Electronic instrumentation called data acquisition systems (DAQs or DASs) acquire data from sensors. They are often extended to be instrumentation and control systems. This is a tutorial presentation of principles of instrumentation that is typically multi-channel, medium to high resolution (12 to 20 bits) and relatively slow – slower than oscilloscopes in sample rate. DAQ principles – mostly basic, with some subtleties and simplifying concepts – and sensors will be emphasized.
To give the presentation a concrete context, the application example will be a rocket test system as used to verify the performance of small rockets fired statically on test stands. Test firings must be sequenced by a controller and a DAS acquires the sensor data. The rocket test control system must be informed about what is going on in the rocket, and the instrumentation subsystem provides it. Sensors convert quantities of interest, such as tank pressures or acceleration, to electrical form. The data acquisition system converts these electrical quantities to digital form compatible with the control computer as input.
Data Acquisition System
The numerical values of measured quantities, or data , are usually converted into digital form for the control computer by a DAS. A typical DAS is shown below.
Sensor waveforms enter antialiasing filters that remove high frequencies from them. This is sometimes necessary to prevent aliasing , the generation of spurious waveforms. A common example of aliasing is the backward rotation of wheel spokes in movies or on television. The successive picture frames of a movie or television signal are not continuous and cause difference (or “beat”) frequencies that result in these spurious images. Spurious waveforms are generated in DASs when the sensor waveforms are not “slowed down” sufficiently to remove the faster changes in them that cause aliasing. Any process that samples a continuous quantity outputs a discrete sequence of values and is subject to aliasing. To avoid it, all frequencies that are at least half the sampling rate are removed by the filters.
The MUX is an analog multiplexer , an electronic switch, like a television channel switch. It is commanded by the microcomputer (μC) to switch to a particular sensor input channel . Each channel is selected in turn for measurement. The PGA is a programmable-gain amplifier . Different sensors require different amounts of amplification of their waveforms, and the PGA gain is commanded by the μC. The A/D converter, or ADC , converts the filtered and amplified analog sensor waveform into a digital count that can be input to the μC.
The number of discrete values of analog input voltage that the ADC can distinguish between is its resolution , measured in the unit of bits. For N bits, the number of different outcomes is 2N . A 12-bit ADC can discriminate between 212 , or 4096, different analog input values. If its full-scale range is 4.096 V, each of the 4096 input levels is 1 mV apart. The 12-bit digital output of the ADC consequently has a resolution of 1 mV per count, or 1 mV per least-significant bit (LSB). The resolution of the ADC is 1 mV/LSB.
The computer further processes the sampled sensor quantities from the ADC, but in digital form. The ADC counts are unprocessed or raw data that must be corrected for offset and gain (slope) errors caused by inaccuracies in both the sensors and analog DAS circuits preceding the ADC. Then the result must be corrected for sensor nonlinearity, if appreciable, and the result accurately scaled.
Sensors typically used in rocket flight or testing are:
- temperature sensors: thermocouples, RTDs, thermistors, solid-state.
- pressure sensors: silicon or sapphire.
- flow sensors: turbine, ultrasonic doppler.
- inertial sensors: rate and vertical gyros, solid-state acceleration and rotational sensors, tilt switches.
- proximity sensors: microswitches.
- electrical sensors: voltage and current sensing.
- cryogenic sensors: cryogenic thermistors.
Most sensors output a voltage that is scaled to the measured quantity, to give a conversion factor (gain) in the form of, for example, V/kN, for a force sensor, or V/o C for a temperature sensor. Voltages occur across two circuit nodes (as “across” quantities). If one node is the zero-volt reference node for the system, or ground , then the sensor output is a voltage relative to ground. Voltages measured on nodes relative to ground are single-ended .
Some sensors have two terminals across which their output voltage appears and neither is ground. These are differential voltages because they are the difference between the voltages on each terminal, as measured with respect to ground. They are also sometimes referred to as “floating”.
Differential outputs are typical for sensors that are part of a common instrumentation circuit called a bridge . The “Sensor Bridge Circuit” diagram below illustrates this for a strain-sensor bridge. The bridge output voltage is the difference in voltages between the AIN+ and AIN– nodes, measured with respect to ground. In other words, a voltmeter with its – input connected to the ground terminal can be used to measure the voltage at AIN–. This reading is subtracted from the voltage measured at AIN+ with respect to the same ground node for the bridge output voltage.
Examples of bridge-based sensors are RTDs (temperature), pressure and strain or force (load cell) sensors. The resistance of these sensors varies with the measured quantity. In the diagram below, a load cell is configured from two strain gages with opposite polarity drives.
Bridge circuits consist of two voltage dividers driven by the bridge supply. Each two-resistance divider is a half-bridge . Bridge output sensitivity is proportional to bridge excitation voltage. For half-bridge sensors, the other half bridge is a divide by 2 voltage divider, consisting of precisely matched equal-value resistors.
The two strain gages are attached to opposite sides of a cantilever, so that when it is flexed, the top strain-gage increases in resistance (by +ε ) and the bottom one decreases (ε ). When not flexed, both sensors have nominally the same resistance and the voltage at AIN+ is half the bridge voltage, Vbr . For zero differential input voltage at zero scale, the other voltage divider consisting of stable, equal-value (R ) resistors divides Vbr in half at AIN-. The output voltage of AIN+ varies around half the bridge voltage, resulting in a bipolar (+/-) differential output.
2-, 3- and 4-Wire Bridges
It is not always necessary to locate the two resistors of the right-side half-bridge at the sensors. For negligible voltage drop in the bridge drive lines, an accurate half-bridge voltage is duplicated on the instrumentation system circuit-board (as Vbr /2) and provided for the AIN- inputs, usually through configuration jumpers on the board. This half-bridge voltage is measured through a dedicated channel and used as an offset for bridge-based sensors. By using the on-board half-bridge, only one sensor output line (AIN+) and the two bridge supply lines need to be run to each sensor bridge.
For full-bridge sensors, both AIN+ and AIN– are run from the sensor, and the bridge voltage is measured on the acquisition board. For negligible voltage drops in the bridge wires, these wiring schemes are satisfactory.
For non-negligible voltage drop in the bridge supply wires, 4-wire sensing is required. Four-wire or Kelvin sensing is the most accurate and uses separate pairs of bridge drive and sense lines.
RTD Temperature Sensors
RTDs (resistance-temperature devices) are based on the repeatable temperature coefficients (TCs) of metals such as platinum. RTDs are somewhat nonlinear and require correction. Standard RTD curves specify resistance as a function of temperature, such as the PT100 (DIN 43760) curve for platinum RTDs. A TC of resistance for two points, at 0o C and 100o C is designated as α :
For the PT100 curve, α = 3.850×10– 3/o C. But α is not constant over the full temperature range. The general RTD equation is:
where R0 is the resistance at 0o C (100 Ω or 1 kΩ) Solving for T ,
From -100o C to +800o C (the workable range for suitably-encapsulated RTDs), 100 Ω RTD resistance varies by about x 6.48, from 60.25 Ω to 390.26 Ω, with a positive TC.
Typical 1 kΩ thin-film RTDs are the Sensing Devices, Inc. (SDI) GR2141 and the Minco S251PF12 (or as thermal ribbons, S17624PF440B). The SDI Pt100/15P has an R0 of 100 Ω and S251PF12 of 1 kΩ.
Unlike load cells, RTD bridges use only one sensor, as shown below, and are suited for single-ended bridge circuits, as shown. AGND is the analog ground, a separate ground connection at the measurement system to the system ground.
Thermocouples are formed when two dissimilar metals are joined, as in a spot-weld. A small voltage will occur across the two metals that varies with junction temperature. K-type (chromel-alumel) or J-type (iron-constantan) thermocouples are common and useful for measuring temperatures too high for RTDs and thermistors.
K-type thermocouples are not as sensitive as J-type but have a higher temperature range. Each connection made to thermocouple wire is another thermocouple sensor. Using copper wire, the copper-chromel and copper-alumel connections form two additional thermocouples. These undesired thermocouples are called reference-junction or cold-junction thermocouples. Their effects must be nulled out by some means of compensation.
By running the thermocouple wire to the instrumentation board connector, the reference junctions will be near the thermocouple processing circuitry and will be at about the same temperature. The cold-junction compensation circuitry measures this temperature and compensates the thermocouple circuit output.
A separate temperature sensor could be used to measure ambient temperature near the cold junctions, and compensation done in the computer.
Thermocouple integrated circuits (ICs) that amplify and cold-junction compensate thermocouple voltages are the Analog Devices Inc. AD595, for K (chromel-alumel) thermocouples, and AD594 output for J (iron-constantan) thermocouples. Their outputs are
To extend the measurement range to the high end of 1250o C (K-type) and 750o (J-type), the output voltage may need to be divided, say, by 3 to accommodate a typical 4.1 V fs range of the ADC.
Ambient temperature can be conveniently sensed by the Analog Devices Inc. (ADI) AD22100 IC, a low-cost, three-pin, silicon-based temperature sensor. It has an analog voltage output of
where VCC is the AD22100 supply voltage. It operates from -50o C to +150o C with +/- 2 % full-scale inaccuracy. This sensor is ratiometric with VCC because the output varies in proportion to it. By operating it from the bridge voltage supply (Vbr ), bridge compensation can be applied to track bridge-voltage drift.
The AD22100 can be two-point calibrated because it is a linear type of transducer. (Then inaccuracy approaches its +/- 1 % nonlinearity specification.)
For easier calibration at somewhat less accuracy, immerse the (electrically insulated) sensor in ice-water and one-point calibrate for 0o C; or measure the temperature at the sensor with another thermometer or another (calibrated) temperature channel. No temperature measurement at all need be taken if its measurement channel is voltage calibrated and the above equations used, though the accuracy will be about +/- 2 %.
The AD22100 is specified to operate at a VCC from 4 V to 6 V and can be powered from the 4.1 V bridge supply. Then the raw-data value from the ADC is
One recommended sensor intended for ambient pressure measurement is the Motorola MPX2202AP, a low-cost, absolute sensing, 200 kPa (29 psi) full-scale, silicon-based pressure sensor. It can be used as a barometer because it senses absolute pressure, and atmospheric pressure can be converted to altitude. It also has sufficient range to sense the dynamic pressure of a typical flight vehicle. It is temperature-compensated and has a nonlinearity of under +/- %.
The MPX2202AP is a complete, compensated bridge circuit with an output ratiometric to the supply voltage. It can be one-point or two-point calibrated. For a 4.1 V bridge supply, at full scale, its output is about 16.4 mV and has a nominal scale factor of 82 μ V/kPa. The zero scale (zs) is at zero pressure, and the offset voltage error is specified at +/- 1 mV.
To calculate the gain required, divide the ADC full-scale input voltage (= Vbr = 4.1 V) by the sensor full-scale output and round down. This results in a gain setting of x100. This provides ample overrange capability for capturing burst-failure data.
Comparable sensors are the Sensym SCX30ANC and TRW Novasensor NPC-410-30-A. Some bridge-based sensors, such as the Motorola MPX4250 (250 kPa fs), operate off the computer +5V supply and have a bridge voltage other than Vbr . Their bridge voltage must be tracked (by measurement through another channel) to compensate bridge sensitivity for maximum accuracy.
An accelerometer suitable for most sounding rockets and other low-g applications, is the ADI ADXL105. It is inexpensive, silicon-based, and has a range of +/-5 g. It can be two-point calibrated by using gravity. In the orientation of maximum acceleration, it is subjected to 1 g. Inverted (rotate 180o ), its input is -1 g. Nominal g0 at the Earth’s surface is 9.806 m/s2 .
Power-Supply Voltages and Currents
Ground-based power supplies or on-board batteries can be sensed, usually through a voltage divider. The advantage of differential voltage measurement channels is their ability to measure “floating” voltages, such as the voltage across a current-sensing resistor in series with the + terminal of the battery.
Typical turbine flow sensors are designed as magnetic-vane flow sensors. Magnets in the turbine vanes rotate past a coil in the body of the sensor and induce a voltage pulse into it as they pass by. Typical pulse amplitudes are at least 50 mV over the flow range of interest. Maximum flow pulse rates are typically 100 Hz to several kHz.
These pulses are usually processed by analog circuitry and converted to computer-compatible digital pulses. These pulses are then input to a counter controlled by the computer. Counts are accumulated over an accurate time interval, usually set by the time-base of the computer. That is, another counter/timer interrupts the computer at regular intervals. Between these interrupts an accurate time interval is established that can be used as a frequency counter time-base. The frequency is calculated as
where N is the number of counts over time interval Δt.
Cryogenic thermistors are highly nonlinear temperature sensors, used to detect the presence of cryogenic fluids. They can be placed at ullage levels in tanks and used to detect when the tank is filled to the ullage volume. They can be placed in the high side of a voltage divider and can drive a digital bit input directly.
One cryogenic thermistor is the Thermometrics, Inc. A105CTP100DE104R thermoprobe. It has a resistance of 100 kΩ at the boiling point of liquid nitrogen (-195.82o C). LOX boils at -183o C. Its resistance at -185o C is 54,322 Ω and at -180o C is 37,081 Ω. But at -100o C, it is 146 Ω. By making the thermoprobe the upper resistor in a divider driven by +5 V, and with a lower resistor value of around 1 kΩ, the divider output can directly drive TTL-level digital computer inputs.
In the next part of this article is DAS system acquisition and processing strategy and calibration.