, that the bit-error ratio will be degraded. An example may explain this: Fig. 1 shows the BER versus LRi, for various values of CIR. The curves in this 72 figure are typical of a radio-relay link (RL) with a capacity of 6 or 8 Mbit/s applying 4FSK-modulation (Frequency Shift Keying between four frequencies). If we assume a fading-free input level L Rx = - 6 0 d B m , the bit-error ratio, BER, Fig 2 Simplified RL network, with the paths a b in a triangular configuration in a quadrangular configuration interference path will be better than 10~8, as long as CIR >15dB. A resulting level L,= -82dBm from all interfering signals at that receiver input meets this requirement with a good margin. If LRx drops to -67dBm, which gives a CIR of 15dB, the BER decreases to 6 - 1 0 6 (instead of <10~e for CIR = oo). If LRx drops just 3dB more, to - 7 0 d B m , CIR decreases to 12dB, and the limit for the planning objective of BER = 1CT3 is reached The influence of interfering signals is thus not noticeable until fading occurs and then as a deterioration of the receiver threshold level, LTe. This is the same as a drecrease of the path's fading margin, M. It is therefore of interest to discuss the correlation between fading of the wanted and the interfering signal Fading correlation between wanted and interfering signal a) When studying mutual interference within an RL network it is easiest always to consider two paths at a time, the disturbed and the disturbing path. These two paths are either part of a triangular configu rat ion, fig. 2a, or of a quadrangular one, fig.2b. Multipath fading Concerning multipath fading, the fading events for the wanted and the interfering signals are uncorrelated for all disturbing connections, whether the paths A - B and A - C or D-E and F-G resp. use the same radio frequency or not. Even in the case of co-channel operation, the fading of the interfering signal travelling from antenna A1 to antenna C, fig. 2a, is uncorrelated with that of the wanted signal A 2 ^ C, provided that the distance between antennas A1 and A2 is at least 150 wavelengths. For short paths, however, i.e. paths having a length of about 10 km and less, multipath fading may be disregarded. This is especially applicable to high radio frequencies (r.f. == 13GHz), as at these frequencies and short path lengths, fading caused by rain will always be the dominating planning parameter. Rain fading In the case of rain fading, the fading events for interfering signals from A1 to C, and from A2 to B, fig. 2a, are always 73 correlated with those for the wanted signals A2—»C or A1 —. B respectively. They can therefore be excluded from the interference considerations. The rain fading for the interfering signals B - ^ A 2 , C — A 1 , D ^ G and F*->-E is basically uncorrelated with the corresponding wanted signals. For short interference paths, however, a certain degree of correlation cannot be e x c l u d e d : If, as an example, the distances A - B , A - C and also B - C are about 5 k m , or less, the probability that a rain cell covers both the wanted path (e.g. B^>A1) and the interference path ( C ^ A 1 ) , or parts of them, has to be taken into account. The shorter the length of the interference path, the higher the degree of correlation. One possible way of taking this into consideration in the interference calculation w o u l d be to decrease gradually the influence of the interfering signal with decreasing length of the effective interference path; the effective interference path being defined as: diet, = d, sin(a|>/2) d, actual length of the interference path in km ip angle between interference and wanted path Fading correlation summary The uncorrelated events for fading of interfering and wanted signals are summarized in table 1. fluence of interference on the overall performance. Two different approaches to such a prediction are discussed here: - Starting from a given receiver input level, LRx, and calculating the influence on the performance - Starting from an allowed interference level, L h at the input of the disturbed receiver, and selecting the proper antennas and/or radio frequencies. Starting from a given input level This approach is illustrated by fig. 1 and fig.2a. Path A - B is assumed to be the disturbed path, and A - C the disturbing path. The length of both is assumed to be 4.5 km. The input level d u r i n g fadingfree time at A1 (and at B) is LR„ Using an RL equipment of the type Ericsson MINILINK18 (ML18), with a built-in antenna of 0.6 m diameter, a value of LRx = - 4 3 . 0 d B m is obtained. The receiver threshold level for BER = 1 0 - 3 and an undisturbed path (CIR = °°) is LTe. This can be taken from the equipment's data sheet. For ML18, the corresponding value is: LTe= - 8 0 . 0 dBm See also fig. 1. The flat-fading margin is t h e n : Interference calculation Table 1 Logical chart for uncorrelated events between wanted and Interfering signals for various fading events yes interfering signal considered no interfering signal not considered The principal work in drawing up a frequency plan is the prediction of the in- The interfering signal C ^ A 1 reaches the receiver A1 via antenna A1 with a level, L,, w h i c h , as an example, c o u l d be: L,= - 9 2 . 0 d B m Path Multipath Wanted/ interfering path length « 1 0 km A-»B/A2-»B A<~B/C->A1 A-*C/A1-»C A<-C/B-*A2 D-»E/F-*E D«-E/G^D F_»G/D->G F«_G/E-»F no no no no no no no no >10km Rain Effective interfering path length s=5km >5km yes yes yes yes yes yes yes yes no partly no partly partly partly partly partly no yes no yes yes yes yes yes If more than one interfering signal has to be considered, L, is the resultant level of n c o m b i n e d individual levels, L,,, namely: n L, = 10lg 2 1 0 L | | / 1 ° i=i Starting from the interfering level, L h of - 9 2 d B m in the diagram in fig. 1, a CIR 74 value of 1 5 d B can be f o u n d at a threshold level, LTel, of - 7 7 d B m . The CIR value for 1 5 d B corresponds t o the difference between these t w o levels: The threshold level has thus decreased by: LTe, is now the new threshold level in the presence of interfering signals: see f i g . 3 for ML18. A, = 0 d B for cochannel operation. t_pi has its previous significance. Formula (1) can be used for interfering sources with a transmitted spectrum similar to that of ML18, see fig.4. The formula applies to the range between BER = 10~ 3 and 10 7, and to degradation figures, D, of not more than 10dB, corres p o n d i n g to CIR = 1 2 d B . For the above example, i.e. for: n=1 A, = 0 (co-channel interference) L,= - 9 2 . 0 d B m This gives a new flat-fading margin, MR, of: To allow for a more handy estimation of the performance degradation, a computer-adapted formula has been introduced: D degradation of the threshold level, LTe, in dB n number of interfering paths K equipment parameter: K = 9 2 d B m for ML18 A, attenuation of the interfering signal in dB in the receiver, dependent on the r.f. spacing between disturbed and disturbing path, and the r.f. and i.f. filters in the receiver assembly, i.e. the same degradation as from the diagram. Differences between the diagram and the f o r m u l a are within the margins of accuracy for the formula. The same procedure has to be repeated to obtain the influence of interfering signals at receiver A2, and, in case of path lengths > 1 0 k m , also for the receivers B and C, see table 1. It should be noted that there is no such symmetry between the interfering levels at the t w o ends of a path as there is for the wanted signal: L,A1 =£ LIB whereas LRxA1 = LRxB This gives us t w o different fading margins, M R , f o r the t w o directions of transmission on the same path. The smaller of these t w o fading margins must be used for the p r e d i c t i o n of the performance. From the above we can conclude that the degradation of the fading margin by interfering signals must be known before the performance prediction can be completed. Fig. 3 Attenuation, A,, of an interfering signal in the receiver vs frequency spacing Another question is how to consider a partial correlation of the rain fading. In table 1 it was stated that a partial correlation could be applied for interference paths < 5 k m . One possibility could be to set the degree of correlation proportional to the effective length of the interfering path, related to 5 km, and 75 to take the same percentage for the degradation of the threshold level and fading margin: If, in the example, the interference path C to A1 has a length of 4.5 km and an angle ip = 90 degrees to the wanted path B - A 1 , the effective interference path would be: This would be 64% of 5 km and decrease the degradation D from 3.0 to 2.0dB. The fading margin, M R , could then be increased to 35.0dB. Fig. 4 Transmitted spectrum from an 8 Mbit/s, 4FSKmodulated radio-relay link The approach discussed above has the following disadvantages: - If radio links already exist in that particular r.f. band and within the geographical area concerned, performance predictions can be carried out only when all data concerning the links involved are known. - Each new RL has an impact on the performance of the existing ones. A check of their performance is necessary. To avoid the new RL degrading the performance of the existing ones to below the planning objectives, stringent requirements for discrimination of the new antennas may be necessary. For each additional link, the requirements will be more stringent or new radio frequencies will have to be used. - An interference level higher than the threshold level, LTe, (CIR = «>) may lock the receiver onto that interference signal in case its " o w n " transmitter at the opposite end breaks down. This connects a subscriber onto a non-authorized conversation or data stream, and in the case of frequency diversity it prevents the link from switching over to the diversity channel. The first two disadvantages can be overcome by allowing for ample fading margin for the first links in a network, to give "space" in their performance for a future degradation. The locking of a receiver onto an interfering signal can be avoided by planning for interference levels: Another method would be to provide the links either with pilot supervision, each link with its own pilot frequency, or to use different scrambling sequences for the various links. Starting from an allowed interference level This approach starts from the assumption of a maximum number of paths which can interfere mutually. In this case, the above formula (1) can be used to determine the maximum interference level contribution, Lh from each interfering path: D, K, A, and n have their previous significance. Assuming 12 interfering paths and an allowed degradation of the t h r e s h o l d level by 1 0 d B for the planned path, we obtain a m a x i m u m level for each interfering signal of: for co-channel interference. The value for LN can be increased by A, in the case of adjacent-channel interference. 76 The critical parameter in this approach is the parameter n, i.e. the number of paths interfering with the wanted signal. The definition interfering path can also be interpreted in the following way: n paths of all interfering paths can be allowed to interfere with a level of l_h. All other paths can be disregarded, provided that they contribute with an interference level of: The planning procedure is as follows: Calculate the planning threshold level for the receiver: and the flat-fading margin: The minimum CIR for each individual interference contribution is: Calculate the attenuation, A,, or the level, Li,, of each interfering signal and compare them with the requirements: - For configurations in accordance with fig. 2a, where A, is determined only by the antenna discrimination, AGA, in the nodal point A, i.e. This applies to the case where nodal point (A) disturbs outstations (B and C). It also applies in the opposite direction, provided that the fading-free input levels at A1 and A2 are equal. Otherwise see below. To keep the costs for the antenna down, it is essential not to use higher fading margins, MFI, than the BER planning objectives require. - For configurations fig. 2b: according to output level of the disturbing transmitter in dBm A0 AGTx AGRx free-space attenuation in dB between the disturbing transmitter and disturbed receiver antenna discrimination in dB for the transmitting antenna do. for the receiving antenna The parameters which can be chosen freely are AGTx, AGRx and LTx. Low LTx requires less AGTx and/or AGRx, but it also decreases the fading margin for its own path. Hence, the fading margin, MF|, should not beset higherthan required by the BER performance objectives. When applying this approach, performance calculation and interference calculation (frequency planning) can be performed more independently of each other. However, during the first stage of a network implementation the performance of the radio links will be a great deal betterthan predicted, as the prediction considers a completely implemented network. The problem is here to estimate the final extension of the network. If the estimation is too optimistic, and the network stays at e.g. 5 links instead of the planned 12 links, the 5 links will have too ample fading margins, and money may have been spent unnecessarily. Planning practice Both approaches to interference calculations have their benefits. The first approach is appropriate (and may give more economical solutions) for lowdensity networks, with only a few radio paths. Possible future extension could be provided for by using other frequency channels within the same r.f. band. In dense networks, e.g. metropolitan areas, where a continuous extension can be expected, the second approach will be the best one. In order to optimize the utilization of radio frequencies, i.e. to use as few different frequencies as possible, the following should be taken into consideration: For all those paths which operate with the same radio frequency, the path which at a repeater station or a nodal 77 Fig. 5 The radio-relay link tower at Ericsson's head office is the nodal point in Ericsson's radio relay network for data communications (station HF in fig. 6) point requires the largest fading margin, MFh (normally the longest path) sets the reference receiver input level (fadingfree time) at that station. For all the other incoming paths, - the receiver input levels at that station should have the same value as the reference level. This can be achieved by decreasing the transmitter output level at the opposite station. (To achieve symmetry between the two directions of transmission, both the transmitting and receiving signal at the opposite end have to be attenuated equally) - the antennas at that nodal station should have the same gain as for the reference path. A trade-off between antenna gain and receiver input level is possible If the above conditions are met, only the discrimination of the antenna in the nodal point (including polarization discrimination) determines how soon (in terms of the angle between two paths) the same frequency can be used again. With a CIRmm of about 24dB (ace. to the example above) and a fading margin M F |=12dB the necessary antenna discrimination would be 36dB. Applying the 0.6m built-in antenna in an ML18 radio link would allow for a reuse of the same r.f. beyond 12 degrees for the same polarization plane, and beyond 5 degrees for orthogonal polarization planes. Fig. 6 and fig. 7 show two networks, one star and one serial network, both implemented with Ericsson's MINILINK. In both cases the same frequencies have been used throughout. The transmitter output levels have been adjusted with attenuators in the outstations. Conclusions When summarizing the various aspects discussed above, some conclusions can be drawn: Correlation between performance prediction and frequency planning The correlation between performance prediction and frequency planning is much closer for digital RL than for analog RL, especially if paths are planned using the first method. Feedback of information between the two planning steps is essential, as the results from one step have an impact on those of the other, even if the latter has been performed some time ago. This may cause some administrative problems, not least if the frequency planning is the responsibility of a centralized, governmental Fig- 6 Example of a star network with optimized receiver input levels at the nodal point Computer centre Remote computer terminals Radio-relay link repeater 5V RF channel number and polarization Station Transmitter output power in dBm HF 15.0 Kl 13.0 KK 6.0 AL -1.0 VH -6.0 TN -11.0 Repeater 11.5 Input levels of the receivers in nodal point HF: L B x =-49.0...-51.2dBm agency, and the performance prediction that of the operating company. Nevertheless, at Ericsson we have taken the consequences and combined our hitherto independent computer programs for performance and interference calculation into one program, with feedback and loops between the various subprograms. Economical considerations Whether we use planning method 1 or 2, we can plan for either one of the following alternaves: - If we plan for a high degradation by interfering signals, e.g. 10dB, the necessary CIR will be low, and the requirements for the antenna discrimination will also be low. This gives small antennas. The high threshold degradation, however, moves the threshold level towards higher values, and thus also the value of the receiver input level during fading-free time. This necessitates a higher transmitter output level. - If we plan for a low degradation, e.g. only 1 dB, the necessary CIR will be high, requiring antennas with high discrimination. The transmitter output level can then be low. The question is where to spend the money: high transmitter output level (and the equivalent cost increase for a solar power device, for example) or large and highly discriminating antennas, which require more expensive masts. Frequency economy Frequency economy means utilizing an RL network with as few frequencies as possible. To achieve this all receivers in an RL station should operate at about the same input level, identical antennas being implied. Controlled transmitter output level When harmful interfering signals are present, the performance of a digital RL during fading is limited by the interfering signal. On the other hand, the Derfor- Fig. 7 An example of a serial network with optimized receiver Input levels In the repeater stations Local exchange Radio relay link repeater 1H RF channel number and polarization Station A B NC D Transmitter output power in dBm 15.0 15.0 12.0 0.0 Receiver input level in dBm -43.8 -43.8/-46.0 -46.0/-47.3 -47.3 mance of the path is always better than BER = 10~8, as long as the receiver input level is just a few dB above the threshold level. If the output level for each transmitter is decreased so that the receiver input level is 4 or 5dB above its threshold level, the level of the interfering signals at that receiver input will also be decreased Fading will be compensated by increasing the transmitter output level. In the conventional way, fading decreases the level of the wanted signal, while that of the combined interfering signals remains constant. If we operate with fading compensation by increased transmitter output level, fading will instead increase the level of interfering signals generated by that transmitter. That results in an increase of only one (or a few) interfering signal(s) at a time at the disturbed receivers. Since the nec- essary CIR will be the same for the combined level in the first case as for the individual level in the second case, the value of the individual interference level may now reach higher values. This can be used either to decrease the requirements for the antennas, or to increase the number of radio links in a particular area, or a combination of both. This approach seems to be the most promising way to improve the economy of radio links, both in terms of costs and frequency utilization. Today's technology should make it possible to design level-controlled transmitters without eating up its benefits. This paper is the result of discussions over a longer period between Mr. P-0 Gustavsson, Ericsson Radio Systems AB, and the author. Mr. Gustavsson has also contributed the degradation formula. Modulation and Switching Using Optical Components in Lithium Bo Lagerstrom and Bjorn Stoltz Collaboration between Ericsson Telecom and RIFAAB as regards integrated optics has led to the manufacture of advanced optical components in lithium niobate (LiNb03). Their applications include optical modulators for Gbit systems and switch arrays for directional coupling of optical signals. The authors describe these and other components and also the relationships between design and application. optical modulation integrated optics manufacture Optic communication systems using single-mode fibres as the transmission medium permit extremely high transmission rates. Optical switches with modulation rates of several GHz are now practicable using electro-optic crystal materials such as lithium niobate (LiNb03). During the last three years Ericsson has built up design knowhow and measuring equipment. CAD systems adapted for optical circuit design are used. The manufacturing engineering has been developed by RIFAAB, and the manufacture produces optical components with a very high performance level. Two applications are of particular interest to optic communications systems: high-speed modulators and switch arrays. Phase modulators constitute a special variant. They are used in coherent systems with the light frequency as the carrier at 300 THz. Such a system permits the transmission capacity of the optical fibre to be exploited to the full. The switch array is used for directional coupling of optical signals. Its operation Fig. 1 Waveguide channels, electrodes and polarization directions in a crystal with a)Z-cut and b) Y-cut Electrode Optical waveguide Electrical field lines is not dependent on the frequency and encoding of the signal that carries the information. Properties of the material Lithium niobate is a negatively uniaxial crystal with high double refraction. By double refraction is meant different refractive indices in different directions. The ordinary refractive index is n0 = 2.221 and the extraordinary ne = 2.145 for a light wavelength of 1.3u.m. The crystal has the interesting property of combining good transmission over a large wavelength range with large electro-optic coefficients. This means that the refractive index is effectively changed by an applied electrical field. It is also possible to manufacture optical single-mode waveguides having low losses by doping the crystal with titanium. Lithium niobate is therefore widely used in the field of integrated optics for the manufacture of opto-components for light wavelengths of between 0.6 and 1.5|im. The change in refractive index with an applied electrical field can be expressed as where n is the refractive index, r the electro-optic coefficient and E the applied electrical field. The largest coefficient, r33, of 30-10~12m/V, is normally used in order to ensure low operating voltages for the components. In this case both the optical and the electrical fields must be orientated along the optical axis (zaxis) of the crystal. Two different cuts of the crystal can thus be used, fig. 1. In one case the crystal is cut with the surface perpendicular to the z axis and the light waveguides along the y axis (Zcut). The light is then TM polarized and the electrodes must be placed on top of the waveguides. In the other case the crystal is cut with the surface parallel to the z axis and the light propagation along the x axis (Y cut). The light is then TE polarized and the electrodes are placed bv the side of the wavenuides. 81 Bo Lagerstrom Ericsson Telecom Telefonaktiebolaget LM Ericsson Bjorn Stoltz RIFAAB Both crystal orientations are used, but Z-cut crystals usually give simpler electrode structures and better coupling to the optical fibres. The components described in this article are manufactured on Z-cut lithium niobate and are polarization-dependent. Design and application An optical switch is usually based on a directional coupler. Fig. 2 shows how the beam is switched between the two adjacent light waveguides. The switching is controlled by electrodes along the interactive area. The design of the electrodes controls the frequency characteristics and efficiency of the component. The two inputs are independent of each other. If the switch is in the cross state with respect to one input, the same state applies to the other input. Fig. 3 An 8x8 array with 64 switch points in a busbar structure. Each switch point is a directional coupler controlled by an operating voltage of approximately 30 V Fig. 2 An optical switch is usually based on a directional coupler. Theoretical simulation of the circuit function is done by means of the beam propagation method (BPM) and shows how the light is switched between the adjacent waveguides The optical waveguides are designed for single-mode transmission, low propagation loss and a light distribution that suits a single-mode fibre. Both the coupling length in the interactive area and the coupling efficiency to the fibre can be optimized by varying the width and depth of the waveguide channel along the direction of propagation. A typical waveguide channel for a wavelength of 1.3 urn is approximately 5^m wide and 3 - 5 urn deep. A theoretical simulation of the circuit function can be performed using a numerical method, the beam propagation method, BPM, which calculates the propagation of the beam in the crystal. The input data for the program consists of production parameters and information concerning the geometry of the circuit. The BPM results include the coupling length in a directional coupler, crosstalk between light channels, bend losses and coupling losses. The optical attenuation is dependent on the length and complexity (curved structures) of the component. The major part of the design work is the designing of the electrodes. Two essential parameters are the products of frequency and length and of operating voltage and length. Both the operating voltage and the frequency are inversely proportional to the length of the component. Typical values: - Frequency times length - Operating voltage times length "lOGHzcm 8Vcm One of the consequences of this type of component is the element of trade-off that exists between operating voltage and bandwidth. A shorter component permits higher frequencies but also re- Fig. 4 A non-blocking 8x8 switch array contains 64 switches. Each directional coupler is 2 mm long and controlled by an operating voltage of approximately 30 V Electrode Optical waveguide Table 1 Number of Inputs or outputs 32 Number of LiNb0 3 8 x 8 chips 3 64 128 256 10 25 65 quires a higher operating voltage, which is difficult to generate. On the other hand the voltage does not set limits for a switch array that works at low switching rates (MHz). Fig. 6 The electrode structure for a directional coupler modulator, in the form of a decoupled transmission line, gives an impendance of 50 ohms Transmission line Electrode Optical waveguide Present-day LiNb0 3 materials make it possible to manufacture components having a length of 60mm. An 8x8 array with an optical through-connection loss of 5 - 7 d B has been manufactured. It contains an array of 64 switch points, fig.4. Each switch point is a directional coupler controlled by a voltage of approximately 30 V. The array is such that a free input can always be connected to a free output without traffic between other inputs and outputs having to be rerouted, so the array offers full accessibility. Several arrays can be used to build up large networks in such a way that full accessibility is retained. The arrays are best arranged in multi-link systems, a structure which is also called a Clos network.2 Table 1 gives the number of LiNb0 3 chips needed to build up large Clos networks. Other structures than arrays and Clos networks may be more suitable in cases where short interruptions in the traffic Fig. 5, right The directional coupler can also be used as a modulator. In order to achieve high modulation frequencies the electrode is designed as a transmission line for microwaves, a travelling wave electrode Fig. 7, far right An optical Mach-Zehnder modulator. The electrode is constructed as a coplanar transmission line, which gives a bandwidth that is limited to approximately 7 GHz by phase disparity between the electrical and optical wave propagation over established circuits are permissible. The directional coupler can also be used as a modulator. The electrode is then designed as a transmission line for microwaves, a travelling wave electrode, in order to achieve high modulation frequencies. The electrodes must be wide and thick so that electrical losses are minimized, which leads to a design with decoupled electrodes in order to ensure low impedance (50ohms) and operating voltages, fig. 6. Using this decoupled electrode design, a component has been manufactured which has a bandwidth of 3.0GHz and a switching voltage of approximately 8V. The bandwidth is limited by the electrical losses in the transmission line and the theoretical limit is approximately 5.5GHz for a decoupled transmission line. An alternative to the described modulator is provided by the Mach-Zehnder modulator. In this the beam is split between two waveguides, and an optical path length difference is introduced by means of an electrical field. This gives rise to constructive and destructive interference - on and off state respectively in the modulator - at the point where the beam waveguides are re- Fig. 8 A push-pull design for the electrode in a MachZehnder modulator ensures a low operating voltage (lower picture). Phase modulation in an interferometric structure (Nlach-Zehnder) gives constructive/destructive interference, i.e. in/off states of the modulator (top picture) Electrode Optical waveguide joined, fig.8. A push-pull construction results in a low operating voltage, 3 - 5 V . In this case the optical geometry permits a coplanar transmission line, which gives an even greater bandwidth. For an electrode of 1 cm the bandwidth is limited to approximately 7GHz by phase displacement between the electrical and optical wave propagation. Fig. 9 The diagram shows how switches S1 and S2 switch the data bit stream and how the highspeed modulator (XM) modulates the beam from the laser. In the passive state the beam passes straight through S1 and S2, the bypass function F Wideband amplifier Optical waveguide Operating voltage Electrode An optical fibre data bus has been designed using opto c o m p o n e n t s in LiNb0 3 . An optical bypass function prevents system cutoff if individual terminals are switched off. It also permits expansion of the data bus without interruptions to the traffic. This bypass function is provided by directional couplers arranged in accordance with fig. 9, giving an overall loss (fibre-chip-fibre) of less than 5 d B . Low-frequency directional couplers (S1 and S2) and a highspeed modulator (XM) for 2.4Gbit/s are integrated on one and the same chip. When the c o m p o n e n t is active the data bit stream is connected via S1 t o a d e t e c tor for demultiplexing and signal processing. The terminal unit (optical chip, detector and laser) repeats the data bit stream continuously and feeds out data generated by the terminal. In the passive state, the bypass f u n c t i o n , data passes straight t h r o u g h S1 and S2. A modified version can be used for bit processing and works as a MUX/DMUX or an interface between high and low data rates. Manufacture Lithium niobate, LiNb0 3 , is available commercially in the form of polished wafers, 3" in diameter and 1 mm thick. Optical c o m p o n e n t s are so large that the chips are usually handled individually. The first step in the process is therefore to cut the wafer into a number of suitable chips, usually 2 - 1 0 per wafer. 84 Fig. 10 Manufacture of LiNbO, waveguides a resist deposited b exposure through a chrome mask c development d evaporation of titanium e resist lifted off f diffusion Waveguides are created by diffusing titanium strips into the crystal. 60-90nm titanium is evaporated on to a photolithographic resist pattern on the chip, fig. 10. When the resist is lifted off, only the titanium that is to form the desired optical waveguide channels remains. The chip is heated in oxygen for five to seven hours in a temperature of 10301050°C, whereby the titanium is diffused into the crystal. The increase in refractive index and the depth of diffusion are controlled by means of the titanium thickness and diffusion parameters. The process is optimized primarily for low bending losses and a high fibre coupling efficiency. After the diffusion the crystal ends are polished to optic quality in order to make possible the coupling to optical fibres. The metal electrodes must be insulated from the optical waveguides in order to minimize the optical losses caused by absorption in the electrodes. This is done by evaporating a 180-260 nm buffer layer of silicon dioxide (Si02) on to the chip surface. This is particularly important for Z-cut crystals, with the electrodes placed directly above the waveguides, and TM polarized light. Theelectrodes are usually made of gold and manufactured through a technique similar to that of the waveguides. High-frequency components require thick electrodes in order to minimize electrical losses. High-frequency electrodes are also made of gold but by means of electroplating. The chip is cov- ered by a thin layer of gold which serves as the cathode for the plating. This is covered by a resist mask in a three-layer process, fig. 11. First a thick (ca. 4um) resist layer is placed on the chip. On top of this a silicon nitride layer of approximately 100 nm is deposited using chemical vapour deposition, CVD. The last layer is a thin (ca. 1 um) photoresist, which is exposed and developed. Next the silicon nitride layer is etched with a freon plasma, with the top resist layer as the mask. Finally the lower, thick resist layer is etched by means of oxygen, with the silicon nitride as the mask. This is a strongly anisotropic method and gives the resists straight and welldefined edges. Electrodes with thicknesses of up to 3 - 4 nm can be manufactured with a separation of a few um. The coupling to optical fibres consists of aligning and optimizing each fibre to the optical waveguide, after which they are bonded. Conclusion This article describes the current state of the research and development in integrated optics at RIFAAB and Ericsson Telecom. The number of applications in optical fibre systems is expected to increase in future. Components that are not polarizationdependent will make LiNb0 3 chips compatible with the technique for presentday single-mode fibres. 6 Fig. 11 Plating of high-frequency electrodes a resist and silicon nitride deposited b exposure c development d vapour etching e plating f resist lift-off and etching of gold contact The growing interest in coherent systems will mean new applications on both the send and receive side in which integrated optics is an essential factor. Further markets are to be found in sensor technology and military applications. The use of integrated optics will also grow as optical amplifiers are developed. In the long run semiconductors are likely to replace certain L i N b 0 3 circuits in applications that require a very high degree of integration. However, lithium niobate components provide better matching for single-mode fibres and will probably be predominant in a large number of applications for many years. Fig. 12 An optical data bus chip assembled with an electrical and optical interface. Low-frequency directional couplers and a high-speed modulator for 2.4 Gbit/s are integrated on the optical chip References 1. Thylen, L : Integrated Optics. Ericsson Rev. 61 (1984):F, pp. 49-52. 2. Clos, C : A Study of Non-Blocking Switching Networks. Bell-System Technical Journal, March 1953 pp 406-424. New Hardware in AXE 10 Urban Hagg and Kjell Persson System AXE 10 was designed with a structure based strictly on function modules. It was anticipated that it would be necessary to add. remove or replace parts of the system in order to adapt it to different applications and different markets, and to modernize its hardware in step with the technical development. The structure is a significant reason for the international success of the system. The authors give brief descriptions of a number of new and modernized units, which are now being introduced in AXE 10. Integrated circuits developed within the Ericsson Group are largely used in these units. telecommunication equipment modules integrated circuits computer architecture The development of AXE 10 has been directed towards providing digital alternatives for all telephony applications. Starting with local exchanges2, AXE 10 now includes tandem and transit exchanges, international exchanges, mobile telephony3, operator facility4, rural network applications, common channel signalling 5 and, before long, digital subscriber lines and nodes in the ISDN (Integrated Services Digital Network)6. Since the introduction of AXE 10 a number of functions have been added in order to meet demands from administrations in over sixty countries. For example, extensive adaptation work has been carried out in order to meet specific requirements as regards signalling Most of this work was implemented in software. Fig. 2 Some of the circuits developed within the Group. From the left: A B C Gate arraty circuits in an arithmetic and logic unit (ALU) in central processor APZ212 Standard cell ICs In an exchange terminal circuit Full custom circuits tor line circuit tunctions Large hardware changes have also been made in the system. At an early stage it became clear that digital switching systems would have great advantages. Digital group switches7 were soon introduced. The next step was to digitalize the subscriber switch. 8 This made it possible to arrange remote subscriber switches, connected via digital line systems. A small subscriber multiplexer9 for 30 subscribers extends the economical application range to include very small subscriber groups. Modern components and new manufacturing engineering have resulted in the development of two new central processors10for AXE10: APZ211, which replaces the earlier APZ 210, and APZ 212 for applications which require very high capacity. These two processors represent highly developed real time processor engineering for telecommunications applications both as regards hardwareand operating systems. The successful introduction of these new units is a tribute to the modular structure of AXE 10. Parts of the system have remained unchanged in spite of the continual hardware improvements. Fig. 1 shows what parts have now been renewed. New components Development in the field of electronic components has been very rapid during the 20 years or so of its existence, and everything points to this trend con- 87 URBAN HAGG KJELL PERSSON Ericsson Telecom Telefonaktiebolaget LM Ericsson Fig. 1 Block diagram of the hardware structure in AXE 10 CP MAU RP SP ST-C ST-R ETC CSR DAM GS LSM RPBC Central processor Maintenance unit Regional processor Support processor Central signal terminal Regional signal terminal Exchange terminal circuit Code sender and receiver Digital announcement machine Group switch Line switch module Bus interface New products Existing products tinuing. Recent years have above all meant greater development of components for special applications. Since its introduction in the world market AXE 10 has continually been rationalized in step with component development. Integrated circuits developed within the Group have now reached such a degree of sophistication that the time has come for their introduction on a broad scale in most of the units in AXE 10. Three different types of internally developed integrated circuits are used in AXE 10: gate array, standard cell and full custom circuits, fig. 2. The choice of type of IC for a certain application is determined by the development time, development and manufacturing costs and production volume. Gate array circuits are primarily used in low-volume products (e.g. cen- tral processors), whereas optimized full custom circuits are used in large-scale manufacture, for exam pie for line circuit functions. Standard cell circuits are used in the intermediate range. Results The objectives of the new units include a reduction of the number of printed board assemblies to about half in a normal local exchange (for approximately 8000 subscribers) and to approximately a third in transit exchange applications (for approximately 6000 junction lines). The comparisons referto fully digital exchanges. Fig.3 illustrates the changes resulting from the new engineering applied to a local exchange. Fig. 3 also shows that the power requirement is not reduced to the same extent as the space requirement. This makes new demands on the construction practice. Relative scale 88 Fig. 3 Comparison of some characteristics of the new and the previously used hardware in AXE 10 Earlier hardware N u m b e r of printed boards New hardware 24 BM N u m b e r of types of p r i n t e d b o a r d assemblies The advantages of the hardware rationalization can be summarized as follows: - Reduced space requirement - Lower power consumption - Increased reliability - Fewer faults - Fewer operator interventions - Fewer types of printed board assemblies - Simplified handling - Fewer spare parts required - Lower handling costs. Fig. 5 Equipment for 128 subscribers in new and previous hardware respectively Cabinet construction practice PSM SDM LSM The new hardware units make greater demands on the ability of the construction practice to dissipate heat. A Processor and switch magazine Subscriber device magazine Line switch module Power consumption Floor space requirement new cabinet construction practice has been developed in order to meet these demands without having to use fan cooling and also to obtain efficient screening against electromagnetic radiation. The cabinet has six shelves for magazines and space for cabling along one side, fig. 4. The new cabinet construction practice has a standard width of 24 BM for magazines (BM = building module = 40.64 mm) and 3 BM for cabling. A similar cabinet of half the width (12BM) is used for I/O units. The cabinet construction practice is used for all new units in AXE 10, and the units are assembled into magazines designed to fit into a total shelf width of 24BM. The changeover to a cabinet construction practice makes it possible to deliver cabinets fully equipped and tested. The cabinet is described in greater detail in another article in this issue of Ericsson Review.1 Subscriber switch Fig. 4 Cabinet construction practice for AXE 10 In AXE 10 the subscriber switching subsystem,8 SSS, constitutes the majorityof the hardware in a local exchange. If the total amount of hardware is to be cut down to any noticeable extent, SSS must therefore be reduced considerably. Previously the equipment for 128 subscribers consisted of two magazines, a processor and switch magazine, PSM, and a subscriber device magazine, SDM. After the introduction of new printed board assemblies one 24BM magazine was sufficient for 128 subscribers. A group of 128 subscribers is called a line switch module, LSM, figs.5 and 6. Fig. 6 An LSM magazine The line circuit board in LSM contains functions for eight subscribers and is called LIC8, fig. 7. Different types of LIC8 are available for different types of subscriber line current feeding. AXE 10 subscriber switch - Digital T-structure 128 subscribers per module 16 modules per group of 2048 subscribers Up to 32 digital links per group Not sensitive to asymmetrical load Two full custom circuits have been developed for line circuit funcions: the subscriber line interface circuit, SLIC, with line circuit functions for high voltages, and the subscriber line audio processing circuit, SLAC, for digital line circuit functions. 11~13 The special regional processor in the subscriber switch, EMRP, has been reduced to two printed board assemblies, partly through the use of memory components with higher capacity. In addition to the hardware changes already mentioned the following improvements are being introduced in SSS: - As an alternative the LSM magazine can be equipped for only 64 or32 subscribers. - An SSS working as a remote unit is equipped with an internal traffic function. This means that a call between two subscribers within the remote unit will occupy neither speech channels to the parent exchange nor multiple positions in the group switch. - Each LSM (128-group) can be equipped with two PCM systems (for 2 or 1.5Mbit/s) instead of one. This increases the traffic capacity so that the maximum traffic per subscriber also meets the requirements of some city exchanges with a good margin. See also the box "AXE 10 subscriber switch". Group switch Fig. 7 Line circuit board for eight subscribers Thegroupswitching subsystem, GSS, in AXE 10 uses time-space-time (T-S-T) switching, i.e. three-stage through-connection. Minor hardware changes in both the time and space switch magazines have led to a considerable reduction in the amount of floor space required for GSS. Compared with pre- Fig. 9 The equipment required for 480 digital circuits using new (green) and earlier (yellow) hardware respectively. The magazine module concept replaces the earlier magazine group ETC RP POU DM Exchange terminal circuit for 30 digital circuits Regional processor Power converter Distribution module vious magazine groups in the low-level construction practice (six shelves) the floor area required has been reduced by about a third. The maximum number of multiple positions in the group switch is 65536 (64x1024). The new group switch units are controlled by the new regional processor. See also the box "AXE 10 group switch". AXE 10 group switch - T-S-T structure Full redundancy 512 multiple positions per module Up to 128 modules (64 k) 25000 erlangs (congestion < 1 0 6 ) All types of network synchronization Fig. 8 Exchange terminal circuit for 30 circuits Exchange terminal for 30 channels Exchange terminal for 24 channels A new exchange terminal circuit, ETC24, for 1.5Mbit/s has been developed for markets that use u-law PCM In order to avoid wastage in the group switch, which handles units of 32 channels (2Mbit/s), four bit streams at 1.5Mbit/s are combined into three bit streams at 2 Mbit/s. This results in a unit for 96 channels. Two such 96-channel systems are installed in a magazine of 9BM, and magazine modules are formed in a way similar to that of ETC30. The exchange terminal circuit ETC30 constitutes the digital interface for 2Mbit/s PCM line systems towards the group switch. Code sender and receiver for MF signalling The ETC for 30 channels now comprises a single printed board assembly as against five previously. The reduction has been achieved by means of standard cell circuits and a special power module for mounting on printed boards, fig. 8. Signalling between exchanges will to an increasing extent consist of common channel signalling, preferably CCITT No. 7. However, it will take long for this system to come into general use. In the meantime a large part of the signalling will require MF equipment for VF signalling. Eight ETC 30 boards occupy a magazine having a width of 9 BM. Two such maga- The new unit for MF signalling, code zines, together with new regional pro- sender and receiver, CSR, is digital and cessors, constitute a magazine module. can thus be connected directly to the The magazine module replaces the pre- digital group switch. viously used concept of magazine group in AXE 10, fig. 9. A magazine module can The use of microprocessor and gate arconsist of anything from a part of one ray circuits has resulted in a unit whose shelf up to several fully equipped cabi- performance is a considerable improvenets. ment on earlier equipment. A9BM magazine holds 16 devices, which can all be The new ETC30 is used for junction used as either transmitter or receiver. lines and remote subscriber switches. Previous equipment comprised four Fig. 9 shows that the space require- transmitters or four receivers and ocments for equipment for digital circuits cupied 18 BM. An eightfold reduction in space has thus been achieved. have been drastically reduced. 91 Fig. 10 Magazine module for 32 code senders/receivers The new code sender/receiver is controlled by the new regional processor. It forms a magazine module in a way similar to ETC, i.e. two CSR magazines together with two RP magazines. The magazine module for CSR can be extended by one or two CSR magazines, fig. 10. Announcement machine Fig. 11 The announcement machine connected to the rest of the hardware DAM RP-A RP-B GS Digital announcement machine Regional processor A Regional processor B Group switch The new digital announcement machine, DAM, in AXE 10 has PCM samples stored in electronic memories. The choice of message and speech channel is dynamically controlled by a microprocessor on the basis of the traffic situation. The announcement machine is controlled by the new regional processor, and the speech channels are connected to the group switch via one to four 2 Mbit/s circuits, fig. 11. The announcement machine occupies a 12 BM magazine and has a capacity of four messages and 128 speech channels. It can store fixed or programmable messages having a length of 32 and 64 seconds respectively. A special variant of the machine can store eight shorter messages. Fig. 12 Regional processor A memory board for one message can be replaced by a printed board assembly for analog/digital conversion whereby an external, analog source (e.g. a tape recorder) can be used to give a message. The announcement machine used hitherto is analog, contains moving parts and is not built up on printed boards. Central processor In systems with low to medium capacity requirements APZ211 is used as the central processor, CP. APZ21210 is used for high capacity requirements. lnbothAPZ211 och APZ212the capacity of the memory components has been changed from 64 kbit to 256kbit, which has resulted in smaller CP dimensions and fewer printed board assemblies for a given application. Regional processor The new regional processor, RP, in AXE 10 includes gate array circuits and new memory components. Previously it was not possible to change a program in RP without replacing hardware. Dynamic RAMs (Random Access Memory) are now being introduced for the RP programs. At the same time software is being developed for the central processor that will aid the loading, functional modification and fault handling in the processs. The new RP is fully compatible with the previous version. The new regional processor occupies only 3BM (5 printed board assemblies, including the power board) and has up to twice the capacity of its predecessor. The memory capacity has been increased considerably. Fig. 12 shows the RP magazine. System for input and output of data The previous input/output subsystem, IOS, is being replaced by four new subsystems: support subsystem, SPS, file management subsystem, FMS, data communation subsystem, DCS, and man-machine communication subsystem, MCS, fig. 13. Unlike its predecessor the new I/O system is not controlled by a regional processor. A new type of processor, the support processor, SP, is being introduced for this purpose. It gives greater flexibility and makes a number of previously developed units available to AXE10. In the new I/O system, discs are the medium used for mass storage. High-capacity disc stores of the Winchester type are used for the system standby copy, logging, buffer storage of charging data etc. Floppy discs are used as a transportable medium for the input and output of programs and data. The disc stores normally hold approximately 50 Mbyte of data each. The floppy discs use industrial standard format and hold approximately 1.2 Mbyte each. Data links for up 92 to 64kbit/s can be included in the I/O system, fig. 14. In addition to the changes described here the hardware used for terminal and alarm interfaces is being rationalized. Fig. 13 New subsystem structure for input and output functions IOS SPS DCS FMS MCS Input Output Subsystem Support Processor Subsystem Data Communication Subsystem File Management Subsystem Man-Machine Communication Subsystem A normal I/O system with full redundancy is m o u n t e d in t w o special cabinets, each with a width of only 12 BM. In addition to VDUs and printers the terminal interfaces in MCS can also c o n nected to a man-machine system, MMS, which includes a personal computer.' 4 Summary A large p r o p o r t i o n of the hardware in AXE 10 has been renewed, to a great extent by integrated circuits developed within the Group. This has resulted in improved performance and simplified handling. The modular structure of the system facilitates introduction of new units. The system software is largely unaffected by the new hardware. This means that the software remains the reliable version that is now in operation in more than 800 exchanges, a fact which ensures high system quality in the new exchanges. References Fig. 14 The hardware structure of the I/O system and its connection to the central processor CP MAU SP Central processor Maintenance unit Support processor I.Hellstrom, B. and Ernmark, D.: Cabinet Construction Practice for Electronic Systems. Ericsson Rev. 63 (1986):2, pp. 42-48. 2. Eklund, M. et al.: AXE 10 - System Description. Ericsson Rev. 53 (1976):2, pp. 70-89. 3. Billstrom, 0. and Troili, B.: A Public Automatic Mobile Telephone System. Ericsson Rev. 57 (1980):1, pp. 26-36. 4. Morell, L.-E.: Operator Position Subsystem in AXE 10. Ericsson Rev. 60 (1983):2, pp. 66-72. 5. Du Rietz, J. and Giertz, H.: CCITTSignalling Subsystem in AXE 10. Ericsson Rev. 59(1982): 2 pp. 100-105. 6. Ericsson Rev. 61 (1984): ISDN. 7. Ericsson, B. and Roos, S,: Digital Group Selector in the AXE 10 System. Ericsson Rev. 55 (1978): 4, pp. 140149. 8. Persson, K. and Sundstrom, S.: Digital Local Exchanges AXE 10. Ericsson Rev. 58 (1981): 3, pp. 102-110. 9. Larsson, C. and Ohlsson, E.: Remote Subscriber Multiplexer, RSM. Ericsson Rev. 60(1983): 2, pp. 58-65. 10. Jonsson, I: Control System for AXE 10. Ericsson Rev. 61 (1984):4, pp. 146155. 11. Bjurel, G., Dudnik, A. and Hjortendal, R.: Development of Line Circuits for AXE 10. Ericsson Rev. 60(1983): 4, pp. 181-185. 12. Eriksson, G. and Svensson, T.: Line Circuit Component SLAC for AXE 10. Ericsson Rev. 60 (1983):4, pp. 186191. 13. Rydin, A. and Sundvall, J.: Line Circuit Component SLIC for AXE 10. Ericsson Rev. 60 (1983):4, pp. 192-200. 14. Backstrom, T. and Lambert, J. : ManMachine Communication in AXE 10. Ericsson Rev. 62(1985):2, pp. 82-92. ERICSSON ISSN 0014-0171 Teletonaktiebolaget LM Ericsson 53786 Liungfbretagen, Orebro 199