In large measurement and control systems (modern cars, for example), which include hundreds of sensors, communication between a central computer and the smart sensors, which are widely scattered, is difficult. It is probable that the sensors are produced by different manufacturers, and that the sensors’ outputs have different formats. In order to design efficient sensor systems the data transfer must ideally be organized in an orderly and reliable fashion. Such systems are called bus systems or networks. In general a bus system consists of a central computer that is connected by a number of wires to a large number of sensors. When a sensor is activated to send information into the central computer, the sensor's address is selected and the sensor will be switched to the digital data line. The central computer can initiate different kinds of tests and recalibrations. Every sensor is connected to the same wires. A dedicated transmission protocol is applied to allow a flexible and an undisturbed data flow.

There are many buses now available and it is still difficult to find the one to suit smart sensors’ requirements. We do not intend to describe in detail all existing sensor buses and network protocols in this book. For more detailed information we recommend references [8] and [178]. The mission of this last chapter is to show the most applicable digital bus interfaces for smart sensor applications as well as to answer what is expedient to do with analog sensors.

10.1 Sensor Buses and Network Protocols

A number of different protocols exist, each having its own interface requirements. The requirements stipulate such parameters as headers, the data-word length and type, the bit rate, the cyclic redundancy check, and many others. Table 10.1 shows distinct smart-sensor network protocols.

There are some interface devices available, for example, Motorola, produces versions of its 68HC705 microcontroller that provides a J1850 automotive-network interface and variations of the company's 68HC05 microcontrollers that incorporate computer-automated network interface functions.

All these interfaces are strongly associated with their field of application. A close examination of existing buses shows that these are designed to fit a special set of requirements, for example domestic, industrial and automotive applications, measuring systems, etc. as shown in Table 10.1. Smart sensors are a new kind of application, therefore the majority of existing digital bus systems cannot be directly applied while maintaining optimum performance.

Table 10.1 Sensor network protocols

Sensor network protocol Developer
  J-1850, J-1939 (CAN)   SAE
  J1567 C2D   SAE (Chrysler)
  J2058 CSC SAE   Chrysler
  J2106 Token Slot   SAE (General Motors)
  CAN   Robert Bosch GmbH
  A-Bus   Volkswagen AG
  D2B   Philips
  MI-Bus   Motorola
  Hart   Rosemount
  DeviceNet, Remote I/O   Allen-Bradley
  Smart Distributed Systems   Honeywell
  SP50 Fieldbus   ISP+World FIP = Fieldbus Foundation
  LonTalk/LonWorks   Echelon Corp
  Profibus DP/PA   DIN (Germany), Siemens
  ASI Bus   ASI Association
  InterBus-S   InterBus-S Club, Phoenix
  Seriplex   Automated Process Control (API Inc)
  SERCOS   VDW (German tool manufacturers assoc)
  IPCA   Pitney Bowes Inc
  HP-IB (IEEE-488)   Hewlett-Packard
  Arcnet   Datapoint
  WorldFIP   WorldFIP
  Filbus   Gespac
Building and office automation
  BACnet   Building Automation Industry
  IBIbus   Intelligent Building Institute
  Batibus   Merlin Gerin (France)
  EIbus   Germany
Home automation
  Smart House   Smart House LP
  CEBus   EIA
  I2C   Philips
University protocols
  Michigan Parallel Standard (MPS)   University of Michigan
  Michigan Serial Standard (MSS)   University of Michigan
  Integrated Smart-Sensor Bus (IS2)   Delft University of Technology
  Time-Triggered Protocol   University of Wien, Austria

The I2C is a very interesting bus from a smart sensor point of view [8]. The topology is rather simple and it has minimum hardware requirements compared to the D2B. But whereas the hardware specification allows simple interfacing, the communication protocol is too rigid. This prompted the design of a new Integrated Smart Sensor bus, IS2 similar to the I2C bus, but with a highly simplified protocol. To keep the complexity of the required electronics to a minimum, only the most elementary functions are implemented in the bus. As in I2C, the IS2 bus requires two lines for communication: a clock line and a data line. An important feature of the IS2 bus is that the data field length is not defined. The transmission can be terminated either by the data master or the data sensor [8].

One of the new industrial examples is the colour digital image sensor MCM20027 from Motorola [179]. This sensor is digitally programmable via the I2C interface.

Another interesting bus from a smart sensor point of view is the Controller Area Network (CAN), which has recently received a wide distribution. It is an advanced serial communication protocol for automotive networking, which efficiently supports distributed real-time control with a very high level of reliability. Originally developed at the beginning of the 1980s by Bosch to simplify the wiring in automobiles, its use has spread to machine and factory automation products. It is also suitable for industrial applications, building automation, railway vehicles and ships. The CAN provides standardized communication objects for process data, service data, network management, synchronization, time-stamping and emergency messages. It is the basis of several sensor buses, such as DeviceNet (Allen-Bradley), Honeywell's SDS or Can Application Layer (CAL) from CAN in Automation, a group of international users and manufacturers, which comprises over 300 companies. CANOpen is a family of profiles based on CAN, which was developed within the CAN in Automation group. The extensive error detection and correction features of the CAN can easily withstand the harsh physical and electrical environments presented by a car. The SDS was developed by Bosch for networking the majority of the distributed electrical devices in an automobile, initially designed to eliminate the large and expensive wiring harnesses in Mercedes.

Other sensors with the CANopen bus are the pressure transducer of COP series with full-scale accuracy up to 0.15% [180] and the microprocessor-controlled CO2 sensor based on the infrared light absorption from Madur Electronics (Austria) [181]. The latter is also available with the M-Bus interface. Dynisco Europe GmbH, STW and BDsensors have also proposed some new pressure sensors with the CAN interface [182].

Today many semiconductor manufacturers offer microprocessors with embedded CAN controllers. For example, the DS80C390 from Dallas Semiconductor, SAK 82C900 and SAE 81C90/91 standalone CAN-controllers, the 16-bit microcontroller C167CR/CS, C164CI and C505CA, C515C8051-compatible microcontrollers from Infineon Technologies Corporation, 68HC05/08/12 microcontrollers from Motorola, μPD78070Y and μPD780948 8-bit family microcontrollers and V850E/IA1/SF1 32-bit family microcontrollers from NEC Electronics Inc [183].

A sensor bus interface for use in generic bus-organized sensor-driven process control systems was developed at the University of Michigan. The sensor-bus interface is microprocessor-controlled. There are parallel and serial bus structures. Both are suitable for distributed control systems.

10.2 Sensor Interface Circuits

10.2.1 Universal transducer interface (UTI)

A common disadvantage of many digital interfaces is that many analog sensors cannot be interfaced in a low-cost way. In order to eliminate this disadvantage, the Universal Transducer Interface (UTI) circuit for different kinds of analog sensors was designed in the Electronics Research Laboratory, Delft University of Technology [184] and is manufactured by Smartec [185]. It is a complete analog front end for low frequency measurement applications, based on a period-modulated oscillator. Sensing elements can be directly connected to the UTI without extra electronics. Only a single reference element, of the same kind as the sensor, is required. The UTI is intended for the low-cost market, for systems where this interface provides an intermediate function between low-cost sensor elements on the one hand and the microcontroller on the other. This interface, which is directly read out by a microcontroller, services the following sensor elements: Pt resistors, thermistors, potentiometers resistors, capacitors, resistive bridges. With some extra electronic circuitry, the UTI can be used to measure voltage (two ICs MAX4560) and current, which makes them suitable for thermopiles, thermocouples and other types of voltage- or current-output analog sensors. The UTI converts low-level signals from an analog sensor to a period-modulated (duty-cycle) microcontroller-compatible time domain signal [186]. When the sensor signal is converted to the time domain, using a period modulator in the UTI, then the micro-controllers do not require a built-in ADC. So, if the UTI is connected to an Intel 87C51FB microcontroller, for example, it is possible to measure the period with a 0.33 μs resolution.

The signal conversion is carried out according to the linear law:


where Si is the analog output sensor's signal, k and Moff are measuring converter parameters that directly influence conversion error. In order to achieve high accuracy the UTI operates in auto-calibration, which is based on a three-phase differential method of measurement. The given method allows removing the error caused by the above parameters. The essence of this method of measurement consists of the measurement of three signals: S1 = 0, S2 = Sref and S3 = Sx (zero, reference and measurand) during one cycle:


The output signal of the UTI has three informative components and is shown in Figure 10.1.


Figure 10.1 Period-modulated output signal of UTI for 3-phase mode

During the first phase Toff, the offset of the complete system is measured. During the second phase Tref, the reference signal is measured and during the last phase Tx, the signal itself is measured. The duration of each phase is proportional to the signal that is measured during that phase. The result is the ratio:


There are 16 different modes with 3–5 phases within one cycle [187]. The connection of capacitors (up to 12 pF, mode 0, 1, 2) to the UTI and time diagram for this mode are shown in Figure 10.2. It is possible to measure multiple capacitances as well as capacitances from 300 pF to 2 pF. Possible applications are liquid level sensors, humidity, position, rotation, movement, displacement sensors. The result can be calculated as follows:


The connection of the thermistor Pt100 (Rx) to the UTI and the time diagram for the modes 5, 6, 7 and 8 is shown in Figure 10.3.

The result can be calculated according to the formula:


The connection of the resistive bridge to the UTI and the time diagram for modes 9 and 10 is shown in Figure 10.4.


Figure 10.2 Measurement circuit for small capacitance and output signal of UTI


Figure 10.3 Measurement circuit for thermistor and output signal of UTI


Figure 10.4 Measurement circuit for resistive bridge and output signal of UTI

Possible applications are pressure sensors and accelerometers. The measuring result can be calculated as follows:


The connection of potentiometers to the UTI and the time diagram for this mode is shown in Figure 10.5.

The measuring result can be calculated according to the following equation:


For the measurement of multiple sensing elements it is possible to use one UTI combined with a multiplexer as shown in Figure 10.6:

Using the power down function of the UTI it is possible to build up a multiple-channel measurement system, because the output impedance of the UTI is very high when the power down is low (PD = 0). The interfacing multiple UTIs for the microcontroller is shown in Figure 10.7. To measure a certain channel, the UTI is set to ‘1’ while others must be set to ‘0’.


Figure 10.5 Measurement circuit for potentiometer measurement and output signal of UTI


Figure 10.6 Multiple sensing elements measurement

The output signal of the UTI can be digitized by counting the number of internal clock cycles in each phase. This sampling introduces a quantization error. The relative standard deviation can be calculated by the following formula:


where ts is the sampling time and Tphase is the phase duration. When the sampling time is 1 μs and the offset frequency is 50 kHz, the standard deviation of the offset phase is 12.5 bits in the fast mode and 15.5 bits in the slow mode. Further improvement of the resolution can be obtained by averaging over several values of the measurand.

Typically, the linearity of the UTI has values between 11 bits and 13 bits, depending on the mode. The UTI is ideal for use in smart microcontroller-based systems. One output-data wire reduces the number of interconnected lines and reduces the number of opto-couples required in isolated systems.


Figure 10.7 Setup for the measurement of multiple channel signal

The microcontroller is used to measure the period-modulated signal from the UTI, to process the measured data and to output digital data to a central computer via the communication interface.

Main technical performances of the UTI are shown in Table 10.2.

10.2.2 Time-to-digital converter (TDC)

Another interesting IC used with quasi-digital smart sensors is the Time-to-Digital Converter (TDC) from Acam-Messelectronic GmbH [188]. It can be used wherever physical quantities need to be digitized for purposes of data processing, with applications including:

Table 10.2 Technical performances of UTI

Parameter Range
Capacitive sensors, pF 0–2, 0–12, up to 300
Platinum resistor Pt100, Pt1000
Thermistor, kΩ 1–25
Resistive bridges with max imbalance ±4% or ±0.25% 250 Ω–10 kΩ
Potentiometers, kΩ 1–50
Resolution, bits 14
Linearity, bits 11–13
Measurement time, ms 10–100
Suppression of interference, Hz 50/60
Temperature range, °C −40– + 85
Power supply, V 2.9–5.5
Current consumption, mA <2.5
  • ultrasonic-based flow and density measurements
  • temperature measurements (Pt100, Pt500)
  • nuclear and high-energy physics
  • laser distance measurement
  • ultrasonic position feedback devices
  • capacitance and resistance measurement
  • frequency and phase measurement (in a few ranges).

By using TDCs the measurement is transferred into time domain signals, making it possible to convert the electronics into a digital single-chip solution.

There is the 2-channel TDC-GP1 and the 8-channel TDC-F1. Some main features of the TDC-GP1 are:

  • two measuring channels with a resolution of approx. 250 ps
  • two measuring ranges: 2 ns–7.6 μs and 60 ns–200 ms
  • four ports to measure capacities, coils or resistors with 16-bit precision and up to 20 000 measurements per second
  • an internal ALU for calibration of the measurement results. A 24-bit multiplication unit enables the results to be scaled
  • low power consumption (10 μA); full battery operation is possible
  • 8-bit processor interface
  • ranges for R, L and C measurements: 100 Ω–>1 MΩ, 10 μH–>10 H, 10 pF–>1 mF accordingly.

The key features of the TDC-F1 are:

  • eight channels with approx. 120 ps resolution
  • optional four channels with approx. 60 ps resolution
  • optional 32-bit channels with approx. 5.7 ns resolution
  • measuring range of approx. 7.6 μs
  • 8-bit I/O interface.

It cannot measure R, L and C parameters or meet the demands of experiments in high-energy physics.


One of the triggers for the growth of smart sensors is the forthcoming sensor bus standard. The standard will spawn both a wide range of smart-sensor ICs and a generation of sensor-to-network interface chips.

Due to available industrial interface circuits, it is possible to convert analog sensors’ signals to the quasi-digital domain easily. The following approach allows an important smart sensor feature: self-adaptation. The interfacing circuits considered above are able to control accuracy and speed using suitable software.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.