Addressing immunity requirements for CAN transceivers
Controller Area Network (CAN) transceivers have been present in cars as a communications backbone for over three decades. Throughout this period, the number of electronic chips as well as the overall transceiver complexity has steadily increased. As a result, multiple electronic subsystems have to coexist within close proximity and perform flawlessly under harsh conditions. This imposes limits on the amount of noise a component can generate (emissions) as well as the amount of noise it must withstand (immunity). In this article, I’ll focus on one of the immunity tests that transceivers must pass before their successful deployment in vehicles.
Electric motors, switching converters, high-current drive stages and oscillators are among the types of noise injectors that can introduce ripple on the power lines. Communication modules such as navigation and internet subsystems as well as external disturbances also add noise, which can conduct or couple onto sensitive electronics and potentially disrupt their behavior. In this context, CAN transceivers are still expected to successfully exchange data; to do this, they must have strong noise immunity.
The International Electrotechnical Commission (IEC) 62132-4 standard was created to harmonize requirements from various equipment manufacturers. It covers the measurement of immunity to directly injected noise, from 1MHz to 1GHz and up to 39dBm (8W) of power. A block diagram of the setup is shown in Figure 1.
The CAN transceiver (device under test) is set up with a 5V, 250kHz, 50% duty cycle square-wave data signal sent to its input. The transceiver’s internal circuits mirror a delayed version of this signal back onto the receive output of the chip. A coupling network then injects common-mode noise into the setup and the receiver output is evaluated against the pass-fail criteria outlined in Table 1. At each frequency, the power level increases up to the maximum test power (39dBm) to see if the transceiver can communicate without any issues. These gradual increases repeat across the frequency band to 1GHz, as shown in Figure 2. If the transceiver has a standby mode, a second test entails configuring it in standby with the receiver output at a static high. Noise is then injected and the device output is checked to see if it successfully remains high and there is no inadvertent state change.
The red line in Figure 2 defines the minimum noise power that the transceiver must withstand. This test is performed both with and without an external common-mode choke (CMC), which can be used to suppress common-mode noise. When using a CMC, the peak power level required to pass is 39dBm (36dBm otherwise). Note that 3dBm corresponds to a 2x power-level delta.
Since CAN is a differential protocol, one way to ensure good performance is to ensure that everything in both CAN pin signal paths is well-matched and balanced. This maximizes the circuits’ common-mode rejection and reduces spurious conversions to differential signals. At peak power, swings can reach up to ±65V. At >100MHz, these signals can easily couple onto adjoining signal traces and wreak havoc unless they are designed to withstand such stresses. Filtering and shielding sensitive signals can ensure that this doesn’t happen.
The IEC 62132-4 test is a stringent metric to measure the immunity of a CAN transceiver. Combined with other electromagnetic compatibility (EMC) metrics such as emissions and electrostatic discharge (ESD), a device that passes this harsh test enables system designers to make smart, cost-effective trade-offs between noise suppression at the source versus immunity at the component level. Good performance in this test is a fairly good predictor of real-world performance and can thus minimize unpleasant surprises later in the automotive design cycle.
Originally published on Embedded's sister site, Planet Analog: "Signal chain basics #125: automotive immunity requirements for CAN transceivers."