The sheer quantity of electronic content in vehicles has risen steadily over the decades. Today, it’s reached the point where our cars, SUVs, and trucks are just as “networked” as our homes and workplaces. Wiring harnesses for multiplexed serial buses in a luxury vehicle can weigh up to 110 lbs.
The response has been development of standards for automotive networking such as Automotive Ethernet. "Automotive Ethernet" can refer to any Ethernet-based networking scheme for in-vehicle electrical systems. It also serves as an umbrella term for BroadR-Reach (or OPEN Alliance BroadR-Reach), 100Base-T1 (the IEEE's 802.3bw-2015), and 1000Base-T1 (IEEE802.3bp). In either case, Automotive Ethernet is specifically tailored to enable faster data communication for in-vehicle networking.
Where there is serial data and communications standards, there is compliance testing to ensure that a system design meets the criteria of the organization that maintains the standard. This article will provide an overview of physical-layer compliance test for Automotive Ethernet.
There are different categories of Automotive Ethernet testing. Test of electrical signaling, which relies on oscilloscopes, is defined in the Physical Media Attachment (PMA) test group. These tests determine whether a product conforms to the electrical transmitter and receiver specifications spelled out in the BroadR-Reach or IEEE 802.3bw/802.3bp specifications.
Other tests focusing on the functionality of the protocol itself are found in the Physical Coding Sublayer (PCS) and PHY Control specifications. This is where you might send a command and verify that the device has responded properly. PCS transmit/receive, state diagrams, encoding/decoding, and scrambling/descrambling are some of the areas touched on by these tests. The test specs also provide recommendations for other elements, such as the common-mode choke, EMC, and the communication channel itself.
About Compliance Testing
Let's now turn to compliance testing in the context of Automotive Ethernet. The 100Base-T1 (IEEE 802.3bw) and 1000Base-T1 (IEEE 802.3bp) specifications include compliance requirements for the PMA, PCS, and PHY Control. But the IEEE's specification does not carry the test specification itself. Traditionally, the University of New Hampshire's InterOperability Laboratory, which hosted the first 100Base-T1 plugfest in November of 2016, has written test documents that describe how to perform compliance testing. Ultimately, it is up to the OEM, PHY vendor, or Tier 1 supplier to work with either a test equipment manufacturer or a test house to determine whether a product complies with the specification.
PHY compliance looms large when you consider that OEMs have lengthy development cycles for a vehicle's electronic control unit (ECU). It is prudent, then, to assure that the PHY chip that the ECU is designed around meets the specification's requirements before it is designed into the system.
Once the PHY chip has been incorporated into the ECU, the system itself must also be tested for compliance. Testing the chip on an evaluation board is one thing but testing it in situ is yet another. Is your ECU board layout correct? Maybe, or maybe not. Does the PHY chip's behavior change in the context of the ECU board? These are questions that are answered by testing the system for compliance.
Bear in mind, though, that compliance with the 100Base-T1 and 1000Base-T1 specifications does not guarantee interoperability. With PCI Express or USB, when your product passes compliance testing, it gets to carry the protocol's logo and you can expect interoperability with all other products that have likewise passed testing. However, while the 100Base-T1 and 1000Base-T1 specifications provide well-defined transmitter requirements, the receiver implementation is left up to the designer.
Six tests comprise the PHY compliance test battery for both BroadR-Reach and 100Base-T1:
- Maximum transmitter output droop
- Transmitter clock frequency
- Transmitter timing master jitter
- Transmitter timing slave jitter
- Transmitter distortion
- Transmitter power spectral density (PSD)
For 100Base-T1 and 1000Base-T1, one more test is added for a total of seven:
- Transmitter peak differential output
The 1000Base-T1 specification also includes an MDI jitter test case that is not in 100Base-T1 or BroadR-Reach.
The test document written by the UNH’s InterOperability Laboratory covers some test equipment requirements. Automotive Ethernet's 100-Mb/s bandwidth calls for an oscilloscope with a bandwidth of at least 1 GHz with a sampling rate of at least 2 GS/s (Figure 1). Teledyne LeCroy recommends a sampling rate of 10 GS/s to ensure 10X oversampling for jitter measurements. With the emergence of the 1000Base-T1 specification, the bandwidth requirement will move up to the 2 GHz range.
Moreover, it's recommended that the oscilloscope used for testing can perform spectral analysis. Failing that, it's recommended to supplement the oscilloscope with a spectrum analyzer.
For the 100Base-T1 transmitter distortion test, you need a sine-wave generator capable of a 5.4 Vpk-pk differential sine wave at a frequency of about 11.11 MHz. Along with that, you'll need two BNC cables and two BNC-to-SMA adapters to get those signals into the Ethernet test fixture. A short Automotive Ethernet cable converts whatever interface is on the DUT's MDI output to an RJ-45 connector. That in turn allows you to plug into the Ethernet test fixture. From there, the signal goes to the oscilloscope. A 1-GHz differential probe accesses the transmit clock, and you'll need a vector network analyzer for the return loss and common-mode tests.
The Five Test Modes
What follows is an overview of the five test modes comprising the transmitter compliance test suite for the 100Base-T1 protocol. The test modes allow for a common pattern to test stressful conditions across all devices. Testing in this fashion offers the best possible odds for achieving true interoperability.
First, a quick digression: Where in a typical Automotive Ethernet channel is the electrical compliance testing to take place? Referring to Figure 2, there is a PHY chip in both the transmit and receive outputs. Also shown is a low-pass filter, common-mode choke, connectors, and wiring. All of this is replicated at both ends of the channel.
Electrical compliance testing is defined at the connector of the transmitter. That would be the point shown in Figure 2 that follows the low-pass filter and common-mode choke. Channel and connector recommendations are defined in another document. But all the electrical compliance testing is to be conducted at the transmitter connector.
With that said, let's turn now to the 100Base-T1 test modes. Again, these test modes are intended to stress devices in ways that are conducive to testing for various issues (Figure 3). Testing all devices with these same test modes gives you a fighting chance to achieve interoperability between them.
The five modes shown in Figure 3 will all comprise different signal conditions, such as long strings of ones or minus ones, or distortion, that wouldn't typically be seen in a real-world data transmission. But they're going to quickly reveal what your implementation can and cannot handle.
Test mode #1 detects transmit droop. The waveform is essentially a square wave, defined in the specification as N +1 symbols followed by N -1 symbols. For this test, the symbol period must be at least 500 ns in duration. At a signal rate of 100 Mb/s, at least 34 symbols will be transmitted in that 500-ns span.
Test mode #2, which quantifies transmit jitter in the DUT's master mode, appears as a sine wave. Thus, it is simply a repeating sequence of -1, +1, -1, +1, and so on. It is essentially a 33.333-MHz clock signal.
To the naked eye, the optional test mode #3 appears indistinguishable from mode #2. The distinction is that while test mode #2 is performed in master mode, test mode #3 is performed with the DUT in slave mode.
Distortion is the object of test mode #4. Mode #4 is a PAM-3 signal with a symbol interval of 15 ns (Figure 4). Note that there is a pattern that repeats every 2047 symbols, a figure that's defined by the polynomial in the test specification.
Finally, test mode #5 for power spectral density and peak differential output is much like mode #4, except mode #4 has that repeating pattern whereas mode #5 is random PAM-3 data with the same symbol interval of 15 ns. Again, there is no discernible difference to the naked eye between modes #4 and #5.
As to the generation of these test mode signals, each PHY silicon vendor incorporates a "backdoor" method of modifying the necessary registers to enter each test mode. These methods are often not publicly available, and the methods vary somewhat from vendor to vendor. The best approach is to ask your PHY vendor how to generate the test modes.
About the author
David Maliniak is Technical Marketing Communications Specialist at Teledyne LeCroy. He blogs, creates support documentation, and writes the occasional technical article. Before joining Teledyne LeCroy in 2012, David was Test and Measurement/EDA Technology Editor at Electronic Design Magazine.