Online Monitoring for Automotive Sub-systems Using 1149.4 C. Jeffrey, A. Lechner & A. Richardson Centre for Microsystems Engineering, Lancaster University, Lancaster, LA1 4YR, UK 1 Abstract This paper investigates the use of an embedded micro controller and the IEEE1149.4 architecture as a facility for online monitoring of automotive systems. 2 Introduction State-of-the-art vehicles have between 20 to 50 processors on board offering increased reliability, safety and performance. The applications range from climate control for passenger comfort to safety critical airbag subsystems. Most of these sub-systems are configured as sense-process-actuate loops to provide enhanced system control. In automotive applications, sensors and actuators are normally placed directly at the source they interact with. In many cases these are in harsh environments (break calliper, engine, exhaust etc.). Currently the associated signal processing and control electronics are placed in benign environments (boot, passenger compartment, off-engine etc.). Locating intelligence directly at the transducer interface is however becoming extremely desirable as sub-system performance increases as signal noise can be reduced, auto/intelligent calibration can be realised, connector count can be reduced and manufacturing costs optimised. Table 1: Electronic environmental conditions in automotive applications Location Typical Continuous Temperature Temperature Range Vibration Level On Engine 140 C -40 to +140 C Up to 15g At Engine (Intake Manifold) 125 C -40 to+125 C Up to 10g Under Hood Near Engine Under Hood Remote Location 120 C -40 to +120 C 3 5g 105 C -40 to +105 C 3 5g Exterior 70 C -40 to +70 C 3 5g Passenger Compartment 80 C -40 to +80 C 3 5g The implications of this however are a paradigm shift in automotive electronic design from board based to fully integrated systems operating in harsh environments. Table 1 provides some examples of the environmental specifications associated with these new systems. Existing practice associated with the use of Fibre Reinforced glass epoxy (FR4) substrates and plastic packaging techniques for automotive electronic control units are unlikely to deliver the performance required of these new systems hence approaches such as the use of bare-dies in aerospace type packaging need to be further investigated The need for integrated test support hardware such as the use of IEEE1149.1 boundary scan [1] will be critical to address production test costs and test quality requirements with further functionality including condition monitoring required for safety critical sub-systems. This paper addresses these challenges by attempting to provide a cost effective solution based on the extension of IEEE 1149.4. 3. Background IEEE 1149.1 Boundary Scan is a digital architecture that enables accessibility and observablity to circuit nodes and interconnections through a Test Access Port (TAP). The IEEE 1149.4 standard for a mixedsignal test bus [1] extends the test access structure for analogue stimulus injection and response evaluation via two busses connected to pins and respectively. Recent research in 1149.4 [2,3] and its implementation and application [4, 5, 6] have shown that measurements for current, voltage level, frequency and phase are supported. However, measurement accuracy is limited due to capacitive and resistive characteristics of the Analogue Boundary Modules (ABM) that also cause degradation of the observed signal. Limitations in passive component measurements (resistance and capacitance) facilitating 1149.4 have been identified [7].
The authors of this paper proposed in [8] to reuse the 1149.4 standard in an online application. The aim is to enable successive monitoring of circuit nodes with minimum area overhead while avoiding an increase in pin count: To assess the feasibility of reusing 1149.4 one needs to 1. Carry out fault simulations to identify critical signal properties for testing (magnitude, frequency). 2. Create an integrated tester that could probe the corresponding circuit nodes and analyse signal properties. It is very important to ensure that the loading of a circuit node will not affect system performance and that the signal is not degraded as it is passes through an ABM to the tester. This work documents the initial experimental results into the technique presented in [8] and outlines the next phase of the work. 4 Automotive Subsystem The demonstrator automotive subsystem is an electric motor drive system. This sub-system gathers analogue data from 8 sources including current monitors, voltage monitors, Hall effect sensors, switch detectors and an accelerometer (see Table 2). The corresponding signal properties that could be verified online are listed in Table 2. Since 1149.4 is not fitted to devices as standard we are using the dual multiplexers from National Semiconductors as an add-on block. Interfacing to the IC signal pins through hardwired devices, as illustrated in Figure 1, emulates 1149.4 compliance. Apart from 1149.4, the tester block in Figure 1 will ultimately be integrated on-chip. The tester controls the configuration of the access structure (via and ) and performs the measurements of signal properties. Table 2: Analogue signal types on electric motor drive system Application Signal Type Hall sensor 11 18mA,10kHz Accelerometer -0.3V to +0.3V, 400Hz Switch Low < 1.53V, High > 3.15V Voltage monitor 0 3V < 1kHz 5 Configuration of Test Architecture Currently the off-chip tester comprises a Motorola M68HC08 micro controller (HC08) for embedded command development. A Solartron 1260 impedance analyser was used for the characterisation of the ABM and a TDS210 oscilloscope was used to analyse the waveforms when the system is online. 5.1 Reconfiguration Algorithm The algorithm used to connect the ABM to the analogue bus is based on controlling the state machine in the IEEE1149.4 compliant device. The two instructions that need to be passed to the s instruction register to facilitate online monitoring are: A0 1149.4 compliant device IC IC 1149.4 compliant device TAP TBIC JTAG development board Ch1 Ch2 Tektronix TDS210 Scope Tester Figure 1: Using devices as an emulation of 1149.4 compliant devices
BYPASS the device: switches the device from a scan chain to a 1-bit bypass register PROBE: enables the scan chain to configure individual ABMs in the scan chain and to place them in certain states (connect to,, high impedance, logic 0 etc). Figure 2 illustrates the different states of the 1149.4 compliant device including the number of clock cycles the HC08 requires to program the relevant vectors. The number of clock cycles for setting the instruction and data registers (Scan IR and Scan DR) depends on the number of devices in the scan chain. The time required for setting registers also depends on the actual data values (transmitting a logic 1 takes 4 extra clock cycles compared with transmitting a logic 0). Reset Scan IR Update IR Scan DR Update DR Run/Test Idle 261 Clock cycles 129 Clock cycles (TMS) + 1396-1400 Clock cycles per 135 Clock cycles 237 Clock cycles/configuration + 2606 Clock cycles per (TDO) + 45 Clock cycles per additional device 34 Clock cycles + Test time Figure 2: Structure of configuration algorithm The configuration algorithm for online monitoring of a set of device nodes predetermined by fault simulation is as follows: RESET: Reset the scan chains. Scan-IR: Load the instruction register where the device-under-test is loaded with PROBE while the others are loaded with BYPASS. Scan-DR: Configure the ABM at the device node to be monitored. Run/Test Idle: Disables the scan chain instruction register and waits for completion of measurements. Go to Scan-DR to configure for remaining device nodes to monitor. Go to Scan-IR to configure for other devices that require their nodes to be monitored. Note that the configuration algorithm provides sequential access to the predetermined set of nodes of a particular device. When all the device s nodes have been monitored, the configuration algorithm steps to the next device in a continuous loop etc. 6 Experimental Results The HC08 was set to run at 2.673MHz and executed the configuration algorithm. The timings shown in Table 3 are based on implementing the monitoring technique alternately on both the muliplexers in the. Hence each device offers the ability to monitor 2 analogue device nodes. Table 3: Configuration times of ABM s using a 2.5MHz microprocessor Number of s Initial configuration (ms) The initial configuration timing in the second column is the time it takes to probe the first circuit node on a new device. The third column denotes the time required to reconfigure to the next node within the same device (switch from monitoring one multiplexer to the other). The fourth column shows the average frequency at which the HC08 can switch between monitoring nodes in the system when using both channels of each. As tabulated, the rate of accessing an individual node decreases with increasing number of nodes to monitor. 6.1 DC Response Configure within a device (ms) Monitoring frequency (Hz) Number of device nodes monitored 1.20132.12996 384 2 2.21912.13176 192 4 3.27692.13356 128 6 4.33472.13536 96 8 5.39252.13716 76 10 A low frequency sinusoid is passed through an ABM to identify maximum and minimum voltages to propagate through without clipping. Figure 3: Maximum and minimum voltage limits of the ABM
As illustrated in Figure 3, the maximum voltage that can be observed through the ABM is 3.92V. The minimum voltage observed was -640mV with an accuracy of +/- 10mV. 6.2 AC Response As the 3dB cut-off frequency of the was unknown, it has been determined through measurements. Figures 4 to 6 show the ABM input sine waves of 200mV amplitude and the signals observed at for different sine wave frequencies. At 50kHz stimulus frequency, no signal degradation in phase and magnitude can be observed. Figure 6: 1MHz input sinusoid and the output of the (3 db point) 6.3 Linearity The Linearity of the probing is tested to see if the waveforms are subject to distortion. The A0 pin of the has its input voltage swept from -0.5V to 5V. The TAP is then configured to connect the A0 pin to AB1. The waveforms are then subtracted from each other, any non zero points on the resulting waveform are considered to be non-linear. The resulting waveforms are show in Figure: 7. Figure 4: 50kHz input sinusoid and the output of the At 200kHz a phase difference becomes observable. Hence one has to account for that difference in phase dependant measurements. Figure: 7 Linearity of measuring a signal through the and ABM The resulting waveform shows that the observed signal is a good representation of the probed signal in the range of 0.5 to 5V. 6.4 Crosstalk Figure 5: 200kHz input sinusoid and the output of the The 3dB cut-off point for the system is shown in Figure 6 at the frequency of 1MHz. The level of crosstalk was measured on the on the when an AC signal was passed through one of the multiplexers on the. The frequency of the input signal was stepped logarithmically and the magnitude of the signal was stepped from 0 to 5V with monitored to characterise any parasitic signal present.
Crosstalk (mv) 0.6 0.5 0.4 0.3 0.2 0.1 0 1 10 100 1000 10000 100000 1E+06 Log Frequency Figure 8: Crosstalk on the It was observed that the magnitude of the crosstalk is independent of input signal magnitudes beyond 0.5V. The voltage independent results are shown in Figure 8. 7 Conclusion This paper shows that it is possible to reuse IEEE 1149.4 as a facility for online parameter monitoring of analogue signals in an automotive environment. The frequency at which the HC08 configures to the relevant ABMs is dependant on the number of 1149.4 compliant devices within the system. The configuration time does not account for the measurement itself. The AC measurements show that the ABM can reproduce signals accurately below 50kHz and can measure non-phase sensitive signals below 200kHz. This suggests that this monitoring method would be appropriate for automotive sensor/actuator systems. The crosstalk on the is quite significant especially if there are many devices connected for online monitoring. However the was not designed for what has been proposed and crosstalk can be greatly reduced if it is a known issue at design time. It is therefore not considered a limitation in future work. This infrastructure is a very good candidate for online monitoring of automotive subsystems and sensor interfaces showing good linearity and frequency response. 8 Further Work The HC08 microprocessor is limited in the speed at which it can configure and does not have sufficient resolution (8 bit) for accurate signal measurements (FFT). The more powerful Shark DSP form Analogue Devices will be used to continue this investigation. For the demonstrator EPB system, further work has to address the selection of device nodes for online monitoring in terms of suitable signal properties and their ability to detect in-field device degradation and reliability failures. Once a set of device nodes is determined, devices will be embedded into the automotive subsystem to implement the online monitoring technique. Acknowledgement This work is being supported by the DREAM project. EPSRC GR/N38271/01. The authors would like to thank National Semiconductor and JTAG Technologies for access to their technology and prototype devices. References 1 IEEE Standard 1149.4-1999, Standard for a Mixed Signal Test Bus, IEEE, USA, 2000 2 S. Sunter, K. Felliter, J.Woo P. McHugh, A general Purpose 1149.4 IC with HF Analogue Test Capabilities, Proc. of International Test Conference, pp. 38 45, 2001 3 U. Kac, F. Novak, Experimental test infrastructure supporting IEEE 1149.4 standard, Poster presentation at European Test Workshop, 2002 4 G. O. Ducoucdray Acevedo, J. Ramirez- Angulo, Innovative Built-In Self-Test Schemes for On-Chip Diagnosis, Compliant with IEEE 1149.4 Mixed Signal Test Bus Standard, Journal of Electronic Testing, vol 19, pp 21-28, 2003 5 J. McDemid, Limited Access Testing: IEEE 1149.4, Proc. of International Test Conference, pp388-395, 1998 6 U. Kac, F. Novak, S. Macek, M. S. Zarnik, Alternative Test Methods Using IEEE 1149.4, Proc. of Design And Test in Eurpoe, pp.463-467 2000 7 I. Duzevik, Preliminary Results of Passive Component Measurement Methods Using and IEEE 1149.4 Compliant Device, http://www.national.com/appinfo/scan/files /duzevik_btw02_paper.pdf 8 C. Jeffrey, A. Lechner & A. Richardson Reconfigurable Circuits for Fault Tolerant Systems: Factors to Consider, Proc. of International Mixed Signal Test Workshop, pp 247-249, 2002