Proceedings of the 2015 2nd International

Symposium on Physics and Technology of Sensors,

8-10th March, 2015, Pune, India

Design and Simulation of Readout Circuit for

Mems Sensor
Bhanuteja G *Veda Sandeep Nagaraja
Dept. of Electronics and Communication Engineering, Dept. of Electronics and Communication Engineering,
Nitte Meenakshi Institute of Technology, Nitte Meenakshi Institute of Technology,
Bengaluru -560064 Bengaluru -560064.
[email protected] [email protected]

Sowmya J S L Pinjare
Dept. of Electronics and Communication Engineering, School of Electronics and Communication Engineering,
Nitte Meenakshi Institute of Technology, Reva University,
Bengaluru -560064 Bengaluru-560064.

Abstract— Output from any MEMS Sensor usually lacks the technology, etc. cannot be fulfilled using conventional
power that can be fed directly to any interface circuit. Hence sensors.
there is a need to add additional circuitry in between the output
of the MEMS sensor and the interface circuit. In this paper the Sensors by themselves cannot be interfaced to the real
authors have compared two combinations of modulator / world; hence there is requirement for readout circuitry for
demodulator circuitry which can be used in the readout circuitry MEMS sensors. Most of the MEMS sensors work on the
for any MEMS sensors. The two combinations are Gilbert principle of capacitive sensing where a small capacitance
multiplier as modulator and demodulator and the second one is change of the order of Femto to Atto farads are sensed by the
Gilbert Multiplier as Modulator and Chopper circuit as interface electronics.
demodulator. A detailed analysis of the two combinations is
presented in the paper. The simulations were carried out using The goal of the readout circuitry is to sense this
gpdk 350nm technology file in Cadence tool. The layout has been capacitance change and convert it into an equivalent voltage.
generated for each block and DRC and back annotation has been The capacitance change produced by the MEMS sensor with
done to note overall circuit delay as 118ns. respect to the physical variable being sensed, is very low.
Hence there is a need to develop signal conditioning circuit to
Keywords—MEMS Sensors, Readout interface, Modulation convert finer capacitance variations to a measurable voltage
Demodulation schemes, Gilbert cell mixer, Chopper, Impedance signal. Numerous methodologies are available to tap the
matching. output from sensors like Resonance method, Oscillator
method, Charge discharge method, Modulation Demodulation
I. INTRODUCTION method [1] etc.
MEMS or Micro-Electro-Mechanical Systems, is
composition of an electro-mechanical and miniaturized In this paper Modulation Demodulation method is used for
mechanical elements i.e. devices and structures that are made developing the readout circuitry, as this method is suitable for
using the micro fabrication techniques. These MEMS devices both high resolution and high bandwidth. In reference [2] the
or structures generally range from 0.02 to 1.0 mm in size. authors have used this technique to sense the acceleration of a
They consists of a central unit that processes data, the MEMS accelerometer. The ASIC based on CTV sensing
microprocessor and several other components which interact methodology is robust to effects of parasitic present in system
with the surroundings such as micro actuators and micro and exhibits a static sensitivity of 30mV/fF for a gain of 10.
sensors. Also the sensitivity exhibited by MS3110 ASIC was 30mV/g
and the sensor integrated by this ASIC was able to detect
A sensor can detect the events or changes in quantities and vibrations as low as 50milli-g. It was found in reference [2]
provides a corresponding output, generally as an electrical or that there is an advantage of using the Modulation-
optical signal. MEMS sensors like gyroscopes, microphones Demodulation methodology over other methods.
have wide variety of applications in automotive, consumers,
industrial etc. Design and fabrication of these sensors with The above method required to modulate the signal using a
high sensitivity and accuracy is a major challenge. Sensors robust modulator circuit. In reference [3] the authors use
made using MEMS are better than conventional sensors due to Gilbert multiplier as a double balanced differential
their: Smaller size (20 micrometres to a millimetre); Lower implementation, which eliminates the Lo to IF and RF top IF
Power consumption; Much greater sensitivity to input feed through. The conversion gain was found to be 9.577dB
variations; Economical due to mass production; Newer after simulation and power gain of 1.96dB. Gilbert as a mixer
applications like in Ear-implants (cochlear implants), space can be used for modulation as well as demodulation purposes
which will be dealt with in detail in this paper.

978-1-4673-8018-8/15/$31.00 ©2015 IEEE 264

Authorized licensed use limited to: Nitte Meenakshi Institute of Technology. Downloaded on April 02,2024 at 09:01:38 UTC from IEEE Xplore. Restrictions apply.
 Having modulated the signal the next stage in the readout The resulting multiplied signal will be,
circuitry is the demodulator circuit. In reference [4] Chopper (3)
is used as demodulator and the authors have used Dickson Where, ω1 and ω2 are frequencies and φ1 and φ2 are phase
type DC-DC voltage converter, which made the condenser components of input signals.
microphone extremely sensitive at remarkably low supply
The gilbert cell mixer is a double-balanced mixer, which
(bias) voltages. The ENL produced was 26-28 dB (A). SPL for
are preferred over single balanced implementations since it
supply voltages of 1.9-2.0 V with no bias voltage applied to
the microphone. From literature it has been found that eliminates LO-to-IF, RF-to-IF feed through which will
Chopper as a demodulator is also a viable option to be used a increase linearity. The Gilbert cell multiplier as modulator is
demodulator circuit in the sensor readout circuit. But there are designed to meet the following specifications
several issues that need to be considered while choosing an Vdd=1.8V, Vss= -1.8V, Input frequency range: 100 kHz,
interface circuitry to be used with MEMS sensor. Clock pulse: 1MHz
One of the issues to be considered is to improve the Assuming all transistors are in saturation region, the drain
sensitivity of the sensor. In [5] the authors have developed a current equation can be written as,
circuitry where in a charge amplifier is used for minimizing ID= 1/2μCox W/L (Vgs- Vtn)2 (4)
the sensitivity losses. But the tradeoff of this method is that it For pull up transistors we have,
increases the noise component in the circuitry. To overcome gm= {2μCox(W/L)Iss}1/2 (5)
this problem the authors have suggested the use of voltage Where,
amplifiers. gm= transconductance, μ= mobility constant, Cox= oxide
Noise component can also be minimised if a balanced thickness, W= width of the device, L= length of the device,
configuration circuit is used as in the case of a differential Iss= Tail current,Vtn= threshold voltage of nmos transistor.
pair. In reference [6], a differential CMOS switch chopper From the gpdk 180nm model file, the following parameters
circuit with low noise, charge injection and clock feed through values were obtained,
for applications in CMOS-MEMS sensors is successfully μ=400 cm2, Tox= 4 × 10-9m, εo= 8.854 ×10-12F/m
demonstrated. The operation of the circuit and its lower noise Hence is obtained as
and offset performance is confirmed with Cadence Cox = 3.9 εo/ Tox = 8.637 ×10-7 F/m2
simulations. The designed circuit is suitable for chopper For the Gilbert multiplier to work as mixer the first six
amplifier blocks implementations, requiring higher chopping transistors must be of same aspect ratios and the current is
frequencies. divided equally in each branch.The tail current can be found
With these options in mind the authors of this paper have from equation (6)
presented a comparison of the four different circuitry for (6)
demodulation purposes. The second section of the paper
presents the design methodology of the Gilbert Cell as a gm is calculated by assuming the voltage gain (Ac) in linear
mixer, followed by the third section which gives all the scale as approximately 2.
simulation results. Towards the end the authors have (7)
concluded the paper with the analysis of their results followed Where, R= output impedance.
by the future scope. Using (6) and (7) in (5), the aspect ratio of pull up transistors
are found. The device aspect ratios are shown in table 1.
II. DESIGN METHODOLOGY
In order to design the readout interface we need to know Table 1: Aspect Ratio of Gilbert Cell Circuitry
the range of the sensing values which will be the output of the Transistors aspect ratios
various MEMS sensors. For this paper the authors considered Transistor name Width(μm) Length(nm)
a set of a clock pulses and sensed signal from MEMS sensors M1 4.2 350
M2 0.70 350
in the form of pure sine wave which is the output of an AC
M3 0.70 350
bridge network. M4 1.4 350
This paper is carried out for an application, where in the M5 1.4 350
M6 1.4 350
signal has to be modulated to protect the low frequency sensed
M7 1.4 350
signal and demodulated back. The Gilbert cell multiplier is M8 3.5 350
generally a mixer which mixes two frequencies to produce an M9 3.5 350
intermediate frequency [7]. This mixer can up convert or The demodulated signal is filtered using low pass filters in
down convert a signal to higher or lower frequencies order to remove high frequency components. For this purpose
respectively. RC filter is used with the cut off frequency calculated using,
The multiplication process starts by multiplying two signals,
(8)
(1) Where, C= filter capacitance. Fig.1 shows the schematic of
(2) designed Gilbert mixer as modulator.

265
Authorized licensed use limited to: Nitte Meenakshi Institute of Technology. Downloaded on April 02,2024 at 09:01:38 UTC from IEEE Xplore. Restrictions apply.
 Fig4. Simulated Waveform of the demodulated signal.
Fig.1. Gilbert Modulator
The simulated waveform of the demodulated signal using
III. SIMULATION AND RESULT ANALYSIS Gilbert mixer is shown in fig. 4. At the output of the
The simulation was carried out using ADEL in CADENCE demodulation signal an RC low pass filter was used to filter
VIRTUOSO EDA tool using gpdk 350nm technology. out all the high frequency components of the demodulated
wave. It can be seen that the filtered demodulated signal is in
A. Modulator Design phase with the modulated signal and has very less or
The schematic of the Gilbert mixer used for modulation is negligible distortion.
shown in fig.1. The respective simulated waveforms are
shown in fig.2. The modulator block was fed with a 100 KHz C. Chopper as Demodulator
sine wave as sensing signal along with the clock pulse with The chopper circuit was taken as the second option for
frequency of 1MHz as carrier signal. demodulation purpose. Chopper stabilization technique has
been widely used in amplifiers for flicker (1/f) noise and
offsets reduction purposes using the principles of modulation
and demodulation. Thus, the functionality and performance of
the modulation and demodulation circuit determines the
realization and attainment of chopper stabilization. The
amplifier designed in [8] can be used for amplifying the
demodulated signal. For this schematic the clock pulse
frequency is of 1MHz and the modulated signal is 100 KHz
PAM wave. Fig.5 shows the schematic of chopper as a
demodulator block.
Fig.2. Simulated waveforms of Gilbert Modulator

From the simulation results it can be seen that the input

100kHz signal has under gone a Pulse Amplitude Modulation
(PAM) without any distortion.
B. Gilbert mixer as Demodulator
For the first method we cascaded gilbert cell along with
another gilbert cell to get back the envelope from the PAM
wave. The schematic of the gilbert cell as demodulator is
shown in figure 3. Here the clock pulse is of 1MHz and
modulated signal is a 100 kHz PAM wave. A RC filter is used Fig.5. Demodulator block using Chopper.
to filter out the high frequency component in the demodulated
signal. The simulated waveforms of chopper as demodulator is shown
in fig.6. In order to filter high frequency component of the
modulated signal simple RC filters are used.

Fig.3. Demodulator block using Gilbert mixer Fig.6. Simulated waveforms of Chopper as demodulator.

266
Authorized licensed use limited to: Nitte Meenakshi Institute of Technology. Downloaded on April 02,2024 at 09:01:38 UTC from IEEE Xplore. Restrictions apply.
A table has been made for various configurations of the demodulation scheme, the combination of gilbert cell as
modulation and demodulation blocks used which is shown in modulator and as well demodulator has better output voltage
table 2. level of 1.506V compared to all other three schemes.
Table 2: Comparison between demodulation schemes. In delay terms, the circuit must have less delay to perform
Gilbert mod – Gilbert Gilbert mod – Chopper effectively. The configuration of gilbert cell as modulator as
demod demod
Parameters
Pmos pull Nmos pull Pmos pull Nmos pull well as demodulator has less propagation delay of 17ns
up up up up without clock circuit connected.
Modulated The overall propagation delay of complete readout circuit
output 1.65V 1.68V 1.65V 1.68V is of 1.835μs (with parasitic) and 1.717μs (without parasitic)
signal level
Demod
and the difference is found to be 118ns.
output 1.41V 1.506V 310mV 320m V
signal level V. FUTURE SCOPE
Differential The complete ASIC implementation of the readout
mod signal 522mv p-p 406mV p-p 522mV p-p 406mV p-p
strength
circuitry can be carried out using pdk 350 nm technology from
Differential AMS technology and its mask file can be integrated with
demod 28.2mV p- 16.82mV p- MEMS capacitive sensors to generate a micro-chip. On chip
100mV p-p 78mV p-p
signal p p clock circuit can be varied dynamically by using variable
strength capacitor (capacitance bank) in place of single capacitance.
Propagation 1.787μs 1.943 μs In this paper assumption has made that sensing signal will
delay diff - - be a pure sine wave, but we can use a capacitance bank to vary
b/w with - 1.77 μs - 1.896 μs capacitance in turn voltage which will act as input signal for
and without = = an op amp in integrator mode for obtaining PAM wave. The
parasitic 17ns 47ns
demodulation circuit can be gilbert cell. The sensing
With reference to the above table, it is found that the mechanism can be even resistive and piezo electric sensing.
configuration in which gilbert cell is used as both modulator ACKNOWLEDGEMENTS
and demodulator has better performance when compared to
other three configurations. The voltage amplifier has to be The authors would like to acknowledge their gratitude to Prof.
used at the output end in gilbert cell as modulator and chopper Navkantha Bhat, Professor CeNSE IISc and his Ph. D student
as demodulator circuit since its output voltage level is of Mr. Thejas for all the guidance and help in carrying out this
300mV whereas other configuration output voltage is of 1.5V. work.
A layout has been generated for whole readout circuit as
REFERENCES
shown in fig 7.
[1] S. Hedge, Thejas and N. Bhat, “Universal capcitance sensor”, Proc. Of
ISSS, 2008
[2] Sambuddha khan, Thejas, N Bhat, “Design and characterization of a
micro machined accelerometer with mechanical amp for intrusion
detection”, IISc, ISSS-2009.
[3] Veda Sandeep Nagaraja and S.L.Pinjare,” Design and Implementation of
Gilbert Cell Multiplier , ANUSANDHANA, Journel of Science and
Engineering Management, , VOL-01, ISSUE-01, MARCH-2012.
[4] Michael Pedersen, Wouter Olthuis, and Piet Bergveld, “ High-
Performance Condenser Microphone with Fully Integrated CMOS
Amplifier and DC–DC Voltage Converter”, JOURNAL OF
MICROELECTROMECHANICAL SYSTEMS, VOL. 7, NO. 4,
DECEMBER 1998
[5] David T. Martin, Jian Liu, Karthik Kadirvel, Robert M. Fox, Mark
Sheplak, and Toshikazu Nishida , “A Micromachined Dual-Backplate
Fig 7: Complete readout circuit layout Capacitive Microphone for Aeroacoustic Measurements” , JOURNAL
After constructing layout, the DRC errors were found to be OF MICROELECTROMECHANICAL SYSTEMS, VOL. 16, NO. 6,
none, and parasitic extraction is performed on the layout to DECEMBER 2007.
obtain parasitic RC components. Back annotation has been [6] N Y Sutri, Dennis, “Low noise, low offset modulator demodulator
circuit for chopper stabilization technique in CMOS MEMS sensors
done for the complete readout circuit, from which the applications”,IEEE, 2014.
difference in propagation delay with parasitic (1.835μs) and [7] Behzad Razavi, “Design of analog CMOS integrated circuits”, McGraw-
without parasitic (1.717μs) in schematic of readout circuit is Hill Education, 2001.
found to be 118ns. [8] Asha K A, Veda S Nagaraj, “Design, simulation and implementation of
readout circuitry for MEMS microphone”, ANUSANDHANA, Journal
IV. CONCLUSION of Science and Engineering Management, Vol-03, Issue-01, June 2014.
The paper gives a comparison between four demodulation
schemes that can be used in readout circuit of a MEMS sensor.
From the table 2 we can see that, out of four configurations for

267
Authorized licensed use limited to: Nitte Meenakshi Institute of Technology. Downloaded on April 02,2024 at 09:01:38 UTC from IEEE Xplore. Restrictions apply.