Ethernet tackles automotive EMI challenge
By Mike Jones, Micrel 1/2/2013 5:23 PM EST
Ethernet has already been widely accepted by the automotive industry as the preferred interface for on-board-diagnostics (OBD) and has been deployed in various car models since 2008. Ethernet provides increased bandwidth speeds over traditional automotive buses, resulting in a reduction in software download times from hours to minutes compared to traditional methods.
This adoption will be accelerated with the introduction of a standardized IP Diagnostics interface, as specified in ISO 13400, using Ethernet as the physical layer.
The expansion of Ethernet-based networks in the automotive industry has continued with next-generation infotainment and driver assistance systems. Originally found in only luxury high-end vehicles, such applications are fast becoming differentiating features on mid-range and even basic models.
These new applications have, in tandem, generated demand for greater system bandwidth, which continues to rapidly rise, as shown in the timeline below:
- 1981: CAN @1000 Kbps
- 2005: FlexRay @ 10 Mbps
- 2001: MOST @ 50 Mbps
- 2008: Ethernet @100 Mbps
- Ethernet @ 1000 Mbps (RTPGE IEEE802.3bp)
Advanced Driver Assistance Systems (ADAS) constitutes one of the fastest growing applications within the automotive market.
Driven by government legislation and a desire for enhanced in-vehicle safety camera sensor networks are becoming commonplace.
By 2017, camera-based module sales are expected to increase to 34 million in total, from 6.1 million in 2010 (IMS Research).
This is in part due to the National Highway Traffic Safety Administration (NHTSA) ruling requiring every car sold in the United States from 2014 be fitted with at least a single rear view camera or sensor.
System costs are significantly lowered using Ethernet connectivity for multi-camera sensor networks in the car. Traditional proprietary methods are making way for open standard Ethernet. This has been reflected in the ISO 17215 Video Communication Interface for Cameras (VCIC) specification, defining Ethernet connectivity for use in vehicle camera and/or sensor applications.
Ethernet is emerging as the preferred network of choice for this new generation of networked vehicles, continuing to be the de facto networking bus for all other markets, thanks to an offering of ample bandwidth and open standardization.
True standardization results in multiple interoperable supplier solutions, rapidly driving down costs.
This is in contrast to current popular infotainment networking technologies, for example MOST®. The proprietary nature of MOST® has often been blamed for failing to deliver the cost needs of car makers, as reported in the Hansen report, November 2008: “Suppliers and car makers think more access to SMSC’s proprietary MOST® technology would lower the cost of MOST® and increase market acceptance. Some are looking at Ethernet as a possible alternative.”
Ethernet provides many advantages over MOST® technology, including increased bandwidths and flexible topology, but probably most importantly lower costs, due to the high volume deployment across multiple markets, accepted supply chain and multiple suppliers.
Challenges for Automotive Ethernet
The needs of more recent Ethernet-based applications, such as infotainment and ADAS, differ distinctly from the current applications, such as on-board-diagnostics in that these are applications operating in real time and while the car is moving.
No matter what the circumstances might be, must meet car manufacturers’ EMI limits.
For Ethernet to be considered for any application in operation whilst the car is moving, it must fully comply with OEM EMI specifications.
Herein lies the challenge: the use of shielded cables would provide a solution to reducing radiated emissions within the car, but is usually undesirable. Shielded cable brings about complications in earth strategies, can adversely affect reliability, and add cost to production. Shielded cables cannot be manufactured in situ using wiring looms during production, but need to be pre-manufactured and purchased. Hence, the ultimate goal would be to operate Standard Ethernet over unshielded cable whilst meeting automotive OEM EMI limits. This solution dramatically reduces cabling costs, by up to 80%, over shielded counterparts, whilst maintaining interoperability with any other standard Ethernet device. The net result is lowest cost cable and silicon, with multiple suppliers.
Handling radiated emissions
Although the use of low cost unshielded cable is clearly desirable, the perception for some time has been that this was not possible with standard Ethernet 100BASE-TX PHY, and more proprietary means were required. When you observe the typical radiated emission characteristics of an Ethernet PHY, in Figure 1, it is easy to understand where this perception originated. The high energy content at the lower frequency band due to the MLT3 coded 65MHz to 80MHz bandwidth typically 10dB to 15dB in excess of automotive OEM limits.
Figure 1. Example of 100Mbps Ethernet Board Radiated Emissions.
After continued investigations into the EMI behavior of 100BASE-TX PHY circuitry, simple techniques have been demonstrated that sufficiently reduce the emissions, meeting automotive manufacturers’ needs. By adding a low pass filter to the transmit front-end, emissions can be reduced whilst still providing an interoperable standard Ethernet solution. The result is that no changes are indeed necessary to the standard Ethernet PHY.
Figure 2 shows the resulting emissions from 100BASE-TX using Micrel’s latest standard Ethernet PHY technology, where such techniques have been applied.
Figure 2. 100Mbps Ethernet PHY Board passing Radiated Emissions Limits
Radiated emissions performance is not the only EMI challenge for in-vehicle Ethernet. Immunity to other electro-magnetic disturbances, both internal and external to the car, needs to be met if Ethernet is going to be considered by the automotive manufacturers. Standard 100BASE-TX Ethernet PHYs have been shown to be robust enough to meet the demands of automotive immunity limits, whilst operating in a reduced emissions configuration.
This is demonstrated in Figure 3, IEC 62132-1/4 Direct Power Injection (DPI) performance, with the same PHY configuration. Here no network errors occur across all frequencies with 39dBm injection limit – equivalent to 106dBuA / 200mA (Level 5) Bulk Current Immunity (ISO 11452-4 stripline) levels.
Figure 3. 100Mbps Ethernet PHY Direct Power Injection Immunity
One of the key differentiators for standards-based solutions is the resulting multiple suppliers. Such competition provides the end user with a greater offering of devices, higher integration, lower power consumption as well as lower costs.
Such techniques have already been proven to be successful by multiple suppliers, working independently, both in performance and successful interoperability with each other and any other IEEE 802.3 Ethernet PHY.
It was recently been announced that two Ethernet PHY vendors have independently demonstrated PHY technology with reduced emissions meeting automotive OEM limits, based on standard IEEE802.3 Ethernet.
Micrel and Marvell have successfully demonstrated interoperability both with each other and, since these PHYs are IEEE Standard, other IEEE 802.3 Ethernet PHYs.
Utilizing standard IEEE 100BASE-TX Ethernet, PHY technology can take advantage of the existing IEEE802.3 ecosystem which has been widely adopted for networking applications, eliminating the need to create any new tooling or support infrastructure including standardization groups.
Perhaps the most significant consequence is the rapid realization of multi-sourced solution portfolios (PHYs, controllers, switches etc.).
Reusing the same silicon in multiple Ethernet applications (in addition to automotive) also leads to economies of scale unlikely to be matched by any proprietary or application specific solutions.
Solutions based on interoperable standard Ethernet provide a simple, inexpensive route for Ethernet semiconductor manufacturers to offer low emission solutions based on their existing 100Base-TX technology.
Any non-Ethernet PHY approach echoes the criticism directed towards MOST technology and the increasing preference towards Ethernet based solutions.
Real-time performance with Ethernet is relatively straightforward with the necessary quality of service (QoS) guaranteed by use of IEEE Audio Video Bridging (AVB) specifications.
The AV Bridging system is based on three specifications:
- IEEE 802.1as Time Synchronization
- IEEE 802.1Qat Stream Reservation
- IEEE 802.1Qav Queuing & Forwarding for AV Bridges
Time synchronization is critical in order to ensure quality audio and video streaming within an Ethernet network. IEEE 802.1as utilizes specific PTP (Precise Time Protocol) packets to provide synchronization across the network to a common system clock source.
Nodes in the network, known as Time-Aware Bridges, will extract timing from the network based on a series of PTP synchronization message exchanges with the master clock source and neighbours.
Micrel recently introduced the new EtherSynch™ family of highly integrated Ethernet devices, supporting both IEEE 1588v2 and IEEE 802.1as time synchronization protocols.
Timing accuracy in the order of sub 100ns can be comfortably achieved; outperforming the IEEE based standards (1us), enabling suitability for control application as well as A/V streaming.
The family of PTP devices is offered in the form of a 3-Port switch with MII or RMII, 2-Port and Single-Port Controllers with 8/16-bit host processor interface.
All devices feature embedded ultra low power PHY transceivers, time precision GPIO in a compact 10mm x 10mm 64-LQFP package.
IEEE 802.1Qat Stream Reservation allows network bandwidth and buffer resources to be reserved for specific traffic schemes using SRP (Stream Reservation Protocol).
IEEE 802.1Qav Queuing and Forwarding methods are based on segregating traffic into isochronous (time critical) and asynchronous (non-time critical) packets and prioritising using the priority class defined in IEEE 802.1p. Egress port buffers are then separated into two or more queues, each allocated to a specific priority class. Isochronous packets will be given the highest priority, while asynchronous packet the lowest. A credit based traffic shaper is defined to smooth the ‘bursty’ nature of video data.
If Ethernet is going to be adapted as a true automotive network bus, then automotive makers will demand a commonly agreed solution that is open and freely accessible.
Talk to industrial control manufacturers and they will certainly agree, based on their experiences with original multiple, non-interoperable field bus standards, now which are based on Ethernet.
Progress is already underway for next generation vehicle networks with the formation of a new IEEE 802.3 Gigabit Ethernet Study Group for Automotive.
Ethernet’s unquestionable success in the Industrial networking sector has proven reliability and quality in an extreme environment.
This marriage of this industrial strength and consumer technology drive provides the perfect physical layer solution for automotive, successfully bridging the gap between lengthy vehicle design cycles and today’s fast moving IP world.
Today, Ethernet has already emerged inside the car to provide an IP-based standard interface for diagnostics and software downloading.
The next step is for Ethernet to form the backbone of the next generation automotive multi-media networks, carrying ‘live’ traffic.
New standards such as IEEE 802.3AVB (Audi-Video Bridging) provide necessary real-time performance, while multiple suppliers already offer proven low EMI emission Ethernet, interoperable both with each other and any other IEEE 802.3 Ethernet PHY.
There is nothing complex about Ethernet technology overall; it is simple, proven and open — the reason for its success.
Cost is a crucial factor in any market and Ethernet has consistently demonstrated the lowest cost of ownership of any network.