1. Signal path of the sensor

    Understand after a simulation signal path, the number system developer can be from various application, catch the sensor data more accurately.

    Main point
    Even it may have different outputs that the same manufacturer is similar to the sensor, and these differences will bring the trouble to system artificer.
    The noise comes from many kinds of reasons, including the CIBOL, radio frequency, thermoelement, even sensor.
    The signal using the sensor to strain the wave rear, must use ADC, will imitate the semaphore and turn a digital signal.
    Can choose to adopt and connect ADC, can adopt the microcontroller of built-in ADC too. There is higher precision to connect ADC, there is a better characteristic in speed and definition.

    Sensor multi-place apply the embedded systems to more and more. Though the industrial products have been using it for making control system all the time for a long time, but consume the apparatus and begin to adopt the sensor now. The manufacturers integrated sensor in the consumer products, in order to create good users experience, if increase the accelerometer in the mobile phone, and add vaporized sensor,etc. into microwave oven. The system artificer who worked only in the digital domain in the past finds now, must have dealings with analog sensor by oneself.


    Fig. 1, the simulation signal path of a sensor can be divided into several grades: Enlarge, filter and digitization.

    Analog signal of the sensor must for the system to use digitizing, and signal take process amplify, filter and digitization ’ Fig. 1) . Every grade usually involves a device around some passive components, by realizing a application correctly. Once do digitization to the signal, can give the control system of the microprocessor of it, or put the data in order, send to a host processor through a communication protocol. The agreement can use the sensor data depending upon need.

    Each sensor has different outcoming signals and ranges. The signal outputted can be voltage, electric current, resistance, electric capacity or frequency, but hardly there is any standard, there is only dedicated industry’s systematic inuse, they. Even it can have different outputs that the same manufacturer is similar to the sensor, and these differences will bring some trouble to system artificer. The artificer meets system’s demands while choosing the sensor. But if the demand present, change during design, sensor take, make modification too then. In addition, one must output slightly different new sensors to the amplifier stage and strain the wave scale to change.

    Most sensors output the electric current or voltage signal of a low level, so a simple resistor network can turn signal of any electric current into a voltage. Some concepts and component selection procedure of simple description of this text.

    Range
    One output of sensor can low to several, can also so as to up to several volts. In order to accomplish accurate digitization, the signal must be enough large for ADC, could read out effectively. Under most situations, the sensor signals all need amplifying. For example one typical the intersection of Model K and thermocouple export as 41μ V / degrees of C, the intersection of design and in need of a degree of C graininess of you, need to do equivalent magnification. So, must consider the definition of ADC, in order to guarantee to enlarge the signal to meeting required graininess.

    Mainly depend on the required type to the choice of the amplifier, for example amplifier of an instrument, difference amplifier, operation amplifier, PGA ’ Programmable gain amplifier) . In addition must also confirm the amplifier required gain magnitude. Amplifier around resistor network ( The tape is feedbacked) Have determined the gain to noise temperature ratio of the amplifier. Under the ideal situation, the maximum gain of the reference amplifier is limitless. Usually presume the gain to noise temperature ratio of PGA for the digital signal of the device. This signal has changed the internal resistor network. The gain to noise temperature ratio of maximum likelihood of a PGA expires by 1/2 for 1‰ of the traditional amplifier, but this section can be accepted under most situations.

    Must also consider another important specification as to the amplifier: Offset voltage. Voltage changed when the offset voltage is a signal passing amplifier. For example, if give a unit gain to noise temperature ratio (the gain to noise temperature ratio is 1) a 500 mV signal Amplifier,offset voltage as 10 mV,there are 510 mV on what has been received output. If the output area of the sensor is 0 to 900 mV, but the system does not need very delicate sensor reading, then this deviation can be ignored. If the range of the sensor is from 450 mV to 550 mV, maybe this deviation is unacceptable. The smaller the offset voltage is, the more expensive the amplifier is. All amplifiers are deflected, but you need to know whether the system can tolerate it. Can reduce or dispel the offset voltage with relevant pair of sampling techniques.

    Filter
    All systems will superpose some noises in the sensor signal. The noise source has all respects, including the CIBOL, radio frequency, thermoelement, even sensor. The signal noise will make the reading of ADC not accurate and unstable, noise level will be strengthened in the amplifier, because the amplifier can amplify the error in signals. The signal noise can be divided into low frequency, high frequency or a certain known frequency. It is the high frequency noise problem to be the usually to be most solved.


    Fig. 2, include amplifier, electric-wave filter and ADC in the signal path of the sensor. Designing to remove the noise in signals of electric-wave filter, limit the bandwidth.

    May filtrate the noise with passive analog filters, electric-wave filters IC and digital filter ’ Fig. 2) . The most common method is passive filtered, this will set up a passive network with resistance, electric capacity and inductance. However, you must design the passive electric-wave filter, and unable to change them briefly. The complexity of filter design may be as great as your required electric-wave filter grade; The design work capacity of an first order Chebyshev electric-wave filter is much smaller than an eight steps Bessel electric-wave filter. So you should confirm the required electric-wave filter order, then the filtering method that reselection adopts.

    Some IC allows you to use the digital programming method, confirms the required electric-wave filter type. These IC sets up electric-wave filter with the internal analogous circuit, and there is the offset voltage correlated to it. They can let you move, quantize filtering step to ADC, too. Digital filter design can be very complicated, but there are many simple and convenient designs that can help to make the higher-order filter. The numerical filtering is an ideal way to remove the noise, however, it usually needs a lot of CPU cycles, have increased the power consumption. The system will usually cause the high frequency noise, so need to adopt the low-pass filter. This kind of electric-wave filter can decay and is higher than the signal part of the alpha cut frequency presumed. Some sensor signals demand to adopt the cascade connected polytype electric-wave filter each other. Have all appointed a fundamental interface circuit in most sensor data lists, but not mentioned the required filtering form. The system artificer must be after understanding required filtering form completely, and then set up the system.

     
    1. pedese posted this