Back in 2012, I wrote about signal conditioning for MEMS and sensors. Amplifiers and architectures have not changed very much since, but there are a few newer devices and design architectural techniques that may be better for MEMS and sensor conditioning in some design cases. I will be including not only analog operational amplifier solutions in this article, but discrete transistor, data converter, microcontroller, and algorithm-based solutions as well. The following are some key applications that use these different architectural sensor conditioning techniques and some of the latest products in interesting topic areas.

Medical: Monitoring sounds in our bodies1

Monitoring sounds in the human body is still an important means of medical diagnosis for doctors. Back in the 19th century, auscultatory1 (from Latin auscultatus , “to listen attentively to”) technology was first used in stethoscopes. Technology has advanced beyond a microphone-type of sensor to piezoelectric sensors in modern day technology.

Most of today’s electronic stethoscopes are designed with a set of configurable filters that have a different frequency response. These filters enable better listening in various areas of the human body, such as the heart (20Hz to 400Hz range), joints, intestines, or the lungs (100Hz to 1200Hz range). Most of these filters are designed as a band-pass with a tunable cut-off frequency. Noise reduction algorithms are frequently employed to reduce interference such as patient movement or ambient noise. Maxim Integrated has a nice block diagram of an electronic stethoscope (Figure 1 ).

Figure 1 Block diagram of a digital stethoscope (Image courtesy of Maxim Integrated)

Stethoscopes can also have mechanically-tunable diaphragms to condition the signal. See this 3M Littmann website and video.

Medical: Wireless ECG for mobile health monitoring2

A wireless wearable device, addressed in Reference 2, is able to measure electrocardiogram (ECG) and respiration rate (RR) through the use of non-contact capacitive-based electrodes. A good analog front-end design is the key architecture element in this design that does the signal conditioning and produces a strong, clean output. Embedded into a wearable vest, active electrodes come in contact with the subject’s chest; a reference electrode is placed directly on the subject’s skin. That reference electrode relays a common-mode input signal back to the subject’s skin, which is the system ground in this architecture. Once the signal has been acquired by the electrodes it is then sent to a differential separation filter (DSF). The DSF is responsible for separating the differential signal into two main signal components:

The entire signal conditioning circuit can be seen in Figure 2 . See Reference 2 for more details about this circuit and the gain and component values used in it.

Figure 2 Analog front end block diagram (Image courtesy of Reference 2)

The instrumentation amplifiers (IAs) in Figure 2 will now begin extracting the ECG signal. High common mode rejection ratio (CMRR) IAs will reject common-mode signals between the two electrodes, thus removing induced noise due to contact or AC interference. INA121s from Texas Instruments (formerly Burr-Brown devices) are FET input devices with high impedance and will amplify the tiny ECG signals. The majority of the system gain will be via the INA121 in order to maximize the CMRR right at the input of the conditioning circuit.

Next, the second-order active low-pass filter (LPF) for the ECG signal has a 100 Hz relatively steep cutoff, which is configured as a Sallen-Key KRC architecture. This is followed by a non-inverting gain of 2 stage and then the necessary anti-aliasing filter in front of the ADC. More details about this AFE and the respiration rate, with similar signal conditioning, and the differential separation filter can be found in Reference 2.

Signal conditioning with chopper amplifiers for MEMS transducers3,5

Capacitive sensing, using MEMS devices, has the advantage of low power consumption, good noise performance and low temperature coefficient. For high resolution sensitivity applications, the sensitivity needed can be less than 1 fF for a MEMS transducer having a nominal capacitance smaller than 100 fF. Such transducers are integrated with the sensing circuitry and thus require a low capacitive load. Signal conditioning is necessary and must be very accurate with low noise. This infers that the input capacitance and noise voltage need to be low as well. In an inertial sensor application, a high sensitivity is necessary at frequencies close to DC. For this reason, chopper amplifiers may be the best choice to remove flicker noise as well as DC offsets.

Dual chopper amplifiers (DCA) will have lower parasitic capacitance and less power consumption than a single chopper amplifier3 . It has been determined that both input noise as well as input parasitic capacitance will affect sensitivity.

In this design, there are two different amplifiers (A2) that are chopped with two different frequencies to remove their flicker noise. In addition, designers added a low-voltage, high-current amplifier (A1) in the first stage. This addition will improve the power consumption and noise floor of the architecture.

The second stage is designed with two parallel paths which improve SNR and provide two gain settings (Figure 3 ).

Figure 3 Schematic of the proposed circuit in Reference 5 where Rbias resistors are pseudo-resistors which achieve higher resistance values (Image courtesy of Reference 5)

The sensor input stage in Figure 3 will support both capacitive and voltage modes. In capacitive mode, the MEMS transducer, CS , is connected to a capacitive bridge which is part of the chopper amplifier circuit. The transducer offsets are removed via a capacitor bank applied to one of the caps in the bridge to match the nominal sensor capacitance.

In voltage mode, the voltage mode mixer is connected to pins IN3 and IN4 while the voltage mode mixer gets enabled to chop the voltage signal. The capacitive path transmission gate disconnects it and the capacitive bridge clocks are disabled.

Figure 3 shows the schematic of the two differential-input pairs going into an adder and on to the Gm-C low-pass filter that adds the signals from the two paths and also removes out-of-band frequency distortion after the signal is chopped to the baseband. Capacitances Cf are 15 pF. See Reference 5 for more details.

Measure resistance6 in a unique signal conditioning environment

Here we will present a signal conditioning method to measure resistance and ultimately the conductivity of water in an insulating tube via capacitively coupled electrodes. Since the electrodes are electrically insulated from the water, measurements are achieved via capacitances formed between the water column and electrode.

This technique solves the electrode contamination and polarization related to a conventional contact-based method of conductivity to measure the water. A major challenge here is the large reactance of the coupling capacitors as compared to the resistance of the water column being measured. The coupling capacitor will also vary over time as well causing another challenge.

Reference 6 demonstrates an auto-balancing signal conditioning method to overcome these challenges in which an output is provided that is directly proportional to the resistance being measured and will be independent of the value of the coupling capacitors.

The measurement probe must have a dielectric coating with two circular metal electrodes (an excitation electrode and a receiver electrode). When the excitation electrode is excited, current will flow from it to the water and that current will be collected by the receiver electrode. If used for conductivity measurement of sea water the entire probe in Figure 4 will be surrounded by sea water and an insulator over the electrode will help avoid direct contact of the electrode with the water surrounding it.

Figure 4 The sensor probe: (a) The conductivity measurement probe side view and (b) The conductivity measurement probe cross-sectional view (Image courtesy of Reference 6)

Figure 4 shows a dielectric present between the excitation electrode and the water column. So, a capacitance Cx 1 is formed between the electrode and the outer surface of the water. Similarly, there is a capacitance formed between the receiver electrode and water, which is indicated as Cx 2 in Figure 5 . The resistance of the water column in between the excitation and receiver electrodes is represented as Rx in the figure. The goal is to create a method that measures Rx and at the same time will not be affected by the values of Cx1 and Cx2 .

Figure 5 An electrical equivalent circuit representation of thesensor probe (Image courtesy of Reference 6)

In Figure 6 , vin is a sinusoidal voltage sourcefeeding into twocurrent-to-voltage converter circuits made up of op amps OA1 and OA2 . vin drives bothcurrent-to-voltage converters, one of which is composed of op amp OA1 whose feedback path consists of the capacitively-coupled conductive measurement probe. vin then also drives the second current-to-voltage converter created by op amp OA2 whose feedback path is composed of a standard capacitance Cs through a voltage-controlled amplifier (VCA) having gain G . The outputs of these two current-to-voltage converters are connected to high pass filters. The outputs of those high pass filters are summed into an inverting summer made from op amp OA3 .

The output of op amp OA3 feeds into the input of a phase detection circuit (PDC) made from comparators OC1 , OC2 , an XOR gate, an SPDT switch, and a Sallen-Key second order low pass filter created by op amp OA4 . The reference input of the PDC is vin to OC2 . The output of the PDC controls the G of the VCA through the integrator implemented with op amp OA5 with its input resistance R8 and feedback capacitance C2 . The current I 01 through the capacitively coupled measurement probe is = v in /R 1A where v in = V m Sin(ωt ). See Figure 6 and see Reference 6 for more design details.

Figure 6 The diagram of an auto-balancing, signal conditioning circuit for a capacitively-coupled measurement probe (Image courtesy of Reference 6)

Latest industry component and software solutions

STMicro recently introduced the TSB712A, a new precision op amp that has very stable parameters over a broad voltage and temperature range. To aid in sensor signal conditioning, this device has a built-in input filter to ensure improved EMI rejection ratio (EMIRR) over a broad frequency range which improves susceptibility in high noise environments such as industrial and automotive applications or systems close to RF equipment. EMIRR, describes the EMI immunity of operational amplifiers. An adverse effect that is common to many op-amps is a change in the offset voltage as a result of RF signal rectification and is defined in the following equation:

EMIRR= 20log(Vin pp /ΔVio )

STMicro also has an eDesign Suite in which you can do some simulation for signal conditioning using STMicro products.

There are so many designs that can use discrete components such as transistors and FETs in signal conditioning. Look at this neat and simple design that adds a low-capacitance JFET to an op amp, which will lower the input capacitance as well as reduce noise in the circuit (Figure 7 ).

Figure 7 A low capacitance JFET added to an op amp circuit to lower input capacitance and reduce noise4 (Image courtesy of Reference 4)

High-impedance input, buffered A to D converter simplifies signal conditioning7

Analog Devices has the LTC2358-18, an 18 bit, 200 kHz, low noise, buffered 8-channel input with simultaneous sampling A to D converter. To assist in signal conditioning of MEMS and sensor inputs, the integrated picoamp-input analog buffers, wide input common mode range, and 128 dB CMRR help minimize the need for external signal conditioning or may not even need it at all in some designs.

The diodes between the inputs and the VCC and VEE supplies provide the necessary ESD protection at the ADC input. This eliminates the need to use external op amp buffers that often have diode protection that turns on during transients, can corrupt the voltage on any filter capacitors at their inputs.

The very high input impedance of the ADC integrated unity gain buffers greatly reduces the input drive requirements and allows designers to include optional RC filters with kΩ impedance and arbitrarily slow time constants for anti-aliasing. Micropower op amps with limited drive capability are also well suited to drive the high impedance analog inputs directly.

A single-ended input drive also adds additional external crosstalk isolation since every other input pin is grounded, or at a low impedance DC source, and serves as a shield between channels. Another nice signal conditioning feature of this ADC is capability to drive an analog input signal above VCC on any channel up to 10mA without affecting conversion results on other channels.

The ADC’s true high impedance analog inputs can accommodate a very wide range of passive or active signal conditioning filters. The buffered ADC inputs have an analog bandwidth of 6 MHz and need no particular bandwidth requirement on external filters. The external input filters can be optimized independent of the ADC to reduce signal chain noise and interference. A common filter configuration is the simple anti-aliasing and noise reducing RC filter with its pole at half the sampling frequency. For example, 100 kHz with R=2.43 kΩ and C=680 pF as shown in Figure 8 .

Figure 8 Filtering single-ended inputs (Image courtesy of Analog Devices)

Mixed-signal AFE and microcontroller for electrochemical and biosensors9

My final observation of a semiconductor solution for MEMS and sensor signal conditioning is the Analog Devices ADuCM355. This IC is a precision analog microcontroller with bio-sensor and chemical sensor interface for such applications as industrial gas sensing, instrumentation, vital signs monitoring, and disease management. 

This solution is the only solution available today that incorporates a dual potentiostat8 and Electrochemical Impedance Spectroscopy (EIS)10 functionality on a single chip. This solution also supports dual potentiostats and more than three sensor electrodes.

Other integrated signal conditioning features are:

See these videos on Analog Devices’ site for more details.

So, you can see that there are a great many signal conditioning techniques and options to enhance the performance of a MEMS and sensor input for your designs. I know that we will continue to see many more innovative ways to enhance signal conditioning in the future and we, at EDN, will continue bring those technical ideas to you in order to meet your design needs.

For more in-depth information on applying the Raspberry Pi in commercial applications, check out these other articles in this AspenCore Special Project:

You must Sign in or Register to post a comment.