CAN physical layer protection hints

Transcription

1 Design CAN physical layer protection hints By Steve Corrigan (Kvaser) Fig. 1: IEC ESD test waveform The transceivers on many industrial CAN networks serve as the last line of defense against unexpected transient electrical events. This may be partly due to the fact that CAN transceiver manufacturers typically list bus protection with high electrostatic discharge (ESD) ratings as a marketing tool for selling devices. However, ESD protection does not provide the protection required by an operating CAN network. Modern semiconductor components, such as low voltage MOSFETs and integrated circuits can be damaged by disturbances as little as 10 V, and their survivability is poor in unprotected environments. System engineers take on a big risk when expecting a CAN transceiver s ESD cells to protect a bus from all transient events. The IEC series of standards on electromagnetic compatibility (EMC) testing and measurement techniques for electronic circuits is widely accepted as reflecting the actual environmental conditions that threaten an operating CAN network. The test and measurement techniques for these conditions are examined in this paper along with circuit protection suggestions against voltage surges and transient voltage spikes. Introduction to CAN protection ESD protection standards were developed to ensure Table 1: IEC ESD voltage test levels Contact discharge Air discharge Level Test voltage [kv] Level Test voltage [kv] 1 ±2 1 ±2 2 ±4 2 ±4 3 ±6 3 ±8 4 ±8 4 ±15 that integrated circuits survive the assembly manufacturing process at the chip level and offer no protection for the system-level CAN network from the transient bursts generated by inductive load switching or the surge of a nearby lightning strike. While many engineers are familiar with ESD standards such as the Human Body Model (HBM), Charged Device Model (CDM) and Machine Model (MM), there is a common misunderstanding among these standards and the IEC ESD standard. The HBM, CDM and MM standards are designed for integrated circuit protection during circuit board assembly, however, IEC testing is for the protection of finished electronic products, not chips. Regardless, it is now often being cited for the protection of chips such as CAN transceiver bus pins, leading many system designers to believe that a CAN network is completely protected because of this feature. The IEC standard clearly defines the unique differences among the common voltage-transient threats to a CAN network beyond ESD events and offers standardized testing in IEC and IEC This paper examines the testing for these transient-voltage events that corrupt data and damage CAN transceivers, then provides protection circuitry suggestions against all damaging events. ESD protection: benefits and liabilities The purpose of the traditional ESD testing of chips in the manufacturing environment is completely different Fig. 2: IEC ESD test-timing sequence than the system level testing of a CAN network. Processes such as the placement of a chip on a circuit board and the soldering process are done in a controlled ESD environment, and limit the level of ESD stress placed on an integrated CAN transceiver circuit. Although ESD protection is more than adequate for this controlled manufacturing environment, it provides no protection for system level voltage transients. The IEC standard relates to the im- 34 CAN Newsletter 1/2010

2 munity requirements and test methods for packaged electrical equipment when subjected to static electricity discharges from operators directly, and from adjacent objects with which the equipment may come in contact. However, a system designer does benefit from the additional protection offered by this standard dur- Festo Fig: 3: Typical ESD Testing ing CAN network construction and maintenance. The standard defines ranges of test levels, which relate to different environmental, installation and maintenance conditions, and establishes corresponding test procedures. As shown in Figures 1 and 2, characteristics of this test are the short rise-time and the short pulse duration of less than 100 ns (basically a low-energy, static pulse). The test defines a minimum of ten single discharges of positive and negative polarity each, with a recommended time interval between discharges of one second. Figure 3 displays the big difference between standards is the amount of peak current and I 2 R power released during an ESD test. Over five times the current is flowing during an IEC ESD test than in a typical HBM test for any of the voltages listed in Table 1. Another notable difference between the standards is the rise-time of the voltage strike. As shown in Figure 1, a typical IEC test strike rise-time is 0,7 ns to 1 ns while the typical HBM strike is 25 ns. In an aver- CANopen and Festo! CANopen automation platforms with IP20/65: Remote I/Os, valve terminals, proportional and servopneumatics, motion control, Safety@Festo all efficiently integrated. For more information: search term CANopen Festo AG & Co. KG Fig. 4: Voltage waveform of an EFT (burst) pulse

3 Design Fig. 5: EFT (burst) pulse timing sequence of a complete test cycle age CAN transceiver, an IEC strike can easily destroy a circuit before the integrated HBM protection has had a chance to engage. Note that IEC contact discharge testing is now considered to be the test methodology of choice since air-discharge testing fails to yield consistent test results in the lab environment. The test procedures of many electronic equipment manufacturers ascribe the ESD tests the lowest priority of all transient immunity tests since a potential ESD occurrence is limited to the handling, installation and maintenance work of modules and CAN wires. During this maintenance, operators typically wear ESD protective clothing and follow ESD discharge procedures prior to any direct contact with the CAN network or modules. IEC : EFT protection A good example of an electrical fast transient (EFT) occurs when a reactive load, such as a motor, solenoid or relay coil, is switched off. The rapidly collapsing magnetic field induces a transient voltage across an inductive load s winding which can be expressed by the formula: V = - L (di/dt) where L is inductance in Henrys and di/dt is the rate of change of current in amps per second. EFTs are caused anytime a switch through which current is flowing is opened. As a switch is opened, arcing occurs between the contacts. At first, low voltage with high frequency arching occurs while contacts are close together. Then, as the contacts become further separated, voltage grows to a higher magnitude and the frequency lowers. When these EFT occur near CAN cables, data is corrupted and node equipments is commonly damaged. Inductive switching transients are known as the silent killers of semiconductor devices since they often occur with no outward indication. The significant features of the IEC Burst test in Figures 4 and 5 are the high voltage amplitude, the short rise-time, the low energy of the transients and the high repetition rate. With a 1- s interval from pulse-front to pulse-front, an EFT burst of 15 ms duration contains at least pulses. When this is multiplied by the number of bursts within a 10 s window yields pulses per 10-s window. Therefore the Fig. 6: Surge voltage and current test waveforms application of six 10-s test windows with a 10 s pause interval results in pulses within 50 s. Since the burst test simulates everyday switching transients, it often has the highest priority of the three transient immunity tests among system designers. Although low in energy content, the vast numbers of transient events that continuously bombard unprotected circuitry often give designers cause for much concern. The specifications of IEC simulate an inductive switching transient-voltage threat having 50-ns wide spikes with amplitudes from 2 kv to 4 kv occurring in 300-ms wide bursts. The test is intended to demonstrate the susceptibility or immunity of electrical and electronic equipment when subjected to electronic disturbances such as those originating from the switching transients generated in the interruption of inductive loads by motor starts and stops or relay contact-bounce. The test is intended to replicate the continuous bombardment of electrical transients that are common to a factory floor. Since the EFT testing does not involve direct contact of conductors and is applied indirectly through a capacitive clamp, it becomes intuitive that the choice of a CAN cable with internal shielding effectively attenuates EFT coupling into the conductors. IEC : Voltage surge protection The electric field strength of a distant lightning strike may be strong enough to cause catastrophic or latent damage to semiconductor equipment. It becomes important to be able to quantify the induced voltage as a function of distance from the strike. Figure 6 displays that these induced voltages 36 CAN Newsletter 1/2010

4 can be quite high, explaining the destruction of equipment from relatively distant lightning flashes. The object of this standard is to establish a common reference for evaluating the immunity of electrical and electronic equipment when subjected to voltage surges. This part of IEC relates to the immunity requirements, test methods, and range of recommended test levels for equipment to unidirectional surges caused by overvoltages that can be caused by switching operations in a power grid and lightning transients. The test equipment output waveforms are specified for open- and short-circuit conditions. The ratio of the open-circuit peak-voltage to the peak short-circuit current in Table 2 is a function of the generator output impedance of 2. Characteristics of this test are the high current due to low generator impedance, and the long pulse duration, which is approximately times longer than the ESD and Burst tests. It becomes readily apparent from Figures 6 and 7 that the critical parameters of a surge protection device are its ability to divert high values of current and clamp the voltage within the short-circuit stand-off voltage range of a CAN transceiver, preferably within the common-mode operating range. CAN transceivers often have short-circuit stand-off protection of ±36 V or more for an unlimited amount of time without damage to the Fig. 7: Surge pulse and timing sequence of a complete test cycle device. Voltage could easily be clamped within this range, however, any bit transmitted during this interval may generate an error frame since all I/O s of these devices typically go into a high-impedance state until the voltage is removed. However, the device is completely unharmed and quickly returns to normal operation. CAN protection circuit suggestions Relying solely on a transceiver s integrated ESD structure when designing a CAN network is a risk. A robust system requires external protection circuitry to absorb the much higher impact of burst and surge transients. Fundamental riskanalysis reveals that the relatively small expense of ad- Port Table 2: Ratio between peak open-circuit voltage and peak short-circuit current Open-circuit peak voltage ±10 % [kv] Short-circuit peak current ±10 [ka] 0,5 0,25 1 0,

5 Design Fig. 8: Transient voltage protection schematic dition board space and the protection circuitry pales in comparison to the possible future expense of system downtime. Figure 8 shows the circuit schematic of a basic CAN node with transient protection. A screw terminal connects the CAN twisted-pair data cable, power and the ground return path to the transceiver. Transient voltage suppressor diodes (TVS) are used to eliminate common-mode transients between CAN_ H and ground, CAN_L and ground, and for differential transients between CAN_H and CAN_L. Transient voltage suppressors are available from several vendors with a low Fig. 9: Common TVS footprints cal TVS response times are in the picoseconds range with power ratings of up to several kilowatts, making them the most effective protection available against ESD, Burst, and surge transients. Care must be taken in the selection of TVS devices. Noting the typical V-I characteristics of a TVS device in Figure 9, a few micro-amps of leakage current (IWM) pass through the device up to the TVS workingmaximum voltage, VWM, while the transient suppressor is in a high-impedance state. Therefore, when selecting a TVS device, make sure that the working-maximum voltage is above or equal to the maximum bus Fundamental risk-analysis reveals that the relatively small expense of addicapacitance, which allows them to be designed-in into every node of a multi-node network without requiring a reduction in data rate. Typivoltage during normal operation, which includes the common-mode operating range of the transceiver, VWM VCAN_H, VCAN_L. At the breakdown voltage, VBR, when a TVS device begins clamping, the device enters a low-impedance state and begins conducting high current (IC). This dynamic impedance change displayed in Figure 10 causes current flowing through the TVS to create a voltage drop known as the clamping voltage, VC, which rises with increasing Fig. 10: TVS operating parameters current. When a transient occurs, the TVS clamps instantly to limit the spike voltage to the clamping voltage while conducting potentially damaging current away from the CAN transceiver. The user must ensure that this clamping voltage does not exceed the maximum voltage stand-off ratings of a CAN transceiver s bus terminals. Most CAN transceivers are designed with a short-circuit stand-off voltage rating of ±36 V or more which provides for a wide margin of clamping. As an example, to comply with the typical - 7 V to 12 V common-mode voltage range of many CAN transceivers, TVS such as those in Figure 10, should have a stand-off voltage VWM 12 V. Depending on the power rating of the TVS chosen, the maximum clamp voltages can range up to 40 V, which falls within the ±36 V short-circuit voltage stand-off range of many CAN transceivers. Conclusion 38 CAN Newsletter 1/2010

6 Design suggestions: Always place the TVS device near the input terminals or connectors to restrict electromagnetic coupling. For a robust board design, use a four-layer PCB with layer stacking in the following order (top-tobottom): Bus signal layer, ground plane, power plane and control signal layer. Note: Adequate bypass capacitors between power and ground provide for transmission line routing over the power plane or ground plane to employ both planes as low-inductance return paths. Route differential signal traces on the top layer to avoid the use of vias (and the inductance) and provide a clean connection between the connector and transceiver. Placing the power plane next to the ground plane creates additional high-frequency bypass capacitance of approximately 100pF/in2. Route control signals on the bottom layer to provide greater routing flexibility as these signals have a greater noise margin to tolerate discontinuities such as vias. Eliminate or minimize all conductive loops including power and ground loops. Never run critical signals near board edges to restrict electromagnetic coupling. Always use shielded, 120 characteristic impedance twisted-pair cable shielding greatly reduces the transient effects of EM coupling. ASCON Sigmadue I/O CANopen Remote Modules Powerful processing capability on board Multifunction for enhanced flexibility Stand alone for real distributed automation tion board space and protection circuitry pales in comparison to the possible future expense of system downtime. Since the EFT test simulates everyday switching transients, it often has the highest priority among system designers of the three transient immunity tests. Although low in energy content, the vast numbers of EFT transient events that continuously bombard unprotected circuitry often give system designers cause for much concern. EFT inductive switching transients are known as the silent killers of semiconductor devices since they often occur with no outward indication. EFT testing does not involve direct contact of conductors, but rather the indirect application through a capacitive clamp, and it becomes intuitive that the choice of a CAN cable with internal shielding effectively attenuates EFT coupling into the conductors. CAN cable such as Belden cable s 3107A is a very good example of a shielded twisted-pair cable for the extremely harsh environments of many CAN applications. info@kvaser.com References and further reading V. A. Rakov, Martin A. Uman, Lightning, physics and effects, 2003, Cambridge University Press, ISBN C.A. Nucci, F. Rachidi, Lightning-Induced Overvoltages, 1999, IEEE Sandia National Laboratories, Electromagnetic Analysis and Testing, 2007 Howard W. Johnson, High- Speed Digital Design, 1993, Prentice-Hall, Inc., ISBN Ivan G. Lawson, Protection Notes, Protek Devices, For more information on industrial CANbus cable visit Ascon Other ASCON Products Controllers, Indicators, Transmitters and Acquisition Systems ASCON spa Bollate (Milano) Italy Tel Fax sigmadue.sales@ascon.it 67 39