IEEE Instrumentation and Measurement Technology Conference Budapest, Hungary, May 21-23, 2001. A Novel Sensor–Bridge-to-Microcontroller Interface A. Custodio 1,2 , R. Bragós 2 , R. Pallàs-Areny 2 1 Dept. de Electrónica, UNEXPO “Antonio José de Sucre” Puerto Ordaz, Final Calle China, Alta Vista Sur, Edif. Ing. Eléctrica-Electrónica, Ciudad Guayana, Venezuela E-mail: custodio@eel.upc.es, custodio2000@terra.es 2 Divisió d’Instrumentació i Bioenginyeria – Dept. d’Enginyeria Electrònica, Universitat Politècnica de Catalunya C/Jordi Girona 1-3, Edifici C4, 08034 Barcelona, Spain Phone +34 93 4016766, Fax +34 93 4016756 E-mail: elerpa@eel.upc.es Abstract – Sensor bridges are usually interfaced to microcontrol- lers by supplying the bridge with a voltage or current and digitizing the resulting voltage or current after being amplified and low-pass filtered. This paper proposes an alternative method to interface a sensor bridge to a microcontroller that does not need any active component between the bridge and the microcontroller. The bridge is considered a network with three inputs and one output. The resistance of each input to the output depends on the measurand. Using each input in turn to charge a capacitor connected to the bridge output yields three different time intervals. For a full bridge (a sensor at each arm), the ratio between the difference between two time intervals and the third time interval yields the fractional resistance change. Two-point calibration reduces zero and gain errors attributable to the electrical parameters of the ports of the microcontroller. The absolute error for a 15 psi (103.4 kPa) pres- sure sensor with 5000 Ω Ω Ω arms and a full-scale output of 125 mV is below 0.05 % of full scale, which is better than 1 LSB for an 11 bit ADC. Keywords – Sensor bridge, microcontroller interfaces, pressure sensor, piezoresistive sensor, two-point calibration. I. INTRODUCTION Sensor bridges are very useful signal conditioners because they can cancel interference such as temperature. Further- more, bridges with two or four active arms have increased sensitivity and are more linear than bridges with a single active arm. Piezoresistive strain gages, for example, are usu- ally arranged in Wheatstone bridges in pressure sensors and load cells [1]. The common method to interface a sensor bridge to a micro- controller is by amplifying the output voltage or current when the bridge is supplied by either a voltage or current, and ap- plying the resulting voltage to an analog-to-digital converter ADC (Fig. 1). In order to achieve a high accuracy, the ampli- fier, antialiasing filter, and ADC must have minimal errors, which requires expensive components. If several sensor bridges share a single ADC, the added multiplexer must have small errors too. Applications requiring a resolution below 10 bits can use commodity analog components but the several integrated circuits needed take printed circuit board area, increase power consumption, and reduce reliability. Because most of the cost of common interfaces is often attributable to the ADC, some low-cost interfaces rely on embedding the information in a time interval rather than in an analog voltage or current subsequently digitized by an ADC. Fig. 1. Classic approach to interface a sensor bridge to a microcontroller: the voltage output is amplified and digitized by an ADC. Fig. 2, for example, shows a circuit that does not need any ADC [2]. The RC network, op amp and comparator (internal or external to the microcontroller) perform the analog-to- digital conversion. The bridge is supplied by a constant volt- age. The analog multiplexer switches the two bridge outputs to a voltage buffer in order to charge a capacitor C 1 through R 1 . The voltage buffer isolates R 1 from the multiplexer and bridge resistance. When the voltage across C 1 becomes larger than that across C 0 (V r /2), the comparator is triggered low, which drives the RA3 port from high to low and, conse- quently, discharges C 1 through R 2 . Once the voltage across C 1 decreases below that across C 0 , the comparator is triggered high, RA3 is set high, and the cycle repeats. The time between transitions is proportional to the bridge output voltage. A counter measures the time it takes to per- form a given number of transitions (say 1024), which are tracked by another counter. After reaching the predetermined number of transitions, the microcontroller switches the multi- plexer to the other bridge output terminal. The elapsed time needed for the same number of transitions to happen is counted, and the subtraction of the respective counting times for each bridge output yields the conversion result. V R 0(1- x) R0(1- x) R0(1+ x) R0(1+ x) IA ADC μC LPF - 0-7803-6646-8/01/$10.00 ©2001 IEEE