Reading time ( words)
Continuing the series of webinars highlighting recently-published chapters of iNEMI’s 2019 Roadmap—the comprehensive guidebook to the future goals and needs of the electronics manufacturing industry—the instalment on optoelectronics examined and discussed data transmission by optical technology over distances from a few millimetres to thousands of kilometres.
The session was hosted by iNEMI Project Manager Steve Payne, who briefly described iNEMI and its mission to forecast and accelerate improvements in electronics manufacturing, before introducing Richard Otte, president and CEO of microelectronics assembly specialist Promex Industries and chair of iNEMI’s optoelectronics technology working group.
Richard Otte explained that the working group represented a global collaboration concerned with the transmission of information utilising photons and that the roadmap chapter was focused on applications that used optical technologies to transmit data from point to point. In decreasing distances, these included telecom, broadband networks, local-area networks, data centres, plastic optical fibres in automotive, rack-to-rack, although not yet in backplanes, on-card, into and out-of package, and on-chip. An additional example was LiFi, a wireless communication technology using light to transmit data and position in free space between devices, avoiding electromagnetic interference issues.
The demand for data transmission capacity in the internet and telecommunications network continued to grow at more than 40% annually. This was a key market driver for optical technologies, which enabled very secure transmission of massive amounts of data over long distances with little power loss, requiring substantially lower energy than electrical transmission and capable of much higher data density. Once an optical signal was launched into an optical fibre, it would propagate with virtually no degradation, whereas electrical signals would be subject to resistance losses, skin-effect losses, and dielectric losses. Fundamentally, optical technologies could out-perform electrical technologies over distances greater than a few millimetres.
How fast could an optical system operate? Otte commented that a limiting factor for the practical bandwidth per channel of many optical data communication applications was, in fact, the semiconductor electronics—the capability of the CMOS technology used to generate the signals—which was difficult to operate at frequencies higher than 28 GHz without employing multiplexing techniques such as pulse amplitude modulation.
Otte summarised the content of the iNEMI optoelectronics chapter with a chart of applications versus time. Applications were listed by distance, from on-chip at about 10mm, through to telecom, which could be thousands of kilometres. Along the timeline, applications were colour-coded in green, yellow, and red to signify whether electrical was dominant. There was no major demand for known optical solutions (green), commercially viable optical solutions were known (yellow), or where no technically viable optical solutions were yet known (red), annotated to indicate limits, specific issues and needs, and what developments were necessary on what timelines to make them viable.
It was clear that optical solutions were desirable for shorter and shorter distances, but with CMOS semiconductor features now in the nanometre range, it was not yet practicable to use optical technology for on-chip applications. “Photons are too big!” Indium phosphide photonic integrated circuits were becoming available, which could combine hundreds of optical functions, and silicon photonic integrated circuits were being developed.
Although optical transmission in backplanes was being explored, it had not yet been widely implemented, largely as a consequence of traditional concepts of backplane architecture using plug-in cards. An equivalent optical system would require a practicable connector design, and critical alignment would be an issue.
There was substantial growth in the market demand for optical technology and installed capacity was increasing annually by about 25%, mainly in active optical cables and transceivers for data centres and local area networks. There was growing interest in using optoelectronics in backplanes, on circuit boards, on-to and off-of chips, even on-chip. As previously mentioned, LiFi offered an alternative “wireless” system.
Summing up the data-com situation, Otte observed that single-mode optical fibre technology-dominated long-haul telecom and distances down to about one kilometre, and data centres used thousands of 1 Gb/s to 100 Gb/s active optical cables with single-mode replacing multi-mode at higher data rates. But single-mode demanded much tighter tolerances in manufacture than the older multi-mode systems, and he was surprised that these were no longer employed in some of the less-demanding applications.
Because of the variety of requirements across the applications, the supply chain was quite broad, with manufacturing mainly in Asia. Faster and lower-cost transceivers were being introduced to the market. Otte reported that although the optical data-com industry was growing, it was a massively competitive business, which was struggling to generate attractive returns.
The roadmap activity had identified several technology needs, gaps, and challenges. Although the gaps differed across the range of applications, there was a generic need to develop improved optical modulation and detection methods to enable more data to be transmitted through existing fibres. Another generic need was for a CMOS-based light source. The addition of indium phosphide to CMOS photonic integrated circuits offered a potential solution.
The cost of packaging optical devices was a significant factor limiting the economics of data transmission over short distances. As a result, electrical methods continued to be used for distances less than one metre, although the need to transmit more data with less power—particularly into and out of packages—was stimulating ongoing interest in optical solutions. These gaps and challenges were already being addressed by industry consortia and government programmes but were difficult problems to resolve.
The critical infrastucture issue was to reduce the cost of packaging optical devices by continuing the development of photonic integrated circuits, reducing the cost of aligning optical fibres to sources, detectors, and waveguides—eliminating hermetic packaging and eliminating the use of optical fibre pigtails for cable termination.
The chapter made several recommendations: developing standards and promoting the wider use of solutions to increase volumes, continuing the development of photonic design software with semiconductor foundries providing standard process design kits to minimise the cost and time to design photonic integrated circuits. It was also recommended that the industry continue to develop multiple supplier agreements for products such as transceivers. A particularly significant recommendation was to offer an education programme to inform engineers and potential adopters of the benefits of optical data communication technology and methods of implementing the technology.
Richard Otte was an exceptionally knowledgeable presenter with a relaxed and informative style derived from more than 50 years in technical electronics manufacturing that made this webinar an enjoyable and enlightening experience. His explanation of the optoelectronics chapter of the 2019 Roadmap provided an excellent demonstration of iNEMI anticipating the product requirements and defining the technology needs of a key market segment.