Department of Science and Technology Institutionen för teknik och naturvetenskap Linköping University Linköpings universitet

gnipökrroN 47 106 nedewS ,gnipökrroN 47 106-ES

LiU-ITN-TEK-G-14/081--SE

Over voltage protection devicefor ROV

David Kantzon

Sebastian Lahti

2014-12-16

LiU-ITN-TEK-G-14/081--SE

Over voltage protection devicefor ROV

Examensarbete utfört i Elektroteknikvid Tekniska högskolan vid

Linköpings universitet

David KantzonSebastian Lahti

Handledare Kjell KarlssonExaminator Lars Backström

Norrköping 2014-12-16

Upphovsrätt

Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare –under en längre tid från publiceringsdatum under förutsättning att inga extra-ordinära omständigheter uppstår.

Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner,skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat förickekommersiell forskning och för undervisning. Överföring av upphovsrättenvid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning avdokumentet kräver upphovsmannens medgivande. För att garantera äktheten,säkerheten och tillgängligheten finns det lösningar av teknisk och administrativart.

Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman iden omfattning som god sed kräver vid användning av dokumentet på ovanbeskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådanform eller i sådant sammanhang som är kränkande för upphovsmannens litteräraeller konstnärliga anseende eller egenart.

För ytterligare information om Linköping University Electronic Press seförlagets hemsida http://www.ep.liu.se/

Copyright

The publishers will keep this document online on the Internet - or its possiblereplacement - for a considerable time from the date of publication barringexceptional circumstances.

The online availability of the document implies a permanent permission foranyone to read, to download, to print out single copies for your own use and touse it unchanged for any non-commercial research and educational purpose.Subsequent transfers of copyright cannot revoke this permission. All other usesof the document are conditional on the consent of the copyright owner. Thepublisher has taken technical and administrative measures to assure authenticity,security and accessibility.

According to intellectual property law the author has the right to bementioned when his/her work is accessed as described above and to be protectedagainst infringement.

For additional information about the Linköping University Electronic Pressand its procedures for publication and for assurance of document integrity,please refer to its WWW home page: http://www.ep.liu.se/

© David Kantzon, Sebastian Lahti

LINKÖPING UNIVERSITY

Abstract

Department of Science and Technology

Bachelor of Science

by David Kantzon

Sebastian Lahti

Supplying power to equipment always imposes a risk of damaging it. This risk is appar-

ent in every application whether it be an industrial or a home appliance. One of these

harmful occurrences is something like lightning which introduces a potentially harm-

ful voltage in the system. To reduce the risk of damage significantly an over-voltage

protection device is needed. Devices which deal with this problem are common in any

electrical equipment and this report deals with the construction and evaluation of one

such unit.

The device or equipment needing protection in this case is an industrial underwater

robot built by Ocean Robotics. In order for this robot to operate safely it needs several

protective measures where an over-voltage protection system is one of them. This system

provides protection against over-voltages appearing on the main cable supplying the

robot, where the input voltage ranges from 200 to 600 volts. As the desired voltage

output range from the over-voltage system is 250 to 350 volts it must be able to handle

significant power in some cases.

Due to the restrictions on functionality given by the contractor a novel way of achieving

the goals was tested. The completed prototype can only dissipate 70 watts of power

which is far from the required 1500 watts which was specified in the sheet of requirements.

However, this system could be used to protect equipment with lower requirements for

power handling capability and with added time and effort it could possibly meet the

requirements for this project as well.

Acknowledgements

Ocean Robotics for providing this opportunity as well as guidance during our work

Lars Backström for being lenient and understanding

Pinkesh Sachdev at Linear Technologies for answering our questions

ii

Contents

Abstract i

Acknowledgements ii

List of Figures v

List of Tables vi

Abbreviations vii

1 Introduction 1

1.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.4 Scope and limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Background Theory 4

2.1 Effects of excessive voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.2 Protective measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.2.1 What needs to be protected? . . . . . . . . . . . . . . . . . . . . . 6

3 Design Overview 7

3.1 Hardware Design and Choices . . . . . . . . . . . . . . . . . . . . . . . . . 7

3.1.1 MOSFETs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.1.2 Digital pot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.1.3 Voltage regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.1.4 STM32 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.1.5 Transceiver Isolation barrier . . . . . . . . . . . . . . . . . . . . . . 11

3.1.6 Measure pin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.1.7 24 Volt input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.1.8 PCB Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.1.9 Component selection . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4 Evaluation 14

4.1 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4.1.1 Transient simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4.1.2 Observing output voltage during a spike . . . . . . . . . . . . . . . 17

iii

Contents iv

4.2 Physical testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.2.1 Operational test . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

4.2.2 Test with load - One MOSFET . . . . . . . . . . . . . . . . . . . . 20

4.2.3 Test with load - Two IGBTs in parallel . . . . . . . . . . . . . . . 21

4.2.4 Test with load - Second test with two IGBTs in parallel . . . . . . 21

4.2.5 Test with load - Alternative connection . . . . . . . . . . . . . . . 22

5 Result and discussion 23

6 Improvements 25

7 Conclusion 27

A Specification of requirements 28

B Circuit diagrams for high speed and regulator components 30

Bibliography 33

List of Figures

3.1 Circuit diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3.2 PCB design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4.1 Simulated circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4.2 Simulated results with 250V clamp . . . . . . . . . . . . . . . . . . . . . . 16

4.3 Simulated results with variable clamping . . . . . . . . . . . . . . . . . . . 17

4.4 Test configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

B.1 Microcontroller and peripheral components . . . . . . . . . . . . . . . . . 31

B.2 Communication interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

B.3 Regulator circuits for microcontroller and communication interface . . . . 32

v

List of Tables

4.1 Table of parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

4.2 Test with single MOSFET . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

4.3 Test with two IGBTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

4.4 Test with two IGBTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

4.5 Two controllers with one IGBT each . . . . . . . . . . . . . . . . . . . . . 22

vi

Abbreviations

LAH List Abbreviations Here

PCB Printed Circuit Board

IGBT InsulatedGate Bipolare Transistor

MOSFET Metal Oxide Semiconductor Field Effect Transistor

ADC Analog to Digital Converter

PSU Power Supply Unit

ROV Remotely Operated Vehicle

SOA Safe Operating Area

MATLAB Matrix Laboratory

vii

Chapter 1

Introduction

In modern marine industry, robots have taken over many of the roles traditionally held

by people. Tasks in deep waters such as exploration and maintenance are better suited

for machines where the risk for loss of lives is eliminated. These underwater vehicles must

function without fault due to the dangers involved in their work e.g. if the equipment

being serviced is an oil well. Malfunction at great depths may also make salvaging the

vehicle difficult or in some cases impossible.

Since the main method of supplying power to machines is by cable, it is also a link in

the chain where errors might occur. The power supply may malfunction but general

operation might also incur undesirable conditions such as switching noise or load dumps

when the machine requests less power than is supplied and so on. Many of these condi-

tions lead to an excessive voltage being supplied to the equipment. These voltages may

come in the form of short spikes or be present during a longer interval of time. If this

is the case then the equipment must be protected so it does not break down due to the

excess energy.

The components diverting this energy, often in the form of heat, must be carefully

selected to provide adequate protection under all conditions.

1.1 Purpose

The purpose of this thesis work is to construct an over-voltage protection device which is

able to handle a host of different harmful conditions. These conditions range from short

1

Introduction 2

spikes to sustained periods of excessive voltage. Some form of measurement capability

and connectivity with the operating system is also desirable. Furthermore the device

must be robust since it will be in use without recurring maintenance.

Design and construction must take available space, separation of high and low power

components as well as heat into consideration.

1.2 Aim

The aim of this project is to design an over-voltage protection device which can dissipate

large amounts of power when the supplied voltage exceeds the desired one by 50V. It

must be able to dissipate 1500W but 3000W or more is desired. Expected voltage

spikes in the system have a dV/dt of 300V/1ms which the device must handle as well

as sustained over-voltage conditions in the range of 5-10 minutes, preferably longer.

The device should have an adjustable clamp voltage ranging from 250 to 350 V. More-

over, the device must be able to operate for at least 4400 hours, preferably 10000 hours,

without malfunction.

These values were obtained from a specification sheet supplied by Ocean Robotics.

1.3 Methodology

Theoretical study of over-voltage protection systems is done to understand conventional

methods and their limitations, AltiumDesigner will be used for designing a schematic

and PCB. The PCB will be ordered from a manufacturer chosen by Ocean Robotics.

Moreover, software such as MATLAB and LTSpice IV are used to design and verify

functionality.

1.4 Scope and limitations

Most of the time spent working on the project was dedicated to the parts required for

functional over-voltage protection. It was decided that any software needed for commu-

nication and monitoring would be developed if the specifications for power dissipation

Introduction 3

were met. In the event that the power requirements were not met any communica-

tion and monitoring would be useless. Many components required for communication

were predetermined by Oceans Robotics since they had previous experience with their

functionality.

Chapter 2

Background Theory

Electronic equipment is dependent on power to operate. While voltage and current

brings life to a circuit it can also cause harm. One of the most common sources of

failure is exceeding the equipments rated working voltage. The most common example

of an over-voltage is lightning. Lightning strikes and other similar types of transient

over-voltages are often referred to as surges. Transient over-voltages have a short dura-

tion of a few microseconds after which they disappear [1].

Surges appear when large amounts of energy are released in a system. This may be

due to the switching of capacitive and inductive elements, connection or disconnection

of power-lines and general switching in and out of other circuit elements. Furthermore,

poorly designed supply- and distribution-grids can increase the susceptibility to distur-

bances.

Transient over-voltages can enter the equipment from the outside through power lines

and other mediums. However, they may also arise from inside. One such example is

the operation of electric motors as they are inductive elements. Moreover, a sudden

decrease in load conditions where more energy is supplied than is needed could cause

an over-voltage[1, p. 2]. Large enough transients may damage or immediately destroy

equipment. Weaker transient over-voltages applied repeatedly over time will also shorten

the lifetime of the equipment. Apart from surges over-voltages can also appear contin-

uously over time .

4

Background Theory 5

2.1 Effects of excessive voltage

When an over-voltage occurs the most common cause of damage is that the power dissi-

pated in a component goes beyond some maximum value. An excessive voltage can also

break down insulators or make a brief connection between conductors separated by air

if the electrical field is sufficiently large [2]. During conduction, large currents may flow

and harm or short circuit devices.

To protect devices from exceeding their rated values components such as a variety of

diodes, varistors, gas discharge tubes and so on are used depending on the applica-

tion. What they all have in common is that until the applied voltage reaches a certain

level they conduct very little current. Once this voltage called the clamping voltage is

reached they rapidly start to conduct larger currents to protect equipment further down

the chain. Another form of protection is called a crowbar device. These devices lower

the voltage to a certain level after the trigger-voltage has been achieved [2].

2.2 Protective measures

Devices such as varistors come with a certain clamping voltage from the manufacturer.

To achieve a higher overall clamping voltage, one can either get a bigger device or con-

nect several devices in series. Furthermore, current-handling capability can be increased

by placing several devices in parallel. The problem with parallelling is that there is

some variance in the clamping voltage between devices. This may lead to clamping at

the wrong voltage and in turn not have all the devices clamping at the same time. The

remedy here is to buy matched sets to ensure that all device parameters are tightly

distributed [3].

Background Theory 6

Thermal fuses are often used to prevent protective devices from conducting over sus-

tained periods of time. If too much current flows through the fuse, it breaks and no

power is supplied to the load after failure.

When the load drops rapidly the device might not be able to lower the input volt-

age fast enough which could lead to too high surges throughout the circuit. When these

surges arise they will destroy the circuit.

Lastly, most conventional methods of protection are not made to withstand continu-

ous periods of over-voltage. When the protective device conducts, heat is generated

which might cause the device to deteriorate and possibly lead to a catastrophic failure.

After seeing a number of excessive voltages the clamping voltage of a device may also

start to change. Changes like these may lead to some devices conducting when they

should not, eventually leading to failure, thus leaving equipment without protection.

2.2.1 What needs to be protected?

The ROV is powered through a so called tether-cable. The cable supplies power from

a six pulse rectifier with an additional twelve pulse rectifier acting as a booster. The

booster circuit adds voltage when the six pulse rectifier is not supplying enough power

by increasing the voltage supplied from the source. Since the rectifier circuit is on a long

cable, the changes that happens in the source will not happen instantaneously. The

rectifier circuit does not have a control loop for continuously supervising the applied

voltage. Because of this the possibility of extended periods with excessive voltage is

increased. This increases the risks of over-voltage in the ROV which is why it needs a

protective circuit.

Chapter 3

Design Overview

This chapter details the choice of components as well as some rudimentary theory needed

to understand these choices.

3.1 Hardware Design and Choices

The Linear Tech LTC4366 is a surge stopper that protects loads from surge transients

and continuous over-voltage. This is done by controlling an external MOSFET effec-

tively using it as a variable resistor. In the case of an over-voltage any excess voltage is

dropped over the MOSFET. Figure 3.1 shows the circuit diagram of the configuration

used in this application. For circuit diagrams of the microcontroller and communications

parts please refer to Appendix B.

The two resistors Rin and Rss allow the circuit to float up with the supply voltage

enabling the device to operate at higher voltages. RFB1 and RFB2 form a resistive di-

vider connected between the OUT and FB pins on the LTC4366. The values on these

two resistors decide at which voltage the device should start clamping.

The reason for choosing the LTC4366 instead of conventional components such as varis-

tors or similar components was because of easier scalability and control over the clamping

voltage. The LTC4366 can operate at any voltage as long as the external components

are sized properly. Furthermore, the clamping voltage can be set with more precision

7

Design Overview 8

compared to the predefined values of discrete components.

During start-up the LTC4366 operates from the voltage applied at the input. When

the voltage is sufficiently high it switches to operating from the output voltage. In this

mode of operation the control voltage to the MOSFETs will need to exceed the output

voltage for it to conduct. This is achieved by an internal charge pump charging the

capacitor Cg to the desired value.

Figure 3.1: Circuit diagram

The difference between the LTC4366-1 and 2 is that the LTC4366-2 has an auto-retry

function after a shutdown due to a fault. On the T imer pin there is an external ca-

pacitor. The size of the capacitor decides the allowed duration of the over-voltage fault

before shutdown. In this case the external capacitor was not used and as such both

versions of the 4366 could be used. The reason for neglecting the timer capacitor was

Design Overview 9

that no shutdown was to be allowed.

All calculations done in order to obtain the values seen in Figure 3.1 were done ac-

cording to the design example in [4, p. 16]

3.1.1 MOSFETs

Choosing the right MOSFET for the application is very important as the transistors must

handle considerable amounts of power. Among the parameters to consider are drain-

source breakdown voltage (BVDSS), drain-source on state resistance (RDSon), threshold

voltage (Vth) and safe operating area. Apart from the aforementioned parameters power

handling capability is also important.

The drain-source breakdown voltage BVDSS denotes the maximum voltage that the

MOSFET can block between its drain and source terminals while in the off state. Op-

eration at voltages above this is not guaranteed and may result in avalanche breakdown

causing large currents to flow through the device [5, p. 12]. BVDSS is important in this

particular application since the full supply voltage will appear over the MOSFET in the

case of an output short or over-voltage condition. Because of this a transistor must be

chosen with a BVDSS rating higher than the maximum supply voltage.

Drain-source on state resistance dictates the resistance that the device presents to the

circuit when in the fully on state. A low RDSon is preferred since this will limit the

voltage drop over the device when conducting large currents and consequently lower the

amount of power dissipated [5, p. 12].

The gate-source voltage at which the MOSFET starts to conduct, Vth, is called the

threshold voltage [5, p. 12]. If the voltage applied is less the transistor will conduct very

little to no current. Vth is also used in calculating the values for some other components

used with the LTC4366. Since there are variations in this voltage between components

of the same type, parallelling them to achieve higher power handling can cause problems.

This difference can cause the load over each MOSFET to differ making one component

dissipate more heat than the others.

Design Overview 10

Safe operating area (SOA) describes under which conditions the device may operate

without breaking down. This includes the drain-source voltage and current as well as

time of operation [5, p. 10]. These values were ignored in this particular application

since continuous power dissipation results in going out of these bounds. What kind of

power which can be dissipated continuously is dependant on cooling and other factors

which must be determined through testing.

More on MOSFET parameters can be found in [5].

In this project IXFN44N100Q3 MOSFETs were chosen because they met the required

voltage and current handling capability. They were also picked in large part due to

a ”bigger is better” mentality as this would result in a larger thermal mass and as

such increased cooling. Another reason for choosing this transistor is because better

transistors were not available at the time. Waiting for these to become available would

have made testing impossible in the time frame of the project.

3.1.2 Digital pot

The clamping voltage is decided by the resistive divider. Increasing RFB1 will lower

the clamping voltage and vice versa. To control the output voltage one could wire

a potentiometer as a variable resistor in series with RFB1. The value of the fixed

resistor then sets the upper limit of the output voltage. Any resistance added by the

potentiometer will lower this limit.

An analog potentiometer works in a testing environment but something that could be

controlled remotely is desired.

3.1.3 Voltage regulators

All voltage regulators used in the project were made by TRACOPOWER. This is in large

part due to the fact that they require no peripheral components to function although

decoupling capacitors were used in some areas of the board. TRACOPOWER devices

were also recommended by Ocean Robotics because of their familiarity with the products

along with the aforementioned lack of requirements for peripheral components.

Design Overview 11

Apart from a 3.3V supply for logic components the transceiver connection needs an

isolated 5V source. For this application a 5V regulator from TRACOPOWER was

chosen. The regulator has a required input voltage of 9-34V.[6]

3.1.4 STM32

The STM32 is a microcontroller whose main function is to handle the communication

between the user and the protection device. It supervises the output voltage and controls

the clamping voltage.

The STM32 is chosen for its interchangeable capabilities, this means that it is possible to

use any STM32 with different properties and still have the same pin-out. It is also in the

specifications received by Ocean Robotics. For this particular project an STM32F303C

was used.

3.1.5 Transceiver Isolation barrier

Communication interfaces such as RS-232 or CAN are susceptible to noise and other

disturbances. While the logic operates at 3.3V there is no telling what voltages may

appear in a cable connecting devices together in a harsh environment. Differences in

potential lead to currents flowing which if sufficiently large could damage or destroy

equipment. Because of these potential differences the data interface needs to be galvani-

cally isolated. Data must be able to cross the isolation barrier while preventing the flow

of current.

A common solution is the use of optocouplers. An optocoupler consists of a light-

emitting source and a photosensitive device which will turn on or off when subjected

to light. Each side has its own power supply that are isolated from each other as well.

Because there is no physical connection between the two sides, no current may flow

between them. Any over-voltage appearing on the transmission line is also blocked.

For the communication interface in this particular application a MAX3313 RS-232

transceiver from Maxim Integrated was used. In order to create an isolated interface an

Design Overview 12

ADUM1201 from Analog Devices was also added. These two devices were chosen be-

cause they had been used by Ocean Robotics in other projects and as such were familiar

with them.

3.1.6 Measure pin

It is desirable to measure the output voltage for logging purposes as well as warning the

user in case of an over-voltage. The measure pin (PB2 pin 20) serves as an input to an

ADC. This input is fed by a voltage divider which provides 0-3.3V depending on the

output voltage of the circuit. The acquired data should then be sent over either RS-232

or CAN.

The two resistors used in the divider were chosen in such a way that they could handle

the high voltages while still providing an acceptable signal for the ADC.

3.1.7 24 Volt input

To be able to use the STM32 without connecting the 300V power supply a 24 Volt power

input was added as a safety and testing measure. This input could also be used in the

event that the resistive divider scheme did not work.

Design Overview 13

3.1.8 PCB Design

Figure 3.2: PCB design

The circuit board was designed by trying to keep the low power components on one

half and the high power component on the other half. Keeping low and high power

components on separate parts of the board also separates high and low speed components

at the same time. This is a common practice when designing circuit boards. The low

power components were also connected to a ground-plane for easier routing.

3.1.9 Component selection

Some devices such as the microcontroller and communication interfaces were chosen

because Ocean Robotics had previous experience with them in other projects. This

simplified the design process as no thought had to be put into what kind of voltages

they had to handle and so on.

Any components in the high voltage part were chosen with respect to power and voltage

handling capability. This in turn imposed some requirements on the physical size.

Chapter 4

Evaluation

The circuit was evaluated via simulation as well as physical testing. This section high-

lights the differences between the two as well as the problems encountered during testing.

4.1 Simulations

As stated in the methodology, LTSpice IV was used to do all the simulations. LTSpice IV

contains models of every device produced by Linear Technologies as well as all common

parts such as resistors, capacitors and so on. This enables the user to focus on actual

development without the need to find all the appropriate models.

4.1.1 Transient simulation

In order to verify that the calculated values of various components were correct they

were simulated. The values of all the components were obtained using formulas from

the LTC4366 datasheet in MATLAB[4].

14

Evaluation 15

Figure 4.1: Simulated circuit

The simulated circuit can be seen in Figure 4.1. Because the MOSFETs used on the

physical board were not available as models, devices with similar characteristics were

used in simulation. Moreover, the simulated circuit uses four devices in parallel whereas

the physical circuit only uses one or two in the various tests. In addition, the various

devices will not present different resistances or have differing threshold voltages because

of ideal models. As such, no information on whether these common occurrences will

cause problems can be found.

In this simulation the input voltage was set to start at 300V. After a period of time a

spike topping out at 600V with a dV/dT of 300V/1ms appears, after which the voltage

settles at 400V. The maximum voltage and dV/dT was obtained from the maximum

expected values in the specification sheet. Furthermore, the output voltage was set to

clamp at 250V using R7 and R1. R7 is equivalent to RFB1 in Figure 3.1 and R1 acts

as the potentiometer. Moreover, with R1 at or close to zero the clamping voltage is

set to 350V with a subsequent lowering of the output voltage as R1 is increased. R1 is

indicated in Figure 4.1 with an arrow.

Evaluation 16

Figure 4.2: Simulated results with 250V clamp

Figure 4.4 shows the curves for the input and output voltages. In the blue curve the

input voltage with the added spike going up to 600V can be seen. The output voltage

is seen in the green curve. The initial part is where the circuit is in its startup phase

after which the voltage stays at the specified clamp voltage.

No information on power dissipation was gathered from this simulation since such pa-

rameters would have to be obtained through physical tests.

Evaluation 17

4.1.2 Observing output voltage during a spike

After verifying that the circuit worked, at least in a virtual environment, a test was done

to see how fast the output voltage increases when a spike is applied. The same circuit

as in the previous test was used but the clamp voltage was set at 350 V by lowering R1.

Input voltage was set to the same waveform as in the previous test.

Figure 4.3: Simulated results with variable clamping

In Figure 4.3 the waveforms of the input and output voltages can be seen. The output

voltage follows the input very closely in the beginning with a small difference possibly

due to the internal resistance of the MOSFETs. As the spike appears the output voltage

ramps up to 350V and stays there even though the input has settled at 400V.

Evaluation 18

4.2 Physical testing

The tests focused on sustained periods of over-voltage. To generate controlled transients

of sufficient magnitude special equipment is needed which was not available. One can

assume that if the device can withstand long periods of over-voltage it should also be

able to withstand shorter spikes.

During testing the two MOSFETs were mounted to a fan-cooled heat sink. Both resistors

responsible for dropping the voltage down to a suitable level for other electronics were

also mounted on a smaller fan-cooled heat sink. To measure the temperature of the

various components a laser measurement tool was positioned approximately 30 cm away.

Figure 4.4: Test configuration

Evaluation 19

4.2.1 Operational test

A first test to determine whether the circuit was at all operational when mounted on

the board was done with a large power resistor as a load. The potentiometer controlling

the clamping voltage was set to allow the maximum clamping voltage of 350V.

Voltage applied to the circuit was slowly ramped up but no significant voltage was

detected at the output. When measuring the voltage over the LTC4366 it appeared that

the supplied voltage was too low. Since the values on Rin and Rss were chosen close to

the maximum allowed values according to calculations found in the data sheet [4, p. 16]

it was decided that these should be lowered. New values were taken from an example

on page 19 in the datasheet which had similar constraints on input and output voltage.

After replacing Rin and Rss the circuit started operating correctly and the voltage

measured across the load resistor followed the input voltage up to the limit set by the

resistor divider. Changing the value of the potentiometer also resulted in changes to the

output voltage.

During this test the two resistors responsible for dividing the output down to levels suit-

able for low power electronics became overheated and subsequently destroyed. Because

of this it was decided that any low power electronics in the circuit should have their own

24V feed line.

Evaluation 20

4.2.2 Test with load - One MOSFET

In this test only one MOSFET was connected to determine what kind of power it could

sink before breaking. The output voltage was set to a certain value while the input was

ramped up in steps to slowly increase the power dissipated by the MOSFET. A 500W

halogen lamp served as a load in this experiment.

To determine the power dissipated P = U · I was utilized where U is the voltage over

the MOSFET (Vin-Vout) and I is the current drawn by the load. Table 4.1 describes the

various parameters.

Vin (V) Input voltageVout (V) Output voltageI (A) Load currentT (C) MOSFET case temperaturePdis (W) Dissipated power

Table 4.1: Table of parameters

Vin (V) Vout (V) I (A) T (C) Pdis (W)

191.1 189.7 1.604 26.7 2.2456

193.2 189.7 1.599 29.5 5.5965

195.3 189.7 1.599 33 8.9544

197.4 189.8 1.598 35.6 12.1448

201.5 189.8 1.598 40 18.6966

205.7 189.8 1.597 49 25.3923

207.7 189.8 1.597 52 28.5863

209.8 189.9 1.597 56 31.7803

214 190 1.598 62 38.352

218.1 190.1 1.599 64 44.772

224.3 190.2 1.599 82 54.5259

228.5 190.3 1.599 89 61.0818

232.7 190.4 1.6 100 67.68

Table 4.2: Test with single MOSFET

The MOSFET could withstand a load of about 70W before overheating as seen in

Table 4.2. Even though the case temperature measured did not exceed the maximum

value, the junction temperature was probably higher inside the device resulting in failure.

Evaluation 21

4.2.3 Test with load - Two IGBTs in parallel

Since one of the two acquired MOSFETs had failed there was no way to test two of them

in parallel. Two IGBTs with similar characteristics were soldered to the board and tested

instead. IGBTs were chosen due to the fact that they are voltage controlled devices.

Akin to the MOSFET they have their characteristic parameters such as threshold voltage

but also share similarities with bipolar transistors. All measurements were conducted

in the same manner as earlier tests.

Vin (V) Vout (V) I (A) T1 (C) T2 (C) Pdis (W)

200 190.4 1.604 31 33.5 15.3984

210.4 190.4 1.604 40.5 40 32.08

200 190.4 1.603 33 34.6 15.3888

218.7 190.4 1.6 67 32 45.28

Table 4.3: Test with two IGBTs

The temperatures shown in Table 4.3 indicate that the load is not shared symmetrically

by the two devices.

4.2.4 Test with load - Second test with two IGBTs in parallel

This test was done in the same manner as the other test conducted with the IGBTs. The

difference here is that the current flowing through each transistor was also measured.

Vin (V) Vout (V) I1 (A) I2 (A) Itot (A) T1 (C) T2 (C) Pdis (W)

191.5 190.4 0.44 1.161 1.601 32 32 1.7611

201.8 190.4 0.433 1.167 1.6 35 36 18.24

210.2 190.4 0.36 1.24 1.6 36.2 52 31.68

218.5 190.4 0.26 1.3404 1.6004 31 70 44.97124

Table 4.4: Test with two IGBTs

Looking at Table 4.4 it is clear that the load is once again not shared symmetrically

across the two transistors.

Evaluation 22

4.2.5 Test with load - Alternative connection

This test was done with two PCBs, where each board had its own controller and IGBT.

The two boards were connected to each other through the FB pins on the controller

circuit. This approach did not solve any issues. The current in the table shows the total

current drawn by the circuit.

Vin (V) Vout (V) I (A) TPCB1 TPCB2 Pdis (W)

201.6 188.8 1.593 26 27 20.3904

209.8 188.3 1.593 27.7 29.8 34.2495

218.1 188.5 1.593 29.6 30.9 47.1528

230.6 200 1.67 60 30 51.102

Table 4.5: Two controllers with one IGBT each

Chapter 5

Result and discussion

This chapter describes some of the problems encountered during the project.

Although the circuit was functional and capable of some continuous power dissipation it

did not meet the specifications. A peak power dissipation of around 70W was achieved

before device failure occurred. This is far from the desired power dissipation of at least

1500W. To achieve this level of protection more transistors need to be added if the cur-

rent design is to be used. Furthermore, whatever the protection scheme used, the device

needs to be bigger.

From the specifications sheet a dimensional limit of around 10x5x4 cm was obtained.

Designing the device with roughly this size in mind was done to avoid any complications

later. These complications might arise if the device did not fit the box in which it was

to share space with other electronics.

The box itself was to be made out of aluminium and fit in the ROV, thus submerged

in water during operation. However, with this in mind, the allotted dimensions for the

protective device are still small considering the amount of power that might have to be

dissipated for extended periods of time. Furthermore, no actual testing with the box

submerged in water could be done since final dimensions depended on other electronics

being added. The improved cooling of a submerged box might have increased the power

handling capability.

23

Results and Discussion 24

The main reason why the circuit does not work as intended is because the actual specifi-

cations of the MOSFETs do not apply when using the transistor in this manner. Using

the MOSFETs in parallel was a bit harder than expected. The reason for this is because

the on-resistance or threshold voltage of each MOSFET might differ. Even though this

difference might be small it could start a snowball effect which eventually leads to one

or more MOSFETs failing leaving devices down the line without protection.

Chapter 6

Improvements

This section explores different improvements or changes that could be made in order to

meet the requirements. Some overlap with the previous chapter do occur but with more

focus on the actual solution rather than the problem.

An idea that was explored in this project was to connect more transistors in parallel

to increase power handling. However, this is complicated by the fact that there exists

variations between devices from the same series. Because of these variations e.g. in the

threshold voltage, the transistors do not share the load symmetrically.

By increasing the number of devices in parallel the effect of the variations in param-

eters should decrease. Even though the load may not be shared symmetrically across

all transistors the difference should still be smaller than before. In addition, a device

with tight parameter distribution could be used to further insure a symmetrical sharing

of the load between devices. However, since the power handling does not scale linearly

with the amount of devices used the exact number needed would have to be determined

experimentally.

Another idea is to use a number of circuits in series. Each circuit would contain the

necessary components such as controller, MOSFET and so on. These would then be

configured so that each step in the circuit clamps at a certain voltage akin to a resistive

ladder. Connecting circuits in this way should remove the issue of different transistors

having differing threshold voltages since each transistor now has its own control circuit.

25

Improvements 26

If the clamping voltage for each circuit increases stepwise the load should also be shared

by all the devices.

Most other components need not be changed from circuit to circuit except the ones

that control the actual clamping voltage. This is where problems arise since controlling

the output voltage from the entire circuit would require controlling a large number of

sub-circuits.

To actually control the clamping voltage a digital potentiometer would be ideal since it

could be set by the micro-controller housed on the same PCB. However, most digital

potentiometers only allow for supply voltages up to 60V. Since the voltage over the

circuit relative ground would be over 200V this would not work. One solution would

be to let the potentiometer along with its regulator float with the supply voltage. This

still leaves the problem of interfacing logic signals with a device driven by 3.3 V. No

matter the configuration, be it series, parallel or a combination of the two, some sort of

automatic control should be added to the power supply. By sensing the current drawn

from the PSU and knowing what kind of voltages appear during these load conditions,

the PSU could increase or decrease the supplied voltage to the ROV. In conjunction

with this a circuit such as the one designed in this project could be used to handle any

over-voltages that might occur from the control system failing or being too slow.

Chapter 7

Conclusion

There are a number of ways to protect circuitry from excessive voltages. All of these

come with their various pros and cons. This work details the design and evaluation of a

protective device using somewhat unconventional technology because of the restrictions

imposed in the sheet of specifications. It might not be the most efficient solution but it

provides the functionality required although at a much too low power handling capability.

All in all, this device could be used to protect circuits from high voltage spikes and

extended over-voltages as long as cooling needs are met. In its current form it is not

suitable for handling power above 70W but with increased available space and cooling

this could very well be achieved.

27

Appendix A

Specification of requirements

28

Appendix A 29

Ocean Robotics International

Address: Teknikringen 7 Tel: +46 (0)761 86 86 30 www.ocean-robotics.com S-583 30 Linköping Fax: +46 (0)13 15 20 66 SWEDEN

Teknisk spec överspänningsskydd Ett överspänningsskydd som ska kunna hantera olika utspänningar genom att vara ställbart och

kunna klippa överspänningsnivåer, även kontinuerliga sådana inom visa gränser.

Inspänningsnivåer

200VDC – 600VDC

Utspänningsnivåer

250VDC – 350VDC, ställbart genom t.ex. mjukvara, spänningsdelare, något annat…

Spänningsspikar

De största spikar som förväntas i systemet har en dV/dt på 300V/1ms. Spikar måste klippas innan de

hunnit bli för stora men ska inte kunna hanteras kontinuerligt.

Effekthantering

Då överspänningen är högst 50VDC över inställd utspänning ska skyddet klara att kontinuerligt sänka

en effekt på:

Krav: 1500W Önskemål: 3000W

Mätning

Spänning ska kontinuerligt mätas, både för skyddet(?) samt för loggningsändamål.

Hårdvara för mjukvara

Ev. mikroprocessorer i systemet ska helst vara av modell STM32F303. Kommunikation ska ske över

isolerade gränssnitt (CAN, RS-232) då kommunikation med omvärlden behövs. Lågspännings-

elektronik bör drivas av den övervakade spänningen.

Kylning

Skyddet ska monteras i en aluminiumlåda, delar som behöver kylning kan skruvas direkt i väggen.

Alltsammans kommer sedan att vara i vatten under drift, så bra kylning finns.

Storlek

Då alltsammans ska monteras på en relativt liten undervattensfarkost krävs att hänsyn tas för att

välja komponenter som är så små som möjligt och förpackningen kan bli så liten som möjligt. För

laborationskort krävs inte liten storlek men det måste beaktas så inga överraskningar kommer då

slutlig produkt ska utvecklas.

Appendix B

Circuit diagrams for high speed

and regulator components

30

Appendix B 31

Figure B.1: Microcontroller and peripheral components

Appendix B 32

Figure B.2: Communication interface

Figure B.3: Regulator circuits for microcontroller and communication interface

Bibliography

[1] Chris J. Salmas Kostas Samaras, Chet Sandberg and Andreas Koulaxouzidis. Elec-

trical surge protection devices for industrial facilities - a tutorial review. Techni-

cal report, IEEE, 2007. URL http://www.electricitysaversplus.com/uploads/

SurgeSupressionTutorial.pdf.

[2] R. Schainker. Effects of temporary overvoltage on residential products - system com-

patibility research project. Technical report, Electric Power Research Institute, 2005.

URL http://www.hvacovervoltage.com/info/EffectsOfOvervoltage.pdf.

[3] AN9768 - Transient Suppression Devices and Principles. URL http:

//www.littelfuse.com/~/media/electronics_technical/application_notes/

varistors/littelfuse_transient_suppression_devices_and_principles_

application_note.pdf.

[4] LTC4366-1/LTC4366-2 High Voltage Surge Stopper, LT 0812 REV C edition. URL

http://cds.linear.com/docs/en/datasheet/436612fd.pdf.

[5] AN11158 - Understanding power MOSFET data sheet parameters, v.3 20130107

edition. URL http://http://www.nxp.com/documents/application_note/

AN11158.pdf.

[6] DC/DC Converters, TMR 1 TMR 1SM Series, 1 Watt, Rev. August 30. 2013

edition. URL http://www.tracopower.com/fileadmin/medien/dokumente/pdf/

datasheets/tmr1.pdf.

33