June 2008

Special Section: Temperature/Pressure

Tiny transducer meets big challenge

High accuracy, miniature pressure transducer proves industry mettle

Fast Forward

  • Industry still demanding accurate, easy-to-install, low-cost measurement system.
  • Researchers reach high transducer accuracy using piezoresistive pressure sensor.
  • Test equipment results show error of measurement at 0.01%.
 
By Steve Carter, Anthony Kurtz, Alexander Ned, and Joseph VanDeWeert

For the past 20 years, industry has used base and cavity pressure measurements for wind tunnel test articles, flight tests, and other critical test areas using electrically scanned pressure modules (ESP), which entails a high cost of the measurement system and associated ESP operation, calibration, and repair. In response, the instrumentation community has demanded a measurement system that is accurate, easy to install, and relatively low cost. One miniature absolute pressure transducer can meet this instrumentation challenge, and test results show benefits in measurement accuracy and cost savings.

ESP modules are relatively complex measurement systems, having a water-cooled enclosure, cooling water lines, pneumatic calibration and control tubes, thermocouples, and heaters. Each ESP module must be temperature controlled in order to function to product performance specifications. The high accuracy transducer design uses MEMS technologies for pressure sensor chip fabrication and packaging and is suitable for applications where size, accuracy, and environmental conditions are critical.

We achieved high transducer accuracy by using a sensitive piezoresistive pressure sensor in conjunction with high-resolution digital electronics for sensor output amplification and correction. We minimized transducer size for reducing transducer impact for ease of installation. The pressure sensor chip has a silicon-on-insulator structure with piezoresistors dielectrically isolated from the deflecting diaphragm. We determined optimal pressure sensor geometry for low pressure (5-500 PSI) measurements through finite element modeling.

Corrections in the sensor output for temperature variations used commercially available digital integrated circuit and a novel algorithm for computing temperature compensation coefficients. The optimized pressure transducer has an accuracy of 0.1% of full-scale output at temperatures up to 250°F. The transducer has a volume of less than 0.2 cubic inches and a bandwidth of over 10 kHz. Independent testing has confirmed the transducer's high accuracy over a wide temperature range (-65°F - 250°F). An upgraded transducer version suitable for 500°F/750°F operability is now under evaluation.

The technology originally saw use in wind-tunnel applications, where typical flight performance studies using hypervelocity wind-tunnel test articles involve the measurement of very low pressures at elevated temperatures. Existing technology allowed for the measurement of these pressures, but they require sacrifices in order to achieve a reasonable level of accuracy. Normal pressure installations require extensive preparation, lengthy tunnel installation, in-situ pressure calibrations, and continuous on-line system verification. As the industry's need for a solution to rapidly install, measure, and transmit flight performance data grows, this solution should connect to existing analog data acquisition systems and accurately measure pressure in a hypersonic test environment (range 0 to 5 psia, temperature 50 to 250°F) in a small-form factor.

temperature3

Sensor design, assembly

The pressure sensor used in this work is of piezoresistive type and is fabricated using silicon-on-insulator technology. The piezoresistors are dielectrically isolated from the deflecting diaphragm by a SiO2 layer. Sensor packaging uses a leadless technology in which high temperature glass-metal frits replace traditional wire-bonded chip-package interconnects. The glass contact component has through holes aligned with the metal contacts of the sensor chip; this is for making electrical connections using the metal-glass frits.

In a leadless sensor packaging technique, the chip mounts onto a header using a high-temperature, non-conductive glass, designed to fire at the same temperature as the glass/metal frit. Once the chip mounts onto the header, only the back-side of the sensor chip is exposed to the pressure media, while the sensing patterns are hermetically protected. This leadless method allows the tip of the transducer to operate at temperatures in excess of 1100°F.

The assembled transducer has a volume of about 0.1 cubic inches and is suitable for rapid installation. We used finite element analysis and material selection to develop and optimize the high-accuracy piezoresistive leadless sensing element. Finite element analysis modeling helped to better understand the existing designs and to fine-tune the new designs to achieve better performance under all environmental conditions. The new sensor designs help improve performance by enabling larger, more linear, and more stable output characteristics for a specific sensor diaphragm thickness. Increasing the sensor's output also created an opportunity to increase the diaphragm thickness for the respective design (while obtaining the same outputs), thus leading to improvements in the overall stability and repeatability of the sensors. The transducer is capable of operating up to 750°F, while being compensated between 50°F to 500°F.

Signal conditioning, electronics design

In order to better interface with standard analog data acquisition systems, we added an amplifier to the piezoresistive sensor to provide output of 0.5 - 4.5 V DC. The amplifier was based on a commercial digital correction integrated circuit with additional electronics for enabling input voltages of 8 - 32 V DC as well as for improved ESD protection.

Digital correction fixes temperature errors from the sensing chip and the amplifier. The signal chain in the amplifier is entirely analog with digital changes to the gain and offset at different temperatures. This allows for a high bandwidth (larger than 10 kHz) signal with no aliasing or other sampling effects. We measure temperatures using the resistance of the Wheatstone bridge itself, which eliminates errors caused by distance between a temperature sensor and the sensing chip.

A new third-order algorithm corrects offset and gain errors. We can measure the temperature coefficient, gain, and offset at five or more temperatures and then curve-fit this data into a third-order correction scheme. We then split this polynomial into 17 evenly spaced piecewise linear regions. Over each of these regions, we linearly adjust the gain and offset with temperature.

We tested the am¬plifier over repeated tempera¬ture cycles between -65°F to 302°F, and they proved to be highly stable (better than 0.01% repeatability).

Transducer testing, benefits

We tested the fully packaged transducers containing the optimized sensing element and the latest high-accuracy electronics during temperature cycling over the temperature range of 50°F to 250°F. One unit was within 0.05% over the entire temperature span. The worst single deviation was 0.09%. The error of measurement with the test equipment at that time was 0.01%. Further tests showed operability from -65°F to 275°F with survivability up to 302°F. Independent transducer testing at Arnold Engineering Development Center (AEDC) and follow-up calibrations at the AEDC Precision Measurement Equipment Lab (PMEL) yielded statistically similar results.

We determined the response time of the transducer using a shock tube. By applying a step-pressure function to the front of the device, it is possible to determine the rise time of the transducer and allows for the computation of the bandwidth.

We also successfully demonstrated the manufactured 5 psia transducer in hypersonic wind tunnels at AEDC. Transfer of calibration coefficients from the AEDC PMEL enabled the new transducer to achieve specified performance without in-situ calibration. Highly reliable measurements did not require periodic tares or electronic adjustments.

Under normal circumstances, you must remove the test article from the wind tunnel after 10 to 20 minutes of testing at Mach 8 to cool the installed ESPs. Since most configurations are about 30 minutes long, data productivity suffers. The new transducers improve productivity because they eliminate the ESP cooling time, operating to temperatures of 250°F with minimal accuracy degradation, but you must control the ESP temperature well within ±5°F to perform near the quoted accuracy.

Eliminating 10 minutes of non-productive ESP cooling time per configuration increases the test pace substantially. Total savings to the test user when using the new transducers on a typical Mach 8 test are nearly $25,000. During a second test entry, new transducers made all base/cavity pressures measurements.

A cooling water line failed during the initial test configuration run, and the temperature of the miniature absolute pressure transducers reached 260°F before we could retract the test article. The temperature of the transducers stabilized at 220°F once we retracted the test article. This temperature would have destroyed an installed ESP (175°F maximum). Unlike ESPs, the miniature absolute pressure transducers are not easily damaged by water and suffered no degradation of measurement accuracy during this event. The cost to replace an ESP is about $8,000. The total AEDC cost avoidance for this single event is estimated at $16,000.

ABOUT THE AUTHORS

Anthony D. Kurtz is chief executive, chairman, and chief scientist at Kulite Semiconductor Products in Leonia, N.J. Contact him at drkurtz@kulite.com. Alexander Ned is vice president of sensor operations at Kulite. Contact him at alexn@kulite.com. Joseph VanDeWeert is technical director of miniature transducers at Kulite. Contact him at joev@kulite.com. Steve Carter is test and measurement IT system architect at Aerospace Testing Alliance.  Contact him at steve.carter@arnold.af.mil.  

RESOURCES