EMBEDDED FIBER-OPTIC COMBUSTION-PRESSURE SENSORS FOR AUTOMOTIVE ENGINES
M. T. WLODARCZYK, T. POORMAN, L. XIA, J. ARNOLD, and T. COLEMAN
A reliable, long-life, and low-cost cylinder pressure sensor is the enabling element of advanced control systems that have potential for significant fuel economy improvements, reduced levels of combustion pollutants, and improved engine reliability and performance. Optrand has developed high-temperature, long-life, and miniature fiber optic pressure sensors suitable for integration into higher-functionality devices such as "smart" ignition systems, fuel injectors, or glow plugs. The design and performance of integrated Optrand sensors are reported for gasoline, diesel, and natural gas engines demonstrating a total error due to non-linearity, hysteresis, and thermal shock below +/-1% and the lifetime of 500 Million pressure cycles.
Cylinder pressure is the fundamental variable that determines a combustion engineís operating state. In particular, combustion pressure information can be used in advanced engine control and monitoring systems, if available continuously and in real-time . Based on cylinder-specific pressure information, closed-loop control applications have been proposed for power balancing in large-bore natural gas engines , lean burn combustion in passenger cars , or stall control in aircraft engines. The most advanced controls allow both fuel delivery and ignition control in each cylinder and during each combustion cycle in what has been termed the Controlled Combustion Engine (CCE) . Combustion pressure, when detected in all cylinders, in addition can provide the most deterministic information about engine performance in the area of engine monitoring.
The two critical combustion attributes that are of interest in engine control are ignition timing and air-fuel charge delivery. Closed-loop spark control techniques provide timing that is optimized in the presence of changing engine and environmental conditions. Such approaches provide optimum performance under all conditions and are able to adapt to varying knocking conditions in each cylinder. Based on cylinder pressure information, air-fuel ratio can be estimated to provide the best possible transient control in individual cylinders during cold-start and in lean-burn region. The latter application is particularly important in direct injection engines that carry a promise of as much as 35% fuel economy improvements combined with significant power and torque increases .
While the benefits of cylinder pressure information for engine control and monitoring are well recognized, commercial implementations of practical systems have been hampered by one overriding factor: the lack of a cost-effective, reliable, and durable combustion pressure sensor. Piezoelectric-quartz pressure transducers that have been used over decades in engine development and calibration  are not suited for implementation in production engines. They are subject to electromagnetic interference (EMI) effects, have limited lifetime, and are unacceptably expensive. Lower cost piezoceramic devices, such as spark plug washers and boss-type sensors, do not offer high accuracy under all engine conditions, are subject to electrical interference problems, and are prone to large temperature errors. In addition, their durability is not sufficient for use in production engines as a consequence of degrading effects of alloy segregation, selective oxidation, and diffusion.
Newly available piezoresistive Silicon combustion pressure sensors have potential for use in production engines . They utilize a low-cost piezoresistive element, which must be maintained below a temperature of 125oC, connected to the sensor diaphragm via a transfer pin. However, these types of sensors require a separate access point into the engineís head as well as special mounting location for water-cooling and non-disruption of oil passages. In modern four-valve engines there is no space available for a separate sensor. Furthermore, piezoresistive sensors are subject to EMI, are large in size, have not shown long lifetime, and are too expensive at present. Due to the above limitations, the piezoresistive sensors have found very limited use in production engines.
Silicon Carbide (SiC) based sensors, such as recently reported by a team from Daimler Benz, target operations at temperatures as high as 500oC . Unlike in Silicon devices where the sensing diaphragm and the processing electronics are spatially separated SiC aim at combining the both functions into a single compact device. However, SiC sensors suffer form a number of limitations of other electronic devices including susceptibility to EMI, lead oxidation, as well as lifetime problems related to long-term diffusion and phase segregation effects at high temperatures. The reported devices demonstrated relatively poor performance in particular in the area of temperature errors as large as 20%. Finally, the SiC material itself is prohibitively expensive and its processing techniques are complex and expensive as well.
In contrast to electronic devices, fiber-optic sensors are potentially very well suited for applications characterized by high temperatures and high levels of EMI encountered in combustion engines . These benefits combined with exceptional durability and very low cost make fiber optical sensors prime candidates for use in automotive production engines. Due to their miniature sizes, resistance to high temperatures, and immunity to EMI, these sensors can be combined with existing engine components such as ignition spark plugs, fuel injectors, or glow plugs. Such multifunctional devices with an embedded pressure sensor offer numerous advantages for practical and low-cost automotive systems not only from the point of view of sensor expense alone but also on the account of minimum total installation and operational cost. An embedded sensor does not require a separate access point into the engine and the device that the sensor is integrated with can be conventionally installed. No additional cable or connector is needed since the pressure sensor information is sent via the existing cable and connector. Connecting multiple non-embedded sensors to the engine controller represents a complex and costly task in engines with a large number of cylinders.
2. SENSOR DESCRIPTION
The low-cost fiber optic sensor developed by Optrand for combustion engine applications consists of three basic components: a sensing head directly exposed to combustion pressure, a fiber optic strand, and an opto-electronic module containing all sensor optical and electronic components, as shown in Fig. 1.
Fig. 1. Schematic sensor block diagram.
The opto-electronic module contains a photodiode, an LED, and a dedicated Application Specific Integrated Circuit (ASIC). Two input electrical pins are for power supply and ground while the output pin is for sensor output and fault diagnostics. The ASIC controls light intensity, amplifies and filters photodiode signal, and provides the auto referencing function. This Optrand patented technique regulates LED light intensity in response to any undesirable environmental conditions that may alter minimum detected light intensity. Baseline light intensity in fiber optic sensors may vary due to optical link transmission fluctuations resulting from connectorsí mechanical and thermal instabilities, fiber bending, light source or detector temperature dependence, or their aging over time. The auto-referencing approach not only corrects for offset drift but sensor gain error as well. A side benefit of the technique, not possible with other combustion pressure sensors, is the availability of sensor health monitoring output. By continuously monitoring the LED current level or its rate of change, one can identify potential sensor failure before it occurs. This ability is particularly important in control applications where sensor failure may cause malfunction or even failure of the controlled device.
In a simple and robust design the sensor head consists of a metal housing with a welded sensing diaphragm, a fiber holding ferrule, and two fibers bonded inside the ferrule, as schematically shown in Fig. 2. Please note that the dimensions shown in Fig. 2 are not to scale.
Fig 2. Sensor Head Construction
The sensing diaphragm is the most critical element of the sensor. It has to be as small as possible so that the embedded sensor occupies smallest amount of space in a device it is integrated with. At present our smallest size sensor has the diaphragm diameter of 2.8 mm. This small diameter creates a significant design challenge due to the simultaneous requirement of large diaphragm deflection (for high signal to noise ratio) and low stresses required for long lifetime. In a typical passenger car application the sensor has to function reliably over hundreds of millions of pressure cycles while in diesel engines the lifetime has to approach 1 Billion cycles. The diaphragm reflectivity must also remain nearly unchanged over the sensor lifetime. To ensure durable operation, the present sensor uses an Optrand patented sculptured, hat-shape diaphragm with varying thickness across its diameter. A high strength alloy (Inconel) has been used as a diaphragm material. This design has been selected so it can meet fatigue life of 1 Billion cycles (possibly more) and the overpressure requirement. Other benefits of the present construction include excellent linearity of the pressure response and reduced sensitivity to direct flame and hot combustion gas effects. In addition, the diaphragm is welded to sensor housing away from the flexing area, thus improving long-term weld stability and strength.
The sensor response to pressure results from the displacement of a diaphragm that in turn changes optical signal transmitted from the sending to the receiving fiber upon reflection from the diaphragm. In a two-fiber design, light intensity collected by the receiving fiber may either decrease or increase with increasing diaphragm deflection. For a given diaphragm displacement due to a full scale pressure change the sensor response can be adjusted by a suitable choice of optical fiber core diameters and numerical apertures, as well as relative position of the fibers in respect to the diaphragm.
3. EMBEDDED DESIGNS
Three types of embedded sensor systems offer the most promise for production engine automotive applications. In the first system, termed PSIcoilTM, the sensor is integrated within a spark plug/ignition coil, as schematically shown in Fig. 3.
Fig. 3 PSIcoil diagram
The PSIcoil consists of three basic components: (1) a miniature fiber optic pressure sensor with its sensing head mounted in a special spark plug, (2) an ignition coil with an embedded fiber optic strand, and (3) an opto-electronic module and a dedicated Application Specific Integrated Circuit (ASIC) mounted on the top of the ignition coil. In the production-engine version, the spark plug, the ignition coil, and the sensor are preferably integrated into a single, non-detachable device. In such a version both the spark plug and the ignition coil are designed differently from the presently used components so the pressure sensing ignition coil/spark plug device meets the automotive industry requirements of performance, long lifetime, reliability, and low cost.
Different versions of an ignition system with an embedded combustion pressure sensor can be used. In the design shown in Fig. 3 the sensor opto-electronics and ASIC are mounted on the top of an ignition coil permanently attached to spark plug. The coil low-voltage supply cable powers the sensor and the combustion pressure output is available through an additional pin of the coil-on-plug electrical connector. Alternatively, the fiber optical strand of the sensor can be broken into two sections, connected to each other via an optical connector. For ease of installation and low cost, the connector is combined with the coil electrical connector. Between the coil and Distributorless Ignition System (DIS) the fiber optic strand is embedded right inside the high voltage cable, together with a carbon strand. Such integration is only possible with a fiber optic cable, due to its non-electrical nature, so that neither the ignition field is perturbed nor the sensor cable is affected by the ignition system EMI. Such a measurement ignition system is the most desirable for applications where an ignition coil may be exposed to temperatures over 125oC, such as for example in push rod engines.
A fuel injector used in direct gasoline or diesel engines is the second device designated for embedding of a fiber-optic combustion pressure sensor. The device, termed PSIjetTM, in addition to a combustion sensor may incorporate a fuel pressure sensor Ė as schematically shown in Fig. 4.
Fig 4. PSIjet Diagram
The resulting product offers the most benefits in direct-injection engines that have been under recent intense development, for both diesel and gasoline fuel applications. Due to significant fuel economy improvements direct injection engines offer superior performance compared to those with conventional injection. Direct Diesel Injection (DDI) engines have a higher baseline thermal efficiency (about 40% peak), 20-35% better fuel efficiency, 10-20% lower CO2 emissions, near- zero evaporative emissions, and low cold-start emissions. Fuel economy improvement of as much as 35% has been recently reported, combined with a simultaneous increase in engine power and torque of 10%, for a Gasoline Direct Injection (GDI) engine. Such a remarkable performance has been realized through a combination of very lean burn combustion (Air-Fuel ratios as high as 40) and a stratified charge inside an engine cylinder.
A key component that is needed in both DDI and GDI engines is an accurate and cost-effective fuel injector. In diesel engines, a new direct injector has to operate at extremely high fuel pressures (as high as 30,000 psi or 2,000 bar), provide accurate and repeatable spray patterns, and need to be precisely timed. In addition, such an injector has to operate for as many as 0.5 Million miles and be of low-cost. In GDI applications the non-lubricating nature of gasoline and critical injector specifications make these injectors difficult and expensive to manufacture. To provide the required performance, reliability, and low-cost, in a preferred design the DDI and GDI injectors integrate two miniature fiber optic pressure sensors. One sensor controls fuel delivery to the injector while the second sensor provides combustion pressure information. Such a "smart" injector does not need to be individually balanced, as currently done, so its price can be significantly lower. Differences caused by manufacturing variability, aging, pressure line fluctuations, or fuel quality can be compensated for by using a closed-loop control of fuel timing, duration, and pressure.
In addition to control functions, the embedded combustion pressure sensor of the smart injector provides real-time information about combustion process including peak pressure, Indicated Mean Effective Pressure, start and end of combustion, and location of peak pressure. Using advanced control algorithms the air-fuel ratio during each cycle is calculated and then controlled by the engine electronic control module. In GDI engines spark timing and its duration are controlled based on the combustion pressure sensor information as well. The outcome is the simultaneous benefits of increased injector reliability, improved engine performance, reduced emissions, and improved fuel economy.
Fig.5. PSIglow Diagram
Finally, the third embedded design, integrating a fiber optic pressure sensor with a glow plug, is shown in Fig. 5. The product, called the PSIglowTM, is intended for diesel and alternative fuel engines, where a glow plug is used. The pressure sensing diaphragm is recessed from the tip of the glow plug so the maximum continuous temperature of the sensor package does not exceed 350° C. Pressure access is provided on the side of the glow plug stem in the form of a series of small holes, which provide an additional functionality of a flame arrester. The sensor interface opto-electronic and ASIC circuitry is embedded into the glow plug body outside of the high temperature zone and connects electrically to the ECU (Electronic Control Unit). Unlike the PSIjet, which is restricted to direct injection engine applications, the PSIglow can be used with both indirect and direct injection engines.
Differently from engine testing and R&D applications, a combustion pressure sensor intended for engine controls does not need to be extremely accurate. The critical features are high reliability, low cost, and long life.
Typically, no individual sensor calibration is needed in the interest to lower the total system cost. However, a high degree of linearity and low-hysteresis are required within a single combustion cycle so a normalization technique can be applied based on the ratio of cylinder pressure during compression and expansion stages of the cycle.
Basic specifications of Optrand pressure sensor system intended for production engine applications are summarized in Tab.1
0 to 200 bar
1.5x pressure range
Linearity, hysteresis, thermal shock:
0.1 Hz to 30 kHz
Minimum Sensor diameter:
Sensor housing continuous temperature:
-40 to 350 oC
Electronics and opto-electronics continuous temperature:
-40 to 135 oC
500 Million cycles
Laboratory testing of the sensor reported here includes linearity and hysteresis comparisons against commercially available piezoelectric sensors, temperature soaking and cycling, and sensitivity dependence on temperature. Fig. 6 demonstrates a comparison of the typical room temperature dynamic pressure response, over a 0 to 70bar pressure range, of Optrandís sensor and a reference water-cooled piezoelectric transducer (PZT) (Kistler Model 6121). The signal to noise ratio (SNR) is approximately 4000 at 15kHz bandwidth.
Figure 6. Dynamic pressure comparison between Optrand AutoPSI sensor and a piezoelectric transducer
Below we present the series of results obtained during short- and long-term tests with both spark plug- and head-mounted fiber optic pressure sensors. In all experiments a high quality piezoelectric transducer was used as a reference directly mounted in the engine head and water-cooled for maximum pressure reading accuracy. Fig. 7 demonstrates the performance of Optrand sensor with 3.8 mm diameter diaphragm (M5x0.5 thread) mounted directly in an engine head.
Fig 7. Comparison between M5 head-mounted sensor (3.8mm diaphragm) against a head mounted piezoelectric reference sensor: a: time-evolution, b: linearity comparison
The above data illustrates a typical performance of a head-mounted sensor. When correctly protected against thermal shock effects, Optrandís uncooled, head-mounted sensors demonstrate a total error, including thermal shock, non-linearity, and hysteresis, ranging from +/-0.4% to +/-1%.
A spark plug-mounted combustion pressure sensor is subject to potential errors compared to a sensor mounted in an engine head. First, a spark plug sensor needs to be extremely small to fit into production spark plugs as small as 10-mm thread diameter. The small sensor size means small diaphragm deflections and a relatively larger thermal shock effect due to hot combustion gasses. Second, a spark plug-mounted sensor does not benefit from the same degree of cooling available to a head-mounted device since the spark plug body temperature is considerably higher than that of a water-cooled engine head. Maximum heat dissipation from the sensor diaphragm is required to maintain lowest possible diaphragm temperature for maximum pressure reading accuracy. Third, unless the sensor diaphragm is flush-mounted in the spark plug a potential for pressure attenuation and phase delay exists. In our present approach we use a short pressure channel machined in a production spark plug to connect the inside chamber of the spark plug to the sensor diaphragm. In most cases the channel can be short so the Helmholtz and standing wave resonances are above 15 kHz targeted for the automotive sensor (required to detect engine knock frequencies). In the production version, however, a custom device is preferred with the pressure sensor flush mounted at the top of the spark plug, as shown in Fig. 3.
Performance data is presented below for two spark plug-mounted sensors, 3.8mm and 2.8 mm in diameter. Fig. 8 shows the performance of a 3.8mm diameter diaphragm mounted in a M4.5 thread sensor. Optrandís PSIplug measuring spark plug was used in the test.
Fig 8. Comparison between a spark plug-mounted 3.8mm diameter (M4.5 thread) sensor and an head-mounted piezoelectirc reference
Fig. 9 illustrates the performance of Optrand smallest to date sensor with 2.8-mm diameter diaphragm. It is to be noted that a full-scale deflection of the sensorís diaphragm is approximately half of the deflection of a 3.8-mm diaphragm, for the same pressure. The result is that the signal to noise ratio of a 2.8-mm sensor is approximately half of that of a 3.8-mm sensor, approximately 2000:1, at 15kHz bandwidth.
Fig 8. Comparison between a spark plug-mounted 2.8mm diameter (M3.5 thread) sensor and an head-mounted piezoelectirc reference
Finally, in Fig. 10 we demonstrate the results obtained on a diesel engine. Optrand sensor was mounted in the head of a single cylinder diesel engine with a reference piezoelectric transducer mounted in the same head in proximity to the optical sensor. It is to be noted that while the excellent performance of Optrand sensor was obtained with a head-installed device, a similar performance is expected for an injector-mounted product. Unlike in a spark plug, a sensor mounted inside a fuel-injector benefits substantially from the cooling effect of the fuel flowing through the injector. As a consequence, the sensor body and average diaphragm temperatures are relatively low resulting in improved performance as compared to that of a plug-mounted sensor.
One of the greatest challenges in developing a combustion pressure sensor for production engine applications is the need to satisfy the sensor lifetime requirement of 10 years or 500 Million pressure cycles. The harsh environment that the sensor is exposed to not only includes combustion temperatures but also vibration, widely fluctuating under-hood temperatures, and the effect of salt spray and engine fluids. We believe that the fiber optic construction of Optrand sensors makes them inherently the most durable. While the longest test to date has lasted for 12000 hours and 800 Million pressure cycles, the lifetime of 20,000 hours and one billion cycles should are expected.
Fig. 11 shows a comparison between Optrandís fiber optic pressure sensor and a reference piezoelectric transducer (Kistler Model 6121) collected approximately after 100 million pressure cycles in a lean-burn, natural gas, stationary engine. For clarity, pressure traces have been shifted vertically.
5. SUMMARY AND CONCLUSIONS
A low-cost and durable in-cylinder pressure sensor is the enabling element for closed-loop combustion controls for significantly more efficient and minimally polluting engines. Optrand has developed a miniature fiber-optic combustion pressure sensor that has all the characteristics required for production engine applications. When embedded into a multifunctional device such as "smart" spark plug/ignition coil, fuel injector, or glow plug, the pressure sensor can be introduced into an engine at minimum cost. In a robust design, the sensor utilizes the principle of light reflection from a flexing metal diaphragm, monitored by two optical fibers. Other sensor components include one LED source, one photodiode detector, and a dedicated ASIC. A smart auto-referencing technique compensates for sensor drift, fiber link fluctuations, temperature effects, and guarantees drift-free operation. In a thermally optimized package the spark plug sensor demonstrates combined non-linearity, hysteresis, and thermal shock errors between +/-0.4% and +/1%. With the proven lifetime of 500 Million pressure cycles and a projected price of $7 to $9, the sensor meets the performance, durability, and cost requirements for production car applications.
. J. D. Powell, "Engine Control Using Cylinder Pressure: Past, Present, and Future", ASME J. of Dynamic Systems, Measurement, and Control, Vol. 115, pp.343-350, June 1993.
. "Cylinder Balancing System for Gas Engines", Diesel & Gas Turbine Worldwide, March 1995
. N. Sugitani, and K. Tsukada, "A combustion pressure sensor for TOYOTA lean burn engine control system," SAE paper no. 930882, presented at the 1993 SAE International Congress and Exposition, Detroit, MI, March, 1993.
. A.K. Oppenheim, "Paving the Way to Controlled Combustion Engines (CCE)," SAE paper no. 951961, presented at the 1995 SAE International Congress and Exposition, Detroit, MI, March, 1993.
. T. Kume, Y. Iwamoto, K. Iida, M. Murakami, K. Akishino, and H. Ando, " SAE paper no. 960600, presented at the 1996 SAE International Congress and Exposition, Detroit, MI, March, 1996.
. A. Randolph, "Cylinder-Pressure-Transducer Mounting Techniques to Maximize Data Accuracy" SAE paper no. 900171, presented at the 1990 SAE International Congress and Exposition, Detroit, MI, February 26 - March 2, 1990.
. H.-J. Kress, J. Marek, M. Mast, O. Schatz, and J. Muchow, "Integrated Silicon Pressure Sensor for Automotive Application with Electronic Trimming" SAE paper no. 950533, presented at the 1995 SAE International Congress and Exposition, Detroit, MI, February 27 - March 2, 1995.
. R. Ziermann, J. von Berg, W. Reichert, E. Obermeier, " A High Temperature Pressure Sensor Width ß-SiC-Piezoresistors on SOI-Substrates," Transducers 97, Chicago, Illinois, USA, June 16 - 19, 1997, Digest of Technical Papers, 1411 Ė 1414.
. T. Poorman, Liangdao Xia, and M. T. Wlodarczyk "Ignition System-Embedded Fiber-Optic Combustion Pressure Sensor for Automotive Engine Control and Monitoring," SAE Paper No. 970853, 1997.