The development of the 2mm Metral connector changed the telecom industry forever.
The connector industry is very interesting, in part, because of exciting inventions developed by engineers – some of which have such an impact that the direction of the industry is changed.
In the late ‘80s, telecom OEMs were outgrowing the capabilities of the 0.100″-pitch connectors then available. They needed more signals to communicate to the backplane and were also moving to higher speeds that required more grounds, which further drove the need for more pins. The European telco industry was quite robust at the time, with large, growing companies in each country, including Ericsson, Siemens, Alcatel, and GPT. The US was dominated at the time by AT&T, which had just spun out Lucent. Nortel and NTT were also strong.
Telecom companies at that time were also connector manufacturers. Alcatel, Ericsson, GPT, Nortel, and AT&T all made their own connectors. Alcatel and Ericsson were using variants of the DIN 41612 standard (96-pin DIN connectors), but others were making complex multi-row connectors on 2.54mm pitch. Nortel made a four-row connector and AT&T made an exotic eight-row 0.125″-pitch connector that straddled the daughtercard with four rows of surface-mount leads on each side. They called this eight-row monster Fastek. You can imagine the manufacturers of telecom equipment were feeling really constrained by the limitations of these connectors and were quite anxious to find a new interconnect solution that would enable more pins per inch on the backplane, more cards per shelf, and lower costs.
AT&T had internal CAD (computer-aided design) standards around the Fastek system that basically allowed design engineers to design boards and electronic circuits, all using a standard backplane connector and backplane grid. AT&T had a unique manufacturing process for backplanes where it inserted individual pins into the backplane with fully automated machines. This process eliminated in-process inventory except raw pins on reels and backplane PCBs. To make system A, engineers programmed the machine to insert pins exactly where needed. They could have multiple pin lengths inserted to facilitate sequenced mating in a much easier way than that of competitors. The flexibility was quite impressive. A backplane-populating machine could make sophisticated backplanes for large switches today and for a simple access backplane tomorrow. AT&T had a plant in Richmond, Va., that utilized more than 10 of these stitching machines, which produced hundreds of backplanes every day. AT&T was so integrated that it made its own connectors, pins, backplanes, and daughtercards. Those were the days!
Both AMP and Berg were working on this challenge simultaneously, and the development centers for both companies happened to be a few blocks apart in s’Hertogenbosch, Netherlands, known to the Dutch as denBosch (the Duke’s woods). There must have been innovation in the drinking water since both companies developed industry-changing designs.
Berg assembled a team to address this opportunity: John Harding, telecom industry manager, shuttled between the major customers gathering requirements, and Hans van Woensel, product engineer at Berg, was assigned the task of inventing a new connector that could double the pin density without sacrificing on reliability or cost. As the concept developed, Harding revisited each of the customers to get their reactions. To address the broadest range of opportunities, Berg decided to develop a system composed of 12mm-long blocks with four rows of contacts on a 2mm x 2mm mating grid. By stacking these modules side by side, Berg was able to provide up to 456 mating pins on a 233mm-tall Eurocard form factor, replacing two 96-pin DIN 41612 connectors then in use.
Berg was able to quickly prototype this concept because the contacts and housings could be tooled quickly and inexpensively. Berg also developed a small stitching module that inserted one contact at a time at high speed. This flexible stitching module enabled them to produce production-quality parts very quickly and to expand capacity by adding as many additional stitching modules as necessary. These modules could stitch receptacles today and headers tomorrow with changeover tooling, raising the bar in terms of time to market and quality of initial parts. This family became the Metral 2mm backplane connector system.
Ericsson was especially forward-looking for the time. Its team envisioned its facility populating daughtercards and backplanes with fully automated machines fed by 12mm connector blocks in tube packaging. They used robot-like pick-and-place machines to place power, signal, and guidance modules as needed for any daughtercard that came along.
Berg also developed power and guidance modules that could be placed in convenient locations to optimize the connector configuration for each card. The team was able to take initial Metral parts to make a prototype that met the requirements of the IEEE 896 working group developing the Futurebus+ standard.
A constant 2mm contact pitch allows many efficient manufacturing techniques. Contacts on a constant 2mm pitch can be mass-inserted into housings continuously, maximizing use of the contacts on strip and minimizing the number of insertion steps. Contacts in right-angle daughtercard connectors can be inserted then bent, a rather efficient manufacturing process. Berg chose to stitch contacts and created standard high-speed stitching machines that could make every variant at high speed and low cost. The cost reduction relative to the 96-pin DIN connectors was substantial – more than 50%. Much less material was used, including metal, plastic, and gold plating. The manufacturing process was also optimized for efficiency.
The telecom customers were delighted. They could see significant cost reductions for their future designs, as well as density and performance improvements. They rapidly endorsed this approach and changed their internal card-cage standards to take advantage. AT&T, moving from 0.125″ connectors to 2mm- (0.79″-) pitch connectors, saw so much of an advantage that they created a new AT&T standard card-cage architecture.
The Metral connector was introduced to the IEEE 896 Futurebus standards committee and won the standard against six-row 0.100″ connector designs. At the time, Augat presented a six-row straddle-mount connector on 0.100″ centers and AMP proposed a right-angle six-row solder-to-board connector. Berg proposed a mix of 4×12 signal modules and six-pin power blocks that was able to meet all of the requirements of the Futurebus specification. Berg partnered with AT&T and Ericsson to present the Metral connector. Part of the deal was that other connector manufacturers were then able to tool Metral connectors royalty-free, a key business decision. This created enough momentum to win the standard and quickly proliferate the number of Metral producers, making this connector one of the dominant standards in the telecom industry. Metral has been tooled by Berg, Molex, AT&T, Ericsson, ITT Cannon, SOURIAU, CECO, and several other suppliers.
A couple of ironies evolved from this saga. First, after its approval by IEEE, Futurebus never achieved commercial success as a standard. It was a seven-legged calf, a term for a standard that incorporates so many features and variants that it becomes totally un-optimized for any of them. The connector, however, was broadly adopted, especially for telecom applications at Ericsson, Alcatel, GPT, and AT&T (that later spun off Lucent). The simplicity and modularity of the connector, and the fact that it was tooled by so many suppliers, made it one of the most cost-effective backplane connectors ever tooled. It is still broadly used today, mostly in telecom access boxes.
The second irony is that, as the telecom industry imploded in the late ‘90s, the connector companies consolidated. Berg acquired the connector operations of Nortel, Ericsson, AT&T, and Alcatel. FCI, who had earlier acquired SOURIAU, acquired Berg. The net result was that FCI acquired nearly all of the Metral capacity.
Special Metral modules have evolved, including high-speed modules with internal shielding, larger modules, cable solutions, and even RF. CECO, a California molding house with special expertise in building tools for long connectors, developed a very nice business in making a better connector than the commodity producers. The FX-2 product line included long “monoblock” housings, dual-beam beryllium copper contacts (lower mating force, higher conductivity), and integrated guide pins to fit customer requirements exactly. IBM, in particular, specified the CECO 2mm FX-2 connectors for its mid-range servers.
Stay tuned: In two weeks, we’ll share the equally significant story about the AMP 2mm Z-Pack HM backplane connector system and some of the breakthrough innovations that further changed the direction of backplane connector technology.
To read the first article in this series, on compliant press-fit pins, click here.