Companies like Motorola are paving the way for true system-on-a-chip (SOC). These new SoCs can not only analyze data, but also measure, analyze and respond to the environment.
Power supply and analog components have been integrated with CMOS microcontroller unit (MCU) for several years. Products have been introduced, such as integrated 68HC05 motor controller and integrated power device with H-b ridge configuration (1990). In 1993, a product called system chip MCU was introduced, which provided the J 1850 interface of the Society of Automotive Engineers, including the physical layer. Based on the combination of power and analog capability of MCU, the chip can withstand 40 V. However, the system input was not included in the previous single-chip design.
What is the latest development that is expected to truly realize the system on chip? It is the ability to combine CMOS and MEMS structures into a process flow. The photo 1 shows a 68HC05 microcontroller, which integrates a 100 kPa pressure sensor on a sil icon chip. One possible application is the side airbag sensor.
The pressure sensor in the automobile door panel can detect the pressure change when the door panel is wrinkled by impact. The ability to program the on-chip microcontroller will enable the automobile manufacturer to embed the control algorithm into the chip. In order to complete the whole system, only one mechanism is needed to start the airbag. This driving ability may be another step in the continuous integration of silicon and electronic/electromechanical systems. This platform provides the first step in the integration of electronics and electromechanical structures, and at the same time, it puts forward several problems that must be solved before low-cost and high-quality products can be mass-produced. One of the problems is testability.
Typical logic circuits have accumulated test data for many years, which can be used as the basis for building the next generation of products. However, for sensors, the previous technology can rarely be reused. The reason is the relative naivety of sensor technology and the uniqueness of each type of sensor. For example, the technology used to measure pressure (thin film sheet with integrated strain gauge) is quite different from the technology used to measure acceleration (forming the detection quality of moving capacitor). Testing techniques are also different. Pressure measurement needs to connect the pressure source to the sensor; Acceleration or vibration detection requires shaking the equipment at some known frequency and strength.
system configuration
In order to develop a proof-of-concept vehicle (see figure 1), the 100 kPa pressure sensor and related analog circuits [1] are integrated into Motorola's standard 8-bit 68HC05 microcontroller core. An analog circuit for signal conditioning, a voltage and current regulator, and a 10 bit analog-to-digital converter and an 8-bit digital-to-analog converter are added to this basic core. For compensation purposes, temperature sensors are also incorporated into the design.
Pressure sensor is related to temperature and has inherent nonlinearity. In order to improve the accuracy of the system, the calibration or conditioning algorithm must be programmed into the microcontroller.
The output of the pressure sensor is regulated by a variable gain and input offset amplifier, which is controlled by a program stored in the MCU. A/ D converter is used to read the output of temperature sensor and pressure sensor. Band gap voltage regulator provides constant voltage for pressure sensor, amplifier and A/D converter. The bandgap current regulator provides a constant current source for the temperature sensor.
Calibration method
The MCU calibrates and compensates the nonlinearity and temperature drift of the pressure sensor. In order to provide maximum accuracy, the A/D input resolution of 10 is selected, and the calculation resolution is set to 16, fixed point. In order to calibrate the range and offset and compensate the nonlinearity of the sensor output, the calibration software performs second-order polynomial correction on the sensor output, as shown below:
vout = c0+c 1Vp+c2vp 2( 1)
Cp = (c0，c 1，c2 ) (2)
Among them:
Vout = calibration output
Vp = uncompensated pressure sensor output
In order to compensate the temperature dependence of Cp, the calibration software uses a second-order polynomial fitting equation to calculate Cp:
c0 = c00 + c0 1Vt + c02 Vt2 (3)
c 1 = c 10+c 1 1Vt+c 12 Vt2(4)
c2 = c20 + c2 1Vt + c22Vt2 (5)
(6)
Among them:
Vt = temperature sensor output
During calibration, CT is read and stored in EPROM. MCU calculates Cp according to the output of temperature sensor, Vt and Ct. Then, Cp is used to calculate the calibrated pressure sensor output using the output Vp of the pressure sensor.
Calibration procedure
The calibration system first adjusts the gain and offset of the amplifier to use the full modulus range. Then check the characteristics of the uncompensated pressure sensor output at several temperature points. At each temperature, the second-order polynomial described in equation 1 is obtained by least square fitting, and the coefficient group Cp is determined. After Cp calculation of all temperature points is completed, Ct is determined by least square fitting of equations 3, 4 and 5 to determine the Cp of temperature points. At present, at least three separate temperature sampling points are needed to complete the fitting calculation.
Figure 2. The uncompensated output of the sensor-based system-on-chip is plotted at four different temperatures.
characteristic
Figure 2 shows the uncompensated sensor output characteristics at different temperatures after adjusting the gain and offset. Based on these data, the calibration system calculates the calibration coefficient and writes it into the on-chip EPROM. The compensation value is rounded to 8 digits. Figure 3 shows the calibration and compensation output of the integrated MCU. Figure 4 shows the error of the expected value. Because the error of 1 bit is 0.4%, the results show that the error is within 0.4% of the full-scale output.
Figure 3. Through testing and calibration at three temperatures, the compensation output of the system on chip is improved.
Test question
This preliminary work raises several questions, including different types of tests, unique test equipment and the need for multiple tests. In order to make low-cost integrated solutions possible, these problems must be solved.
Integrating the physical measurement function into the already complex mixed-mode analog-digital chip increases the demand for additional types of tests. The physical medium to be tested must be applied to the device and the response must be measured. Measuring the response to physical stimuli is not
Figure 4. At 30°C and 85°C, the bit error of the compensated output is within 1 bit.
Standard test of semiconductor industry, especially at various temperatures. Standard equipment can test the digital and analog parts of the chip, but the application of physical stimulation and the process of heating and cooling the device under test quickly and accurately promote the demand for improved and unique tester. These testers are unique and not standard. Therefore, the tester represents a large part of the final unit cost.
Not only the tester is expensive, but also the production capacity is limited. This increases the cost of each component because the depreciation cost allocated to each device increases. The cost is further increased due to the need for multiple tests. Remember, at least three different temperatures are used to test each device to determine the output characteristics of the sensor in the whole temperature range. These values are then used to derive the compensation algorithm and load it into the on-chip EPROM. In order to complete this cycle, the temperature of the device is tested again to prove its accuracy. Therefore, not only a special tester is needed, but it also becomes a bottleneck, because it is necessary to use the tester twice to complete each device-once to measure the characteristics and once to verify the results.
Future direction
Finding a way to reduce the test cost is one of the keys to realize low-cost integrated sensor and MCU. Promising ideas include:
Thorough characterization design
Limit working temperature
Limit the accuracy
Programming MCU to acquire data during testing.
Load the test and compensation algorithm into MCU before testing.
Since this is the first proof-of-concept device, further characterization can provide a way to limit the amount of temperature required for compensation. Limiting the operating temperature range can also reduce the number of temperatures required for compensation testing. The data shown in Figure 3 shows that the accuracy is 5% in the range of 5 C to 25 C. Another potential cost reduction measure is to use the programmability of MCU to record data during testing. By storing the compensation program in onchi p EPROM before the test and then recording the uncompensated output in EPROM during the test, it is possible to develop an algorithm that can pass the test at one time in the whole temperature range.
If there is no breakthrough in reducing the test cost of this new integrated sensor and MCU, system designers may be limited to continuing to use the current solution-independent MCU and sensor.
-
All DS 18B20 sensors used for multipoint temperature measurement are connected to a single chip microcomputer on an IO bus to collect temperature data in turn. If the system has a large number of sensors, the processing time of temperature data by single chip microcomputer is obviously prolonged, so the cycle of alternate testing becomes longer. This paper puts forward a new method to group DS 18B20 reasonably, and takes some measures in software. Therefore, the speed of alternate testing is obviously improved.
-