Which connector where?

Regular travellers know that a suitable adaptor is a 'must have' companion; it's an everyday example of how something relatively innocuous, like a wall socket, can differ from country to country. A prime example of standardisation is DIN41612 (now IEC60603), which not only defines different contact sizes and widths as structural shapes for a mating interface, but also determines the numbers of poles, male-female connector interfaces, contact configuration and contact surfaces, as well as the requirement levels and quality levels. But this seemingly detailed definition has weaknesses, because it has been necessary to develop new standards and measuring systems to enable verification of the requirement level. The original definition of the mating interface did not include the actual contact structure on which the cautious user was forced to test crosstalk. This shows that a standard cannot be pushed to infinity. A brief look at contact spring principles shows that, if a connector has two redundant contact points, these can be arranged opposite each other. Normal contact forces compensate each other and the connector system functions without the support of the housing. The disadvantage is the double sided refinement of the blade contact and poorer signal integrity at higher frequencies. Both disadvantages can be prevented by single sided contact mating, with the quadrupling of contact reliability arranged through separate resilient legs next to each other or – because of contact density – behind each other. We have reached the stage where, at high data rates, the influence of the plugging ranges needed for contact reliability is being examined with regard to undesirable resonances. Having looked at connector standardisation, we now focus on interface standardisation. An interface has to be described via the signal flow – how the signals run, are evaluated and reassembled. For signalling, measurable physical values have to be determined with regard to how they appear in whatever medium. Finally, signal assignment has to be determined in relation to a defined connector. This applies to copper (number of contacts), as well as fibre (core diameter and modes) and radio waves (carrier frequencies and modulation). It becomes apparent that, although connector standardisation has its difficulties, interface standardisation is more complex. Interface standards Using Fibre Channel as an example, there are four fibre interfaces and three copper based interfaces for shorter distances. Other interfaces are being worked on and the 40Gbit interface is likely to be standardised shortly. Why are we looking at this? Because it would be a never ending task to describe all possible connectors and their signal assignments for all interface standards. Looking at the OSI model (fig 1), it can be seen that, in addition to the direct fibre channel, which can be executed in copper as well as fibre, it is also possible to transmit data via Ethernet (over copper or fibre) and via InfiniBand interfaces. In brief, there are some 15 transmission variations. If we compare the OSI model with TCP/IP – the basis for all internet traffic – we can appreciate why there are often problems getting internet access at home. Apart from the electrical interface and the box, signal processing in the transportation layer makes a substantial contribution to overall success. Because going from interface standards to connectors is complex, let us try to deduce the use of connectors in standard interfaces from a selection of connectors. This is an 'assignment attempt'. We refer to it as an 'assignment attempt', as it is nearly impossible to give a comprehensive account with the large number of application cases, as well as the environmental conditions, mechanical demands and restrictions. Let us begin with I/O connectors for the two best known mating interfaces: D-Sub and RJ connectors. D-Sub serial and parallel connectors proved popular for Profibus and printer applications, with the smaller Micro-D format becoming the de facto standard interface for vga monitors. Flat ribbon D-Sub connectors were used for IEEE485 based measuring systems, as well as parallel printers and telephone/network (E1/T1) equipment. RJ connectors, in their various guises, are primarily used on telephones and networking equipment; the popular RJ45 format is used for ISDN, 10BaseT, 100BaseT and GbE networks. USB, DVI, HDMI and DisplayPort connectors have generally gained acceptance for desktop computers and notebooks, while circular, non polarised connectors are used for audio connections and external power supply units. For industrial applications, M8 and M12 circular connectors are widely accepted as the standard interface, largely due to the required IP67 protection rating, the concepts of which go back to the 'Kleintuchel' connectors of the 1960s. For computers, there are module connectors, used mainly for modems, network cards and wireless cards, as well as for removable memories, such as the SD card. There are internal connectors, mostly where pcbs are located perpendicularly. We make the distinction here between direct pcb connectors, the mating parts of which are gold fingers on the board to be connected, and two piece connectors. The direct pcb connector is difficult for connector manufacturers because the pcb manufacturer is responsible for the mating connector and the end product manufacturer has to guarantee the connection. However, pcb manufacturers now know how to handle gold fingers and the 'shaving brush' effect is a thing of the past. The records are incomplete with regard to two part pcb connectors; there are many approaches and the records show the most common and most successful connectors. Ten years ago, there was an attempt to standardise everything. After that, developments proceeded too quickly for the committees. Since 2005, air has become the de facto dielectric by virtue of the synthetic substances of the carrier material becoming frequency dependent and shielding measures – embedded in plastics – have not displayed the success in relation to lower crosstalk of adjacent signal pairs. For backplanes, connectors are generally selected by large telecom equipment manufacturers and network equipment providers according to platform related specifications. Accordingly, standardisation approaches are only found in the area of embedded computers. In addition to connector selection by data rate (which also depends on the frequency equalisation measures employed), mechanical dimensions are also important because there are often cheaper connector groups located on the same pcb due to the larger number of control signals from the components. Finally, there are 'mezzanine' card connectors for separable internal connections – as parallel board or 90° designs – in use with baby boards, which are stacked on the components as modules. What card gap (for parallel board connectors) is needed with how many signals for what data rates is crucial. There are no standards here and the designer often cannot find secondary sources. The world of connectors remains 'special'; there are millions of applications and hundreds of thousands of products. The fact there is no matching product for a specific application means the domain of connectors remains an area requiring intensive consultation. Herbert Endres is technology marketing director for Molex