Catto / Topic Map · Electronic Measurements I Level 4
Electronic Measurements I · 120 h · Topic 8 of 11

Signal conditioning and sensors

A Pt100 connected with two wires and 10 m of thin cable reads 4.5 °C too high. The sensor is accurate; the cables introduce the error.

Pt100 Thermocouple Strain gauges Instrumentation amplifier Four-wire

01From the phenomenon to the number

To measure a temperature, a force or a pressure, a sensor converts the quantity into an electrical signal: a resistance that changes, a voltage of microvolts, a current. That signal can almost never be connected directly to the converter. Signal conditioning is everything between the two: exciting the sensor, amplifying, filtering, isolating, compensating and linearizing. The errors of that chain are usually larger than those of the sensor itself. Sensors and actuators are presented in the technician program, in sensors and actuators.

02Temperature sensors

SensorPrincipleTypical range and accuracyConditioning
Pt100 (RTD)Platinum resistance, 100 Ω at 0 °C, rises 0.385 Ω/°C−200 to 600 °C, ±0.3 °C (class B at 0 °C)Excitation current, three or four wires
ThermocoupleSeebeck voltage between two metals: about 41 µV/°C for type K−200 to 1250 °C (K), ±2 °CCold-junction compensation and high amplification
NTC thermistorResistance that falls sharply with temperature−40 to 150 °C, very sensitiveDivider and Steinhart-Hart linearization
Integrated (LM35 and similar)Junction voltage of transistors, already calibrated−55 to 150 °C, ±0.5 °CAlmost none: 10 mV/°C

The resistance of a Pt100 follows the Callendar-Van Dusen equation, which for positive temperatures is:

\[ R(T) = R_0\left(1 + A\,T + B\,T^2\right) \] With \( R_0 = 100\ \Omega \), \( A = 3.9083 \times 10^{-3} \) and \( B = -5.775 \times 10^{-7} \) according to IEC 60751. At 100 °C it gives 138.51 Ω. Assuming a linear relationship with 0.385%/°C gives zero error at 0 and 100 °C, but almost 9 °C at 300 °C.

03Strain gauges and bridges

A strain gauge is a resistor that stretches with the part it is bonded to. Its relative change is proportional to the strain \( \varepsilon \), with a gauge factor close to 2. Strains are in thousandths, so the change in resistance is tiny, and it is measured in a bridge:

\[ \frac{\Delta R}{R} = GF\cdot\varepsilon \] \[ V_o \approx V_{exc}\,\frac{GF\,\varepsilon}{4} \] For a quarter bridge, with a single active gauge. With 1000 microstrain, a factor of 2 and 5 V of excitation, the output is 2.5 mV. With four active gauges, four times more, and the effect of temperature cancels.

Load cells and piezoresistive pressure sensors are full bridges specified in millivolts per volt of excitation: a 2 mV/V cell excited with 10 V gives 20 mV at full load. That is why they are measured ratiometrically: the converter uses the same excitation voltage as its reference, and its variations cancel.

04The instrumentation amplifier

The output of a bridge is a difference of millivolts sitting on a common voltage of several volts. An ordinary operational amplifier would also amplify that common voltage and the errors of its resistors. The three-op-amp instrumentation amplifier has a high-impedance differential input, a gain set with a single resistor and a common-mode rejection of 100 dB or more:

\[ G = 1 + \frac{2 R}{R_G} \] \( R \) are the internal resistors of the first stage. In integrated circuits the formula is given directly, for example \( G = 1 + 50\ \text{k}\Omega/R_G \).

Before the converter goes an anti-aliasing low-pass filter, and in industrial environments, an isolator that breaks the ground connection between the sensor and the system.

05Excitation, cables and self-heating

A two-wire Pt100 adds the resistance of the cables to that of the sensor: 10 m of thin cable has almost 1 Ω per conductor, and 1.7 Ω is equivalent to 4.5 °C of error. The three-wire connection compensates the cables in a bridge, assuming they are equal; the four-wire one carries the current through two and measures the voltage through the other two, with no current, and eliminates the error completely.

The excitation current heats the sensor, which then measures a little more than the ambient. With a self-heating coefficient of 0.4 °C/mW, 1 mA in a Pt100 produces 0.06 °C of error; 5 mA, 1.4 °C. More current gives more signal, but not a better measurement.

Lab · the chain of a Pt100

A Pt100 10 m from the equipment, with cable of 0.84 Ω per conductor, excited with a constant current. An instrumentation amplifier takes the 0 to 300 °C range to the 4.096 V of a 12-bit ADC. The graph shows the error of each source over the whole range; the dot, the chosen temperature.

06Other non-electrical parameters

QuantityUsual sensorOutput
Force and weightLoad cell with strain gaugesBridge of 1 to 3 mV/V
PressurePiezoresistive diaphragmBridge, or 4 to 20 mA already conditioned
PositionPotentiometer, LVDT, optical encoderVoltage, modulated AC or pulses
IlluminationPhotodiodeCurrent, with a transimpedance amplifier
HumidityCapacitiveCapacitance, or digital over I²C

In industry, many transmitters deliver a current of 4 to 20 mA: the current is not lost in long cables, and the zero at 4 mA makes it possible to detect a cut cable, which would give 0 mA.

07In the laboratory

Activity 1 · Pt100 with two, three and four wires

Measure a Pt100 in ice water and in boiling water with the three connections, through 10 m of cable. Calculate the error of each.

Activity 2 · A scale

Condition a load cell with an instrumentation amplifier and a 24-bit ADC, calibrate with two known weights and measure the repeatability.

Activity 3 · Cold junction

Measure with a type K thermocouple and an amplifier without compensation, and then with a cold-junction compensation IC. Compare with a reference thermometer.

08Common mistakes

  • Measuring a two-wire Pt100 with long cables. Each ohm of cable is 2.6 °C.
  • Raising the excitation to get more signal. Self-heating grows.
  • Forgetting the cold junction of a thermocouple. The reading is shifted by the temperature of the terminals.
  • Amplifying a bridge with a simple op amp. The common-mode voltage and the mismatched resistors add errors larger than the signal.

09Self-assessment

What resistance does a Pt100 have at 50 °C?

\( 100\,(1 + 0.195 - 0.00144) \approx 119.40\ \Omega \).

What output does a quarter bridge give with a gauge factor of 2, at 500 µε and 10 V of excitation?

\( 10 \cdot 2 \cdot 500 \cdot 10^{-6} / 4 = 2.5 \) mV.

What gain resistor gives G = 101 with the formula 1 + 50 kΩ / RG?

500 Ω.

Why do industrial transmitters use 4 to 20 mA and not 0 to 20 mA?

To distinguish the minimum of the scale from a cut cable, and so that the transmitter can be powered from the 4 mA of the loop.

10Further reading

  • Ramón Pallás Areny. Sensores y acondicionadores de señal (in Spanish). 4th ed., Marcombo, 2003. The reference in Spanish: each type of sensor with its conditioning circuit.
  • Jacob Fraden. Handbook of Modern Sensors. 5th ed., Springer, 2016. The physics of all sensors and their interfaces.
  • Walt Kester (ed.). Sensor Signal Conditioning. Analog Devices, 1999. Bridges, instrumentation amplifiers and temperature conditioning with real circuits.
Development of the topic “Signal conditioning and sensors” of Electronic Measurements I (Level 4), based on the curriculum of the UTN Electronic Engineering program, 2023 curriculum — Ordinance No. 1849 of the UTN Higher Council. Back to the Topic Map · catto.ar