Remote Three-Wire RTDs Can Multiplex Accurately To Measurement System
To maintain accuracy when multiplexing several three-wire resistive temperature detectors (RTDs) to a measurement system, multiplexer error voltages must be eliminated. These voltages are produced by the RTD excitation current flowing through the on resistance of the multiplexer. With a little forethought, such voltages can be eradicated. Because the RTDs are remote from the measurement system, designs must account for the errors caused by the ohmic drops along the long wires. The circuit shown here demonstrates how any number of three-wire RTDs, the most popular form of RTD, can be monitored without introducing more errors (see the figure).
The figure shows the measurement of eight RTDs. Each remote RTD to be measured has four single-pole, single-throw (SPST) CMOS switches associated with it: SW1a through SW1d for RTD#1, all the way up to SW8a through SW8d for RTD#8. The four switches associated with any single RTD are local to the analog-to-digital converter (ADC) and are either all open or all closed. The ADG711 contains four SPST switches per package. Therefore, one ADG711 package is dedicated to each RTD to be measured.
Only a single RTD can be measured at once, so each RTD must have an individual control line to operate its set of switches. To keep the figure as simple as possible, these control lines are not shown. However, all four switches inside the ADG711 have their control inputs tied together and are controlled by the measurement system's microprocessor. In addition, the RTDs are considered remote, so the wiring resistances between each remote RTD and the measurement system are represented by lumped resistances, RL1a, RL1b and RL1c for RTD#1, and so on.
The AD7783 24-bit, low-power, sigma-delta ADC includes two matched current sources—each of a nominal value of 200 µA—that can eliminate both switch and wiring induced errors. These current sources match each other to within 1% at 25°C and track each other very closely (typically better than 1 ppm/°C) over temperature. This level of matching would typically produce a 1% error in any reading, but the AD7783 allows the swapping of current sources to eliminate this error. Taking two readings of the RTD in succession while switching the excitation currents between measurements reduces this error source to zero. Input signal IPIN, not shown in the figure, controls the swapping of the two current sources.
It takes two conversions to produce an accurate measurement. For example, to measure the value of RTD#1, its control line closes all of the SW1 switches. The current sources are switched so that IEXC1 flows out of package pin IOUT1 and through SW1a to excite RTD#1. Compensating for the wiring losses incurred (IEXC1 * RL1a), a matched current source, IEXC2, flows out of package pin IOUT2 and through SW1d. As long as RL1a equals RL1b, the wiring losses will cancel out and not be seen by the measurement system.
To avoid seeing any losses locally across the switches (typically the on resistance is 2.5 Ω with a 5-V power supply), the measuring system uses SW1b and SW1c to monitor the RTD voltage on the remote side of the switches. The very high input-impedance buffer of the AD7783 precludes any current flow through these two monitoring switches. The voltage reference is ratiometric, as it's the product of (IEXC1 + IEXC2) and RREF. The conversion result is stored temporarily.
For the second conversion, the current sources are switched so IEXC1 flows out of IOUT2 and IEXC2 flows out of IOUT1. This second conversion result is added to the previous conversion result to provide a composite. This can be seen as a dimensionless fraction reflecting the ratio of the RTD value to the precision reference resistor, RREF, value.
With the input voltage range of the AD7783 set to unity and a precision reference resistor value of exactly 6.25 kΩ, an RTD#1 value of 800 Ω will produce a fraction equal to 0.128 and an RTD#2 value of 945 Ω will produce a fraction equal to 0.1512.