Automotive Communication at the Speed of Light

High-speed infotainment networks in cars were first introduced by premium carmakers like Audi, BMW and Daimler. In recent years high-volume carmakers like Toyota, Volkswagen and Hyundai have applied the technology. It is possible Chinese manufacturers will be next. More than 10 years after the implementation of the first MOST25 systems at BMW, the new generation of optical MOST150 systems running at 150MBit/s are going to be introduced by Audi in 2012.

[ARM TechCon 2012, the largest ARM design ecosystem under one roof, is Oct. 30 - Nov. 1 in Santa Clara. Click here to learn more]

One of the success factors for MOST is the optical physical layer that has advantages compared to electrical wires, specifically EMI immunity and low weight. Since plastic fiber cable is not a metallic conductor, communication links using this technology are not affected by the many electromagnetic devices and noise sources in a car. This is a key advantage for carmakers as it makes most debugging unnecessary and thereby reduces development time.



Figure 1: MOST150 transmitter AFBR-1150 and receiver AFBR-2150 from Avago Technologies in clear mold package

Critical components for the smooth operation of a MOST150 network are the optical transmitters and receivers. The transmitter sends out bit patterns in the form of light-pulses over light-guiding plastic fibers to the receiver, a photosensitive element that transfers the optical bit pattern into the electrical domain. The clear mold package devices from Avago Technologies shown in figure 1 are using conventional LEDs optimized for speed and thereby avoid strong temperature compensations needed for many resonant cavity LEDs (RCLEDs). This feature and the high-performance integrated driver IC are responsible for the dynamic signal integrity performance.

While MOST25 devices only have 4 electric pins, MOST150 components can be easily identified by their use of 7 pins. The increase in pins is required to provide more functionality and better signal integrity, including differential data signals. While the existing transmitters and receivers are designed for conventional wave soldering, an upgraded version for reflow pin-in-paste soldering is under development and was scheduled for release in late summer 2012. This will help to reduce the manufacturing costs of MOST150.

In addition to the different hardware, the dynamic performance requirements of MOST25 and MOST150 are much different. In the following paragraphs we describe the dynamic characteristics and the new measurement methods for them in more detail.

Characterization and Testing of MOST150 vs. MOST25

A comprehensive characterization has been performed for the MOST150 optical components that was much more sophisticated than for MOST25. The MOST150 oPHY sub-specification [1] has been derived from MOST25 specification. However, there are some significant differences, especially regarding specification of the signal performance of the specification points SP1 thru SP4, which are describing the input and output points of a single optical MOST link (see also Figure 3).

MOST25 signals are specified by pulse width variation, average pulse width distortion, data dependent and uncorrelated jitter (jitter quantifies the undesired timing variation of the signal edges). In contrast, MOST150 signals are specified by transferred jitter and alignment jitter defined by nominal eye masks as shown in Figure 2. Alignment jitter has to be accounted for in each single node in a MOST ring in order to correctly receive the incoming signal. Most of the jitter, except for the transferred jitter, is eliminated by the network interface controller (NIC). Transferred jitter, however, is accumulating along the entire ring and can be measured with a special jitter filter as specified [1].

Because of the higher data rate, signal jitter is much more critical for MOST150 compared to MOST25 since one unit interval (UI) is just 3.29ns (vs. 20.3ns for MOST25). Within a device characterization, it is more suitable to perform a parametric total jitter measurement using adequate tools instead of a pure eye mask measurement, which takes a very long time for the required bit error rate of 10-9, and still delivers only pass/fail information.



Figure 2: Eye mask tables/diagrams for different specification points

Eye masks are specified for SP1, SP2 and SP4, but for SP3 (Rx input) no particular mask is specified. This has to be calculated by SP2 worst case condition and the POF transfer function:




Worst-case performance testing

All tests regarding signal integrity have to be performed using the MOST150 stress pattern provided by the MOST Cooperation [1]. The pattern is about 32kbits long and reflects a worst case scenario regarding the data sequence. However, this pattern run on either an arbitrary waveform generator (AWG) or a digital pattern generator does not yet reflect the actual worst case without jitter being added. This can only be achieved by applying worst case signals to SP1 or SP3, thus signals which marginally violate the specified eye masks for SP1 or SP3. Since no particular eye mask is specified for SP3, a worst case SP2 signal has to be generated and in addition, a worst case optical fiber transmission has to be regarded.

Influence of Optical Fiber

Because of different fiber lengths and multiple optical connectors which may appear on a single optical transmission line in a car, the standard MOST150 receiver is specified for a wide optical input power range from -2dBm down to -22dBm at SP3 [1]. High power levels will be achieved in the case of very short fiber cables, while lower power levels are caused by longer fibers and inline connectors. This also has to be taken into account for the measurements over the entire input power range. Typical fiber characteristics are described in detail by the MOST Application Note POF Transfer Function [2]. This document contains the results of extensive measurements and theoretical considerations. The formulas within this document, also including the mentioned POF transfer function (1) can be used for a mathematical calculation of an SP3 waveform.

Evaluation of a Worst Case Test Signal

A suitable signal is composed of different contributors: Beside the digital pattern itself, uncorrelated jitter (RJ), data dependent jitter (DDJ), duty cycle distortion (DCD), slow signal edges and overshoots (OS) have varying influence on signal distortion. In order to find a signal as close as possible to worst case, it is required to calculate multiple waveforms based on the MOST150 stress pattern but with different contributors of the mentioned effects.

For the receiver evaluation, the following aspects must be addressed: For real measurements, the calculation of the POF transfer function will not be used. A long fiber with a mode mixer (or a short fiber in case of high optical power) has to be utilized instead.


Figure 3: Test Setups for Transmitter and Receiver

The test setup for dynamic signal performance (Figure 3) according to the MOST150 oPHY Compliance Measurement Guideline [3] is used. For the Rx, the setup consists of an AWG, an analog 650nm transmitter (EOC), an optical attenuator (ATT), a long fiber with a mode mixer, the DUT and a serial data analyzer (SDA). An optical input power level, which shows the most critical jitter behavior, has to be set. Usually this is the low specification limit of the optical power range. For the Avago receiver AFBR-2150, we found that a mixture of RJ and especially DDJ has the highest impact on signal integrity, while the influences of DCD, OS and slow edges have almost no effect on the output. This may of course be different for other receivers. Figure 4 shows an extract of some signals with different input jitter modes and their effects to the output jitter. The eye diagrams of SP2 and SP4 for different cases are shown in Figure 5. The undistorted MOST150 stress pattern is indicated as “clean” in Figures 4 and 5.

Although the receiver is the more critical device related to link jitter, the same evaluation procedure is recommended for the transmitter. However this is easier to perform, since the calculated jitter signal waveform can be directly applied to SP1. The test setup is also shown in Figure 3.


Figure 4: Total Jitter SP4 vs. SP2 for different Jitter contributors


Figure 5: SP2 and SP4 Signals with Different Jitter Modes

Full Characterization

The explained evaluation will deliver signals for transmitter and receiver, which are actually very close to the worst case. In the last step, these patterns have to be utilized for the full device characterization in order to guarantee proper performance over all possible variations of conditions. Those are temperature T, supply voltage VCC, optical input power Popt3 (for the Rx only), LVDS voltage levels (for the Tx only) and input signal jitter. As a result of the full characterization, the Avago components AFBR-1150/2150 were found to have excellent dynamic performance over the entire ranges of the mentioned conditions.

The next level: GBit/s and beyond

The 150MBit/s speed of MOST150 gives ample bandwidth for today’s applications like data, video and audio. MOST150 will also be able to support the performance requirements for driver assistance applications that are gaining popularity. Certain applications such as uncompressed video or consumer interfaces (i.e. USB3.0 or Thunderbolt) require higher transmission speed in the range of up to 10 GBit/s. A next-generation MOST system that supports such speeds will likely introduce a new physical layer implementing glass fiber components due to the higher bandwidth than POF. Using Avago’s experience and proven technology of high-speed parallel transceivers running at up to 120GBit/s, first evaluations with VCSEL based transceivers are being conducted. Both from a reliability and performance point of view the use of VCSEL-based transceivers is expected to be feasible for future automotive applications.

Literature
[1] MOST150 PHY Automotive Physical Layer Sub-Spec. Rev. 1.1, 05/20/2010
[2] Application Note POF Transfer Function Rev. 1.0, 01/2009
[3] MOST150 PHY Compliance Measurement Guideline Rev 0.9, 06/2009

Markus Stich is working in the R&D department of Avago Technologies in Regensburg, Germany, where he is developing fiber optic components since 2005. He has been involved in the development and characterization of the MOST150 FOTs.

Bernd Luecke is head of product management and marketing for optical fiber components for industrial and automotive applications at Avago Technologies in Regensburg, Germany.