System Level Specifications

In the context depicted in Chapter 1, it clearly appeared that the potentials of plastic waveguide systems could be leveraged in multiple markets, which are already addressed by conventional copper cables or optical fibers. However, introducing such a disruptive technology is not benign from an industrial point of view. Although the similarities with optical fibers have been highlighted in Chapter 2, the huge wavelength scaling and radically different processing are likely to raise new industrial issues (while probably solve other ones).

Because of this intermediate Technology Readiness Level, currently between two and four, we can assume that the developments of this technology will first focus on existing solutions or standards. Even if the introduction of dedicated standards would be more appropriate and / or optimized to plastic waveguides characteristics, replacing copper cables or optical fibers with proven business prospects is a much less risky approach from an economical point of view. With these preliminary economical and industrial conclusions in mind, the most relevant applications for plastic waveguides are presented in Figure 3-1.

The idea of replacing passive copper cables by systems containing a plastic waveguide channel as well as necessary mmW TRx at each end may seems troublesome. These latter actually need some energy to operate. However, in most cases, this point may not be a difficulty because host connectors usually have power delivery capabilities. Typically, a few Watts may be available and could be leveraged by the envisioned solution. In that sense, a plastic waveguide system is an active cable, meaning that the system has to take energy from its hosts in order to process transmitted signals according to the characteristics of its transmission channel. In fact, active cables are already found in Thunderbolt 1 and Thunderbolt 2 cables as well as in Direct Attach cables for datacenters applications. While signal processing in these latter solutions is realized in the analog or baseband domain, plastic waveguides demand dramatically different signals. To account for the high-pass or band-pass characteristics actually observed in plastic waveguides and reported in Chapter 2, a frequency transposition is required so that the information contained in transmitted signals is used to modulate a suitable mmW carrier frequency. These concepts will be detailed in section 3.2.

Note that, one more time, the analogy with optical fibers is natural since lasers also provide appropriate carriers for the transparency window (or bandwidth) of the fibers. Because of this common principle, the compatibility with applications dominated by optical solutions is even more apparent.

Figure 3-1: Plastic waveguides applications have been classified in three main categories.

To address successfully the variety of the markets presented in Figure 3-1, dedicated electrical interfaces should be integrated along with the mmW circuits. Implications are manifold. In addition to connector integration, specific digital IPs are to be integrated to mimic the response of a traditional connectivity, especially for the replacement of passive low-pass cables. Considering the mmW TRx, one can wonder whether it is possible to rationalize developments using the very same mmW IP in a “one size fits all” approach. An electrical performance analysis is proposed in Table 3-1 based on the most promising standards for each of the three identified segments.

Beginning with the consumer market, the widespread USB standard has been logically selected. The freshly released USB 3.2 standard defines four possible use cases to maximize data rates up to 10 Gb/s for the first generation and 20 Gb/s for the second one. Note that such data rates strongly limit maximum lengths to 2 m and 1 m respectively. As regards the professional market, there is no denying that it is dominated by datacenters applications.

However, it is noticeably fragmented with a large variety of standards, each one being implemented using many different physical layers. In this context, 10GbE SFP+ DAC connectivity is already widely deployed for intra-rack connections and is therefore an immense opportunity. AOC solutions are still pre-existent competitors up to 100 m while passive DAC cables are low power but only cover distances up to 5 – 7 m. Besides, anticipating future bandwidth requirements, migration to QSFP+ 40GbE is possible in the long run.

Currently, these high-speed standards are also mainly in use for inter-rack connectivity. With the same limitations, it is then relevant to address 40GbE QSFP+ DAC cables. Note that higher performance links (essentially 100 Gb/s and coming 400 Gb/s) have not been selected here because of inappropriate length X data rate products.

Professional connectivity

(datacenters, connected industries, medical equipment…)

Consumer connectivity

(USB, Thunderbolt, HDMI, DVI…)

Automotive connectivity

(Vehicular automation, infotainment,…)

Digital Interface

& PHY layer Data rate

(per lane) Parallel

lanes Operation Length Conso.

(per end)

Table 3-1: Comparison of the most promising standards for consumer, professional and automotive markets.

Finally, the automotive market will transition from 1 Gb/s copper cables to higher data-rate solutions in the coming years to sustain increasingly more autonomous vehicles. This important market is looking for a cost-effective, lightweight, low-power and EMI-friendly connectivity. Because of these combined constraints, leveraging on conventional 10GbE solutions is not satisfying so that dedicated standards had to be introduced (2.5GBase-T1, 5GBase-T1, 10GBase-T1). Note that higher data rates rely on reduced baseband bandwidth for EMI constraints… at the expense of multiple lanes in parallel. For most point-to-point connections, 2.5 Gb/s is enough but higher throughputs would enable point-to-multipoint connections, thus saving weight and costs.

As a conclusion from Table 3-1, it should be possible to address the most promising applications with a unique mmW system based on plastic waveguide technology, which (ideal) specifications are given in Table 3-2. It is the purpose of this chapter to investigate the feasibility of such a system and to discuss system portioning while still mentioning relevant limitations based on the state of the art.

Data rate

(per lane) Parallel

lanes Operation Length Conso.

(per end) Energy efficiency

10 Gb/s 4 Full-duplex ≈ 15 m ≈ 500 mW ≈ 0.83

pJ/b/m Table 3-2: Summary of the ideal performances of the mmW plastic waveguide system.