Beat the automotive data-net bottleneck with new ICs, topologies
Today's automotive designers have an opportunity wrapped in a major challenge: customers want more connectivity and graphics - and are apparently willing to pay for it. Drivers and passengers want basic operating information, of course, but they also want real-time maps, entertainment, and information. If they have learned to live with and, in fact, love those multiple screens in their home and office environment, why should they also not want the same in their mobile automotive world?
The reality is that this sort of connectivity and display has gone from being "nice to have" to "must have" with today's younger buyers. It doesn't matter if you call it infotainment, smart driver interface, networked vehicle, vehicle connectivity via the cloud, or anything else: it's becoming a standard accessory on all but the low-end vehicles.
But there's a technical problem. Issues which are manageable in the fixed home/office situation are very different in the automotive world. Supporting multiple screens is not just matter of computing "horsepower". It also brings complex cabling, connectors, power dissipation, and signal integrity concerns - all of which are at a premium and severely constrained in the automotive world.
Technically, there are three areas that must be addressed: the graphics processing core(s); the multiple displays; and the high-performance interconnecting network between the graphics engine and the displays. Consider that a vehicle will have an instrument cluster, a front passenger-side display, a central console, possibly a heads-up display [HUD], a rear backup camera, and even a rear-seat monitor camera. The overriding challenge is this: how do you effectively support four, five, or more displays and cameras from a single, centrally located graphics-rendering and image-processing core?
The obvious solution of using large, standard PC cables and interfaces such as HDMI will not work, for several reasons. First, there's the sheer bulk of routing these cables. Signal integrity in the noisy auto environment is a problem, especially over the distances. Plus, there's the cost and weight of the copper cabling, and the connectors themselves are too large and awkward. Reliability and performance in the harsh auto environment is also inadequate.
Fortunately, there is a solution that leverages the existing mass-consumer market technology, by adapting it in a cost- and performance-effective way for the automotive application and environment. It begins with the ability of today's video subsystems to support 3D streaming video, with separate images for right eye and left eye.
But by leveraging the stereo video format and HDMI interface using a proprietary technique, Inova Semiconductors (Munich), in conjunction with leading IC vendors such as Analog Devices, Inc. (Norwood, MA) and others, solves both the bandwidth issue and the physical connectivity problem.
Instead of 19 wires of the HDMI cable, this interface backbone requires just one four-wire cable, with differential signaling via two pairs, arranged as a "star quad" with 100Ω nominal impedance. The 1-Gbps APIX link was introduced in 2007, Figure 1, and has been proven effective and reliable.
Of course, it appears that users can never have enough speed, so the second-generation AIPX2 reaches 3 Gbps (Figure 2); it is backward compatible with the original APIX1 and has been in production since last year.
AIPX2 is not just a speculative concept it is being used in automobiles on the road today. It was demonstrated at Electronica in November 2012 by Inova at Analog Devices' booth, where it transmitted two uncompressed high-definition video streams, multi-channel audio, and 100Mbits/s Ethernet data over a single 4-wire, shielded, twisted-pair cable. The demonstration featured video conferencing and touch surfaces in remote displays via standard Ethernet protocols and the integration of real-time user interfaces and applications using HDMI connectivity to smart phones.