Low-Noise Resistive Sensing Interface for Integrated

Eddy Current Based Non-Destructive Testing System

David Oliveira, Diogo Miguel Caetano and Jorge Fernandes
[email protected], [email protected], [email protected]

Instituto Superior Técnico, Lisboa, Portugal

November 2016

Abstract preparation. EC evaluation is an intuitive method.

A novel implementation of a pseudo-resistor There are different EC probes, each one specific for
with a boostrapping technique is proposed. different evaluation purposes. Most of the EC NDT
The pseudo-resistor is used to implement systems use an emitter, where an alternating cur-
a controllable resistor on the feedback of a rent generates an alternating magnetic field on its
high-pass amplifier with a low set-up time. surroundings, and a pickup receiver to sense the
This high-pass amplifier is implemented on a magnetic field. The target NDT system is shown in
Non-Destructive Testing system that is used Figure 1.
for detection of superficial and internal flaws
on conduction materials, using magnetoresis-
tive sensors. With this pseudo-resistor im-
plementation, it is possible to vary the resis-
tance between kΩ and GΩ by changing the
VSG voltage of the PMOS transistors that
comprises the pseudo-resistor. This resis-
tance variation can be used to vary the cut-
off frequency of the high-pass filter between
Hz and MHz, accordingly to the material
under test. To reduce the set-up time be-
tween each magnetoresistive sensor reading, Figure 1: EC NDT System [3].
the pseudo-resistor resistance is switched to
a resistance in the order of kΩ, resulting in The depth of penetration to be analysed depends
a set-up time equal to 0.3 µs. This circuit is on the magnetic field frequency, the material elec-
designed in 0.35 µm AMS technology. trical conductivity and the magnetic permeability.
1. Introduction For the materials of interest, the magnetic field fre-
quency is in the range [100 Hz, 10 MHz]. The depth
Magnetoresistive (MR) sensors are resistive sensors
of penetration is described by:
that present variation on the electrical resistance as
1
a function of the magnetic field. They are used δ=√ (1)
in a wide range of areas from biomedical analy- πf µ 0 µr σ

sis systems to industrial applications, such as Non- where: f - Frequency of the magnetic field; µ0 -
Destructive Testing (NDT) based on Eddy Currents Magnetic permeability of air; µ - Magnetic perme-
(EC) evaluation. In NDT based on EC evaluation, ability of the material; σ - Electrical conductivity
an advantage of using MR sensors is their higher of the material.
SNR compared with common coil-based probes [1]. In [2], a system for reading MR sensors is pre-
NDT is a set of techniques used to characterize sented. This system has the main objective of de-
and detect defects and flaws on certain materials, tecting defects of conductive materials under test
without changing their characteristics and utility. and it has to be a NDT system. The architecture
These techniques can be applied to reduce the cost of the system is divided into two main elements:
of operation, increase the safety of use and ensure the Emitter System and the Receiver System. The
the reliability of industrial components. EC evalua- Emitter System comprises an Emitter Coil and a
tion is a NDT method that requires minimal surface driving generator at a frequency f1 (frequency of

1
the magnetic signal). The Receiver System con- for a 32-sensor array. The number of sensors can
tains a Sensor Array, an Application-Specific Inte- be scaled. The ASIC has the following functions:
grated Circuit (ASIC) that drives the sensors with a drive the magnetic sensors; multiplex the signals,
frequency f2 , and some interface circuits. The pur- amplify/filter the signals and cancel the electrical
pose of the Sensor Array is to acquire the signals sensor drive signal. The ASIC allows DC+AC cur-
produced by the currents induced by the Emitter rent input signals. Each sensor should be followed
Coil. The interface circuits generate an analogue by a band-pass preamplifier (centered on f1 , f3 =f1 -
output and digital synchronization signals. f2 or f4 =f1 +f2 ) with bandwidth fb . The DC and
Although this system is already implemented, the f2 components have to be filtered at the output,
there are some aspects with room for improve- since they do not hold information on the sensor re-
ment, including the way the sensors are biased sistance value. In Figure 3, the simplified circuit of
and switched and the type of amplification em- each block of the ASIC is presented. As explained
ployed. This work is focused on an Operational before, the focus of this work is the amplifier Pre-
Transconductance Amplifier (OTA). This amplifier amp and the HP filter at its input.
(Pre-amp) is implemented in the ASIC as part of
the NDT system receiver. This amplifier is preceded
by a first-order RC High-Pass (HP) filter. This HP
filter has to present a cut-off frequency of at least
50 Hz, since the frequencies of interest are in the
range [100 Hz, 10 MHz]. The amplifier set-up time
has to be the minimum possible, in order to switch
between the sensors with minimum time overhead.
Our objective is to achieve a set-up time between
0.1 µs and 1 µs.
The main challenge is to have a low resistance, in Figure 2: ASIC top diagram [2].
order to reduce the set-up time to a value between
0.1 µs and 1 µs. A resistor that presents a resistance
between 1.5 kΩ and 15 kΩ and a very high resis-
tance (320 MΩ or higher) to guarantee that the HP
filter cut-off frequency is at the desired frequency is
necessary. Therefore, it is possible to conclude that
two significantly different resistances are needed to
accomplish the specifications of the pole frequency
and the set-up time. By analysing the equations
that give the set-up time and the cut-off frequency,
it is possible to conclude that the set-up time and
the HP filter cut-off frequency have a different rela- Figure 3: Circuit (block) simplified schematic [2].
tion with the resistance. The set-up time equation
gives the amount of time required for the system to
reach 99.9 % of its final value. 2.2. Amplifier Pre-amp and High-Pass Filter
As explained before, the focus of this work is the
amplifier Pre-amp and the HP filter at its input.
1
tset−up = 6.9RC ; fcut−of f = (2) In Figure 4, the HP filter and the feedback circuit
2πRC of the amplifier Pre-amp can be visualized, which
A variable resistor can be implemented with one is the part of the circuit that is the core of this
or several MOS transistors, which is known as a work. The amplifier Pre-amp has a Folded-Cascode
pseudo-resistor. A pseudo-resistor implementation topology. This amplifier is part of a two stage out-
is a solution that meets our specifications, since it put amplifier. The output amplifier drives a capac-
can be controlled to present a high or low resis- itance around 1 pF. With an output voltage of 2 V
tance. Therefore, our work will be focused on im- and a maximum frequency of 10 MHz, it leads to a
plementing a pseudo-resistor on the feedback of the slew-rate of 125 V/µs, and an output current of 1
amplifier Pre-amp. mA [3].

2. Background 2.3. MOS Pseudo-resistor

2.1. Full System ASIC A pseudo-resistor implementation using one or sev-
In [2], the ASIC that is part of the NDT system eral MOS transistors working as switches can be
receiver is presented. The ASIC drives the sen- used to implement a variable resistor. Taking ad-
sors and it has 8 sensors per amplifier and 4 blocks vantage of the transistors operating regions and

2
 Figure 4: Pre-amp and HP filter.

controlling their gate voltage with a DC voltage, it

is possible to use a pseudo-resistor implementation
to have a variable resistor. A variable resistor im- Figure 6: Fixed-VGS implementations.
plementation is our main topic in this work, since
with transistors in triode region a low resistance
is obtained, creating a low resistance path on the tween 0 and 1.65 V.
feedback of the amplifier, which is very important 16
10

to achieve our main goal of implement an amplifier 14

10
A
C

with a low set-up time. With transistors operat- 12

10

ing in sub-threshold region a high resistance is ob-

Resistance[Ω]
10
10

tained, allowing also to vary the cut-off frequency 8

10

of the HP filter accordingly to the material under 6

10

test. This means that the pseudo-resistor can be 4

10

tuned to the bandwidth.

2
10
−1.5 −1 −0.5 0 0.5 1 1.5
∆V [V ]

Several pseudo-resistors implementations are pro-

posed in [4], which are possible solutions to imple- Figure 7: Implementation A and C resistance.
ment the resistor of the HP filter. Three variable-
VGS pseudo-resistor implementations are shown in
Figure 5, while three pseudo-resistor implementa- 10
16

tions with fixed-VGS are presented in Figure 6. 10

14

Vtune = 0 V

10
12 Vtune = 0.18 V
Vtune = 0.37 V
Resistance[Ω]

Vtune = 0.55 V
10
10
Vtune = 0.73 V
Vtune = 0.92 V
8
10
Vtune = 1.1 V
Vtune = 1.28 V
10
6
Vtune = 1.47 V
Vtune = 1.65 V
4
10
−1.5 −1 −0.5 0 0.5 1 1.5
∆V [V ]

Figure 8: Implementation B resistance.

With these simulation results, it is possible to

conclude that with these variable-VGS implementa-
tions high resistances are achieved, which is one of
our main goals, in order to have a pole frequency at
50 Hz.
For the fixed-VGS implementations, the resis-
Figure 5: Variable-VGS implementations. tance is simulated with different values for the ideal
voltage sources (0.2 V, 0.4 V and 0.6 V, for exam-
In order to evaluate the behaviour of the pre- ple), in order to evaluate how the resistance varies
sented implementations in the NDT system, it is with that voltage. That voltage defines the volt-
necessary to simulate and obtain the resistance un- age between the gate and the source of the tran-
der a situation identical with the conditions of our sistors. In Figure 9, the simulation results for the
NDT system. Therefore, one of the nodes (VOU T ) implementation D are presented. In Figure 10, the
of the target resistor is fixed at 1.65 V, while the simulation results for the implementation E are pre-
other node (VIN ) is swept between 0 and 3.3 V. The sented. Finally, in Figure 11, the simulation results
simulations results for the variable-VGS implemen- for the implementation F are shown. With these
tations (A and C) are shown in Figure 7, while for simulations results, it is possible to conclude that
the implementation B the simulations results are with the fixed-VGS implementations, as with the
presented in Figure 8, for different Vtune values be- variable-VGS implementations, it is possible to have

3
a pseudo-resistor that presents a resistance that is time between 0.1 µs and 1 µs with a unique pseudo-
in the MΩ range to have a cut-off frequency equal resistor implementation. In [5], a fast settling tech-
to 50 Hz. With that cut-off frequency, the MR sen- nique is proposed, which is a technique that allows
sors readings can be performed for the materials of the control of a pseudo-resistor implemented with
interest, since they have a magnetic field frequency the PMOS transistors to achieve a low set-up time.
in the range [100 Hz, 10 MHz]. For ∆|V | < 0.5 V, it This technique connects the gate of the transistors
is necessary that the pseudo-resistor resistance does to ground, in order to enable the fast settling tech-
not change to a value lower than 320 MΩ, in order nique.
to maintain the cut-off frequency at least below 50
Hz and not compromise the MR sensors readings. 3. Implementation
3.1. Pseudo-resistor Implementation
10
15
By using PMOS transistors, the bulk terminal is
VGS= 0.2 V ; VSG= 0.2 V
10
14

13
VGS= 0.4 V ; VSG= 0.4 V
VGS= 0.6 V ; VSG= 0.6 V
available to connect to the source or to the supply
10

10
12 voltage, which can be an advantage to minimize the
leakage currents. The leakage currents can have a
Resistance[Ω]

11
10

10
10

10
9
significant influence in high resistances, since the
variations in the current can have a significant influ-
8
10

7
10

10
6

5
ence on the resistance. A fixed-VGS pseudo-resistor
10
−1.5 −1 −0.5 0
∆V [V ]
0.5 1 1.5

implementation inspired in the fixed-VGS imple-

mentations presented before is proposed, which is
Figure 9: Implementation D resistance. shown in Figure 12.

An increase in the VGS voltage (VSG for PMOS

transistors) causes a decrease on the resistance,
which can be an improvement to our objective of
having a resistance of 320 MΩ or higher. This
improvement allows to adjust the bandwidth with
resistances smaller than 320 MΩ accordingly to Figure 12: Fixed-VSG implementation with PMOS
the material under test, obtaining a variable band- transistors.
width.
As with the other implementations, a simula-
14
10 tion is performed to evaluate the behaviour of this
VGS= 0.2 V ; VSG= 0.2 V

pseudo-resistor under similar conditions of the tar-

13
10 VGS= 0.4 V ; VSG= 0.4 V
12
VGS= 0.6 V ; VSG= 0.6 V
10

11
10 get NDT system. One of the nodes (VOU T ) of the
Resistance[Ω]

10
10

9
10
pseudo-resistor is fixed at 1.65 V and the other node
8
10 (VIN ) is swept between 0 and 3.3 V. The simulation
7
10

6
results with different VSG values (0.5 V, 0.7 V and
10

5
10
−1.5 −1 −0.5
0.9 V) with four PMOS transistors in series are pre-
0 0.5 1 1.5
∆V [V ]
sented in Figure 13.

Figure 10: Implementation E resistance. 10

10
VSG= 0.5 V
VSG= 0.7 V
9
10
VSG= 0.9 V

8
10
Resistance[Ω]

14
10
7
10

13
10

VGS= 0.2 V ; VSG= 0.2 V 6

10
12
10 VGS= 0.4 V ; VSG= 0.4 V
VGS= 0.6 V ; VSG= 0.6 V 5
Resistance[Ω]

10
11
10

4
10 10
10 −1.5 −1 −0.5 0 0.5 1 1.5
∆V [V ]
9
10

8
10

7
10
−1.5 −1 −0.5 0 0.5 1 1.5
Figure 13: Simulation results with pseudo-resistor
∆V [V ]
implemented with PMOS transistors (bulk con-
nected to the source).
Figure 11: Implementation F resistance.
With 0 V applied to the gate of the PMOS tran-
Concerning the set-up time, it is necessary that sistors, the resistance obtained is low, as shown
the pseudo-resistor implementation that presents a in Figure 14. The values in the order of kΩ that
320 MΩ (or higher) resistance presents also a re- are needed are achieved, which are values that are
sistance between 1.5 kΩ and 15 kΩ. This change needed for reducing the set-up time. The imple-
allows to achieve the objective of having a set-up mentation with four PMOS transistors presents a

4
resistance higher than what is desirable to theoret- the voltage at the gate of the PMOS transistors of
ically reduce the set-up time to a value between 0.1 the pseudo-resistor is equal to the voltage VT U N E .
µs and 1 µs. Although, the difference is not signifi- Therefore, it is possible to use this circuit to re-
cant and the current that flows through the pseudo- place the voltage source between the source and the
resistor in the NDT system is higher, reducing the gate of the PMOS transistors that are part of the
resistance. pseudo-resistor. With this circuit, it is possible to
4
vary the resistance by changing the voltage VT U N E
x 10

and, consequently, vary the pole frequency. Taking

6.5

5.5
advantage of the node B is grounded in phase 1,
5

this circuit can be used to connect the gate of the

Resistance[Ω]

4.5

3.5
PMOS transistors to ground, in order to obtain a
3
low resistance.
2.5

2
In order to provide the voltage VT U N E to the cir-
1.5
−1.5 −1 −0.5 cuit with a bootstrapping technique, our choice is
0
∆V [V ]
0.5 1 1.5

to implement a DAC with a voltage divider archi-

N
Figure 14: Simulation result with gate of the PMOS tecture. This DAC architecture consists of 2 equal
transistors connected to ground (bulk connected to value resistors followed by a matrix of switches and
the source). a voltage buffer. In this case, a 6-bit DAC with 1kΩ
resistors is used. This architecture uses a buffer, in
order to have a low output impedance.
In order to implement the pseudo-resistor in the
NDT system, it is necessary to modify its topology. 3.3. High-Pass Filter Resistor
The ideal voltage sources have to be replaced for a With the circuit represented in Figure 15 and a 6-
circuit that is capable of generating a voltage varia- bit voltage divider DAC, it is possible to have a
tion between the source and the gate of the PMOS final implementation of the pseudo-resistor, as rep-
transistors that are part of the pseudo-resistor im- resented in Figure 16. The proposed pseudo-resistor
plementation. implementation has a circuit with a bootstrapping
technique that defines the VSG voltage of the PMOS
3.2. Bootstrapping
transistors. That voltage can be varied with the
A circuit (see Figure 15) inspired in the circuit pre-
voltage provided by the 6-bit DAC between 0 V and
sented in [6] is proposed.
1.65 V. When the capacitor C of the circuit with a
bootstrapping technique is charging, the node B is
grounded, allowing to have a resistance in the order
of kΩ, which reduces the set-up time. The HP fil-
ter resistor is implemented with two equal devices
in series. Therefore, in total are used four PMOS
transistors and four equal circuits with a bootstrap-
ping technique with a shared 6-bit DAC to generate
the voltage VT U N E .

Figure 15: Proposed circuit with a bootstrapping Figure 16: Pseudo-resistor implementation using a
technique. circuit with a bootstrapping technique.
The proposed circuit works as follows: in phase 1,
N1 and N2 turn on, the capacitor C is charged with 3.4. Single Block Diagram
VT U N E − vcm, while the transmission gate creates Each block of the full system ASIC (see Figure
a high impedance path between the nodes A and 2) has three main components: a current source
B (node B connected to ground); in phase 2, N1 (CURRENT SOURCE), a block to select which
and N2 turn off, while the transmission gate con- sensor is read (MR MUX) and an amplifier (PRE-
nects the node A to the node B, which allows that AMP ) with a HP filter at the input, which is the

5
main focus of our work. The MR sensors are fab- the amplifier Pre-amp output voltage (blue) set-up
ricated separately (1 kΩ with ±10% variation) and time is shown in Figure 18.
wire bonded to the ASIC, while RCOM is an exter-
nal 1 kΩ resistor. One MR sensor (MR1 to MR8) is 3.5
nSync

selected by the MR MUX accordingly to the signals 3

VOUT

S < 1 : 8 > and N S < 1 : 8 >, which are generated 2.5

by an auxiliary circuit that is controlled by a clock

V oltage[V ]
2
X: 2.501e−06 X: 2.786e−06
Y: 1.65 Y: 1.65

signal (nSync). The selected sensor is driven with 1.5

a precision low noise current source. 1

0.5

As mentioned before, one of the nodes of the 0

0 0.5 1 1.5 2 2.5 3 3.5 4

pseudo-resistor is fixed at 1.65 V (negative input t[s] x 10

−6

of amplifier Pre-amp) and the other node should

tend to 1.65 V (amplifier Pre-amp output), yet volt- Figure 18: Simulation result concerning amplifier
age variations may occur with the pseudo-resistor Pre-amp output voltage set-up time.
on the feeback of the amplifier Pre-amp. When
the clock signal nSync commutation 0 00 →0 10 oc- With the marked points in the simulation result,
curs, the amplifier Pre-amp output voltage presents the set-up time of the amplifier Pre-amp output
a variation due to charges stored on the nodes of voltage can be obtained, resulting a set-up time
the pseudo-resistor implementation. Those charges equal to 0.296 µs, which represents a value that is
cause a variation on the current that flow through within our objectives.
the pseudo-resistor, resulting in the amplifier out-
put voltage variation. This variation take, approxi- 4.2. Transient Analysis with DC Input
mately, 10 milliseconds (this value increases as the In order to evaluate the achievable cut-off frequency
resistance increases) to the amplifier output voltage range with our pseudo-resistor implementation, a
settles to 1.65 V as supposed. In order to minimize simulation with all the codes of the 6-bit output
the time that the amplifier Pre-amp output volt- voltage DAC to obtain the resistance for each code
age tends to 1.65 V, four switches are connected in is performed. The simulation results for fourteen
parallel with the PMOS transistors of the pseudo- codes, which presented the better results, for a clock
resistor. When the clock signal nSync commutation signal nSync period with a DC input are shown in
occurs, the switches are closed for a period around Figure 19.
300 nanoseconds to create a low resistance path be-
tween the amplifier Pre-amp negative input and the
10

output voltage. The configuration of the amplifier

10
100101

X: 0.07641 100110

Pre-amp with the pseudo-resistor implementation

Y: 1.231e+09
9
10
100111

101000

on its feedback is shown in Figure 17. 8

10
101001

101010
Resistance[Ω]

101011

101100
7
10 101101

101110

101111
6
10
110000

110001
X: 0.07431
Y: 3.526e+05
110010
5
10
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
t[s]

Figure 19: Resistance with clock signal nSync = ’1’.

For these codes, a resistance almost constant for

all the clock signal nSync period is obtained, de-
spite that for few milliseconds after nSync commu-
Figure 17: Amplifier Pre-amp with HP filter.
tation 0 00 →0 10 the resistance presents a variation
caused by variation in the output voltage ampli-
4. Results fier Pre-amp, which can prejudice the settling time.
4.1. Amplifier Pre-amp Set-up Time With the marked points referring a minimum and
By reducing the set-up time is possible to reduce a maximum resistance obtained, a maximum and
the time needed to switch between the MR sensors, minimum cut-off frequency is determined, respec-
since the purpose of the ASIC is to quickly address tively. Therefore, the equation 2 is used to de-
the 32 sensors of the array. Therefore, if a reduction termine fcut−of f . With Rmin = 352.6 kΩ results
on the set-up time at each commutation occurs, a in a maximum cut-off frequency equal to 45 kHz,
reduction in the set-up time of the whole NDT sys- while with Rmax = 1.231 GΩ a minimum cut-off
tem is ensured. The simulation result concerning frequency equal to 13 Hz is obtained.

6
4.3. Amplifier Pre-amp Bandwidth
A simulation result presenting the AC frequency
response of the HP amplifier is shown in Figure
20. Theoretically, the cut-off frequency is lower
with small VSG voltages, since the pseudo-resistor
presents a very high resistance. Increasing the VSG
voltage of the PMOS transistors that are part of the
pseudo-resistor and consequently, decreasing the re-
sistance, the cut-off frequency increases.
Figure 22: Pre-amp amplifier AC response.
50

0
VSG = 0.03 V
VSG = 0.2 V
AC Response [dB]

VSG = 0.25 V
−50
VSG = 0.32 V
VSG = 0.37 V
VSG = 0.42 V
a logic level ’1’. This can be explained by a charge
redistribution when the commutation occurs, since
−100
VSG = 0.49 V
VSG = 0.59 V
VSG = 0.71 V
−150
VSG = 0.8 V
VSG = 1 V
it is necessary to take into account the capacitances
−200
10
−3
10
−2
10
−1
10
0
10
1
10
2
10
3
10
4
10
5
V
10
SG
6
= 1.2 V
10
7
associated with the devices to predict the devices
Frequency [Hz]
AC behaviour [7].

Figure 20: Amplifier Pre-amp bandwidth. A Monte Carlo simulation with 500 iterations was
executed, using an AC input signal with 50 Hz,
Analysing the AC frequency response for each analysing at each iteration the amplifier Pre-amp
VSG voltage, it is possible to conclude that the cut- output voltage, in order to evaluate the effects of the
off frequency has nine orders of magnitude, between Monte Carlo simulation variations. The simulations
mHz and MHz, which corresponds to resistances be- results allowed to conclude that 90% of the itera-
tween T Ω and kΩ, respectively. tions are similar with the simulation results shown
in Figure 21, while for 10% of the iterations the DC
4.4. Transient Analysis with AC Input shift referred before is more problematic. The DC
With an AC input signal with a certain frequency, shift can result in a signal that is not stabilized, pre-
especially with a frequency equal to 50 Hz, the senting variations through the clock signal period,
pseudo-resistor has to present at least one resistance or result in a DC shift high enough to saturate the
that allows the MR sensors readings. Therefore, a amplifier Pre-amp, which is prejudicial to the MR
simulation result with an AC input with a frequency sensors readings.
equal to 50 Hz is presented in Figure 21, for two dif-
ferent codes, resulting in two different resistances.
5. Conclusions
3.5

3
The HP filter resistor was implemented with a
2.5
pseudo-resistor, which is an implementation that
uses PMOS transistors, which means that it can be
V oltage[V ]

1.5 implemented on-chip, presenting resistances in the

1
Clock signal
Input
orders desired. A pseudo-resistor implementation
0.5 MR sensor
Output (110011) that allows varying its resistance by changing the
Output (001101)
0
0 0.02 0.04 0.06 0.08 0.1
t[s]
0.12 0.14 0.16 0.18 0.2
VSG voltage of the PMOS transistors using a cir-
cuit with a bootstrapping technique was proposed.
Figure 21: Transient analysis with AC input for two With that variation, our objective of having a cut-
different resistances. off frequency equal to 50 Hz was improved, present-
ing a variable bandwidth, which means that we can
In Figure 21, it is possible to visualize that for change the pseudo-resistor resistance to vary the
’001101’ the amplifier Pre-amp output voltage is cut-off frequency, accordingly to the material under
only a DC signal. For ’110011’, the amplifier Pre- test. This improvement does not have influence in
amp output voltage has an AC component. This the reduction of the set-up time to a value between
means that for the first code the cut-off frequency 0.1 µs and 1 µs, since a set-up time equal to 0.3
is above 50 Hz and for the second code the cut-off µs is achieved. A variable resistor that presented
frequency is below 50 Hz, as represented in Figure good results with a variation between 352.6 kΩ and
22. 1.231 GΩ is presented, which results in a variation
Despite the good results presented, the amplifier in the HP cut-off frequency between 13 Hz and 45
Pre-amp output voltage suffers a shift on its DC kHz, approximately, as confirmed by the simulation
voltage when the clock signal nSync commutes to results.

7
References
[1] T. Costa, “Integrated circuits for interfacing
magnetoresistive sensors,” Ph.D. dissertation,
Instituto Superior Tècnico.
[2] D. Caetano, M. Piedade, J. Fernandes,
T. Costa, and L. Rosado, “A CMOS ASIC
for Precise Reading of a Magnetoresistive Sen-
sor Array for NDT,” Circuits and Systems (IS-
CAS), 2015.
[3] D. Caetano, F. Rabuske, D. Oliveira,
T. Rabuske, J. Fernandes, and M. Piedade,
“Fast Settling VGA for Eddy Currents Non-
Destructive Testing with an Array of Magneto
Resistors,” PRIME (Conference on PhD Re-
search in Microelectronics and Electronics),
2016.
[4] H. Kassiri, K. Abdelhalim, and R. Genov, “Low-
distortion Super-Gohm Subthreshold-MOS Re-
sistors for CMOS Neural Amplifiers,” Biomedi-
cal Circuits and Systems Conference (BioCAS),
2013.
[5] M. A. B. Atalf, C. Zhang, and J. Yoo, “A
16-Channel Patient Specific Seizure Onset and
Termination Detection SoC With Impedance-
Adaptive Transcranial Electrical Stimulator,”
Journal of Solid-State Circuits (JSSC), vol. 50,
November 2015.
[6] T. Rabuske, J. Fernandes, S. Nooshabadi,
C. Rodrigues, and F. Rabuske, “A 749nW
1MSps 8-bit SAR ADC at 0.5V Employing
Boosted Switches,” Electronics, Circuits and
Systems (ICECS), pp. 500–503, December 2012.
[7] B. Razavi, Design of Analog CMOS Integrated
Circuits. McGRAW-HILL INTERNATIONAL
EDITION.