Dedicated impedance sensors with reduced influence of undesired physical effects

Transcription

1 Dedicated impedance sensors with reduced influence of undesired physical effects Gerard C.M. Meijer, Xiujun Li, Zu-Yao Chang and Blagoy P. Iliev Delft University of Technology (TUDelft), Delft Institute of Microsystems and Nanoelectronics (DIMES), Faculty EEMCS, Mekelweg 4, 2628 CD Delft, The Netherlands Tel.: Fax: Abstract This paper discusses the problems of impedance sensors in their application. Four case studies of industrial sensors are presented and show how a systematic approach can reduce the effects of undesired physical influences. It is shown that in a first case of ideal environmental conditions, with capacitive sensors even with simple interface electronics a very high precision can be obtained. A second case study shows that when the sensing elements have resistive leakage, the related problems can easily be solved by modifying the interface circuits. The third case study concerns capacitive detection of water content in soil. It is shown that the presence of salt in water creates measurement problems related to physical conditions and phenomena. With a special probe construction and dedicated electronics, the water content in soil with high salinity can be detected. However, because of physical effects such as skin and proximity effects, the water content is only measured for the local environment of the probe. Finally, the physical problems of a blood-impedance sensor are discussed. Also in this case, application of dedicated measurement techniques and a special probe construction enables to solve the problems. A number of experimental results are given. Keywords: Impedance sensors, capacitive sensors, smart sensor systems, impedance measurement interface 1. Introduction The complexity of impedance measurements is due to the complexity of the material to be characterized, the probe construction, and frequency and waveform of the excitation signals. Therefore, to optimize impedance measurements, dedicated setups and probes are required, and measurement techniques and excitation signals have to be designed for the best performance for specific applications. In this paper this is demonstrated for a variety of impedance (capacitive and resistive) sensors, in which capacitances represent the measurand and have to be measured in presence of disturbing parasitic resistors and capacitors. A case study for simple capacitive elements shows, that with simple electronic circuits, a very high performance can be obtained [1 2, 3]. However, three other case studies demonstrate that in many applications, undesired physical conditions can seriously degrade the accuracy. A systematic approach to such problems is required. Section 3 discusses such an approach for the case, that contamination and condensation cause parasitic resistances. It is shown that, in addition to the application of active guarding and two-port technique, rapid discharge methods can solve the related measurement problems [4]. Section 4 shows that, when the shunting resistances are very low, that modifying the probe construction and applying dedicated impedance electronics can yield a useful solution. Section 5 discusses the more complex problems of measuring the impedance of blood with a smart catheter. 2. Simple capacitive sensing elements Even with rather simple circuits, sensor capacitances in full-scale ranges from 2 pf to 300 pf can be measured with an accuracy of 14 bit in a measurement time of 100 ms. As an 4-297

2 Fig. 1. Basic circuit diagram of the UTI system for capacitive-sensor. example, Fig. 1 shows the interface circuit applied in the Universal Transducer Interface of Smartec [2]. The sensor capacitance C x is connected in the so-called two-port configuration, so that parasitic capacitances of the connecting wires do not influence the accuracy [1]. In this system, also autocalibration [1] is applied, which makes the system immune to additive and multiplicative parameters of the interface. The square-wave excitation signal for the sensor capacitor and the reference capacitor C ref is generated with a pair of switches, which are controlled by a relaxation oscillator in the system. The frequency of the excitation signal can be as low as a few tens of khz. Because of this, the power consumption of such an interface is rather moderate. The power consumption of the whole system, including the applied microcontroller, can be less than 100 mw. Amongst other applications, capacitive sensors are used, for instance, in positioning systems of optical lenses in wafer steppers. When these sensors are applied in the feedback loop of control systems, then for stability reasons, the measurement speed should be much higher than that of the UTI. In [3] it is shown that, at the cost of higher power dissipation (1 W), a resolution of 14 bits can be obtained in a measurement time of only 64 µs. In case of extreme precision requirements the capacitive sensor should be designed as an integral part of the positioning system. Usually, the environmental conditions in wafer steppers are tightly controlled. In many other industrial applications, this is not the case, which requires an adapted approach to cope with the resulting physical problems. This will be shown in the next two sections. 3. Leaky capacitive sensing elements The circuit of Fig. 1 has been designed for capacitive sensors that are free from leakage. Due to the low frequency of the excitation signal (about 10 to 50 khz), a shunting leakage conductance of, for instance, 1 μs can cause a considerable measurement error. In capacitive sensors for industrial application, such as humidity sensors, a main point of concern is that of parasitic resistances caused by contamination/condensation, and parasitic wiring capacitances. I ch C off 1 S 1 G s S 3 2 C int V int V ex S 2 C s C p1 C p2 1 S 4 Comp V comp 2 Capacitive sensor OPAMP Fig. 2. The schematic diagram of the interface, with improved immunity for shunting conductance

3 In this section, it is shown that application of the so-called two-port technique and rapid discharge methods can solve the related measurement problems. In [4], Li and Meijer present an interface circuit (Fig. 2), in which the effect of a shunting conductance G s is significantly reduced. This circuit is a modification to that of Fig. 1. The improvement is achieved by minimizing the DC voltage over G s and by transferring the charge of the sensor capacitance C s as fast as possible, so that no charge is lost. In [4] it is reported that, for a sensor capacitance of about 2 pf, a shunting admittance of 1 μs causes an error of only 0.4%. In some application the shunting conductance can be so large, that the method discussed above cannot help anymore. In that case, as will be shown in the next two sections, other methods are available to solve the related problems. 4. Capacitive detection in conductive material Using capacitive sensors for water-content detection in soil, a capacitance value C x (Fig. 3(a)) represents the measurand. Due to the unknown salinity of the water a large shunting conductance G x disturbs the measurement accuracy [5, 6, 7]. To reduce this effect, a high excitation frequency (20 MHz) has to be used. However, when using high-frequencies, skin and proximity effects introduce the problem that the current will flow along the surface and not anymore in the bulk of the material to be tested. To solve this problem, we designed a small probe (Fig. 3(b)), which enables local measurements in the neighbourhood of the probe electrodes [6]. With a series of such electrodes, an impression can be obtained of the watercontent profile over the depth of the material. Figure 3(c) shows a block diagram of the measurement interface, which is based on the principles applied in RLC measurement instruments [8]. In the block diagram of Fig. 3(c), two AD9951 chips from Analog Devices, designated DDS1 (Direct Digital Synthesizer) and DDS2, are shown. The output signal of DDS1 provides the 20MHz-sinusoidal excitation voltage for the impedance Z x. The output signal of DDS2 is a reference signal, that is transferred to the upper input of the AD8302 Gain-Phase chip. The voltages U z and U i across Z x and R i, respectively, are multiplexed amplified and transferred to the lower input of the AD8302 Gain-Phase chip. From the amplitude ratios U DDS2 /U z and U DDS2 /U i and the phase differences of the input signals of the AD8302 Gain-Phase chip, the values of the unknown Z x with impedance Z and phase θ can be derived. Next, from these measurements, the measurand C x and its shunt conductor G x can be found [7]. With this interface system, water-content can be derived from the capacitive impedance component, even in the case that the conductivity of the water is high and not well-known. Its performance has highly been improved with an added calibration procedure of the gain-phase detector and using so-called open/short compensation. Experiments with the presented measurement interface, show that, when measuring a capacitance C x at a full-scale range of 30 pf, shunting resistors of 820 Ω and 47 Ω cause full-scale errors which amount to only 1% and 1.5%, respectively. R par C par L par imp. analyser two electrodes G x C x sharp tip to cut into soil (a) (b) (c) Fig. 3. (a) Electrical model for a probe in salty water. (b) Probe with segmented electrodes. (c) Basic structure of the measurement interface

4 5. Impedance sensors for sensing blood viscosity The most complex impedance sensor dealt with in this paper concerns a characterisation system for blood impedance. This system is applied in a smart catheter for in-vivo characterisation of blood-impedance for various frequencies from 20 khz to 1.2 MHz. From the measurement results, important blood properties, such as blood viscosity, hematocrit and other parameters can be extracted [9]. The physical problems of the measurements are related to the non-newtonian behaviour and the emulsion character of blood. Moreover, the parasitic capacitances of connecting wires and the formation of bio-layers create challenging measurement problems. Blood is a suspension of red cells, white cells, and platelets in plasma. A simplified three-element circuit model describes its properties (Fig. 4) [9, 10]. The concentration of red cells is called hematocrit H t and in this electrical structure is represented by R p. The resistor R i represents the cell interior resistance, while C m is a measure for the cell membrane capacitance. The total complex impedance measured by a sensor includes the polarization effect of the electrodes ( ) too [9]. Though R p is an accurate measure of H t and therefore good indicator for the viscosity, it does not reflect fully viscosity because of the non-newtonian characteristics of blood. The non-newtonian behavior of blood causes a higher viscosity at lower shear rate. The flow-dependency of the shape of red-blood cell also affects the values of impedance components and the correlation between those values and blood viscosity. The best correlation is found at relative low shear rates. For this reason the measurements are synchronized with the T-wave in the ECG signal. Taking these effects into account, the value of R p can be considered to be a measure for H t while the combination of R p and C m is an indirect measure for the viscosity. Figure 4(c) shows the block diagram of the A B (a) R p R C i m C D A1 (b) Us Rc A3 Triax1 Sensors Triax2 F1 A4 Triax3 A5 DDS Coax1 Rt A2 A6 A7 (c) Uz F2 F3 Log Ampl Log Ampl Phase Temperature ECG Signal A/D Phase (Uphs) Fig. 4. (a) Three-element model of blood including electrode polarization impedance, (b) Catheter for in-vivo characterization of blood in the right atrium, (c) Block diagram of the HemoCard Vision interface electronics, (Courtesy of Martil Instruments). Gain (Ug) upc RS232 PC/PDA 4-300

5 interface electronics [9]. The surface roughness of the sensor electrodes appeared to be critical for the biolayer growth. By a special preparation of the surface, imperfections have been eliminated and a biolayer growth completely avoided. 6. Conclusion This paper shows that impedance sensors have to be designed as overall systems in their physical environment. A study of the physical environment should result in a characterization of sensing elements and their parasitics, which magnitudes should be taken into account when designing the interface system. For simple capacitive elements a very high accuracy (16 bits for a full range of 2 pf and a measurement time of tens of ms) can be obtained with lowpower interface circuits in which powerful measurement techniques, such as autocalibration, chopping and two-port techniques have been applied. When leakage resistance of the sensor capacitance degrades the system performance, modified circuits can still yield an acceptable accuracy. However, when the shunting conductance dominates in the sensor admittance, then more complex measurement techniques have to be applied. This has been demonstrated for two case studies. A first case study concerns a water-content sensor system, in which the capacitive impedance component represents the water content of artificial soil. A sinusoidal signal with a frequency of about 20 MHz has been applied. Because of skin and proximity effects, special impedance probes are required, which spatial range is limited to the direct environment. With such probes, water-content can be measured for conductivity up to 2 S/m. A second case-study concerns an in-vivo monitoring system for blood impedance. It is shown, that for characterization of suspensions, such as blood, requires a rather complex measurement approach, using a range of signal frequencies and synchronization with the ECG signal. Such a system can be used to monitor blood viscosity, in-vivo. References 1. G.C.M. Meijer. Interface Electronics and Measurement Techniques for Smart Sensor Systems. In: G.C.M. Meijer (edt.) Smart Sensor Systems. Wiley, Chichester, UK Smartec, data sheet Universal Sensor Interface UTI R. Nojdelov, S. Nihtianov. A Fast Interface for Low-value Capacitive Sensors with Improved Accuracy. Proc. IEEE I2MTC 8, 2008, Vancouver, Canada. pp X. Li, G.C.M. Meijer. An accurate interface for capacitive sensors, IEEE Trans. Instrum. Meas. 2002, vol. 51, No B.P. Iliev, G.C.M. Meijer. An impedance-measurement system for electrical characterization of rockwool substrates. Proc. ISA/IEEE SICON , pp Z. Chang, B.P. Iliev, J.F. de Groot, G.C.M. Meijer. Extending the limits of a capacitive soil-water-content measurement. IEEE Trans. Instr. Meas. Dec. 2007, pp Z. Chang, B.P. Iliev, G.C.M. Meijer. A smart interface system for measuring the impedance (C x, R x ) of soil at a signal frequency of 20 MHz. Proc. IEEE Sensors. Atlanta, USA. Oct. 2007, pp A. Bate. Modern impedance measurement techniques. Electronics World. December B.P. Iliev, G.A.M. Pop, G.C.M. Meijer. In-vivo blood characterization system. Proc. IMTC2006. Sorento, Italy. 2006, pp Zu-Yao Chang, G.A.M. Pop, G.C.M. Meijer. A Comparison of Two- and Four-Electrode Techniques to Characterize Blood Impedance for the Frequency Range of 100 Hz to 100 MHz. IEEE Transactions on Biomedical Engineering. March 2008, vol. 55, issue 3, pp