Circuitry related to different types and applications of transducers has already been described in various sections. This section summarizes the basic systems required and alternative methods of readout, recording, or display. Our starting point is the strength and type of signal generated by the transducer itself. Normally it will need amplification (al though not necessarily so) and possibly filtering to eliminate phasing errors. The resulting dc signal output can then be used to give a readout by an analog instrument, or, alternatively, it can be subjected to signal processing as required (ill. 29-1).
The most common requirement in signal processing is converting a dc analog signal into a digital signal. Three basic methods that may be employed are: - successive approximation - voltage-to-frequency conversion - voltage-to-time conversion SUCCESSIVE APPROXIMATION DIGITIZING This form of analog-to-digital converter employs a digital divider to feed back to the amplifier known increments of reference voltage in fixed sequence. At each step the amplifier then decides whether that increment is switched on, or ignored, so that a voltage is built up approximately to the value of the input. A block diagram of this form of working is shown in ill. 29-2. The actual circuitry involved can be quite complex, incorporating automatic sequencing and capable of very high conversion speeds. Accuracy of conversion depends on the accuracy and stability of the reference obtained from the digital divider.
VOLTAGE-TO-FREQUENCY CONVERSION With voltage-to-frequency conversion the converter produces a train of pulses with a frequency proportional to the voltage in put. Pulses are accumulated in a counter for a fixed period, and at the end of that period the total is readout. Figure 29-3 shows a block diagram of this type of converter.
An integrator produces a voltage ramp whose rate of rise varies with the output from the amplifier. As the ramp reaches a defined level, the gate opens and allows the clock to trigger the pulse generator. The pulse generator then resets the integrator and a new ramp commences. Thus, a train of pulses is produced that are filtered to give a dc level, and this is fed back to the input amplifier where it's compared with the unknown. The loop therefore adjusts itself so that the pulse frequency gives the correct feedback to balance the input. The process is continuous, unlike successive approximation, and the pulse frequency will follow changes in the unknown. Pulse amplitude is derived by using a reference voltage to define the output of the pulse generator. Thus, this reference is indirectly compared with the unknown input. The width of the pulses is defined by the clock oscillator. This could be crystal con trolled, but in fact a simple LC oscillator can be used because the system can be made self-compensating for variations in clock frequency. On a start command the variable pulses are allowed to enter the display counter, and clock pulses are gated to the timer counter. The capacity of the timer and the clock frequency are arranged so that a pulse is issued from the timer at the end of the required count period. This is used to shut down the display counter whose total is decoded and displayed. Should the clock drift to a higher frequency, the filtered dc fed back to the input will tend to fall; this will cause more pulses to be generated by the converter loop. But, since the clock frequency has increased, the court period will be shorter. Thus, the display counter will accumulate pulses faster, but for less time. Hence, the displayed reading is unaltered. This technique is particularly suitable for measuring small voltages and is readily adapted to provide a range of sensitivities. VOLTAGE-TO-TIME CONVERSION For voltage-to-time conversion the input is allowed to charge a capacitor for a fixed time. With the input removed, a reference voltage is then used to discharge the capacitor, the time taken for discharge being a measure of the input. In order to achieve linear characteristics the capacitor is associated with an integrator, the working of which is as follows. When the integrator is connected to the input its output “ramps up” at a rate that is directly proportional to the value of the input. After a fixed time the switch changes over and connects the reference in place of the input. it's arranged that the reference voltage is of opposite polarity to that of the input, so that the integrator output now “ramps down” at a fixed rate that is deter mined by the value of the reference. The time taken to complete ramp-down is a direct measure of the unknown. The ramp-down time can be measured, and the ramp-up time defined. A block design of this form of circuit is shown in ill. 29-4.
The sequence starts switching the integrator to the input by the control. As soon as the integrator leaves zero, the gate is opened and clock pulses pass into the counter. After a time that is dependent upon the clock frequency and the counter “full-house,” the counter fills and sends a pulse back to control. This causes the in put to be exchanged for the reference, and ramp-down commences. Meanwhile, clock pulses are still fed to the counter, which has started again from zero. As the end of ramp-down is reached, the gate closes, and the counter now holds a total that is a measure of the unknown input. This is decoded and passed to the display. Because the unknown input is applied to an integrator, the voltage stored on the integrator capacitor at the end of ramp-up is directly proportional to the integral of the input. Thus, if the input contains a periodic component such as 50 Hz, the integral of this component is zero when the ramp-up time is equal to its period (20 cms for 50 Hz). The result is similar to that obtained with the V to I arrangement, in that there is inherent rejection of series-mode interference. With the ramp-up time fixed at 20 ms, if the mains’ supply varies about 50 Hz, as commonly occurs, the rejection will be less than ideal. But as the ramp-up time is controlled by the clock, the clock frequency can be adjusted, so that ramp-up always lasts for one mains period. One DVM achieves this by comparing the ramp-up time with the mains and putting appropriate corrections into the clock by means of a digital servo. This technique permits very thorough isolation of the input circuits, while the rest of the voltmeter is grounded. The isolation and series-mode rejection can be so effective that with some products interference can virtually be ignored. DATA LOGGING Recording measurement results digitally on magnetic tape is also the basis of data logging. Basically this technique samples the parameters at regular intervals and records such data for subsequent offline input to a microprocessor, computer, or other data processing equipment. Such a system is shown in block form in ill. 29-5. it's composed of a transducer to provide the signal input and a signal conditioner or processor to convert the input signal to a form acceptable to the data logger, the output from which is passed to the data processor or analyzer.
Normally if more than one transducer is involved, a separate conditioner is required for each transducer. Outputs from the conditioners are then scanned by the data logger at preset intervals (that is, sampled sequentially), and the results are recorded in digital form on magnetic tape. For fast systems, inputs may have to be sampled and held at the beginning of the scan so that recorded in formation for a scan all corresponds to the same point in time. Thus, for optimum flexibility the signal-conditioning system needs to be programmable. Another major requirement of the signal-conditioning system is intelligence. Many logging applications require some inputs to be logged only as background information, whereas others are logged more frequently, or to be logged only when input signals exceed preset levels. With the availability of today’s large-scale integrated circuits, it's fairly easy to design this level of intelligence into a conditioning unit accommodated on a single printed circuit board. The final step in the data-signal-handling procedure is that of transferring the data from the recording medium to the appropriate analysis system, which normally takes the form of either a main frame computer, which also performs many other tasks, or a dedicated minicomputer or microprocessor-controlled system. In the former case a suitable media-conversion system is usually required, capable of reading the logger’s cassette or cartridge and outputting the data in suitable format for input to the mainframe. The hardware involved in such a system could well be comparable with that required for the complete dedicated analyzer but with much simpler software programming. COMPUTERS and ANALYZERS Computers are readily capable of calculating data directly from transducer signal readout and can present almost any type of in formation required. They can also employ many different techniques of filtering, windowing, zooming, and averaging, or they can extract individual frequency components such as modal parameters or graphical modeling all at very high speeds. Modern real-time analyzers can, in fact, sample at rates of greater than 250,000 times per second. They represent a highly specialized subject, however, that is outside the scope of this guide. Simple Techniques Analog-to-digital converters are available in the form of inexpensive IC chips, requiring only a minimum of additional components to produce a working circuit. There are numerous ICs of this type produced (AID converters), many of which are designed to interface directly with a microcomputer data lens. There are also IC AID converters that incorporate a display driver for powering an LED readout. Because these incorporate specific circuits and pin numbering, general circuit designs can be given. To use them, purchase a type for which an application diagram is available. Various other useful forms of converters are also available as IC chips. Among the most useful are frequency-to-voltage converters. Many of these are designed to interface directly with specific types of transducers, such as magnetic variable-reluctance pickups, for presenting the signal frequency received directly as an analog voltage output. This arrangement could be used as a simple meter-reading tachometer. Many chips of this type contain far more features than are necessary for simple readout but are not necessarily much more expensive than less complex types. The fact that they are more complex, too, probably means that they are more sensitive and stable. it's simply a case of using only the actual parts of the IC circuitry required for the job, which again virtually necessitates having an application diagram for the particular chip chosen. PREV: Vibration
Measurement and Monitoring |
Updated: Thursday, March 19, 2009 1:09 PST