April 2004


SENSOR TECHNOLOGY AND DESIGN
Table of Contents

A Single-Chip Pressure Sensor
for the Automotive Market

Combining silicon micromachining designs and processes with advanced mixed-signal CMOS circuitry overcomes many of the early limitations of single-chip sensors. The result is a high-performance single-chip piezoresistive pressure sensor with an amplified and calibrated output.

figure

The automotive industry has identified a need for small, robust, and accurate sensor systems to monitor manifold absolute pressure and tire pressure. The high-performance single-chip sensor described here integrates the sensing element, signal processing, and packaging to produce a device with amplified and calibrated output. Ease of automated manufacture keeps the price within limits that have been defined by automakers.

Background
Piezoresistive pressure sensors need signal-conditioning circuitry for interchangeability and for compatibility with most electronic control systems. Because each sensor element has its own “personality,” and is affected by both pressure and temperature, signal conditioning typically requires a number of correction coefficients. These include calibration for offset, full-scale variations to make the units electrically interchangeable, and provision for temperature compensation for offset and full scale. Additionally, linearity correction, diagnostics, and filtering are sometimes required.

Conventional methods of marrying signal-conditioning electronics with sensing elements have resulted in technical and commercial tradeoffs between product cost and performance. Monolithic approaches have reduced both cost and size, but the fabrication prücess places restrictions on the circuit and sensor design that compromise performance. Hybrid techniques use either a dedicated ASIC for
figure
Figure 1. The single-chip co-integrated piezoresistive pressure sensor contains onchip CMOS circuitry for signal conditioning, EEPROM for storage of calibration coefficients, and a MEMS sensor element.
signal conditioning or a discrete circuit approach. Although the benefits include a performance better than that of monolithic devices as well as the flexibility to adapt new designs with simple component changes, these hybrids generally are larger, cost more, and require additional assembly steps that raise concerns about reliability. Two-chip solutions may also laxk the third-order temperature compensation and linearity correction increasingly required by automotive manufacturers.

A Co-Integrated Technique
 Fabrication. The co-integrated pressure sensor (see Figure 1) was a collaborative effort among MEMS sensor designers and mixed-signal IC designers. Based on a standard 0.65 mB mixed-signal CMOS process (see Figure 2), the chip size is 10 mm2. The circuitry performs calibration for offset and sensitivity, in addition to multi-order temperature compensation and error correction, and stores the calibration coefficients in onchip EEPROM.

figure
Figure 2. A SEM cutaway view of the sensor shows the CMOS circuitry and etched pressure-sensing silicon diaphragm.

The manufacturing process flow has been designed to allow all standard CMOS steps to occur at the front end of fabrication. The micromachining steps—silicon etch that forms the pressure-sensitive diaphragm, anodic bonding of a glass substrate for absolutý pressure configurations, probing, wafer dicing, and inspection—are performed after the circuitry is completed. Readily available, standard P-type implanted sensing elements eliminate the need to vary the process to incorporate a special extra implant. The resulting dies (see Figure 3) are then either shipped to die customers or assembled into a variety of packages.

figure
Figure 3. In this photomicrograph, the sensor area is the square in the center of the chip. Onchip CMOS circuitry performs signal conditioning, calibration, and temperature compensation. The calibration and compensation values are stored in EEPROM.

Packaging is completed by overmolding in a 16-pin SOIC cavity package. One approach to packaging uses a patented injection molding
figure
Figure 4. A patented injection-molding technique exposes the diaphragm area of the sensor chip. Stresses from die attachment wire bonding and molding are compensated by programming the trim parameters in the final package.
technique that exposes the diaphragm area of the sensor chip (see Figure 4). Any stresses from die attachment, wire bonding, and molding are compensated by programming the trim parameters in the final package.

The result is a high-performance part that serves as a foundation for other configurations, including gel-filling of the cavity. Various ports may also be welded onto the package, and other packaging techniques are also possible.

 Electronics. The circuitry (see Figure 5) includes all the functions for signal conditioning and calibration, such as preamplification, offset correction, span calibration, multi-order temperature compensation, and multi-order nonlinearity correction
Click for larger image
Click for larger image
Figure 5. The sensor circuitry includes all the functions for signal conditioning and calibration.
for pressure and for the temperature coefficients. Incorporating a DSP-based correction engine provides much better accuracy than is possible with a 2-chip unit (see Figure 6). For example, many 2-chip solutions have good accuracy over ranges near room temperature, but then exhibit increasing error as temperature approaches extremes, due to either imperfections in the electronics or higher-order nonlinearities in the sensing element at the extremes. With the co-integrated sensor, the error curves can remain the same over the entire temperature range.

The system uses an 11-bit A/D converter with 8× oversampling for an effective 14-bit data conversion resolution. Data are then processed through an onboard DSP where they are corrected acccording to calibration coefficients stored in the onboard EEPROM. The corrected signal is then fed to a 12-bit D/A converter that drives the output amplifier. The amplifier has been designed to drive >2 nF, as needed for EMF suppression in the automotive environment.

Figure 6.
Comparison of Various Signal Conditioning and Error Correction Approaches
Parameter Two-Chip Thick-Film Laser Two-Chip Thin-Film Laser One-Chip Thick-Film Laser One-Chip Analog Correction with EEPROM One-Chip Digital Correction with EEPROM
Circuit complexity Low Low to moderate Moderate to high High High
Process complexity Low Moderate Moderate to high Moderate to high Moderate to high
Component costs Highest High Low Low Low
Test complexity High High High Low Low
Test cost Very high High High Low Low
Package size Large Good Small Small Small
Package cost Very high High Low Low Low
Pressure range adaptability Good Good Moderate Moderate Moderate
Trim after gel and/or packaging No No No Yes Yes
Ability to do multi-order correction No No No Yes, with difficulty Yes
Accuracy Low High Moderate Moderate Very high
Resolution High High High High Moderately high

Calibration
Candidate technologies that can be used to store the calibration and compensation values for each sensor include:

  • Thick- or thin-film resistors that are laser trimmed after device testing to achieve temperature compensation and calibration
  • Electronic trimming, in which the compensation and calibration values are electronically programmed into the chip, either with a one-time programmable technology such as fuses, Zener diodes, or onchip trimmable resistors, or a reprogrammable technology such as EEPROM

With EEPROM and other reprogrammable technologies, coefficients can be programmed into the device multiple times for real-time programming during manufacture or after assembly based on the test data for that unit. Further, the parts can be programmed after encapsulation, in contrast to the laser-trim technologies.

For calibration, the sensor is measured at multiple temperatures and pressures. Uncorrected data from the A/D converter are read out through a digital I/O for each data point. The external calibration computer then determines the minimum error and loads the order of the pressure and temperature corrections and their coefficients into EEPROM. The calibration can then be verified to determine accuracy as desired (see Figure 7 and Figures 8 and 9).

figure
Figure 7. These data show the signal processor’s ability to compensate pressure nonlinearity over a temperature range of –20ºC to 110ºC.

figure
Figure 8. The signal processor can also compensate to better than 0.2% for a temperature coefficient of zero error, even when the sensor exhibits a second-order nonlinearity in the temperature coefficient of zero.

figure
Figure 9. As can be seen in a plot of the corrected temperature coefficient of span, the –20% change with increasing temperature of the uncompensated part gets corrected to close to zero. Total error over the compensated range for the system is <0.25%, including initial zero and full-scale gain-set errors.

The algorithm can be adapted to easily correct pressure nonlinearities of various orders, and can facilitate corrections necessitated by temperature-dependent pressure nonlinearities resulting from gelling or other media interfaces.

After programming, the EEPROM is electronically locked as a completed module or further trimmed after system assembly. Because EEPROM allows the trim parameters to be changed after initial programming, the trim values can be optimized during test or trimmed again after assembly of the sensor package into the system (for example, for serialization information or fine offset correction).

Summary
A new single-chip pressure sensor, fabricated by means of a process flow compatible with standard 0.65 mm CMOS and MEMS technology, combines all the signal conditioning and pressure-sensing functions on the same chip, including the temperature and linearity correction required by the auto industry. This co-integrated unit offers improved performance while ýeducing size and cost compared to hybrid configurations. Onchip EEPROM stores the calibration and compensation values, and test results show a dramatic improvement in device perfom.-ance over the uncompensated result. Co-integration technology can be extended to other devices as well, including accelerometers, gyros, and other sensor types.

Acknowledgments
This sensor system was developed as a collaborative project by the Design and Processing Groups at Silicon Microstructures, Inc., and Elmos AG. The authors wish to acknowledge the contribution of each member of the team that made this device possible, and the work of Andreas Nebeling, Ignas van Dommelen, and Jurgen Raben of Eurasem, Nimegen, The Netherlands, on the overmolded package approach.



 
E-NEWSLETTERS
SUBSCRIBE NOW!
Sensors Weekly
  What's New
  Product Picks





Questex Media
Home | Contact Us | Advertise
© 2009 Questex Media Group, Inc.. All rights reserved.
Reproduction in whole or in part is prohibited.
Please send any technical comments or questions to our webmaster.