Measurement Techniques, Vol 40, No 10, 1997
MEASUREMENT MODULAR
CONVERTER
F O R S E N S O R S I G N A L S IN
SYSTEMS
UDC 621.317.76
Yu. K. Blokin-Mechtalin
We consider a facility for measurement of signals from different types of sensors; the system complies with the CAMAC and VME standards, and it is designed to increase measurement accuracy. We present the basic characteristics of the system.
The measurement information system of wind tunnels makes use of various types of sensors: tensiometric, thermometric, potentiometric, sensors generating FM output signals, etc. As a rule, modern systems are modular, and based on standard interfaces (CAMAC, VME, VXI). The link between the signal convertors and the system bus efficiently uses structural and algorithmic methods for increasing accuracy. The greatest measurement accuracy, highest resolution, and broadest dynamic range is provided by measurement convertors that are constructed with the method of balanced charges. Devices designed with this method include, in particular, high-accuracy pulse-compensation voltage-to-frequency convertors [1]. There are frequency measuring instruments that can be programmed to meet accuracy and response requirements [2]. These devices have been used to develop an integrating instrument for conversion of signals from various types of sensors for modular systems. The convertor contains the following components (Fig. 1): a differential instrumentation amplifier DIA with programmable gain; a voltage-to-frequency converter VFC and a frequency-to-code converter FCC; voltage and current sources VS and CS; a reference signal generator RS; a scaling circuit SU; a communication controller CU with a strain sensor SS; a circuit IU for separation of the analog and digital pairs of signals; and a decoder D for decoding signals from the bus master. For the DIA we used an M2USI~0084 hybrid chip that is based on a classical scheme using three operational amplifiers; this circuit permits selection of any of six gains (1, 10, 100, 200, 500, and 1000). The basic characteristics of the DIA (noise level, thermal drift in bias voltage, gain, attenuation of sin-phase noise) correspond to the requirements of measurements for aerodynamic experiments. There is a symmetric RC-filter at the input of the amplifier to suppress high-frequency noise. The voltage-to-frequency converter VFC is based on an M2PNM1781 hybrid chip. The circuit contains an integrator with a time constant of r = RIC a source for bias voltage U b with switchable polarity, a source of balancing current Io with a polarity switch, I0 = +_Ub/R 2, and a logic switch. Depending on the sign of U b, the input voltage for the VFC is converted to the 0-10 V range. To deal with a sign-varying signal from the strain gauge, there is a scaling unit that supports an inputsignal range of 0 to +_5 V. The scaling unit can also compensate for the signal from the initial equilibrium state in the strain gauge. The supply voltage U s for the strain gauge is drawn from the bias-voltage generator, which permits polarity switching. Changes in the polarity of Usg during measurement make it possible to reduce error due to drift in the amplifier, thermal effects, and other additive noise. Since the input voltage U x is promotional to Usg, there is no influence from U b on the result of conversion. Errors due to changes in the resistance of supply lines are eliminated by using a four-conductor connection between the strain gauge and the voltage source. The frequency at the output of the VFC is given by the relation R2Ux E fx = R'~--'~b 0O,
(1)
where Fo is the frequency of the timing signals, and Q = 2 is their duty factor. Translated from Izmeritel'naya Tekhnika, No. 11, pp. 15-17, November, 1997 0543-1972/97/4011-1047518.00 9
Plenum Publishing Corporation
1047
~
U~
Usg
fr
MMe
E]
KS
l
L I'--'-
IN
Controllin~
I
Bus Rs
R0
I
Fig. 1. Block diagram of signal converter. Keeping in mind that U x = 2Uberkag, where e r = Ar/r is the relative incremental resistance of the strain gauge, 2U b = Usg is the voltage supplied to the bridge, kag is the gain of the DIA, we can write 2QkagR2/R 1 = K and represent relation (1) in the form f = e, KQ.
(2)
It follows from (2) that the output frequency is independent of the voltage supplied to the sensor. When we account for the bias voltage Ubt for the DIA, we can represent (2) in the form
Summing pulses at the frequency fx with different polarities of Usg (Ub) and dividing by 2, we obtain a result that is not influenced by Ubi. The frequency encoder FCC is based on the principle of comparing the measured frequency fx and a clock signal f0 from a quartz oscillator. The results of frequency measurement are averaged over a time interval tx equal to the duration of some number of pulses at the measured frequency. The time t x and the period of the measured frequency T• is found from the number of pulses at the measured (nx) and clock (no) frequencies and the known period T O = f o - t : t = n T = noTQ: T =noTo l n .
(3)
The numbers of pulses no and n x are set in programmable counters. The relative error in measurement of the period T x due to quantization error is
where fir is the error in the clock oscillator, which can be neglected. Using relations (3) and (4), we can measure T x with constant, given error over the operating range, changing n x during measurement as a function of the current value of T x. To do this, we make measurements in two stages. In the first we program the minimum number of pulses (nx)mi n (the minimum measurement interval) to obtain a coarse estimate for the value of T x from 1048
0,05 0.04-
0,030,02-
0,01" 0,01 0.02
I
[
I
I
0,02 0,04
0,05 0,1
0,1 0,2
0,2 0,4
0.5 t x, sec 1,0 t i , sec
Fig. 2. Relationship between converter error and time. (3). In the second stage, where we obtain the specified accuracy, we program the required number of pulses nxt, as determined from (4). The measurement time also remains constant, i.e., the following conditions are satisfied:
n x r Tx
-
const; t = n r T ~ = const,
where 6Tx is the given error in measurement of the period and T x is the current value of the period. For a given time tx or given error (STx we can perform measurements in one step. Here nx is determined from (3) or (4), respectively. Pulses at the output frequency fx of the VFC have constant duration equal to the duration of pulses from the clock oscillator Fo. Only the following frequency fx changes as a function of the converted voltage and its sign. Since the input signal for the VFC must be treated as a meander, the frequency fx is first divided by a binary counter (by n = 2, 4, 8, or 16). The reduction in the upper limit for the measured frequency (increase in the period T x = n/fx) makes it possible to increase the averaging time tx and reduce the measurement error at lower frequencies. The range of the FCC is 0.1-1130 kHz, and the absolute error in frequency measurement is 0.1 Hz. When formula (2) is used, we can write the digital equivalent N x of a measured quantity e r that is obtained after transformation of the frequency fx/n over a given time tx as tx Nx =
=
rK
0tx/n.
0
The conversion coefficient for the instrument k = Ux/N x is determined by means of internal calibration signals (U k = + 2 5 mV). The calibration-signal generator RS is a star voltage divider. In order to increase the accuracy with which the conversion coefficient is determined, the divider is supplied with the voltage Vsg obtained, as noted above, from U b. The sign of U k is established by changing the polarity of the voltage applied to the input of the divider by an electronic switch. The converter is calibrated at three points, null on the scale, which corresponds to short circuiting of the DIA inputs to ground, and + U k. The converter is equipped with a circuit for monitoring the connection to the strain gauge. For this purpose one of the arms in the bridge can be electronically shunted by a precision resistor. The result is application of a control signal to the input of the converter that is close to the operating range Ux = +25 mV. The operating mode (null, calibrate, test, measure) is chosen by means of an electronic switch. In addition to a bridge strain gauge (SS), the converter can be used with other sensors. A potentiometric sensor (PS) can be attached to the converter either by way of a bridge or directly to the VFC. Resistance thermometers and singte strain resistors Rx can be attached to the current source CS. Voltage for the current supply is provided by U b. The measurement circuit for sensors includes a four-conductor circuit for reducing the influence of changes in resistance. The measured signal is U x = Ut) - IR s = Ub(R s - Ro)/Ro = IAR, where I = Ub/Ro is the current supplied to the sensor; Ro is the initial resistance of the sensor. Sensors with frequency output FS are connected through the galvanic isolator IU directly to FCC, and supplied with a constant voltage U s from a separate power supply.
1049
Figure 2 shows the relationship between measurement time t i and the error in the converter % Here t i covers two integration cycles t k with positive and negative voltages + Usg. The time t i = 2t k is chosen to allow for dynamic components in the measured signal and accuracy requirements, The structural and algorithmic features of the design provide the following advantages: broad dynamic range and high resolution, which, in particular, eliminate the need to balance the bridge, which would lead to additional error; elimination of error due to instability in the supply voltage; reduction of error due to drift of the amplifier null, thermal effects, and other additive noise; this results from changes in the polarity of the sensor supply voltage during measurement; reduction of error due to long-term instability in the conversion coefficient of the instrument that results from calibration directly before measurements are made; the ability of the instrument to adapt to accuracy requirements by variation in the integration time.
Basic Converter Characteristics: measurement ranges: +10, +25, +50 mV, +0.5, +5 V; gain: 500, 200, 100, 10, 1; supply voltages: +_6, + 9 , + i2 V. integration time: 0.01-4 sec; conversion error no greater than +0.05%, reducible by increasing integration time; compliance with CAMAC and VME standards. The use of such converters in measurement systems makes it possible to increase the measurement accuracy for signals from sensors used in wind turmels.
REFERENCES 1.
2.
1050
V. I. Voroskolevskii and N. Ya. Pinigin, Voltage-to-Frequency Converters and Their Application for Measurement and Control [in Russian], l~nergoatomizdat, Moscow (1994). Yu. K. Blokin-Mechtalin, V. M. Vlasenko, and L. F. Nazarova, Patent No. 2018173 RF; Izobret., No. 15 (1994).