Water-Level Measurement Design Trade-Offs using MEMS Pressure Sensors Michael Nelson & Michelle Clifford

Transcription

1 Water-Level Measurement Design Trade-Offs using MEMS Pressure Sensors BY Michael Nelson & Michelle Clifford Freescale Semiconductor, Inc., All rights reserved.

2 MEMS pressure sensors are a superior alternative to electro-mechanical switches in water-level measurement systems, such as washing machines. This paper will briefly compare the reliability, resolution, and accuracy benefits of both solutions, and will discuss, in detail, all the different steps needed to complete an industrial-grade waterlevel system design. We will go into an in-depth analysis of all the major trade-offs related to using MEMS devices. Particular attention will be paid to the inaccuracies due to temperature coefficients of offset and sensitivity, to temperature hysteresis effects, and to overall ageing and drift. Software accuracy-improvement techniques such as one-point autozero, and two-point calibrations will be reviewed, based on a variety of different application conditions. A water-level measurement system designer will find answers to all the questions that are not typically answered in device specifications and data sheets. Introduction MEMS pressure sensors provide greater reliability, resolution and accuracy for many industrial, automotive and consumer applications. During the past 20 years they have gone through extensive improvement providing on-chip compensation and amplification, improved packaging, and increased performance. Now, with a cost that is comparable to a full mechanical switch solution along with a comprehensive understanding of the tradeoffs, MEMS pressure sensors can provide a solution for water level sensing for the appliance applications. Selecting a pressure sensor is achieved with a first level of design decisions such as integration, pressure range, sensitivity and accuracy. The second level design decisions are the types of software compensation for the device variation of offset and sensitivity, temperature hysteresis effects, and device aging and drift. The MPXV4006G was selected as solution for measuring a water level, so that the device inaccuracies and software accuracy improvement techniques can be discussed. Description of the IPS Device The MPXV4006G series of piezoresistive transducers utilize an implanted strain gauge with advanced micromachining techniques for highly sensitive pressure monitoring of water levels from 0 to 600 mmh20. The inclusion of thin-film and bipolar processing provide an accurate, high level analog output signal that is proportional to the applied pressure. This integration provides on-chip temperature compensation and calibration for improved accuracy and temperature stability. Further application enhancements can be obtained by interfacing the unit with microcontroller processing for even higher accuracy resolution at low pressures. A typical gauge configured pressure sensor is shown in Figure 1. The die is bonded to the package using an RTV donut to allow pressure transfer to the chip cavity.

3 A flourosilicone gel protects the topside die surface and wire bonds from the environment. The gel allows the pressure to be transmitted to the Silicon diaphragm. Figure 1 Cross-sectional representation of the MPXV4006G pressure sensor. The MPXV4006G operational transfer function is presented in Figure 2. This shows the linear relationship between applied pressure and the resulting Vout. The accuracy or error budget of the device associated with temperature affects is comprised of the following components. - Temperature Hysteresis: Output deviation at any temperature within the operating temperature range, after the temperature is cycled to and from the minimum or maximum operating points, with zero differential pressure applied. - TcSpan; Output deviation over the temperature range of 10 to 60C, relative to 25C. - TcOffset: Output deviation with minimum rated pressure applied, over the temperature range of 10 to 60C, relative to 25C. The temperature affects are determined by several material parameters. Among these are the doping concentration of the transducer, Thin film stresses that are temperature dependant and transferred to the membrane, and the stress associated with the flourosilicon gel die coat as it ages.

4 Figure 2 MPXV4006G series transfer function showing the output voltage as a function of the differential pressure. Improved Water-Level Monitoring Application A pressure sensor water level measurement system can provide the functionality of a mechanical switch for water level detection but also offer additional functionality beyond recognizing logic on or off signals between discrete trip points. A pressure sensor can provide both static and dynamic measurements. The static pressure measurement can replace the functionality of a mechanical switch where logic on or off signals are sent to a controller to turn a pump on when the tub level reaches empty and shuts the pump off when the tub is full. Since the pressure sensor can continuously monitor the pressure changes, multiple discrete trip points can be programmed into the system using only one pressure sensor. Due to the continuous monitoring capability of the pressure sensor, dynamic pressure measurements can be performed, enabling both leak detection and tube blockage detection as additional features. A microcontroller samples the pressure at a known sample rate calculating the pressure change to determine the rate of the tub filling or emptying. In addition, the pressure sensor system requires few components, reducing the overall system cost. The pressure sensor only requires a low cost microcontroller with an analog to digital converter. Depending on the resolution required a low cost 8-bit or 10- bit microcontroller, e.g. a microcontroller from Freescale s Nitron family, can be selected. Experimental Procedure Low Pressure test analysis was conducted using Freescale Semiconductors PSEC (Pressure Sensor Electrical Characterization) tester. A total of 84 units were tested over the temperature range of 0 to 60 degrees C. The units were segregated into 12 unit runs to minimize potential test induced variation in the low pressure regime. The data was then compared with the Ideal Transfer Function Vout = Vs*[(0.1533*P) ]. Figure 3 shows the MPXV4006 deviation from the ideality plot. It should be noted that the negative slope seen on these samples are not representative. This data shows the accuracy that one would expect without Auto Zero calibration of the units.

5 Figure 3 MPXV4006 delta ideality plot showing accuracy of the device. Use of an Auto Zero can greatly improve the accuracy of the MPXV4006 at temperatures around the Auto Zero point. Freescale Semiconductor App. Note AN1636 provides greater detail on the implementation of Auto Zero for Integrated Pressure Sensors. The basis behind Auto Zero is to sample the pressure at zero reference pressure (atmosphere) and subtract this error from the sensor output voltage. As the pressure sensors point of operation deviates from the auto zero point, increased variation will be seen. Figure 4 presents this relationship for MPX4006 devices that received an Auto degrees C, 0mm(H20). Figure 4 MPXV4006 devices with Auto degrees C, 0mm(H20)

6 The ability to conduct a two point calibration in addition to temperature monitoring further improves the accuracy of the pressure sensor. Figure 5 presents the MPXV4006 delta ideality curve with calibration points at 25 and 60 degrees C, and with temperature feedback to a microcontroller. Figure 5 MPXV4006 devices with 2 point calibration and temperature feedback. Resolution and Accuracy Discussion The resolution of the system depends on the pressure sensor and the ADC, however the accuracy depends on the software compensation. The MPXV4006G has a sensitivity of 766mV/kPa, as specified in the datasheet. This corresponds to a sensitivity of 7.512mV/mm H mv/kpa x [1kPa / mmH20] = 7.512mV/mm H20 If the ADC is 10-bits, then a resolution of 0.65 mm/bit can be achieved since each step will recognize a 4.88mV change. However, if the ADC is 8-bits than each step change will correspond to 19.53mV/bit, so the resulting resolution would be 2.6 mmh20/bit. 8 bit: mv/bit x [7.512mV/mmH20] = 0.65 mm H20 / bit 10 bit: 4.88 mv/bit x [7.512mV/mmH20] = 2.6 mm H20 / bit 12 bit: 5V/4096bits x [7.512mV/mmH20] = mm H20 / bit

7 A low resolution can be achieved by selecting the appropriate devices, but software compensation enables the required accuracy to be achieved. Inaccuracies from device offset variation and TC offset can provide water height errors. However, with a one point calibration, most of the offset variation can be removed, as seen in Figure. A two point calibration at the minimum and maximum pressure can eliminate sensitivity, offset, and span errors. In order to remove the TC offset and TC span errors, a two point calibration needs to be performed at the operating temperature range. These values can then be stored in the flash of the microcontroller. Using temperature feedback, the microcontroller can compensate for the changes in TC and TC span during operation substituting the values stored in the flash into the transfer function of the device. Offset calibration should be performed at 0mm H20, but also when the temperature of the device is near the temperature it will be at during operation when the electronics are hot. Before the calibration values are stored, a minimum amount of error checking should be performed. The offset variation and TC offset should be used to determine if the sampled offset fit in those parameters to ensure that an accurate sample has been made. Media compatibility discussion The media of a water-level monitoring system, such as a washing machine is humid from evaporated water and detergent. Since the media recommended for pressure sensors is dry air, extensive testing was performed to ensure that there were no adverse effects on the sensor performance or long-term reliability. A bath consisted of tap water (ph = 6.86) mixed with Laundri Destainer bleach (8.4% sodium hypochlorite) at a concentration of 0.96 oz. per gallon of tap water was used. To accommodate with the media requirements, a sensing tubes were placed into the water with the distance varying between 3-9 inches for the low testing, and inches for the high distance testing. The other side of the tube, while providing a distance barrier from the media, was connected to a pressure sensor which monitored the pressure in the tube corresponding the change in the tub level. The results of the media testing found that all devices were within the specification after the exposure. As seen in Figure 6, the devices at the high distance location showed less of an offset shift, and lower standard deviation than the devices at the lower location. It was also found that by increasing the distance between the device and the chemical improves the reliability of the device.

8 Figure 6 Boxplot of the offsets of pressure sensors in a controlled environment, a high location (18-24in.) from the tub, and a low location (3-9in.) from the tub. Summary MEMS pressure sensors can replace the functionality of electro-mechanical switches in water-level measurement systems while providing additional functionality. Using calibration with a minimum amount of software, the required reliability, resolution and accuracy for an industrial-grade water-level system design can be achieved. A one point calibration at zero pressure, can removed offset variation. A two point calibration a two known pressures can remove sensitivity, aging and drift variation. Adding a calibration reading taken at system operation temperatures can provide coefficients of TC span for additional software compensate during operation.