Resistance thermometers, also called resistance temperature detectors (RTDs), are very common sensors used for temperature measurement. Their reliability, ruggedness, wide range, and value make them popular in the process industry and research laboratories alike. A wide range of instruments, PLC I/O systems, data acquisition, and control systems are designed to interface with these sensors and perform actions based on their measured value. Testing of such systems requires a sensor to be simulated, where hardware-in-the-loop testing is mandatory. Often the validation of such systems will end up being a complex issue if not properly planned.
An RTD element is essentially a variable resistor whose output is based on its temperature. In order to read this resistance, a constant amount of current is pumped into the circuit by the RTD signal conditioner, and the resulting voltage is read back. The amount of this test current varies between 100μ A – 1mA. Larger test currents result in self-heating of the sensor element (by the virtue of I2R law), and hence corresponding error. At the same time, too low a test current will generate weak voltages across the sensor element, which can get easily corrupted by electrical noise. Hence, the users will choose the test current optimized for a particular application. In some cases, a pulsating type excitation is also used, where short pulses of current are pumped into the sensing element, and the output is logged within the interval of the pulse.
Classical methods of sensor simulation use a simple variable decade resistance box. The output resistance is controlled manually, as per the requirement of the system. This type of arrangement may be sufficient for small laboratory experiments, but not suitable for automated test procedures. This type of simulator suffers from coarse resolution, temperature instability, poor repeatability, and limitations associated with lack of programmability.
Digitally controlled potentiometers offer programmability to some extent, but still have many of the limitations of their analog cousins, including reduced accuracy, poor repeatability, and self-heating. In addition, many of them do not support the test current ranges used in the process industry. Their control is cumbersome, since the programmer has to calculate the equivalent resistance of the element at every temperature point. These calculations are often complex since the resistance vs. temperature curves are not only nonlinear, but also vary slightly based on standards.
Another popular method uses a string of resistors switched in and out of the circuit in a binary weighing method to produce the required resistance value. The least value of the resistor arm used in this string decides the resolution of such circuit. Issues associated with this approach include resolution/granularity in which a Pt-100 type RTD sensor (arguably the most common type) changes its resistance by approximately 0.394 ohms per every °C change in temperature. In order to faithfully simulate temperature with a resolution of 0.5 °C, the simulation system should be capable of faithfully reproducing ~200m ohms of resistance change. Achieving this using switch resistor arms is difficult, not only since it is difficult to get an accurate resistance at that small value, but also due to individual contact resistances of switching relays, which vary between 0.1Ω to 1Ω. Since the number of relays shunting the arms varies based on the absolute value of simulated resistance, it is difficult to reproduce ΔR of 200mΩ.
Due to the number of individual resistors used, the channel density of such systems is drastically reduced. They are not only expensive (per channel cost), but also support very few types of RTD simulation. Thus, multichannel modules have limitations in the range of temperatures they simulate.
Since the tolerance of resistors used in the switching circuit can’t be tightly matched and controlled, it is typically difficult to achieve accuracy values better than 1 °C. Aging of resistors, temperature coefficient of resistance, and contact resistance variations of shunting relays affect the stability and repeatability of switch resistance ladder type circuits. The switched resistance ladder type circuits also suffer from glitches that are generated due to chattering of various shunt relay contacts before settling to a new value. This chattering causes resistance value to jump wildly and could appear as “thermal shocks” to the device under test, and thus cause false triggers.
Due to the coarse resolution and chattering of relay contacts, it is almost impossible to generate smooth ramps when using switched resistance ladder type circuits. Insertion of large capacitors between the simulation terminals could arrest the glitches to some extent; however, this will affect the settling time and response time of the simulator. Since there are no programmable components in the switched resistance ladder type circuits, they must be calibrated externally, and the calibration coefficient must be handled in the application test software. This could increase the programming complexity and load of application software development.
The latest technology in electronics and control systems helps resolve the issues associated with switched resistance ladder type circuits by using a different technique. In this method, the resistance is simulated by generating a controlled opposite voltage against the user supplied current. RTD simulators implement an advanced, solid-state servo mechanism to actively simulate resistance values. Each channel continuously measures the test current and generates a voltage opposing this flow of current across a MOSFET.
This method produces “bounce-less” changes in resistance and smooth ramps at the required rate of change. Aging and ambient temperature variation effects are nullified, as the resistance value is continuously monitored in a closed-loop and corrected for errors. The onboard processor calculates the engineering values into equivalent resistance values, on the fly. This also reduces the burden on the application software. Since this intelligence is embedded into the card itself, it can also handle the calibration, like any other smart instrument.
This work was done by VTI Instruments Corporation, Irvine, CA. For more information, Click Here.