Department of Physics and Technology

Design and Experimental Investigation of Charge

Amplifiers for Ultrasonic Transducers
—
Svein Kristian Esp Hansen
FYS-3921 Master’s Thesis in Electrical Engineering
June 2014
Abstract

Amplifiers are used in all types of electrical circuits to boost signal and
there is a huge variety in designs used for different applications. For ul-
trasonic applications our group has previously used commercial available
transimpedance amplifiers that converts a current to a voltage, but these
amplifiers have a linear response over its frequency range. To preserve as
much information as possible for lower frequencies a flat frequency response
is preferable.

The main goal of this thesis were to design an amplifier that has a
reasonable flat frequency response within the desired frequency range, from
1 to 20 MHz, while also having an acceptable amplification. To achieve this
we have taken a brief look at opamps and amplifier design in a very general
way, and the theory behind amplifier design. We have then used this theory
to design two charge amplifiers we believe will work satisfactory with an
ultrasonic transducer.We have simulated these amplifier circuits with a
SPICE program, built them and then tested them experimentally.

Our single stage amplifier gave us lower amplification then the current
amplifier, but we are still inside the acceptable amplification range. The
charge amplifier has a frequency response that matches the original signal
much more closely then the current amplifier and our amplifier is less
expose to high frequency noise.

i
ii
Acknowledgements

I would like to thank my supervisors Frank Melandsø and Svein Jacobsen

for their help and suggestions during my work with this thesis. I would also
like to thank Sveinung Olsen for his help and expertise with practical
circuit design.

Svein Kristian Hansen

Tromsø, June 2014

iii
iv
Contents

Abstract i

Acknowledgements iii

1 Introduction 1
1.1 Sound and Ultrasound . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Ultrasound Applications . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Medical Imaging . . . . . . . . . . . . . . . . . . . . . 4
Medical Ultrasound . . . . . . . . . . . . . . . . . . . . 5
1.2.2 Non-Destructive Testing . . . . . . . . . . . . . . . . . 6
1.3 Ultrasound Sensors . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 Piezoelectric Sensors . . . . . . . . . . . . . . . . . . . 7

2 Theory 9
2.1 Sensor Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Operational Amplifiers . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Amplifier Design . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.1 Inverting Amplifier Design . . . . . . . . . . . . . . . . 13
Charge Amplifiers . . . . . . . . . . . . . . . . . . . . . 14
Amplifier Gain . . . . . . . . . . . . . . . . . . . . . . 15
Frequency Response . . . . . . . . . . . . . . . . . . . 16

3 Simulations 21
3.1 Opamp Selection . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2 Simulation method . . . . . . . . . . . . . . . . . . . . . . . . 22

v
 3.2.1 Software . . . . . . . . . . . . . . . . . . . . . . . . . . 22
SPICE Models . . . . . . . . . . . . . . . . . . . . . . 23
3.2.2 SPICE Sensor Model . . . . . . . . . . . . . . . . . . . 25
3.2.3 Simulation types . . . . . . . . . . . . . . . . . . . . . 25
3.3 Single Stage Charge Amplifier . . . . . . . . . . . . . . . . . . 26
3.3.1 Gain and Bandwidth . . . . . . . . . . . . . . . . . . . 27
3.3.2 Transient Analysis . . . . . . . . . . . . . . . . . . . . 31
3.4 Dual Stage Amplifier . . . . . . . . . . . . . . . . . . . . . . . 33
3.4.1 Gain and Bandwidth . . . . . . . . . . . . . . . . . . . 35
3.4.2 Transient Analysis . . . . . . . . . . . . . . . . . . . . 35

4 Experimental Investigation 39
4.1 Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 General Amplifier Overview . . . . . . . . . . . . . . . . . . . 40
4.3 Power Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.4 Terminated Input Analysis . . . . . . . . . . . . . . . . . . . . 43
4.5 Amplification and Bandwidth Test . . . . . . . . . . . . . . . 46
4.5.1 Test with Signal Generator . . . . . . . . . . . . . . . . 46
4.5.2 Test with Ultrasound Transducer . . . . . . . . . . . . 48

5 Conclusion and Further Work 57

A Simulation Plots 59

List of Figures 67

List of Tables 69

vi
Chapter 1

Introduction

In this chapter we will take a brief look at sound waves and ultrasound in
general terms, and how we can measure an ultrasonic signal with a
piezoelectric sensor. In the next chapter we will look at how to amplify
such signals.

1.1 Sound and Ultrasound

Sound is longitudinal waves that propagate through a medium such as air,
water or fat/tissue in a human body. We can also think of sound as
vibrations in a medium. As sound needs a medium to propagate through,
there will be no sound in places with absolute vacuum such as in outer
space. How fast these waves move through the medium is dependent on
how dense the medium is, and especially for gasses the temperature of the
medium.

Air (0 ◦ C) 343 m/s

Helium 1005 m/s
Water 1440 m/s
Sea water 1560 m/s
Concrete ≈ 3000 m/s
Aluminium ≈ 5100 m/s

Table 1.1: Speed of sound in different media [5]

1
To help describe sound we will use some common terms in relation to
sound, loudness and pitch. Loudness is the energy or intensity of the sound
and is most often measured in decibel (dB).

I
β(in dB) = 10 log
I0

Here I is the intensity of the sound and I0 is the reference intensity. The
intensity is commonly called the amplitude of the signal and the letter A is
used as its symbol.
Pitch is related to the frequency of the sound measured, and can either be
measured as temporal frequency, in hertz (common symbol f), or as angular
frequency, radians per second (common symbol ω). They are both tied to
the signals period T. The relationship between period, temporal frequency
and angular frequency is:

1
f =
T
2π
ω =
T
ω = 2πf

2
 Figure 1.1: A sine function with period and amplitude marked in

In Figure 1.1 we can see the period and amplitude in relation to a sine
function.

We usually associate sound with what we can hear. This is sound waves
moving through air and that have a frequency that lets the human ear
sense them. These sound waves have frequencies roughly between 20 Hz
and 20 kHz.

Figure 1.2: The soundspectrum from infrasound to ultrasound1

1
Taken from http://en.wikipedia.org/wiki/Ultrasound

3
We can then from Figure 1.2 define ultrasound as waves with a frequency
that is above what the human ear can hear (20 kHz). Different ultrasound
frequencies have different applications. For our application in this thesis we
will however not go as low as 20 kHz, but instead look at sound waves with
frequencies in the 1 MHz to 20 MHz range. Sound waves in this band have
a wide range of applications, from medical imaging to non-destructive
testing of objects [5][6].

1.2 Ultrasound Applications

The use of sound waves has been present for many different application
over the years. From the technique of counting seconds between a lightning
flash and the sound of thunder to determine how far away the storm is to
the more advanced techniques used in sonar and ultrasound imaging. What
these techniques have in common is the measure of distance (range). When
we know how fast something moves and we know the travel time, we can
calculate the distance. The example above with a thunderstorm is one of
the most basic applications, and something we all learn as children. The
storm sends out both light and sound in all directions, and by using the fact
that light has almost no travel time over the distance we can use the speed
of sound in air to determine how far away the storm is. Other applications
require more equipment and the calculations are more complex, but
fortunately we can also gather more information from these processes.

1.2.1 Medical Imaging

Medical imaging is the field of using different techniques and types of
technology to collect information about a specific part of the human body,
and use this to provide a diagnosis of some disease. Among the different
techniques commonly used we find X-ray, MRI, mammography and
ultrasound. The different techniques are used to diagnose different parts of
the body, X-rays will show you a broken bone, but give little information
about soft tissue. Mammography is used in detection of breast cancer, and

4
ultrasound is most often used to look at hearts, lungs and fetus.

Medical Ultrasound

To diagnose the body we use something called a pulse-echo technique. This

technique applies an ultrasound pulse, with a frequency in the 1 - 10 MHz
(and sometimes higher) range, that is emitted into the body. As this pulse
propagates through the body and hit areas with organs/tissue changes,
parts of the pulse will be reflected back to the emitting transducer. The
time between emittance and when the reflected pulse is detected will give
the distance from the transducer to the reflecting object [5]. By applying
ultrasound to a larger area of the body we can create a 2D image with
outlines of organs or areas were the tissue are different from its surrounding
area. Ultrasound pulses with different frequencies penetrates the body to
different depths, lower frequencies will penetrate deeper, and for
applications like fetus imaging that require us to go deep into the body we
use the lower part of the ultrasound frequency band.

Figure 1.3: Ultrasound image of a human fetus

Our desired frequency band in this thesis goes from 1 to 20 MHz and will

5
therefore include a wide range of depths and diagnostic applications.

1.2.2 Non-Destructive Testing

Non-destructive testing (NDT) is used to test or examine an object without

altering that object in any way. There exists a big variety of techniques
used for NDT, ranging from magnetic and radiographic testing to thermal
and ultrasound testing. NDT are used to verify the integrity of welds or to
look for cracks in the structure of a jet engine, among many other things.
Using sound is probably one of the oldest techniques to determine the
condition of any solid object, giving root to sayings like "sound as a bell".
Like medical imaging, NDT with ultrasound also applies a pulse-echo
technique, but instead of looking for organs we are now looking for e.g.
bubbles of air or cracks inside solid objects [6].

Figure 1.4: Ultrasound test of weld2

2
Taken from http://en.wikipedia.org/wiki/Ultrasonic_testing

6
1.3 Ultrasound Sensors
Detecting ultrasonic sound can either be done with a ultrasound transducer
that can both detect and emit ultrasound pulses, or with a ultrasonic
sensor that can only detect ultrasound pulses. An ultrasound emitter can
be as simple as a dog whistle and an ultrasound sensor can be a special
microphone made for higher frequencies. However, more advance
transducers and sensors are most often made out of a piezoelectric material.

1.3.1 Piezoelectric Sensors

A piezoelectric sensor is a device that when exposed to changes in one of its
surrounding conditions (pressure, strain, force or acceleration) will produce
a current. This works similar to the human ear, were a sound wave hits the
eardrum and makes it vibrate. Or when you experience a sudden change in
the atmospheric pressure around you. In the same way, when ultrasonic
waves hit the piezoelectric sensor we will get a current in the piezoelectric
material that we can detect. The piezoelectric sensors utilize the
piezoelectric effect, where a current is produced in a material as a response
to an outer mechanical effect. A measurable current will often be produced
if the material is distorted by as little as 0.1% of its original size/placement.
However, not all materials produce a piezoelectric effect, which is most
often found in crystals and some types of ceramics. As this effect is
bidirectional, a piezoelectric material that is exposed to an electrical field
will change its physical size (by approximately 0.1%). If the material is
exposed to a changing electrical field with the right frequency, the changes
in size will produce a sound wave.

7
8
Chapter 2

Theory

In this chapter we will take a look at operational amplifiers in general, and

then study how to design an amplifier to our needs using an operational
amplifier as the basic building block in our amplifier circuit.

2.1 Sensor Model

We need a model for our piezoelectric sensor to use with the electrical
circuit. The actual sensors we are going to use with our circuit is the
FDT1-028K and FDT1-052K [13] from Measurement Specialties. These
sensors consist of a thin piezoelectric film attached to a flexible lead with a
connector.

9
 Figure 2.1: FDT Series Piezoelectric sensor1

There are two types of equivalent circuits; either a current/charge source in

parallel with a capacitance, or a voltage source in series with a capacitance,
see Figure 2.2.

Figure 2.2: Equivalent models of a simplified piezoelectric sensor

In our calculations we choose to use the simplified model that consists of a

voltage source in series with a capacitance as we are looking at connecting
our circuit to signal sources. The increased voltage control makes this our
preferred model over the current source that supplies a constant current.
Since we do not have a good estimate off the capacitance value in this
model, we will make the assumption that the sensor will not limit the
amplifiers bandwidth. We will therefore adjust this capacitance value in the
1
Taken from http://www.meas-spec.com/product/t_product.aspx?id=2482

10
pico and nano Farad range and see how that affects the circuit.

2.2 Operational Amplifiers

While there are lots to be said about the history of operational amplifiers
(opamps for short) and their uses in different applications, we will only
provide a brief here overview before moving on to specific amplifier designs
suited for our use. We will also take a brief look at the ideal opamp and
how it differs from real opamps as this is useful when designing amplifiers.
Opamps were first developed in the early to mid 1940s as vacuum tube
opamps. As time and technology progressed after the second world war
opamps were made smaller and more compact until we in the 1960s started
to see the integrated circuit opamps we have today. While the op amp itself
is a complex circuit, it is today designed as an integrated circuit (IC) that
lets us use it as a normal component in our circuit design. The opamp is a
differential input, single-ended output amplifier [12].

Figure 2.3: The Opamp symbol with its connectors

Looking at Figure 2.3 we see both these input terminals (V+ and V− ), the
output terminal (Vout ) and also two power terminals (VS+ and VS− ). While
many circuit designs will show the opamp without the power terminals
(ideal opamps are drawn with input and output terminals only), five
connectors is the minimum number we will find on any real opamp.

11
The ideal opamp has certain attributes that are important for their
operation, and helps with circuit analysis [12].

• Infinite Differential Gain

• Zero Common Mode Gain

• Zero Offset Voltage

• Zero Bias Current

• High Input Impedance

• Low Source Impedance

Another name for differential gain (often denoted A) is open-loop gain [1]
that will differ from the closed loop gain (G) we get by connecting a
feedback circuit. We can also add infinite bandwidth to our list, as the
ideal opamp will have the same gain for all frequencies from 0 Hz (DC) to
∞. Real opamps will on the other hand have a frequency band specified in
their datasheet. While we almost always will connect either V+ or V− to
ground, the combination of infinite differential gain and zero common mode
gain mean that if we have no signal on our input (same potential as
ground) we will have no amplification, but any signal will have a high
amplification. This is achieved by the opamp by sensing the difference
between V+ and V− and forcing this difference (multiplied by the gain) at
Vout . The facts that we have zero bias current and high input impedance
mean that there is no current going into the opamp on either V+ or V− and
we can control the behaviour with our feedback circuit.

2.3 Amplifier Design

There are two basic ways to design the external feedback circuit for an
opamp. Either with a non-inverting gain stage or with an inverting gain

12
stage. There is also a configuration with a differential gain stage, but this is
more rarely used compared to other designs and will therefore not be
discussed here.

Figure 2.4: Non-inverting(left) and inverting(right) amplifier designs

Note that we in Figure 2.4 are using the general impedances Zf and Zg as
they can be either resistors, capacitors or inductors (or a combination of
those).

2.3.1 Inverting Amplifier Design

The main difference between the non-inverting and the inverting design
(other than the inverted gain) is that the inverting amplifier has a relatively
low input impedance at Vin and this provides a finite load for our
source [12]. The low input impedance comes however as a trade-off against
gain. While we in theory can adjust the gain over a large area, there are
practical limitations to how small we can make Zg .

We can define the closed loop gain as G = vvout

in
. By using the attributes of
the ideal opamp and Ohms law we can find an expression for the closed

13
loop gain G for the non-inverting amplifier in Figure 2.4 (Left panel).

Vout
V+ − V− = A
=0
Vin −V− Vin
i− = |Zg |
=
|Zg |
Vin
Vout = V− − i− |Zf | = 0 − |Zf |
|Zg |
Vout |Zf |
G= Vin
=− (2.1)
|Zg |

Using the same analysis we can also find that non-inverting amplifier has a
|Z +Z |
gain of G = f|Zg | g .

Charge Amplifiers

The charge amplifier is a variation of the inverting amplifier with a charge

source instead of a voltage source. It is however important to note that the
charge amplifier does not amplify the charge itself, but instead produce a
voltage signal. In that way, we can also look at the charge amplifier as a
charge-to-voltage converter.

Figure 2.5: Basic charge amplifier

14
Looking at Figure 2.5 we will have some sort of charge source connected to
Vin and some type of sensor to measure the signal connected to Vout . The
charge presented on Vin will be transferred to Cf and we get an output
voltage at Vout that is proportional to the charge on Vin divided by the
capacitance Cf [12]. The resistor Rg is a small resistor that functions as
protection for the inverting input on the opamp, while the resistor Rf
functions as a DC/low frequency pathway protecting Cf from saturation.
This also makes the circuit perform as an active low-pass filter. We can now
apply the same circuit analysis we used on the ideal inverting opamp using
the absolute value of complex impedances. We also have the requirement
that all frequencies and resistor/capacitor values are real and positive.
Comparing Figure 2.4 with Figure 2.5 we can say that Zf = ZRf k ZCf .

Amplifier Gain

Zg = Rg (2.2)
ZRf = Rf
1
ZCf =
2πjf Cf
1 1 1
= +
Zf ZRf ZCf
1 1 1 + 2πjf Rf Cf
= + 2πjf Cf =
Zf Rf Rf
Rf
Zf =
1 + 2πjf Rf Cf
p
|Zf | = Zf Zf ∗
Rf
|Zf | = p (2.3)
1 + 2πf Rf Cf

Combining (2.1) with (2.2) and (2.3) we get

15
 Rf
√
1+2πf Rf Cf
G = −
Rg
Rf
G = −p (2.4)
1 + 2πf Rf Cf Rg

Using (2.4), we can examine how different variables affect the overall gain
off the circuit. We keep all but one variables at a fixed value and observe
how changes in this variable affect the circuit.
We use the following values as our baseline:

f = 1 MHz
Cf = 10 pF
Rf = 100 kΩ
Rg = 100 Ω
Table 2.1: Components base values used for theoretical investigation

Studying Figure 2.6 we observe the following:

• Increased frequency gives lower gain.

• Lower capacitance (Cf ) gives higher gain.

• Higher Rf gives higher gain.

• Higher Rg gives lower gain.

This behaviour is as expected and in accordance with (2.4).

Frequency Response
1
To study the amplifiers frequency response, define ZC = sC and ZR = R,
we use these definitions to get a transfer function H(s).
To derive our frequency response we start by looking at an active low-pass
filter.

16
Figure 2.6: Variables effect on Gain

Figure 2.7: Active low-pass filter

17
Looking back at Figure 2.4 we get

1 1 1
= + 1
Zf Rf sCf
1 Rf sCf + 1
= (2.5)
Zf Rf

And we have the transfer function

Zf
H(s) = −
Rg
Rf 1
H(s) = − (2.6)
Rg Rf sCf + 1

Studying (2.6) we can see that we have a pole for s = − Rf1Cf . If we let
s → jω we get the cutoff radian frequency ω0 = Rf1Cf or in the frequency
domain fcutof f = 2πR1f Cf .
We can build an active high-pass filter using the circuit in Figure 2.8.

Figure 2.8: Active high-pass filter

18
Using the same method as we used for the active low-pass filter we get

1
Zg = Rg +
sCs
Rg sCs + 1
Zg = (2.7)
sCs
Rf
H(s) = −
Zg
Rf Rg sCs
H(s) = − (2.8)
Rg Rg sCs + 1

With a pole at s = − Rg2Cs and we get the cutoff point ω0 = 1

Rg Cs
or
fcutof f = 2πR1g Cs .

Figure 2.9: Charge amp with sensor, band-pass filter

If we add the simplified sensor model together with the active low-pass
filters we get Figure 2.9, from which we can identify the feedback loop (Cf
and Rf ) as the components of an active low-pass filter and the input (Cs
and Rg ) as an active high-pass filter. Overall this is a band-pass filter. We

19
can then get the total transfer function for this filter.

Zf
H(s) = −
Zg
Rf Rg sCs 1
H(s) = − (2.9)
Rg Rg sCs + 1 Rf sCg + 1

We can identify two poles for this transfer function, s = − Rf1Cf and
s = − Rg2Cs . This gives us the two cutoff frequencies

1
f cl = (2.10)
2πRf Cf
1
f ch = (2.11)
2πRg Cs

Using the previous baseline values for components, we get a lower cutoff f cl
= 159 kHz. We have not looked at specific values for Cs yet, but using
(2.11) we find that the highest value for Cs that gives us 20 MHz max
frequency is around 80 pF. We should therefore make sure that Cs <80 pF.

20
Chapter 3

Simulations

The basic design for our amplifier has been described in the previous
chapter and we have taken a look at the basic theory behind how we can
express the gain and manipulate it to suite our needs. In this chapter, we
will focus on confirming the theoretical results by simulating the circuit
with different op amps. We will also be fine tuning the other components in
the circuit to give us the best gain and bandwidth possible. The baseline
value we used for creating Figure 2.6 looks to be in the correct range and
we will continue to use those as a starting point.

3.1 Opamp Selection

Together with photodiodes, piezoelectric sensors belong to a type of sensors
that are called high impedance sensors. These sensors produce a very small
current and are sensitive to distortion. It is therefore important to choose
an opamp with a very high input impedance to match the sensor, as well as
low bias current, low noise and high gain [12]. To achieve the high input
impedance and low bias current opamps with JFET input stages are
common for these applications. OPA656 [7] and OPA657 [9] from Texas
Instruments are two opamps that fulfil these requirements. In their
datasheets both are recommended for photoelectric applications and as
test/measurement front ends. Photoelectric sensors are in many ways

21
similar to our ultrasonic sensors. We will also compare these to the current
feedback amplifier AD8007 [3] from Analog Devices. While not having a
JFET input stage and therefore not designed specifically for high
impedance sensors this is a high speed amplifier with ultralow distortion
that we think will work well with our circuit.
To start we will compare the input stage on three opamps.

OPA656 OPA657 AD8007

Input impedance 1 TΩ 1 TΩ 4 MΩ
Input capacitance 0.7 pF 0.7 pF 1 pF
Bias current 2 pA 2 pA 4 µA
Table 3.1: Input characteristics of selected opamps

As we can see in Table 3.1, the input stage on OPA656 and OPA657 has
the same characteristics while AD8007 having a lower input impedance will
draw more current and therefore divert more from the ideal opamp.
Looking at other attributes we can see some differences between the
OPA656 and OPA657.

OPA656 OPA657
Open Loop Gain 65 dB 70 dB
Table 3.2: OPA656 and OPA657 Open loop gain

3.2 Simulation method

3.2.1 Software
The electrical circuit are simulated using the SPICE (Simulation Program
with Integrated Circuit Emphasis) software LTspice1 from Linear
Technology. SPICE is an open source analog electronic circuit simulator2
that works as the engine behind many software simulation packages used
for creating analog electronic circuits. LTspice is a free software package
1
http://www.linear.com/designtools/software/
2
http://en.wikipedia.org/wiki/SPICE

22
that in addition to normal passive circuit components also have a library of
op amps and other IC components from Linear Technology. As SPICE is
the industry standard for simulating analog components almost every
manufacturer provides SPICE models for their components along with
datasheets. We choose to use LTspice as our simulation software because it
is free for everyone to use, and because it has a large user community that
can help new users and also provide a large number of component packages
from different producers. To use components from other producers, like
Analog Devices and Texas Instruments, LTspice let us import third party
SPICE models.

SPICE Models

Some of the third party models available to us will work in LTspice without
any modifications, while other models will have different pin orders or other
differences that do not make them directly compatible with LTspice. By
reading the information provided in a SPICE model file and comparing that
information to the models in LTspice we can see if we have to make any
changes. To see how this work we can look at the opamp2 symbol file
opamp2.asy that is a standar symbol used in LTspice. This symbol have all
the normal pins for a opamp, two inputs, two power supply pins and an
output.

23
 PINATTR PinName In+
PINATTR SpiceOrder 1
PIN -32 48 NONE 0
PINATTR PinName In-
PINATTR SpiceOrder 2
PIN 0 32 NONE 0
PINATTR PinName V+
PINATTR SpiceOrder 3
PIN 0 96 NONE 0
PINATTR PinName V-
PINATTR SpiceOrder 4
PIN 32 64 NONE 0
PINATTR PinName OUT
PINATTR SpiceOrder 5

Figure 3.1: Part of opamp2.asy

Here we can see that the pin In+ corresponds to the first pin in the spice
model, that In- corresponds to the second pin in the spice model etc.
Looking at two different opamps we can see how the pins are listed in their
spice models.
* CONNECTIONS:
* Non-Inverting Input
* | Inverting Input
* | | Positive Power Supply
* | | | Negative Power Supply
* | | | | Output
* | | | | |
.SUBCKT OPA656 + - V+ V- Out

Figure 3.2: Part of opa656.lib

Comparing the pin order in Figure 3.1 to the pin order in Figure 3.2 we see
that the order is the same and that we can use this symbol together with
this opamp without having to do any modification.

24
 * CONNECTIONS:
* Non-Inverting Input
* | Inverting Input
* | | Output
* | | | Positive Supply
* | | | | Negative Supply
* | | | | |
* | | | | |
* | | | | |
.SUBCKT OPA657 + - Out V+ V-

Figure 3.3: Part of opa657.lib

Comparing Figure 3.3 to Figure 3.2 we can see that they do not have the
same pin order. In Figure 3.3, the OUT pin have position 3 instead of
position 5. To use this spice model we would then have to modify the part
of opamp2.asy shown in Figure 3.1 and give PinName OUT the SouceOrder
3 etc. We will then save the modified symbol file with a new name and
adding this symbol to our symbol library to make it applicable.

3.2.2 SPICE Sensor Model

While there exist complex SPICE models for different types of piezoelectric
sensors, most often a simplified equivalent circuit is used. We will therefore
use the same simplified model in our simulations as we used for the
theoretical analysis.

3.2.3 Simulation types

We will conduct two different types of simulations to determine the
different properties of our circuit. First we make a small signal AC analysis
to determine the frequency response and gain of the circuit. When we are
satisfied with those we will do a transient analysis to see whether the
circuit is stable for our use, and to see how the circuit reacts to a general
input signals.

25
3.3 Single Stage Charge Amplifier
In this section we look at the circuit shown in Figure 2.9 and study how our
three different opamps perform. We will also look at how changing the
values of our passive components affects the circuit, and how sensitive the
circuit is to these changes.

Figure 3.4: Charge amplifier circuit, with OPA656 drawn in LTspice

We start out with the circuit shown in Figure 3.4 and have used the
following values as our baseline values for all simulations.

Cs = 20 pF
Cl = 1 pF
Rs = 100 Ω
Rf = 100 kΩ
Cf = 10 pF
Rl = 50 Ω
Table 3.3: Component base values used for simulation

These values are the same as we used for our theoretical work in the last
chapter, see Table 2.1. In addition we have chosen Cs as 20 pF and Cl as 1
pF to simulate the capacitance in the sensor and in the transmission line
leading from the sensor to the amplifiers input. The load resistance is
chosen as 50 Ω as we will connect the output to an oscilloscope that has a

26
50 Ω connection. As all of our amplifiers have 5V as their normal supply
voltage we will use this for all of our simulations.

3.3.1 Gain and Bandwidth

To get a good overview of how the different opamps perform with a range
of passive components we will be simulating a wide range of values for our
passive components. We will only list the main results in this section and
the plots that make up our dataset will be included in Appendix A.
We will be simulating using the following passive component values by
changing one component at the time and let the rest stay at the values
listed in Table 3.3.

• Cf : 1, 2, 3, 4, 5, 6, 7, 8, 9 and 10 pF

• Cs : 10, 20, 30, 40, 50, 100, 200 and 500 pF

• Rf : 1 kΩ, 10 kΩ, 100 kΩ, 1 MΩ and 10 MΩ

• Rs : 0, 10, 50, 100, 200, 500 and 1000 Ω

27
 Figure 3.5: Example values from Figure A.1

In figure Figure 3.5 we have three example values taken from our dataset.
We observe that for the lowest Cf give a small dip for lower frequencies,
but the largest amplification. For higher Cf values we see a flatter response
but lower amplification.
From the complete dataset (see Figure A.1 to Figure A.12) we observe the
following characteristics:

• Smaller Cf gives higher gain at the cost of flatness (in reality there
are also limits to how small we can make it).

• The gain is only flat over our whole frequency range for Cs ≈ 10 - 50
pF.

• Rf ≈ 100 kΩ - 10 MΩ gives a flat gain over the whole frequency range.

• If Rs is above 50 Ω we observe a steep drop in gain for higher

frequencies.

28
We get the gain as high as 40dB with Cf = 1 pF for the OPA656/657
opamps. However, we want to avoid capacitances in this range, as they are
in the same order as parasite capacitances we will find in our circuit. The
capacitance values also give a dip in amplification for lower frequencies. A
feedback capacitance of ≈ 5 pF will give us a decent gain and still be
outside the range of any parasite effects.
As described in Section 2.1, we are using a simplified model for our sensor,
the simulations for Cs is therefore those that have the highest level of
uncertainty. What we can read from the simulations, however, is that for a
wide range of capacitances we get a relatively flat gain, but if the sensor
capacitance is too high, we get higher gain for the lower frequencies then we
get for the higher ones.
While Rf ’s main goal in the circuit is to provide a DC pathway in the
feedback loop we can clearly see that a value between 10 kΩ and 10 MΩ is
needed to keep the gain acceptably flat.
Ideally, we want a resistor Rs to protect the opamps input stage against
high surge voltage. However we see that any values over 50 Ω give us an
exponential dampening.
Using our base values we then compare all three opamps, both inside our
desired frequency range and in a wider band.
Studying Figure 3.6 we see that they all have almost matching gain inside 0
- 20 MHz, but when we look frequencies all the way up to 2 GHz we see
some clear differences. As we do not want to amplify high frequency noise
we choose an opamp that have a steep fall after 20 MHz.

• AD8007 has the flattest response all the way up to 2 GHz.

• OPA656 has a large spike around 800 MHz.

• OPA657 has a small spike around 400 MHz.

While the OPA656 have a large spike around 800 MHz this frequency is so
far outside our working frequency range that it should not present any
issues in regards to normal operation. The OPA656 also have a lower gain
then the OPA657 up to around 500 MHz. We think that the OPA656 is the

29
 Figure 3.6: Comparison of OPA656, OPA657 and AD8007

amplifier that are best suited for our application. We will therefore focus
our simulations on the OPA656 for the reminder of this section.
We can also see from the top part of Figure 3.6 that our average
amplification inside the 1-20 MHz band is ≈20 dB. We will therefore set
our -3dB point to 17 dB amplification. Looking at Figure 3.7 we see that
we exceeds 17 dB at 160 kHz and stays above it until 70 MHz. The lower
limit corresponds with fcl = 159 kHz that we calculated in Chapter 2. This
give us a -3 dB bandwidth of 68.2 MHz.

30
 Figure 3.7: -3 dB Bandwidth

3.3.2 Transient Analysis

As we have decided on OPA656 as our preferred amplifier for the single
stage setup, we look closer at some of the circuit characteristics. We use the
transient analysis to see how the circuit reacts to a specified input signal,
and to observe the shape of our output signal. Our input signal for this
analysis is a continuous sine with the following properties:

• Amplitude: 0.5 V

• DC offset: 0 V

• Frequencies: 1 - 20 MHz

In Figure 3.8 we can see a typical response of the transient analysis. Here
we have a sine function with a frequency of 1 MHz. As we are using an
inverted configurations we see that Vin and Vout are inverted, as expected.
We can also see that Vout have a build up time of about two periods. After

31
 Figure 3.8: Typical output from transient analysis with OPA656

the first two periods Vout looks stable, but closer inspection show that there
are small variations between the maxima. By increasing Vin , we see that
the circuit is unable to drive the output above ± 3.725 V and that a input
signal with a to high amplitude will give signal clipping.
As we know that the amplification will decrease as the frequency increase
we want to study the output from the transient analysis over the whole
frequency range. We do this by measuring the maxima and minima of
output on the first periods after the build up time.
We can in Figure 3.9 see that the amplification decrease as frequency
increase. We also see that the circuit is not completely symmetric, the
amplifier can not drive V- as far as V+.

32
 Figure 3.9: Transient analysis for OPA656, F = 1 - 20 MHz

Based on the simulation we choose the following final values for our single
stage amplifier. These values are designed to give us both the bandwidth
and amplification we desire.

Component Value
AMP1 OPA656
RG1 100 Ω
RF 1 100 kΩ
CF 1 5 pF
Table 3.4: Choosen values for one stage amplifier

3.4 Dual Stage Amplifier

A big drawback with our one stage amplifier design, and the three opamps
we have tried, is that the overall gain feels too low unless we make Cf so
small that the parasite capacitances in our circuit will be a major factor.

33
We want an overall gain of say 40-50 dB, and the best we realistically can
hope for with a single stage solution is 20 dB. To increase the gain to an
acceptable level we therefore suggest a dual stage solution were the first
stage is the charge amplifier discussed in Section 3.3. As this first stage also
acts as a charge-to-voltage converter we can then add a simple voltage
amplifier as our second stage to boost the voltage amplification. We should
also be able to use a none JFET opamp in the second stage as a voltage
amplifier should not need the same attributes as a charge amplifier. We will
however still use the same three opamps for the second stage, and see how
different combinations of these three opamps affect the overall gain.

Figure 3.10: Dual stage amplifier circuit with two OPA656s, drawn in LTspice

Comparing Figure 3.4 to Figure 3.10, we see a few components added tp

the latter variant in addition to the second stage with its feedback resistor.
The resistor Rg is added between the non-inverting output and ground on
opamp U1 and is matched to resistor Rf . Between the two stages we have
added the resistor Rp and the capacitance Cp . Opamp U2 have the resistor
Rf2 as feedback.

34
 Figure 3.11: Dual stage amplifier opamp comparison

3.4.1 Gain and Bandwidth

By studying the different configurations in Figure 3.11 we observe that
some of the configurations with OPA657 get a small peak around 18 MHz
while the rest have a rather flat response. Outside the 1 - 20 MHz range
they all follow each other closely up to 400 MHz were they start to drop at
different rates. Most have fallen below 0 dB at 800 MHz. Overall OPA656
+ OPA657 should give us the best performance both inside and outside 1 -
20 MHz.

3.4.2 Transient Analysis

As we did for the single OPA656 configuration we will perform a transient
analysis for the OPA656 + OPA657 circuit. Similar to the single stage
setup we have problems with signal clipping and deforming of the sine
function if we try to bring the output to high. While we could bring the
single OPA656 to 3.75 V we get signal clipping as early as +2.8 V and - 2.1
V with the dual stage setup.

35
 Figure 3.12: Signal Clipping during LTspice transient analysis

Figure 3.12 shows that we have both clipping and unsymmetrical behaviour
with an input sine with an amplitude as low as 0.07 V. To avoid clipping
during our simulations we will use a sine with the following properties for
our dual stage configuration:

• Amplitude: 10 mV

• DC offset: 0 V

• Frequencies: 1 - 20 MHz

After the simulation we choose the following values for our dual stage
amplifier

36
Figure 3.13: Typical output from transient analysis with OPA656 + OPA657

Component Value
AMP2 OPA656
AMP3 OPA657
RG2 100 Ω
RF 2 100 kΩ
CF 2 5 pF
RP 50 Ω
CP 15 nF
RF 3 1 kΩ
RGN D 100 kΩ
Table 3.5: Choosen values for two dual amplifier

37
38
Chapter 4

Experimental Investigation

In this chapter we will conduct an experimental investigation of both the

single stage and dual stage amplifier circuits we already simulated using
LTspice.

4.1 Equipment
To investigate and characterise the amplifier circuit we will be using the
following equipment.

Type Producer Model Name

Oscilloscope Agilent DSO-X 3024A
Power Supply Agilent E3631A
Signal Generator Agilent 33220A
Digital Multimeter UNI-T UT58B
Current Amplifier FEMTO Messtechnik DHPCA-100
Table 4.1: Equipment list for experimental investigation

In addition to our measuring equipment we will be using the DHPCA-100

current amplifier as our reference amplifier and compare our results to what
we achieve with this amplifier. This is a high speed variable gain amplifier
our group have used for ultrasound applications in the past.

39
 Figure 4.1: DHPCA-100 Current Amplifier

4.2 General Amplifier Overview

To investigate both the single stage and dual stage designs we have built
both circuits on the same PCB to save cost. To design our PCB we have
used the free PCB design software EAGLE from CadSoft1 .
To see a bigger version of Figure 4.2 see Figure A.13. As some of the values
we choose in Table 3.4 and Table 3.5 are outside the standard E24 series for
component values2 , we choose the closest values when building the circuit.

1
http://www.cadsoftusa.com/
2
http://en.wikipedia.org/wiki/Preferred_number

40
 Component Value
AMP1 OPA656
AMP2 OPA656
AMP3 OPA657
RG1 100 Ω
RG2 100 Ω
RF 1 100 kΩ
RF 2 100 kΩ
CF 1 4.7 pF
CF 2 4.7 pF
RP 51 Ω
CP 15 nF
RF 3 1 kΩ
RGN D 100 kΩ
C+ 10 µF
C− 10 µF
C1+ 100 nF
C1− 100 nF
C2+ 100 nF
C2− 100 nF
C3+ 100 nF
C3− 100 nF
Table 4.2: Final build values for both amplifiers

41
 Figure 4.2: Amplifier circuit drawn in Eagle

In Table 4.2 the number 1 in relation to a component name means that the
component is related to the single stage amplifier circuit. The number 2
means that the components are related to the first opamp in the dual stage
circuit and the number 3 means that the component are related to the
second opamp in the dual stage circuit. We have also added the buffer
capacitors C+ and C− close to were we connect the power supply to the
board and the capacitors C±1 to C±3 close to the opamps.

Figure 4.3: Image of the amplifier PCB with mounted components and cables
connected (left) and PCB mounted inside RFI box (right)

The amplifier PCB are mounted inside a RFI shielding box to minimize
noise and interference from outside sources.

42
4.3 Power Supply
To confirm that we are supplying the amplifiers with the correct supply
voltage we began our experimental investigation by setting the power
supply to ± 5V and connecting it to the amplifier. We then used a
multimeter to measure the voltage out from the power supply and into each
amplifier. To compensate we are adjusting the output of the power supply

Source Voltage
Power Supply + 4.99 V
Power Supply - -4.97 V
Amp1 + 4.98 V
Amp1 - -4.97 V
Amp2 + 4.97 V
Amp2 - - 4.96 V
Amp3 + 4.97 V
Amp3 - - 4.96 V
Table 4.3: Supply voltage measured on the amplifiers

to ±5.03 V, this give us ±5 V into each amplifier.

4.4 Terminated Input Analysis

To characterize the noise generated by the amplifier circuit we are
connecting the amplifier to its power supply, terminating the input by
shorting the signal and ground pins and then measuring the output on the
oscilloscope. An ideal opamp should have zero common mode gain, and by
terminating/shorting the input terminal we can test the commone mode
rejection. For an ideal opamp/amplifier circuit we should have zero output
signal. We are studying three parameters for each of the two amplifier
circuits, the peak-to-peak Vp−p voltage to find the maximum amplitude of
the noise, the average noise amplitude Vrms and also the Fourier transform
of the noise. This will both show the noise level as well as the frequency
components the noise consists off. The FFT is done on the oscilloscope
using a Hanning Window.

43
 Single Stage Dual Stage
Vp−p 44 mV 6.5 V
Vrms 7.62 mV 3V
Table 4.4: Amplifier Noise Level

Figure 4.4: Amplifier Output with terminated input terminals

44
 Figure 4.5: Fourier Transform of Figure 4.4

Looking first at Figure 4.4 we see that the single stage amplifier (top part)
has a negative DC offset to the noise. The noise also seems to have a
regular pattern. For the dual stage amplifier (bottom part) we can clearly
see that the amplifier goes into saturation even with terminated inputs, and
are therefore only showing us a square function. The signal we see for the
dual stage amplifier is not noise, but a result of oscillations and feedback in
the circuit bringing the amplifier into saturation.
Studying Figure 4.5 we see that the dual stage configuration (right side in
Figure 4.5) have some very dominating low frequency components with
levels up to 20 dBv. These are the components that builds the square pulse
we see in Figure 4.4. Looking at the single stage frequency response on the
left side in Figure 4.5 we see that the lower frequency are more dominating
then the higher frequencies, but they are much less dominating then for the
dual amplifier and only go as high as -30 dBv. We can also look at the
noise level when the oscilloscope averages the input, this will filter away
some of the random noise.

45
 Single Stage Dual Stage
Vp−p 34 mV 6.32 V
Vrms 7.93 mV 3V
Table 4.5: Amplifier Noise Level with averaging over 1024 samples

Comparing Table 4.4 and Table 4.5 we see a small drop in peak-to-peak
noise. We do also see a small increase in Vrms instead of a drop. This points
to the noise not being random and that we have some oscillations also for
the single stage amplifier As the dual stage amplifier are in saturation we
don’t see much change here, this is as expected under the circumstances.

4.5 Amplification and Bandwidth Test

We are going to test the amplification and bandwidth in two different ways.
First we will use a setup with a signal generator and a capacitor to create a
setup that matches our simulation, after that we are going to use a setup
with a ultrasound transducer and the sensor described in Section 2.1. We
will only preform these tests on the single stage amplifier as we observed
that even with terminated inputs the dual stage amplifier were in
saturation.

4.5.1 Test with Signal Generator

Here we connected our signal generator to the signal input on the amplifier
via a capacitor. As we did our simulations with a 20 pF capacitor we
choose a 22 pF capacitor for this test to get as close to the simulation as
possible. We then generated a sinus with constant amplitude and let the
frequency vary over our desired frequency range. We have then plotted the
measured peak to peak voltage coming out of the amplifier.

46
 Figure 4.6: Amplification output from single stage amplifier

Mean Max Min

Output 373.6 mV 442 mV 283 mV
Gain 7.47 8.84 5.66
Gain (dB) 17.47 dB 18.93 dB 15.06 dB
Table 4.6: Key data from Figure 4.6 with 22 pF capacitor

Looking at Figure 4.6 we see that a capacitor close to the value we chose
during our initial simulation (20 pF) gives the flattest response over the
whole frequency range. By increasing this capacitor we can achieve greater
amplification at the cost of flatness. We also see that we are just inside -3
dB from the mean amplification, not having the same bandwidth we had in
our simulations.
We tried to do the same measurements for the dual stage amplifier, but the
fact that the amplifier is in saturation from background noise alone
prevents us from getting any meaningful data.

47
4.5.2 Test with Ultrasound Transducer
To test the amplifier in a more realistic setting we mounted two ultrasound
transducers to an acrylic block that are 5 mm thick. We let one transducer
act as sender and the other as receiver. By connecting the transducer
acting as a sender to a signal generator we can create a pulse signal and
send it through the acrylic block.

Figure 4.7: Acrylic block with ultrasound transducers

Figure 4.7 shows the two ultrasonic transducers from MSI we are using. We
see one transducer mounted on each side of the block, and they can both be
used as either transceiver or receiver. An issue with this setup is that the
cables going into the sensors are 160 mm long on each side and without
shielding. The cables might therefore pick up a lot of noise from the
surroundings.

48
 Figure 4.8: Signal pulse created by the signal generator

We drive the transducer with a 60 ns wide pulse, see Figure 4.8. We also
observe that the pulse have some noise directly from the signal generator.

49
Figure 4.9: Unamplified pulse measured after passing through transducers

In Figure 4.9 we have the pulse from Figure 4.8 passing through the
ultrasound transducers and the acrylic block in Figure 4.7 without any
amplification. The signal is measured with 256 samples averaging on the
oscilloscope. Studying Figure 4.9 we see three distinct responses, one at the
begging, one after 2 µs and one after 7 µs. The first response is created
from the electrical field we get when we shoot the signal pulse into the
acrylic block and it does not contain any information for us. The response
around 2 µs is our first acoustic response, and is the result of the pulse
travelling from the first transducer through the block and hitting the
second (receiving) transducer. Some of this sound will be reflected at the
edge of the acrylic block, travel back to the transmitting transducer, be
reflected at this edge and then travel back to the receiving transducer
again. This result is what we see at the third response, and this is our
second acoustic response.

50
 Figure 4.10: Averaged and non average pulse after charge amplifier

In Figure 4.10 we see the difference between a averaged and a non averaged
signal passing through the acrylic block and being amplified by our single
stage charge amplifier. We have adjusted the axis of the figure to start at
the first acoustic response. In the top part of Figure 4.10 (with averaging)
we can see the first three acoustic responses, one response more then we
could without any amplification. In the bottom part (not averaged) we can
see the first two acoustic responses, but the third are below the noise floor.

51
 Figure 4.11: Averaged and non average pulse after current amplifier

Figure 4.11 is the same as Figure 4.10 with the signal being amplified by
the DHPCA-100 instead of our charge amplifier. While the top parts in
both figures have the same form we see that the current amplifier have a
larger amplification. We have however problems identifying the third
acoustic response with this amplifier. The increased high frequency noise on
the current amplifier are expected as this amplifier both have a larger
bandwidth then the charge amplifier as well as having a linear frequency
response that favours higher frequencies over lower frequencies. When we
compare the bottom parts of Figure 4.10 and Figure 4.11 we see that the
current amplifier have a more noisy signal and that we can not clearly make
out the second acoustic response in the noise. We can also compare the
unaveraged signals in Figure 4.10 and Figure 4.11 to Figure 4.9 and see
that the signal form looks the same, but with lower amplitude.
When looking for information in the received signal we are mainly after the
information contained in the first acoustic responses. To gather this
information we take the time signals in the top part of Figure 4.10 and

52
Figure 4.11, and multiply them with a window that zeros out the signal
outside the first acoustic responses. We then take the Fourier Transform of
this signals and look at they frequency components. In some situation we
also want the Fourier transform of the second acoustic response, this have
been created by placing a window in the same way we did for the first
acoustic response

Figure 4.12: Fourier Transform of the first acoustic response from the charge
amplifier and the unamplified signal

53
Figure 4.13: Fourier Transform of the first acoustic response from the current
amplifier and the unamplified signal

54
Comparing the frequency response of the charge amplifier and the current
amplifier to the frequency response of the unamplified pulse (Figure 4.12
and Figure 4.13) we see that the charge amplifier favours lower frequencies
more then the current amplifier. The charge amplifier has its centre
frequency at ≈15 MHz, the same as the unamplified pulse, while the
current amplifier has its at ≈30 MHz. This makes our charge amplifier a
better fit for this signal. The current amplifier have a much wider band,
but are also picking up more noise. We see a clear shoulder at 8-10 MHz for
the charge amplifier that may indicate some non linear amplification.

Figure 4.14: Fourier Transform of the second acoustic response

Figure 4.14 show us the Fourier transforms of the second acoustic response
from both the charge and current amplifier. We see that both are shifted
down in frequency, but that the charge amplifier still covers our desired
frequency band best.

55
56
Chapter 5

Conclusion and Further Work

We have designed, simulated and built two amplifiers, and tested them
experimentally. The single stage amplifier worked reasonably well to our
exception while the dual stage amplifier failed due to oscillations driving
the amplifier into saturation. We have also compared the single stage
amplifier to a commercial current amplifier.

The test we did on the single stage amplifier with a signal generator and a
capacitor shows the amplification to be 2.5 dB lower then what we found in
our simulations and a lot of the bandwidth had been lost as well. We could
not test the amplifier with sines that had a frequency above 20 MHz as this
also were the maximum frequency on our signal generator. The tests we did
with the ultrasound transducer shows that our amplifier seems to favour
lower frequencies more then the previously used current amplifier. Our
charge amplifier are also affected less by high frequency noise compared to
the current amplifier. We would have liked to do more tests on ultrasound
transducers under more realistic circumstances.

The dual stage amplifiers issue with early saturation were hinted at by the
transient analysis we did as we had to lower our input signal to avoid signal
clipping. A systematic theoretical analysis should be make of the dual stage
amplifier. The second stage of the amplifier should also be redesigned using

57
a different opamp. We believe that a normal low noise opamp made for
voltage amplification will work better in this position. The opamp should
also be able to handle higher supply voltage then ±5 V so we don’t go into
saturation that fast. While changing the second opamp probably wont solve
the issue with oscillations it will put the oscillations inside the amplifiers
voltage range. The use of opamps with higher supply voltage will also
require a redesign of our PCB as it now need to handle two different power
supplies running at different voltages. A study to show were the oscillations
occur should be conducted, and solutions to remove the oscillations
provided.

58
Appendix A

Simulation Plots

Figure A.1: Simulation of Cf values with OPA656

59
Figure A.2: Simulation of Cf values with OPA657

Figure A.3: Simulation of Cf values with AD8007

60
Figure A.4: Simulation of Cs values with OPA656

Figure A.5: Simulation of Cs values with OPA657

61
Figure A.6: Simulation of Cs values with AD8007

Figure A.7: Simulation of Rf values with OPA656

62
Figure A.8: Simulation of Rf values with OPA657

Figure A.9: Simulation of Rf values with AD8007

63
Figure A.10: Simulation of Rs values with OPA656

Figure A.11: Simulation of Rs values with OPA657

64
Figure A.12: Simulation of Rs values with AD8007

65
Figure A.13: Amplifier circuits drawn in Eagle
66
List of Figures

1.1 A sine function with period and amplitude marked in . . . . . 3

1
1.2 The soundspectrum from infrasound to ultrasound . . . . . . 3
1.3 Ultrasound image of a human fetus . . . . . . . . . . . . . . . 5
2
1.4 Ultrasound test of weld . . . . . . . . . . . . . . . . . . . . . 6

2.1 FDT Series Piezoelectric sensor3 . . . . . . . . . . . . . . . . . 10

2.2 Equivalent models of a simplified piezoelectric sensor . . . . . 10
2.3 The Opamp symbol with its connectors . . . . . . . . . . . . . 11
2.4 Non-inverting(left) and inverting(right) amplifier designs . . . 13
2.5 Basic charge amplifier . . . . . . . . . . . . . . . . . . . . . . 14
2.6 Variables effect on Gain . . . . . . . . . . . . . . . . . . . . . 17
2.7 Active low-pass filter . . . . . . . . . . . . . . . . . . . . . . . 17
2.8 Active high-pass filter . . . . . . . . . . . . . . . . . . . . . . . 18
2.9 Charge amp with sensor, band-pass filter . . . . . . . . . . . . 19

3.1 Part of opamp2.asy . . . . . . . . . . . . . . . . . . . . . . . . 24

3.2 Part of opa656.lib . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Part of opa657.lib . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.4 Charge amplifier circuit, with OPA656 drawn in LTspice . . . 26
3.5 Example values from Figure A.1 . . . . . . . . . . . . . . . . . 28
3.6 Comparison of OPA656, OPA657 and AD8007 . . . . . . . . . 30
3.7 -3 dB Bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.8 Typical output from transient analysis with OPA656 . . . . . 32
3.9 Transient analysis for OPA656, F = 1 - 20 MHz . . . . . . . . 33
3.10 Dual stage amplifier circuit with two OPA656s, drawn in LTspice 34
3.11 Dual stage amplifier opamp comparison . . . . . . . . . . . . . 35

67
3.12 Signal Clipping during LTspice transient analysis . . . . . . . 36
3.13 Typical output from transient analysis with OPA656 + OPA657 37

4.1 DHPCA-100 Current Amplifier . . . . . . . . . . . . . . . . . 40

4.2 Amplifier circuit drawn in Eagle . . . . . . . . . . . . . . . . . 42
4.3 Image of the amplifier PCB with mounted components and
cables connected (left) and PCB mounted inside RFI box (right) 42
4.4 Amplifier Output with terminated input terminals . . . . . . . 44
4.5 Fourier Transform of Figure 4.4 . . . . . . . . . . . . . . . . . 45
4.6 Amplification output from single stage amplifier . . . . . . . . 47
4.7 Acrylic block with ultrasound transducers . . . . . . . . . . . 48
4.8 Signal pulse created by the signal generator . . . . . . . . . . 49
4.9 Unamplified pulse measured after passing through transducers 50
4.10 Averaged and non average pulse after charge amplifier . . . . . 51
4.11 Averaged and non average pulse after current amplifier . . . . 52
4.12 Fourier Transform of the first acoustic response from the charge
amplifier and the unamplified signal . . . . . . . . . . . . . . . 53
4.13 Fourier Transform of the first acoustic response from the cur-
rent amplifier and the unamplified signal . . . . . . . . . . . . 54
4.14 Fourier Transform of the second acoustic response . . . . . . . 55

A.1 Simulation of Cf values with OPA656 . . . . . . . . . . . . . . 59

A.2 Simulation of Cf values with OPA657 . . . . . . . . . . . . . . 60
A.3 Simulation of Cf values with AD8007 . . . . . . . . . . . . . . 60
A.4 Simulation of Cs values with OPA656 . . . . . . . . . . . . . . 61
A.5 Simulation of Cs values with OPA657 . . . . . . . . . . . . . . 61
A.6 Simulation of Cs values with AD8007 . . . . . . . . . . . . . . 62
A.7 Simulation of Rf values with OPA656 . . . . . . . . . . . . . . 62
A.8 Simulation of Rf values with OPA657 . . . . . . . . . . . . . . 63
A.9 Simulation of Rf values with AD8007 . . . . . . . . . . . . . . 63
A.10 Simulation of Rs values with OPA656 . . . . . . . . . . . . . . 64
A.11 Simulation of Rs values with OPA657 . . . . . . . . . . . . . . 64
A.12 Simulation of Rs values with AD8007 . . . . . . . . . . . . . . 65
A.13 Amplifier circuits drawn in Eagle . . . . . . . . . . . . . . . . 66

68
List of Tables

1.1 Speed of sound in different media [5] . . . . . . . . . . . . . . 1

2.1 Components base values used for theoretical investigation . . . 16

3.1 Input characteristics of selected opamps . . . . . . . . . . . . 22

3.2 OPA656 and OPA657 Open loop gain . . . . . . . . . . . . . . 22
3.3 Component base values used for simulation . . . . . . . . . . . 26
3.4 Choosen values for one stage amplifier . . . . . . . . . . . . . 33
3.5 Choosen values for two dual amplifier . . . . . . . . . . . . . . 37

4.1 Equipment list for experimental investigation . . . . . . . . . . 39

4.2 Final build values for both amplifiers . . . . . . . . . . . . . . 41
4.3 Supply voltage measured on the amplifiers . . . . . . . . . . . 43
4.4 Amplifier Noise Level . . . . . . . . . . . . . . . . . . . . . . . 44
4.5 Amplifier Noise Level with averaging over 1024 samples . . . . 46
4.6 Key data from Figure 4.6 with 22 pF capacitor . . . . . . . . . 47

69
70
Bibliography

[1] Adel S. Sedra and Kenneth C. Smith. Microelectronic Circuits. 5th

edition, 2004.

[2] John P Bentley. Principles of Measurement Systems. 4th edition, 2005.

[3] Analog Devices. AD8007 Datasheet. http://www.analog.com/

static/imported-files/data_sheets/AD8007_8008.pdf.

[4] Sergio Franco. Design with Operational Amplifiers and Analog

Integrated Circuits. 4th edition, 2014.

[5] Douglas C. Giancoli. Physics: Principles with Applications. 6th

edition, 2005.

[6] Charles Hellier. Handbook of Nondestructive Evaluation. 2001.

[7] Texas Instruments. OPA656 Datasheet.

http://www.ti.com/lit/ds/symlink/opa656.pdf.

[8] Texas Instruments. OPA656 PSpice Model.

http://www.ti.com/lit/zip/sbom136.

[9] Texas Instruments. OPA657 Datasheet.

http://www.ti.com/lit/ds/symlink/opa657.pdf.

[10] Texas Instruments. OPA657 PSpice Model.

http://www.ti.com/lit/zip/sbom137.

[11] James H. McClellan, Ronald W. Schafer, and Mark A. Yoder. Signal

Processing First. 2003.

71
[12] Walt Jung. Op Amp Applications Handbook. 2005.

[13] Measurement Specialties. FDT Series Piezo Sensor.

http://www.meas-spec.com/product/t_product.aspx?id=2482.

72