|
|
 Part 3: Design Techniques for High-Temperature Applications Sensors That Can Take the HeatAs companies see the opportunity for increased revenue arising from the development of high-temperature-tolerant electronics, more of the information about the technology will be publicized, and designers won’t have to reinvent the wheel. Jay Goetz, The previous two parts of this series discussed the availability (or lack thereof) of high-temperature-tolerant components and materials, as well as some of the packaging issues dealt with by designers. In this final part of the series, we look at design issues, particularly those that affect performance when you’re interfacing electronics with sensors in extreme and widely varying temperatures. Electromigration The second part of this series showed how electromigration is one of the leading factors that caused the failure of electronics exposed to high temperatures. Electromigration of metals wears out low-temperature electronics much more quickly than electronics designed for lower current densities. For example, you can expect a 1% failure rate for components that have operated for a year at 160ºC. The same failure rate can be expected within 21/2 months of operation at 200ºC, whereas an HTMOS component will survive more than 10 years of continuous operation at 200ºC. But with the right circuit design, you can reduce the effects of electromigration. Try one of the following strategies: • Lower the current density by reducing the output current drive. Be warned—this may not work for ICs not designed using high-temperature electromigration rules because of the fixed nature of internal currents in an IC. • Use power duty cycling. This is effective for systems that don’t require constant operation (e.g., a memory gauge, which is a downhole device that logs data over a period of time into nonvolatile memory). Because electromigration depends on current density in the on state, power cycling these devices allows them to last longer. • Reduce IC current density by using components at the lower end of their specified rail voltages.
Circuit Design Techniques Eliminating Electronics Thermal Drift. At 225ºC, thermal drift can amount to hundreds of microvolts of offset from room temperature. Manufacturers of precision op amps sometimes advertise drifts of 1 µV/ºC or better, but they’re not specified above 125ºC. There are four common approaches to removing electronics thermal drift: 1. Use a high gauge-factor sensor 2. Use chopper-stabilized amplifiers 3. Use autozeroed amplifiers 4. Compensate by inverse component TC effect Consider an example of signal conditioning an unamplified Wheatstone bridge sensor. The input offset drift of the HT1104 operational amplifier is typically 10µV/ºC, and the stability is 100 µV/year. To create an instrumentation amplifier for a millivolt pressure bridge sensor, you have to do something about the relatively large drift. An example of devices used in Approach 1 would be silicon-based piezoresistive sensors, which are high gauge-factor sensors with outputs of up to 1 V full scale. Approach 2, chopper stabilization, eliminates the offset voltage rather than compensating for it. It also eliminates the long-term (stability) drift. Chopper stabilization requires analog switches to produce the chopping action as shown in Figure 10. Finally, Approach 4 seems to be the least complicated from a hardware standpoint. It does, however, require significant testing to correctly characterize the electronics drift. Sensor Interfaces. Piezoresistive and thermal resistive sensors that are dielectrically isolated and metal thin/thick film sensors (e.g., magnetoresistive and PRT devices) are suitable for use above 200ºC. Temperature-Induced Sensor Gain Errors. Sensor outputs are usually amplified and converted to digital form using such components as instrumentation amplifiers and A/D converters, which can be made ratiometric to power supply effects. Electronic offset drift can be taken care of by chopper stabilization or autozeroing techniques. Often, however, the sensors themselves have large drift caused by temperature coefficient of gain (TCS) and temperature coefficient of offset (TCO).
To compensate for these errors, high-temperature-tolerant signal processing should include a way of scaling the sensor output by using a temperature-sensitive mechanism, such as the following: 1. Compensating the sensor TCS with a component having an equal but opposite temperature coefficient. 2. Using constant current excitation for sensors whose output scales with current and that follow the relationship temperature coefficient of bridge resistance (TCR) = –TCS. 3. Using an integral gain-set resistor (Rgainset located on the sensor die) to control the gain of an external amplifier for sensors that have TCRgainset = –TCS. 4. Measuring the temperature with a temperature sensor and digitally compensating the sensor using a microprocessor. 5. Ratiometric compensation in which a sensor gain monitor is used to correct the TCS. 6. Reducing the input range of the sensor by nulling the input signal using external feedback techniques. This technique is sometimes called closed loop feedback, but it’s not easy to do with some types of sensors. Approach 1. This is achieved by using a positive or negative temperature-sensing resistor to control the excitation voltage of the sensor. This approach is useful with bridge-type sensors whose output scales with excitation voltage. Approach 2. Some sensors have a TCR approximately equal to and opposite their TCS. Magnetoresistive sensors and some piezoresistive sensors are of this type. This kind of sensor can be compensated by using a current source controlled by the TCR of the bridge (see Figure 11)—a voltage source can be used as well.
Approach 3. The sensor is compensated by using an external amplifier with a gain-set resistor incorporated on the sensor die (see Figure 12). The gain-set resistance must have the relationship of TCR ~ –TCS and can also be doped in such a way that sensor nonlinearities are compensated. Approach 4. Measuring temperature for digital compensation of the sensor can be done with PRT thermometers, on-die semiconductor/thin film temperature sensors, or a half bridge made from sputtered CrSi film/diffused resistor elements. Approaches 5 and 6. The ratiometric or closed loop feedback methods of reducing TCS are more desirable for sensors with significant nonlinearity in temperature be cause they reduce or eliminate the effects of temperature on gain. In ratiometric compensation, the sensor gain is monitored and used ratiometrically to control the sensor excitation. This approach usually requires two sensors. Temperature-Induced Sensor Offset Errors. Thermally induced offsets (TCO) can also be a problem with sensors. The problem can be exacerbated if the offsets are not well controlled but vary randomly from sensor to sensor. Direct characterization and correction are typically used to compensate for these types of errors. In some cases, the offsets can be correlated with other phenomenon (e.g., initial offsets), but this effect must be borne out by statistical analysis of many sensors over many fabrication lots. Magnetoresistive sensors can use a trick called set/reset (see Figure 13), in which the polarity of the sensor is reversed continually, and the resulting bridge output is then synchronously demodulated to remove common mode offsets from the signal. AC-coupled, chopper-stabilized amplifiers can be used to block sensor offset errors.
High-Impedance Amplifier. Some high-temperature sensing applications require high-impedance circuits. A high-impedance operational amplifier can be used to amplify signals from piezoelectric accelerometers. In this case, low-leakage ceramic board materials and careful layout of sensitive signal lines are recommended to maintain signal integrity. Power-Circuit Design. When designing these types of circuits, you encounter unique problems at high temperatures: • Self-heating at high temperatures is intensified for MOSFETS by increasing channel resistance and by on-state voltage for bipolar transistors. • Reverse leakage current increases with temperature, dissipating more power. • Capacitors used in these circuits are the most difficult components to find. Large valued or low equivalent series resistance capacitors are especially hard to find for high temperatures.
In these environments, traditional electronic tools often can operate for only several hours, but the companies drilling the wells want to extend the limit to thousands of hours. To meet these requirements, designers have tried component screening and expensive cooling flasks. Even though screening has increased operational life at temperatures below 200°C, they are still limited to several tens of hours. For higher temperatures, flasks can provide only several hours of use before components are damaged by extreme heat. Downhole drilling operations use three main groups of tools: • Logging tools contain sensors that gather snapshot information about the quality and production capability of the well. Typically, these are either surface-powered (wireline) or self-powered steel cable connected (slickline) tools. • Measurement while drilling (MWD) tools contain sensors that measure drill orientation and material formation of nearby rock. Some companies are developing sophisticated real-time directional drilling rigs to avoid having to pull the bit up for readjustment of the drilling angle when a new direction is desired. • Permanent gauges contain sensors that measure long-term quality and productivity, which are installed in producing wells and are monitored at the surface periodically over several years’ time. Most geothermal wells don’t use MWD tools and must use flasked logging tools because of the high temperatures. In 1999, Sandia Laboratory, which has been promoting instrumented drilling and logging operations, designed and demonstrated a flaskless PT (pressure/temperature) tool, which for the first time contained electronics that operated successfully above 300°C . But you can counter design problems that occur at high temperatures by taking one or a combination of the following steps: • At elevated temperatures, derate conventional components. For example, an IRF044 60 V, 30 A MOSFET at Tj = 25ºC is derated to 0 V or 0 A at 175ºC case temperature. Similarly, a MUR5015 ultrafast diode rated at 150 V and 50 A at room temperature is derated to 0 at 175ºC. These are representative of a few power components that have been screened for operation at temperatures as high as 200ºC. Other devices operated at these temperatures include the Harris RFH75N05E MOSFET and TA9796 IGBT and the Motorola MBR3535 and 1N3891 diodes. • Make power control chips from SOI-fabricated gate arrays rated at 225ºC. Conventional chips have been tested at 200ºC and have failed. • Use devices from the high-temperature, linear regulator family HTLREG, which is a collection of SOI-fabricated ICs from Honeywell. These 0.3 A devices come in ±15, 10, and 5 VDC versions. • Use one of a few discrete transistors, which can operate at temperatures as high as 200ºC. • Use SOI-based, N-channel MOSFET 1.0 A devices, which are available from Honeywell. These have breakdown voltages in the 70 V range and R-ON of about 1 ž at 250ºC, with junction temperature allowed to go as high as 300ºC. • Incorporate NPO dielectric ceramic caps, which work well at high temperatures. X7R dielectric caps also work, but they have a large temperature coefficient of value. Recent research efforts have resulted in better high-temperature capacitor dielectrics. • Use SiC diodes (which are available from a few suppliers) alongside derated low-temperature parts to reduce losses and heating as a result of switch-mode operation. Soft-switching techniques (e.g., zero voltage or zero current switched PWM) can be used to reduce switching losses and to avoid the stress (and thus temperature rise) of both high current and high voltage at the power semiconductors (see Figure 14).
Communications Circuits. High-temperature wired communications protocols are often proprietary, sometimes with bidirectional signals and power on a single line to reduce wire penetration into high-temperature environments. Standard 4–20 mA interfaces are sometimes used because of their noise immunity and long transmission lengths, and if the instrument can be powered on <3–4 mA, it doesn’t require additional power lines. For higher data rates over shorter distances, some designers have adopted military or automotive buses (e.g., the Mil-1553 synchronous communications bus or the CAN bus). Manufacturers have made high-temperature protocol controllers for these buses using HTMOS gate arrays and microcontrollers fabricated in SOI.
In 1999, the Defense Advanced Projects agency, under the management of the Air Force, completed a program called High Temperature Distributed Control Systems (HiTeC) to demonstrate high-temperature-tolerant electronics in turbine engine and industrial applications. Several leading aerospace, jet engine, controls, and electronics companies participated in the The full authority digital electronic control is freed from tedious loop closure and signal conditioning tasks, which it performed in the past. Although the engine case temperature approaches 560°C at high speeds and high altitudes, there are still sites inside the engine where electronic controls can operate efficiently at 300°C, the upper limit for specialized high-temperature-tolerant silicon electronics. Conventional uncooled electronics cannot be used here because of severe component derating above 120°C and shortened operating life as a result of temperature-induced wearout. Electronics located in the most hostile environments must use a wider bandgap semiconductor (e.g., SiC) when it is mature enough. Optical buses are being evaluated for high-temperature measurements and have good noise immunity. However, more re search is needed to find fiber that can withstand the high temperatures and corrosive environments found in some applications.
A low-leakage charge amplifier can be implemented as a single-chip HTMOS ASIC design (see Figure 15). This device is used to interface data acquisition systems with remote high-impedance charge-generating accelerometers. Multichip Modules, Custom Gate Arrays, and ASICs. Downhole measurements require reliable, rugged data acquisition systems. The equipment’s mission life may only be hours or days, but the price of failure can far exceed the cost of the instrumentation. Multichip modules (MCMs) containing highly integrated circuits reduce interconnections and improve system reliability (see Figure 16). Custom gate arrays and ASICs can solve many packaging and interconnection problems (which become reliability problems) and reduce power consumption and size. For example, a customized mixed-mode ASIC can combine on one chip all the functionality of the control system shown in Figure 16. The tradeoff is in the nonrecurring engineering effort required to develop such parts (see Table 8) Summary In this final part of the series, I’ve discussed many of the challenges associated with the design of high-temperature-tolerant circuits and systems, and I have proposed solutions and techniques that can be used to overcome them. Because of limited product development, the design of high-temperature-tolerant systems is still not well understood, and useful design information is sparse. For now, most design knowledge exists in the research labs of larger companies and universities, with few external consultants available to help. As more companies move to meet the demand for high-temperature-tolerant products, more information will be available on the design process. Then designers won’t have to reinvent the wheel, as they often do now.
For Further Reading Brusius, P. 1998. “Some Reliability Aspects of High-Temperature IC’s,” Proc 4th Annual High Temperature Electronics Conference, Albuquerque NM. Caruso, M. May/October 1998. “A New Perspective on Magnetic Field Sensing,” Proc Sensors Expo. Day, J., and M. Roach, 1998. “Ceramic Dielectric Performance under High-Temperature Life Test,” Proc 4th Annual High-Temperature Electronics Confer ence, Albuquerque NM. Final Report, HiTeC Program Monitored by U.S. Air Force, Wright Laboratory, Conducted by the HiTeC Consortium (AlliedSignal, Boeing, Honeywell, Moog, Ormet, Parker, Rockwell, United Technologies, and The University of Maryland). Led by United Technologies Research Center, August 1999. Goetz, J., and H. Middleton. May 1999. “Designing Sensor-Based Systems for High-Temperature Environments,” Proc Sensors Expo. Grzybowski, R. 1998. “Long-Term Behavior of Passive Components for High-Temperature Applications—An Update,” Proc 4th Annual High Temperature Electronics Conference, Albuquerque NM 1998. Grzbowski, R., and B. Gingerich. June 1998. “High-Temperature Integrated Circuits and Passive Components for Commercial and Military Applications,” Proc ASME Turbo Expo, Stockholm Sweden. High-Temperature Electronics, Ed. F. McCluskey, R. Grzybowski, and T. Podlesak, CRC Press, 1997. HITEN Regional Seminar—Europe. High-Temperature Electronics. December 16, 1998. HITEN Report-1997, www.hiten.com, London England. Lewis, T. 1998. “Military Aircraft Turbine Engine Electronics and Requirements,” Proc 4th Annual High Temperature Electronics Conference, Albuquerque NM. Naefe, J., W. Johnson, and R. Grzybowski. 1998. “High-Temperature Storage and Thermal Cycling Studies of Heraeus-Cermalloy Thick Film and Dale Power Wirewound Resistors,” Proc 4th Annual High Temperature Electronics Confer ence, Albuquerque NM. Naefe, J., W. Johnson, and R. Grzybowski. 1998. “High-Temperature Storage and Thermal Shock Studies of Passive Component Attach Materials,” Proc 4th Annual High-Temperature Electronics Conference, Albuquerque NM. Normann, R., and B. Livesay. 1998. “Geothermal High Temperature Instru mentation Applications,” Proc 4th Annual High Temperature Electronics Confer ence, Albuquerque NM. Acknowledgments The author wishes to acknowledge the support he received from several individuals with Honeywell, including Phil Brusius, David Wick, Bruce Ohme, Dick Spiel berger, Ben Gingerich, and Connie Krauth, as well as from Randy Norman with Sandia Laboratory, Rich Grzybowski with UTRC, and Tim Lewis with the U.S. Air Force. Editor’s Note The figures, equations, and tables in Part 3 are numbered consecutively from those in Part 1. The information in this series of articles will be presented at Sensors Expo Detroit, September 19–21, 2000. Jay Goetz is a Consulting Engineer. He can be reached at sj.goetz@worldnet.att.net. For inquiries on information or products mentioned in this article, contact Honeywell at 800-323-8295 or visit its Web site at www.ssec.honeywell.com. |
||||||||||||||||||||||||||||||||||||||||||||||||
|
|
joint venture. On June 25, in the world’s first demonstration of distributed control using high-temperature-tolerant electronics, a smart actuator controlled the position of the first stage variable stator vanes of a military test engine. To accomplish this, ICs and smart actuator electronic modules were developed by Honeywell, AlliedSignal (both now Honeywell International), Boeing, and United Technologies. The technical approach involved the use of electronic control systems to replace existing mechanical and hydraulic systems.
