Cape Peninsula University of Technology

Digital Knowledge
Cape Technikon Teses & Dissertations Teses & Dissertations
1-1-2000
Automatic frequency control of an induction
furnace
Irshad Khan
Cape Technikon
Follow this and additional works at: htp://dk.cput.ac.za/td_ctech
Tis Text is brought to you for free and open access by the Teses & Dissertations at Digital Knowledge. It has been accepted for inclusion in Cape
Technikon Teses & Dissertations by an authorized administrator of Digital Knowledge. For more information, please contact barendsc@cput.ac.za.
Recommended Citation
Khan, Irshad, "Automatic frequency control of an induction furnace" (2000). Cape Technikon Teses & Dissertations. Paper 58.
htp://dk.cput.ac.za/td_ctech/58
AUTOMATIC FREQUENCY CONTROL 0.1' AN
INDUCTION FURNACE
!'resellled by
IRSHADKHAN
This thesis is submitted in fulfillment of the requirements for the degree of
MAGISTER TECHNOLOGIAE
in the department of
ELECTRICAL ENGINEERING
at the
CAPE TECHNIKON
Supervised by: Prof. J Tapson and Prof. B Mortimer
December 2000
ACKNOWLEDGEMENTS
I thank my creator and my parents for gIVing me the strength to achieve this
qualification. I would also like to thank my brother Azeem whose support and
encouragement will always be appreciatcd.
A big thank you goes out to Dr 10nathan Tapson, my Thesis Supervisor, to whom I
will always be grateful to, for his sound advice, guidance and confidence in my
ability.
Thank you also due to Mr A 1 Capmbcll , Prof. B Mortimer and Mr S Hamin ftlr their
support and advice throughout this projcct
A special thanks also goes out to thc following pcople who provided valuable advice
and support towards this project:
Universitv of CaDe Town:
Colleagues: Dr lan de Vries and ]e\on Da\ies
Principle Technical Officer: Mr. Stc\cn Schrire
Departmental assistants: Mr Albcrt Martin and Mr. P. Daniels
Lecturing Staff: Prof. J. R. Grccne. Prof. 1. Bell.
11
SYNOPSIS
The development of an automatic frequency control system for a miniature high
frequency induction furnace is described.
A background study into the fields of induction-heating, resonance, power electronic
resonant converters and phase locked-loops are performed with relevance to this
research. An analysis of the resonant load circuit is performed by means of a
combination of measurement and numerical simulations. The study of the load
behavior and power source is used as a :001 to aid effective implementation of the
automatic frequency control system. This Jimulation data is used to detenninc the
operating frequency range of the RLL system.
A background study is performed in whieh several frequency-control schemes for
power electronic converters are investigated. A brief summary, in which the basic
requirements for a frequency control system with regards to this research are
presented.
Two revisions of the Automatic Frequency Control system (RLL) were implemented,
on the induction furnace. Experimental results on both systems (Revl and Rev2),
illustrating the necessity for frequency control are also presented.
Future suggestions for optimizing the loop performance are presented. Further steps in
the developmental process of the miniature high frequencj induction furnace are also
discussed.
III
TABLE OF CONTENTS
ACKNOWLEDGEMENTS i
SYNOPSIS ii
LIST OF ILLUSTRATIONS vi
I. Figures vi
2. Tables vii
CHAPTER I 1
INTRODUCTION I
CHAPTER 2 4
previous work 4
2.1 Previous Induction Heating Reseal ch 4
2.2 Background Study 6
2.2.1. Current Source Inverter using SIT's for Induction Heating
Applications 6
2.2.2 Discussion 7
2.3.1 High Power Ultrasound for Industrial applications g
2.3.2 Discussion 8
2.4.1 Half-Bridge Inverter for Induction Heating Applications <)
2.4.2 Discussion 10
2.5.1 PWM Inverter Control Circuitry for induction Heating 10
2.6 Summary II
Signal Conditioning 11
Stability 11
Speed 11
Protection 12
Initialization Procedure 13
CHAPTER 3 14
AN INTRODUCTION TO INDUCTION-HEATING AND PHASE LOCKED-
LOOPS 14
3.1 Background 14
3.2 Basics of Induction Heating .. 14
3.3 Hysteresis and Eddy-Current Loss .. 17
IV
3.4 Power Source 18
3.5 Choice of Frequency 20
3.6 Eddy current stirring 22
3.7 Resonance 23
3.7.1 Parallel Resonance 24
3.8 Phase locked-loops 27
3.8.1 Loop Fundamentals 28
3.8.2 Phase Detector 28
3.8.2.1 4-Quadrant Multiplier 29
3.8.2.2 Switch type phase-detectors 30
3.8.2.3 Triangular phase detectors 32
3.8.2.4 XOR Phase Detector.. 33
3.8.2.5 R-S Latch 33
3.8.3 Loop filter 33
3.8.3.1 Passive loop filter 34
3.8.3.2 Active loop filter. 34
3.8.3.2.1 Intcgrator and lead filter 34
3.8.4 Voltage Controlled Oscillator (VCO) 35
CHAPTER 4 36
IMPLEMENTATION OF AUTOMATIC FREQUENCY CONTROL 36
4.1 System description and operation 36
4.2 Loading Effect 37
4.3 Load circuit 38
4.3.1 Unloaded hcating coil 39
4.3.2 Copper work-piece 40
4.3.3 Steel work-piece 40
4.2 Concept of resonance locking 41
4.3 Resonance locking methodology 43
4.3.1 Signal Measurement 44
4.4 Control circuit implementation 45
4.4.1 RLL revision I 45
4.4.2 RLL revision 2 .
4.4.3 Discussion .
. 47
. 48
444 A '-L k ..., 49
.. ntl oc protectIOn eIrcuItrj .
v
CHAPTER 5 50
EXPERIMENTAL RESULTS 50
5.1 Revision 1 51
5.2 Revision 2 53
CHAPTER 6 56
CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK 56
6.1 CONCLUSIONS 56
6.2 RECOMMENDATIONS FOR FUTURE WORK 58
REFERENCES 62
APPENDICES I
APPENDIX B: Schematics IV
1. Schematic layout of induction furnccc IV
2. Automatic frequency control system Revision I V
3. Automatic frequency control system Revision 2 VI
APPENDIX C: Technical Data .
IRFF460 ENHANCEMENT-MODE POWER MOSFET .
IR2113 HIGH AND LOW SIDE MOSFET DRIVER .
AD734 HIGH-SPEED ANALOG MULTIPLIER .
VCA 610 AUTOMATIC GAIN CONTROL IC .
CD4046 CMOS PLL lC .
DG301 ANALOG SWITCH .
FIGURES
LIST OF ILLUSTRATIONS
VI
2.1 Simplified block diagram showing the layout of the induction billet 6
heater and its control circuit.
2.2 Simplified block diagram of ultrasonic power supply and control system 8
2.3 Block diagram showing system components of Half-bridge inverter 9
2.4
Block diagram showing frequency control system 9
2.5 Simplified block diagram of ultrasonic power source and frequency 10
control circuit
3.3
Layout of the high frequency power source 18
3.4
Half-bridge voltage-fed inverter topology 19
3.5 Full-bridge voltage-fed inverter topology 19
3.6 Current-fed full-bridge inverter topology 19
3.7 Cycloconverter or AC - AC converter 19
3.8 Current-fed chopper or quarter bridge 19
3.9 The variation in penetration depth in a gold work-piece over a frequency 22
range of 100kHz.
3.10 Frequency response curve of a resonant circuit 23
3.lla General representation of parallel resonant circuit 24
3.11 b Equivalent representation as seen by source 24
3.12 Frequency characteristic of a parallel resonant circuit 26
3.13 Block diagram of basic PLL confi/o,'uration 28
3.14 Sinusoidal characteristic of analog phase detector 30
3.15 Basic switch type P.D configuration 30
3.16 Switching waveforms illustrating phase detector opp ation 31
3.17 Sinusoidal characteristic of analog phase detector 32
3.18 Triangular phase detector characteristic 32
3.19 XOR phase detector characteristic
"7
~ -
3.20 R-S latch phase detector characteristic 32
3.21 Simplified representation of active loop filter 34
4.1 Schematic layout of Induction Furnace 36
4.2 Idealized equivalent circuit of induction heating load 38
4.3 Equivalent circuit for unloaded coil 39
4.4 Impedance characteristic for unloaded coil 39
4.5
4.6
4.7
4.8
4.9
4.10
4.11
4.12
4.13
4.14
4.15
5.1
5.2
5.3
5.4
5.5
5.6
5.7
Equivalent circuit for loaded coil (copper work-piece)
Impedance characteristic for loaded coil (copper work-piece)
Equivalent circuit for loaded coil (steel work-piece)
Impedance characteristic for loaded coil (steel work-piece)
Combined phase characteristic
Combined frequency response
Basic current-fed inverter configuration
Ideal waveforms for driving load circuit
RLL revision 1
RLL revision 2
Block diagram offrequency control system with anti-lock protection
Capacitively reactive tank circuit beiL
s
by the inverter
Gate voltage and transformed inverter midpoint voltage waveforms
locked 90 out of phase by loop 2.
Inverter operating with RLL rev. 1
Inverter operating with RLL rev. 2
Operation of zero-crossing detector and MOSFET drain-source voltage
Tank circuit driven at its natural resonant frequency
Heating cycle of a steel work-piece
vu
40
40
40
40
41
42
43
43
45
47
49
50
51
52
53
54
54
55
2. TABLES
3.1 Coupling effieieneies for several metals at room temperature 16
3.2 Dimensions of the gold work-piece to be melted in the induction furnace 21
4.1 Resonant frequencies for various metals at room temperature 37
Introduction
CHAPTER 1
INTRODUCTION
Induction heating is an important enabling technology in the platinum and gold
jewellery manufacturing industry. This industry is the key to adding value to the
mineral extraction industry of South Africa. Platinum jewellery has fast become a
growing trend in South Africa, the United States and most of Europe'. The nature of
this metal with its high melting point and difficulty in working however does not
easily lend itself to usage by the average jeweller using oxy-gas torches and silica
investment casts. The first reason for using this is that the high melting point of
platinum alloys dissolves the usual refractory materials, causing contamination of the
melt, and results in poor finished products. The second reason is the special mixture
of oxy-gas and hydrogen required for melting platinum is often expensive and
requires skilled labour to operate
2
This research is aimed at the jewellery manufacturing industry. Local jewellers seek a
suitable alternative to the conventional blowtorch
2
Induction-heating provides a faster
and cleaner melt than the conventional blowtoreh, producing a high purity
homogenous alloy brought about by the inherent stirring action of the induced eddy
currents. Commercially available large-scale induction furnaces do exist for the
jewellery manufacturing industry most of which are imported from Gcrnlany, Italy
and the USA. Price ranges of such systems vary between R75 000 and R245 000 per
unit. The technology to work platinum has been limited to t r ~ few who have imported
expensive units allowing them to monopolise the indu.itry with the aid of these
technological benetits.
The first aim of this research is therefore to provide the South African jewellery
industry with a low cost induction furnace capable of melting and alloying small
quantities of precious metal such as gold, silver and platinum.
The second aim of this research is to provide a laboratory standard induction furnace
capable of electrically heating any metal for experimental purposes. Thi, application
Introduction 2
would encourage research into the advancement of materials and metallurgical
research at tertiary institutions
3
.
The advent of solid state power sources for induction heating has enabled conversion
efficiencies of up to 93 % due to low switching losses and good high frequency
eoupling
4
Solid state power sources used to drive induction-heating loads are usually
very efficient, provided that the load is driven at its natural resonant frequency5 This
allows zero voltage (ZYS) and or zero current (ZCS) switching of the converter,
resulting in reduced power losses in the semiconductor switches
5
.
6
Another advantage
of driving a load at resonance is to enable an input power factor close to unity
allowing minimal KYA consumption
6
These parameters enable high conversion
efficiencies due to reduced switching losses i! the powcr source
7
.
8
.
9
The components of an induction furnace can be broken down into thrce main
categories namely:
I) Load circuit,
2) Power source and
3) Frequency control circuit
The basic load circuit and power source has already been developed in part during a
previous research project
lO
The current research project however, focuscs on the
development of a novel frequency control system for the induction furnace.
The induction-heating load forms part of a resonant tank circuit with a Q, which
varies from 3 to 18. The power source is used to drive this tank circuit at its resonant
frequency. The metal which is to be heated (work-p' xe), is situated inside a
refractory crucible, which is placed inside the heating coil. When the coil is loaded a
resulting shift in the resonant freqUency of the tank circuit occurs. This shi ft in
resonant frequency is directly related to the loading effect, which depends on the
resistivity of the work-piece and the efficiency of coupling between the work-piece
and the coil!!. This shift is compensated for by manually adjusting the driving
frequency of the power source to the new load resonant frequency. During a heating
cycle, an IIlcrease in work-piece temperature causes an increase work-piece
resistivity, whieh in turn also causes a shiti in the resonant irequenev of the tank
circuit. When melting metals by induction, it is predicted that a further shin In
Introduction 3
resonant frequency also occurs at the instant of melting. This phenomenon is
attributed to the fact that the resistivity of a liquid metal differs from that of a solid
metal. When dealing with magnetic metals, frequency shifts also occurs during a
heating cycle
I2

5
When heating a ferromagnetic material (eg. steel), the relative
permeability of that metal decreases with an increase in temperature, which causes a
large shift in resonant frequency when the metal is heated through its Curie point.
All of the factors mentioned above should be considered when heating and melting
various metals by induction.
A problem therefore exists when different metals are placed in the heating coil,
because it would require the operator of the induction furnace to manually tune the
system for maximum power and efficiency thoughout the process. This situation is
undesirable because human intervention is not always as accurate and reliable as
automatic control. An example of this situation occurs when heating a high melting
point metal such as platinum. This process requires continuous maximum power
transfer at all times. Incorrect manual tuning of the driving frequency could result in
the freezing of the precious metal at the instant of pouring, due to insufficient heating
above the metals melting point. The system also becomes less complicated to use,
once automatic frequency control is implemented. The system proposed for this
research is one that would automatically search for and operate at the natural load
resonant frequency, and continuously track this resonant frequency during the heating
cycle. This system will be referred to as the Automatic Frequency Control system
(AFC) or the Resonant Locked Loop control circuit (RLL).
This thesis describes the actual implementation of a no el Automatic Frequency
Control circuit to the existing prototype induction furnace. A brief study of previous
frequency control systems are presented and used as design guidelines. The control
circuit implementation is tested on the prototype induction furnace and results are
presented to verify its stability under power conditions.
Previous Work
CHAPTER 2
PREVIOUS WORK
2.1 PREVIOUS INDUCTION HEATING RESEARCH
4
This project was first undertaken at the University of Cape Town in 1995 by Dave
Dean, an engineer from the Materials Engineering Department. The project was
unsuccessful due to an inadequate inverter circuit. The design was reported to cause
continuous MOSFET destruction.
11
This was duc to incorrcct gating of the inverter
switches which caused cross-conduction (MOSFET's switchcd on in the same
inverter leg) resulting in large short-circuit currents in the inverter. The DC bus
voltage for the inverter was then stepped down to about 40 volts resulting in far
higher required MOSFET current ratings.
A second attempt was undertaken as a BTech project in 1996 by Leon Bardenhorst."
The gold work-piece could not be heated to more than 300"C due to dcvicc failurc in
thc inverter circuit. It was found that this was due to incorrect gating of the
MOSFET's and poor inverter layout. Another major problem encountered was
matching transformer saturation due to incorrect design.
A third attempt was undertaken by Marcello Bartolini as a BTech project in 1997
11
. It
was reported that a melting time of 30seconds was achieved for a gold work-piccc.
An input power of 4kW was used, and instability of the system during thc heating
cycle was reported. This problem was mainly due to poor' esonant circuit design and
inadequate load matching.
The previous three attempts all followed the same apprc)1ch, which utilized a Voltage-
Fed inverter driving a series resonant load. An extensive literature study was
performed into the contemporary topologies used for modem induction heating,
before a decision was made for this project. The Loughborough Uni\Crsity Institute of
Technology have performed extensive research into high frequency inductiun heating
power sources employing power MOSFET's ",.D It was clearly pointed out in this
publication that the Current-Fed topology driving a parallel resonant load circuit had
Prerious Work 5
proven to be far supenor In performance and operation, than the three prevIOus
attempts, utilizing the Voltage-Fed inverter. It was at this point that a decision was
made to employ a Current-Fed full-bridge load-resonant topology.
The fourth attempt was undertaken by the present author as a BTech project in 1998.
A working prototype system using a Current-Fed inverter was developed and
conclusive experimental results were presented
4
It was found that the system
operated efficiently off a single-phase supply, drawing less than 900W of input power
to melt 30g of 24-karat gold in less than 26 seconds.
The development of the basic IkW, 100 hHz switch-mode inverter (Current-Fed)
employing power MOSFET's was described i:. the research dissertation. Some good
design guidelines for the construction of the switch-mode inverter, which are crucial
at elevated operating frequencies, were also presented
4
The induction-heating load
formed part of a parallel resonant circuit and the development of this load circuit was
described. The achievement of these results, was mainly attributed to the sound
construction of the power source and careful design of the induction-heating resonant
load circuit.
The results and conclusions to the research described above have motivated research
into further development of the miniature high frequency induction fumace. The
initial system operated in open loop frequency control, which required the user to
manually tunc the operating frequency of the inverter to the natural resonant
frequency of thc load by monitoring the wave shape of the driving voltage across thc
load. The inverter switching frequency was manually tunc: to achieve zero voltage
transition (ZVT) switching in the power source. Failure to do so would result in a
mismatch between the driving frequency and the natural resonance of the load. This
would produce a fall off in inverter efficiency, and maximum power would not be
transferred to the load. A temporary solution was provided by adjustment of the load
circuit bandwidth to compensate for changes in load operating frequency. This proved
to be disadvantageous because the system eftlciency was not constant over the entire
operating range. It was therefore apparent that closed loop frequency control was
needed in order for the system to operate at maximum possible efficiency :ll all times.
Previous Work 6
The problem of frequency control is often encountered when driving loads of dynamic
resonant nature, such as resonant induction heating loads.
Investigations into the operation of several frequency control schemes were therefore
conducted as a basis for the current research. These various frequency control
techniques were studied in detail and discussions are presented to analyse each
system's overall perfonnance. A conclusive summary is also presented, in which the
fundamental requirements of a frequency control system are discussed as a basis for
the current research.
2.2 BACKGROUND STUDY
The work to be discussed concentrates on the ircquency control of resonant loads for
various induction-heating applications as well as a high power ultrasound application.
2.2.1. Current Source Inverter using SIT's for Induction Hcating Applications

Hlgn Power
Current Scurce
Inverter
Parallel Resonant
load Cif CUll
~ InducMnHeaLng
Load
lO:JPl
V,...., I.'JA.::

ADC
53 S4 51,52
LOOP2

ADC
Control LogiC V " C ~
Gate Drtve

CirCUits
PLl Ccnl:":;i
C,rcIA

ROM ~ OAC
GpLmurr.
Phase
;'''91,:
A control scheme was implemented for a 130 kHz, 7.5 kW full-bridge inverter for the
induction heating of iron billets by Akagi, el af 14. The simplitied bloek diagram of
the system is shown in figure 2.1. A current-fed topology was used to drive the
induction-heating load, whieh fonned part of a parallel resonant circuit. The
frequency control scheme employed essentially the switching of the SITs (static
Previous Work 7
induction transistors) at zero voltage in order to maximize converter efficiency. This
control was realized by employing two digital phase-locked loops. Optimal firing
phase angle control values for the SITs which were a function of the average load
current and the RMS load voltage, were pre-caleulated and stored in a 64 kbyte ROM
table. The average load output current and RMS output voltage were used as offset
addresses for the ROM table, which then gave the optimal phase angle to be used.
Loopl controlled the ON timing of the SIT switches in order to maintain a fixed
phase relationship between the load current and load voltage over the entire operating
frequency range. Loop2 provided zero voltage switching by locking the off timing of
the SIT switches to the load voltage.
2.2.2 Discussion
The system was reported to have perfo1Tl1ed well with an estimated inverter efficiency
of 95%. The response time of the system presented was limited by the following
factors:
Conversions of the output voltage and current from an analogue quantity to a
digital value.
Accessing data from the ROM table to produce the optimal phase angle value.
Digital to analog conversion of the optimal phase angle value to be synthesised by
the PLL circuit.
All conversions (0 - A and A - D) had a resolution of only eight bits which limited
the accuracy of the optimal phase angle control scheme. The EMI and induced noise
generated by the switching of the power source could have an adverse effect on the
operation of the frequency control schcme employed. The start-up sequence of this
system was achieved by first manually tuning the inverter switching frequcncy (by
adjusting the VCO) to the resonant frequency of the load. Once resonance was
achieved. the user would then switch to automatic operation. Thus no automatic start-
up was achieved.
Previol/s Work
2.3.1 High Power Ultrasound for Industrial applications
8
Clock
I
Half-Bridge
Inverter
i Tcaosduce,
1Resonant Load
VI.OAO
.;.
FIlter A
!LOAO
i
..
FIlter B
vco f.ol
I
,
I
..
Phase detector
FIg 2.2 S,mplified Block a,agr..rn of ultraso lie po""e, s.upply and control "'ys.tern
The research involved the driving of an ultrasonic transducer (Tonpiitz) at a power of
approximately 1.5 kW by Veldhuizen
ls
A simplified block diagram is shown in
figure 2.2. This project was aimed at ultrasonic eleaning applications. A half-bridge
voltage fed inverter was employed to supply the necessary RF power to the transducer
(load).
The transducer fonned a complex high Q resonant circuit (predominantly sencs
resonant) which required frequency control of the power source in order to lock to a
specific resonant mode in the transducer, hence delivering maximum power to the
load. This was achieved by locking the driving voltage and current to the load in
phase over the specified operating frequency range of approximately 20 - 40 kHz.
The frequency canIro! system employed the monolithic 4046 PLL le. Operating
mode 2 of the PLL, which utilized the R-S latch phase detector, was employed.
2.3.2 Discussion
The system was reported to be unstable due to poor loop filter design. The system was
very susceptibleta noise and EM! which cased the frequency control loop to lose lock
at high power levels. Special noise shielding techniques were employed to ensure
operation. The driving current signal to the load was embedded in noise due the
measuring technique emplvyed, and special filter circuitry was designed for signal
conditioning purposes (filter A and filter B). The susceptibility of the loop to noise
and EMl is characteristic of edge-triggered devices such as the R-S latch. The current
filtering circuitry employed matched 2
nd
order passive tilters on both tl'e output
Previous Work
9
voltage (VLOAD) and output current (I
LOAD
) in order to minimize phase errors over the
operating frequency range of the frequency control system. The signal conditioning
circuitry proved to be a critical factor in the design of this frequency control circuit. It
was therefore concluded that a frequency control circuit should have a low
susceptibility to noise and EMl produced by the power source, if reliable operation
was to be guaranteed.
2.4.1 Half-Bridge Inverter for Induction Heating Applications
I I I ser;espa:l
I Half-Bridge I .. Resonant
1
1llnverter - -,InductionHealing
Load
---- ~ , - -
I
1
T 1-' I
PLL System ~ -
Manual
Overide
J
- VL'JAD.,. - - ~ Loop Filter
Inverter
VCO
Fig. 2.3 Block diagram showtng system layout
Fig 24 Block diagram shoi'lmg frequency CDnlrol system
This research involved the development of a 6 kW, 50 - 150 kHz half - bridge IGBT
inverter for the heating of carbon steel billets above Curie temperature (780
n
C) for
industrial heating applications by Kamli
7
A simplified block diagram is shown in
figure 2.3. The frequency control circuit is shown in figure 2.4. The load circuit
formed a combined series parallel conllguration. which was driven by the imertcr. A
frequency control circuit was employed to track changes in the resonant load circuit.
This was realized by employing the monolithic 4046 PLL IC in operation mode It.
The control of the inverter was achieved by locking the control signals to the IGBT
switches, in phase with the zero crossing points of the load voltage. A simple PLL
control circuit was implemented to realize this control. A passive second order loop
filter was employed to provide the error voltage proportional to the phase di ITerence
between the two input signals. The system was started up by manually adjusting the
vca to resonance before switching over to automatic lock operation.
Previous Work 10
2.4.2 Discussion
The system operated well over the entire operating range and experimental results
showing the tracking operation of the control circuit were presented. The passive
second order loop filter employed does not provide for effective tracking and capture
operation resulting in a finite static phase error due to its low loop gain
l6
2.5.1 PWM Inverter Control Circuitry for induction Heating
---,
I
Series
Resonant
InductionHeating
Load
:.... ---llOAO-
,-r---r-
I I
r ~ ~
_._-.-
Full-Bridge
Inverter
I - ~
I
I
I
. ~
~ -
, I
PWM Controller : ~ - -1 PLL System
I
L- _
Fig. 2.5: Simplified block diagram showing the power source
and frequency control circuit
The research involved the development of a frequency control circuit for a 4kW,
70kHz, full-bridge Voltage Sourcc Inverter for induction heating applications by Ho'.
The basic system is shown in figure 2.5. This systcm was used to hcat carbon stcel
billets past their Curie temperature. A phase shifted PWM controller (UC 3825) was
also employed to generate the switching signals for inverter with the necessary dead
timings between transitions to prevcnt cross conduction of the power source. The
4046 PLL operating in mode Il was employed to achievc ~ e r o voltage switching of
the power source and ensure maximuill PO\\'Cr transfer to thc load at all times.
Frequency control of the power source was achieved by locking the measured output
voltage and current to the load in phase, over the operating frequency range. Thc
phase error produced by the type Il phase dctector was filtcred by a first order passiv'e
low pass filter and provided the DC reference voltage tor clocking the phasc shi fted
PWM controller. The system was started up manually by v'arying the driving
frequency of the clocking circuit until lock in operation occurred.
Previous Work
2.6 SUMMARY
I I
It can be concluded from the previous work discussed that the following problems
exist with frequency control circuits in resonant mode power sources:
Signal Conditioning
All the types of frequency control methods studied thus far reqUIre somc sort of
current and or voltage measurement technique in order to detect resonance. This
occurs when the load voltage and load current are in phase. A problem exists within
high frequency converters when measuring output currents. This is due high
frequency oscillations often being superimposed on the actual measurcd signal
17
Special signal filtering techniques are often required to "clean up" the signal, making
it compatible with standard analog and digital circuitry to be implemented for the
control stage. Passive filtering circuits have a finite frequency and phase response
over their operating range and often require a relatively narrow bandwidth to achieve
optimal filtering at the fundamental frequency. Thesc practicalitics often limit the
operating range of the system and produce phase shifts around its stable operating
point.
Stability
High frequency power sources are generators of electromagnetic interfcrencc (EMl).
The intense magnetic field produced in thc induction-heating coil is also a generator
of high frequency power radiation. These factors make operation of low-power
analogue control-circuitry difficult due to noise EMl produccd by the high power
circuits. Special noise shielding techniques arc usually reqUired to makc these low
power circuits immune to EM!. Digital circuits are relatively immune to EMl and
pose a feasible solution provided that the operating speed does not limit thc
performance of the system.
Speed
The natural resonant frequency of the load circuit is altered when the inductance of
the heating-coil changes. This change occurs by virtue of the loading effect produced
by the work-piece due to factors such as different conductivities, differellt coupling
distances from the surface of the work-piece to the inner coil surface, and the
Previous Work 12
changing relative penneability of the work-piece
6
. These changes occur whenever a
different work-piece is inserted into the coil and so the resonance point can never be
exactly the same. Another factor which alters the inductance of the coil, is when a
ferromagnetic work-piece (such as steel) is heated through its Curie point. The Curie
transition (approximately 780C for steel) causes the material to lose its magnetic
properties resulting in the relative penneability being reduced to unity from several
hundred at room temperature. This transition results in a rapid increase in the
penetration depth of the induced eddy currents. The work-piece is no longer a good
conductor of magnetic flux (due to Curie u, =1) and the amount of flux cutting the
work-piece changes, resulting in a dramatic decrease in the inductance of the coil.
This results in an increase in the natural resonant frequency of the tank circuit"'
Frequency changes in the load circuit can also occur when heating non-ferrous metals
past their melting points. A change in phase (from solid to liquid) in the metal results
in a change in the metal's conductivity which influences the inductance of the coil by
virtue of the magnetic field produced in it. The rate of change offrequency in the load
is detennined by the rate at which power is being delivered to the work-piece. Ideally
for efficient melting systems thc idea is to deposit energy into the work-picce at a rate
faster than what can be dissipated by the work-piece by virtue of its thennal
conductivit/
s
The control system to bc implemcnted should thcrcfore be able track
fast changes in the load resonant frcqucncy, maintaining lock at all times.
Protection
When operating the high frequency power source, a loss of lock in the frequency
control system could produce catastrophic results. If the system loses lock and drivcs
the inverter to a frequency away from the natural resonance of the load, the power
losses in the semiconductor increase rapidly and semiconductor failure could result
due to excessive power dissipation. These power losses are brought about by the loss
of zero voltage transition switching and the conduction of the integral body-diodes in
the switching elements, at operation away from resonance. A subsystem is therefore
necessary to detect a loss of lock in the frequency control circuit. It should then
attempt to force the system back into lock operation as quickly as possible or provide
a trip signal to the DC bus or isolate the load trom the Inverter by shutting down the
gate-drive signals to the inverter-bridge in the event of a malfunction occurring.
Previous Work 13
Initialization Procedure
All the induction heating frequency control systems studied thus far are started up
manually by tuning the vea to the resonant frequency before switching to automatic
frequency control. This drawback is due to the limited capture range of the PLL
control system employed, which makes automatic frequency control from start up
problematic. The frequency control circuits studied thus far also all employ passive
low-pass filtering techniques. Passive loop filters are undesirable in some systems
because of the static phase error produced by low loop gain. Low loop gain also
results in poor tracking operation
l
". Passive filters also have a limited capture rangc
due to their large bandwidth. Two crucial parameters of first order loops viz.: loop-
gain and loop bandwidth, cannot be independently adjusted and therefore do not allow
for effective operation at all times
l6
Active loop filters (e.g. 2
nd
order PI controller)
however provide much better tracking capability, capture range and minimal static
phase error compared to the passive type. These factors are essential for automatic
start up operation as well as good overall performance and will be investigated for this
research.
Ini/iallnvestigations
CHAPTER 3
14
AN INTRODUCTION TO INDUCTION-HEATING AND PHASE
LOCKED-LOOPS
3.1 BACKGROUND
Electromagnetic induction, the basis of all induction heating, was first discovered by
the "father" of induction, Michael Faraday in 1832. With his induced emf theory he
proved that currents could be induced in a elosed secondary circuit as a result of
varying the current in a neighboring primary circuit. The essential feature was a
change in the magnetic flux linkage with the elvsed secondary circuit, produced by an
alternating current in the primary. In 1927, almost a century later, the first medium
frequency induction furnace was developed by the Electric Furnace Company (EFCO)
and since then, the number and size of heating installations have grown steadil/
9
3.2 BASICS OF INDUCTION HEATING
Induction heating utilizes three main effects: electromagnetic induction, skin effect
and heat transfer. The heating is caused by the Joule heating effect when an
electrically conductive object called the work-piece, is placed in an alternating
magnetic field". This alternating magnetic field is set up in the water-cooled
induction coil. The induction heating coil and work-piece can be visualized as a
transformer with primary turns (work-coil) and a short-circuited secondary turn
(work-piece) 19 \Vhen alternating current flows in the primary, volt ages are induced in
the secondary which cause currents to flow in it and these Currents tend to cancel the
flux that produces them, according to Lenz's law
l
'. The frequency of these induced
Eddy currents in the work-piece is determined by the frequency of 'hc power source.
These eddy currents are induced into a peripheral layer of the work-piece known as
the skin-depth (0) or penetration depth which is characteristic of current tlow at high
frequency due to skin effect is given by:
. !P
0:::: --
V>,,"cf
/3.11
Initial Investigations
Where:
15
o= penetration depth
P = resistivity of work-piece
f = frequency of eddy currents
f! = permeability of work-piece which in this case is the same as free space, since the
work-piece is non-magnetic.
The skin depth is roughly where the current density has fallen to about one third its
surface value. The current density falls off from the surface to the center of the work-
piece and the rate of decrease is higher at higher frequencies 19 It is also dependent on
two properties of the material, i.e., resistivity and relative permeabiliti
H
Both the
penetration depth in the work-piece and the work-coil depend on the three parameters
shown in equation 3.1. The ideal situation is to maintain a good efficiency of coupling
between the coil and work-piece to ensure maximum power transfer. Coupling
efficiency is a measure of the amount of power transferred between the coil and work-
piece. The efficiency of coupling in this case is dependent on the resistivity of the coil
and that of the work-piece and is given by equation 3.2.
Where:
I
'1 ~ - ~ = =
1 +) Pc
P"f.!"
(3.2)
T] = coupling efficiency between the coil and work-piece;
Pc = electrical resistivity of the heating coil (which is usually oof! copper tubing)
p" = electrical resistivity of the work-piece
f!" = relative permeability of the work-piece
Initial Investigations 16
When deviating from the idealized concept of equation 3.2, the concept of coupling
efficiency is related to the term known as the coupling factor in conventional
transformer theory. In both cases the idea is to keep the primary and the secondary
closely wound or closely coupled to reduce flux leakage between the primary and
secondary windings, hence improving the power transfer
l9
. In induction heating, the
heating coil is considered to be the primary, and the work-piece is considered to be a
short-circuited secondary winding of a transformer.
Practical factors affecting coupling efficiency include:
Geometry of work-piece, which improves for a tightly packed, solid work-piecc
and decreases for a loosely packed work-piece due to leakagc flux.
Geometry of the heating coil, which improves for a closely wound coil around the
work-piece. Other factors also concerncd with geomctry are thc length of the coil
and the number of coil turns.
The material used for the heating coil. The higher the coil conductivity, thc lower
the l2R losses in the coil, hence the morc power transfcrred to the work-pieec.
Another important factor to be considered is the fact that matcrials such as gold,
copper and silver have relatively low resistivities at room temperaturc, which onec
again results in a low coupling efficiency at startup. Examples of coupling effieieneies
at room temperature are:
Metal
Resistivity(pzu
0
cl Efficiency (TJ)
Platinum 0.106 uOm
71.56 %
Gold 0.023 uOm
53.97 %
Copper 0.Gi673 uOm 50%
Silver 0.016 uOm 49.44 %
Table 3.1: The coupling elliciencit..'5 for sl:\cral metal:> are shown. In accordance to 2 it is c\iJcnt that
Jem resistivity metals rC:'iult in poor coupling cfticiencic-i at room IClllperaturc.
Equation 3.2 is the idealised condition and should be treat cd with care, but it gives a
broad-brush idea of what controls the coupling cfIicieney. If for example, onc
considers a matcrial with high resistivity and pcrnleability such as steel, an efficiency
Initial Investigations 17
approaching 100% can be achieved, but copper with a low resistivity, where the root
term (equation 3.2) approaches unity, has an efficiency of about 50%. This formula
applies for simple coils and is not valid for multi-layer coils where the coil current is
not limited to the skin depth
l9
. The efficiency increases during the heating cycle due
to fact that the resistivity of the work-piece increases with temperature as shown in
equation 3.3. The resistivity of the coil is kept constant by passing cooling water
through it thereby also keeping the losses in the coil to a minimum.
The heating of ferro-magnetic materials poses a special problem because of the Curic
point. Above the Curie temperature the relative permeability of the material reduccs
to unity, which results in a large increase in sk.n depth.
(3.3)
Where:
Po =The resistivity at any temperature e,
a20 = the temperature coefficient of resistance at a temperature of 20
D
C,
PI = the resistivity at temperature el.
3.3 HYSTERESIS AND EDDY-CURRENT LOSS
In conventional induction heating of magnetic materials such as steel, the heating is
caused by eddy-current losses that produce l2R heating and hysteresis losses.
Hysteresis loss is defined as the friction between molecules when the material is
magnetized first in one direction and then in the other. [he molecules may be
regarded as small magnets, which turn around with each reversal of direction of the
magnetic field
2
". Therefore in ferro-mat,'l1etic materials hysteresis lo"s improves thc
~ u c t i o n heating eftlciency. It is therefore concluded that for a material such as gold.
the heat generated in the work-piece can only be due to eddy-current loss sincc thesc
materials are non-magnetic.
Initial Investigations
3.4 POWER SOURCE
18
Induction heating power supplies are frequency changers that convert utility line
frequency (50Hz) power to the desired single-phase power at the frequency required
by the induction heating process
s
. The rectifier portion of the power supply converts
the single-phase line frequency input to DC, and the inverter portion changes the DC
to single-phase high frequency (100kHz) AC. This is illustrated in figure 3.3:
AC
-A
RECTIFIER

INVERTER

HEATING
~ '
SO Hz
v
AC-DC DC-AC
v
LOAD
Figu.-e 3.3: Layout of the high frequency power source showing the converter, im"erter and hc:.tting
Inverter circuits use solid switching devices such as thyristors (SCRs) and transistors.
For high power and lower frequencies, large thyristors are commonly used. For low
power or frequencies above 25kHz, transistors are used bccause of their ability to bc
turned on and off very fast with low switching losses
9
Vacuum tubc oscillators have
been used extensively for many years at frequencies above 300kHz. However, tube
oscillators have a low conversion efficiency of 55 to 60% compared to 85 to 93% for
inverters using transistors. Power vacuum tubes have a limited life of typically 2000
to 4000 hours and are therefore a costly maintenance item') The high voltage (over
10000 volts) required for tube operation is more dangerous than the 1000 volts or less
present in typical transistorized inverters. These negative features of tube oscillators
have brought about a dramatic move toward use of transis,Jrized power supplies in
heat treating applications that require a frequency of less than IMHz
9
Induction
heating power supplies utilize various techniques to produce the high frequency
alternating current. Various topologies are:
Initial Investigations 19
Half-bridge voltage-fed inverter topology (Figure 3.4)
Full-bridge voltage-fed inverter topology (Figure 3.5)
Current-fed full-bridge inverter topology (Figure 3.6)
Cycloconverter or AC - AC converter (Figure 3.7)
Current-fed chopper or quarter bridge (Figure 3.8)

Cf
\"
RL Lr Cr I
DC
Supply
1_C_f ---,\ 52
+I--T---r-
*01 \Sl
DC Cf j I
Supply . 1
04 $4
1
D3
D2
Lr RL
Figure 3.4: Half-bridge voltage-fed
inverter topology
Figure 3.5: Full-bridge \"oltage-fed inverter
topology
'"
RD YELLC',; SLTJE
J
C"

\@
I j
I , .
c , , 0 0
C"
,
\ ,.
\'"
\


,

G
0'
I
I
I
I
0
I
I
I
\'"
' .
D' i:: 02 I 03
* 0'
:i 06
T I f
I
C,

j
DC
Supply
Figure 3.6: Current-fed full-hridge io\encr
topulogy
Figure Cyc!lJCuo\crtcr or AC - AC comcrter
Figure 3.8: Curn:nt-tt:d
dWrrl'r or quarter hrJJg..o
Initial Investigations
3.5 CHOICE OF FREQUENCY
20
Frequency is a very important parameter in induction heating because it is the primary
control over the depth of current penetration and therefore the depth of heating
5
Frequency is also important in the design of induction heating power supplies because
the power components must be rated to operate at the specified frequency. Due to
reduced switching losses at elevated switching frequencies (up to lMHz),
enhancement-mode power MOSFETs have become an important component in high
frequency power sources for induction heating
13
For effective induction heating, the
frequency of the alternating magnetic field in the work-coil is of paramount
importance and is given by:
6.45 P
fld
2
Where:
fc ~ critical frequency
p ~ the electrical resistivity of the work-piece ("Dm)
d ~ the diameter of the work-piece (m)
!.1 ~ the permeability of the work-piece (Hm-')_
(3.4)
Equation 3.4 is defined as the critical frequency below which, a loss of heating would
occur due to field cancellation in the work-piece. The critical frequency is calculated
at a ratio of work-piece diameter to penetration depth (d/iS) > 4_5. Where a free choice
of frequency exists, it should be chosen greater than or equal tJ !c.!
Initial Investigations 21
Equation 3.5 shows the power loss per unit area in the work-piece written in tenns of
current density (Ji). Equation 3.6 shows the power loss per unit area in tenns of the
applied field Hs
2
at the surface of the work-pieceI
8
From equation 3.7 the relationship
between the power density (P) and the penetration depth can be seen. Equation 3.8
shows the relationship between the penetration depth and the applied frequency,
which is derived from equation 3.1. Equating equations 3.7 and 3.8 yields equation
3.9 which illustrates the relationship between the power density in the work-piece and
the applied frequency. It is therefore concluded that for a given work-piece and a free
choice of frequency, it is always advantageous to increase the frequency I8.
(3.5)
p=pHs
cS
(3.6)
I
Pce-
(3.7)
0
:>
PcefJ
(3.9)
I
80-
(3.8)
8
The gold work-piece has the following parameters:
Diameter O.Ol m
Length 0.013 m
Resistivity ( P200C) 0.024 uOm
Resistivity ( PlUMoc) melting point O.lnuOm
Permeability (f-l0) 4nx10' Hmi
Table 3.2: D i m ~ n s i o n s of the gold work-riece to b<.: mdl<.:d lO rh!." IOdUC!loo fumac!."
Initiul Investigations
E
0.35
-5
0.33
-5
0.31
~
0.29
"C
=
0.27
.S
"
0.25
~
"
0.23
=
~
""
0.21
;;0
0.19
50 60 70 80 90 100 110 120 130 140 150
III Applied frequency (kllz)
Figure 3.9: The variation in penetration depth in a gold \wrk-piccc over a
frequency range of lOOkllz.
22
In an induction heating application, the penetration depth (0) of the induced current in
the work-piece is inversely proportional to the applied frequency (equation 3.1 and
figure 3.9). It is common practice in most induction heating applications to make the
penetration depth (0), much smaller than the diameter (d) of the work-piece
l
'. I'). '''. 21.
The gold work-piece diameter is determined by the inner diameter of the crucible,
which in this application was chosen to be IOmm (table 3.2). The penetration depths
in the gold work-piece at room temperature over a range of frequencies (50kH/-
150kHz) are shown in figure 3.9.
3.6 EDDY CURRENT STIRRING
A unique feature in induction heating is the automatic stirring of the molten metal.
This movement is the result of the interaction of the magnet': fields of the currents in
the coil and work-piece
12
This effect is:

Proportional to the square of the applIed field (ampere-turns):

Inversely proportional to the density of the molten metal;
Inversely proportional to the applied frequency of the magnetic field;
I
. h d' fh' h d II "19 01 0'
mportant In t e pro uetlon 0 19 -gra ea 0YS . ' - . --,
lnitia/Investigations
3.7 RESONANCE
23
Resonance is the study of the frequency response of a particular circuit. The resonant
circuit is a combination of R, Land C elements having a frequency response
characteristic similar to figure 3.10
23
.
Av, I
--_. __.-
- ~ f
Figure 3.10: Response curvc of a rcson::mt circuit
It is evident that at a certain frequency f,. the response of the circuit in figure 3.10 is a
maximum. This behavior is classified as resonance. Resonance can be defined as the
point at which maximum response occurs in a circuit. The response can he in tenns of
voltage (V) or current (I) depending on what typc of resonance circuit is heing
analysed. The frequency f,. at which maximum response occurs is defined as the point
at which the reactive components in thc resonant circuit are equal and opposite (XL ~
XC)2J. 24. f,. can be defined in terms of the circuit elements such as inductance and
capacitance (L and C) and is given in equation 3.10. This project deals with
characteristic response and basic analysis of a pal'allel resonant (tank) circuit.
Where:
f,. ~ resonant frequency in Hz
L = inductance in Henries
C = capacitance in Farads.
I
/, = - ~ =
2rr..JLC
(3.10)
Initial Investigatiom'
3.7.1 Parallel Resonance
24
I
. : ~ -
1
---_._-
>-
1
~ , q ; - F
1
Ir
I
L
= QLlr I, = QLl
r
~ XL
c)
i
Rp=QL'R, XLP=X
L
I
~ R
L
I
Xc
Figure 3.11a: General representation Figure 3.11b: Equivaknt representation as seen by the source
The following analysis is based on the assumption that the quality factor (Q) > 10.
Figure 3.11 a shows the general representation of the parallel resonant circuit. The
circuit is modeled with an ideal current source (l) and the source impedance is
assumed to be infinite. ZTP is defined as the input impedance to the tank circuit. Xc is
defined as the capacitive reactance of the tank circuit and XI. is dctined as the
inductive reactance of the coil. RI. is defined as the resistance of the coil. In induction
heating the work-coil and work-piece are modelled as a series R. L circuit as shown in
figure 3.11 a. The quality factor (Q) which exists in all resonance circuits. is defined as
the ratio between the reactive power and real power present in the circuit. The Q is
determined by the coil and is given b/':
Where:
QL = Quality factor of the coil
X
LP
= inductive reactance of coil
R
L
= resistance of the coil
x//-'
R,
(3.11)
Figure 3.11b shows the equivalent representation of the tank circuit as seen by the
source. X
LP
is defined as the total inductive reactance of the coil at resonance. Xc is
defined as the capacitive reactance of the tank circuit at resonance. At resonance the
inductive and capacitive reactances cancel and the resistance of the coil is transfonncd
from R
L
to Rp by the ratio, Q
L
2
R
L
as shown in figure 3.11 bC). Rp is the impedance
which the source secs at resonance. Assuming the tank circuit has a Q of 10 it can hc
seen that Rp is of the order of IOOR
L
.
Initial Investigations 25
It is therefore evident that the Q acts as an impedance transformer and explains why
parallel resonant circuits have maximum impedance at resonance, with an impedance
response curve similar to that of figure 3.1 O. Since impedance transformation occurs
in parallel resonance it follows that current transformation occurs in the reactive
branches of the tank circuit. It can be seen from figure 3.11 b that if IT is the total
current entering the tank circuit, the current in the reactive branches XL and Xc are
given bi
3
:
(3.12a) (3.12b)
Where
I
L
= current in the inductor
Ic =current in the capacitor
IT = total current into the resonant circuit
QL = quality factor of the resonant circuit
Thc bandwidth of the tank circuit is given b1
3
:
EH"
Where
B IV= bandwidth of the tank circuit in Hz;
f,. = the resonant frequency, at which maximum impedance occurs in Hz;
Qp = The quality factor of the tank circuit (Qp = QLJ
(3.13)
Capacitive
Initiallnvestigatiul1s
3.7.2 Frequency characteristic of a parallel resonant circuit
A 9 ~ 9v-9i
+90
0
Inductive
ot-
I
---'--
Wo
I
,
,
_90
0
T
Figure 3.12: Frequency characteristic vf a paralld resonant circuil.
At frequencies below resonance ((I),,) the inductive reactance of thc
circuit shunts the circulating current, making the lmd inducti\"c in
naturc. Like\l,"isc at frequencies above 0)" rho..: JoaJ Oo..:COlllCS
capacitivcly reactive.
26
Figure 3.12 shows the frequency characteristic of a parallel resonant circuit. 0 in
figure 3.12 shows the phase relation between the voltage and current as a function of
frequency. The voltage leads the current at frequencies below resonance (,j,,), where
the inductor impedance is lower than the capacitor impedance, and hence the inductor
current dominates
25
At frequencies above resonance, the capacitor impedance is
lower and the voltage lags the current, with the voltage phase angle approaching -90".
It is therefore evident that a parallel resonant circuit has a lagging power factor at
frequencies below resonance and a leading power factor at frequencies above
resonance. At the resonant frequency (wo), both the voltage and current arc in phase
and the input power factor to the tank circuit is therefore unity. Most resonant
converters in induction heating applications operate by driving the load circuit at its
resonant frequencyl3 7. 5. 9, 10 This has the advantage of providing r ~ d u c e d switching
losses (due to zero voltage and or zero current switching) and thereby, a high
operating efficiency in the power-source.The advent of solid state com'erters have
therefore led to increasing interest in the development of RF power sources for
induction heating. Unlike RF val\'es solid state sw itches cannot tolerate mismatched
load circuits too well at high frequency making these supplies unstable and
inefficient. Driving a mismatched load brings about the conduction of the integral
Initial Investigations 27
body diode within the MOSFET. This results in a conduction power loss as well as
reverse recovery loss when the load current is commutated from one half-cycle to the
next. Phase locked-loop circuits applied to solid state power sources ensures efficient
switching by holding the load at resonance. This concept applied to induction heating
has only recently been discovered. The earliest publication of PLL control for solid
state power sources are dated back to 1985. Since then there have becn only a few
publications on this topic over the past one and a half decades making research into
this field challenging and every contribution significant.
The implementation of the PLL control system was viewed from a power clectronics
perspective. A basic understanding of the oplration of phase locked-loops from the
telecommunications perspective are prescnted a..d the concepts then adapted for the
power electronics discipline.
3.8 PHASE LOCKED-LOOPS
A phase locked loop is a circuit, which synchronizes the signal from an oscillator with
a second input signal, called the reference so that they operate at the same frcquency
with a fixed phase relationship between the two
l
(,. Phase-locked loops arc often used
because they provide filtering to the phase or frequency of a signal that is similar to
what is provided to voltage or current wavefoffi1s by ordinary electronic tilters
l
".
fnitiallnvestigations 28
Phase locked-loops find wide application in the areas such as communications,
wireless systems, digital circuit's etc. The first description of phase locked loops was
published by de Bellescize in 1932 on work involving the synchronous reception of
radio signals.
()
P.D
G, S
... -'
+ A
Loop Filter
vco
rig 3 13: Schematic bloc]" diagram of b SIC I'LL configuration
While the concept of phase locking has been in use for more than half a century,
monolithic implementation of PLL's have become possible only in the last twenty
years and popular in the last ten.
3.8.1 Loop Fundamentals
The basic loop consists of a phase detector (P.D), a loop filter and a voltage-
controlled oscillator (VeO). With the input signal to the loop having a phase of O,(t)
and an output of Oo(t), the following assumptions are made:
The loop is locked
Thc phase dctcctor has a linear characteristic
3.8.2 Phase Detector
The ideal phase detector output voltage is proportional to the phasc difference
between its inputs, i.e.:
Vd (3.14)
where Kd IS the phasc detector gam factor with dim"nsions of \olts p"r radian
(V/rad).
Initial Investigations 29
Many different types of phase detectors exist, all performing essentially the function
of multiplication in a typical PLL system. For the purposes of this project only the
following three types will be discussed:
3.8.2.1 4-Quadrant Multiplier
A multiplier acts as a phase detector (P.D) through thc trigonometric identity:
sin Acos B= [sin(A - B) +sin(A +B)J
2
(3.15)
The term "4-Quadrant" refers to the ability 0: the multiplier to handle both positive
and negative values at both of its inputs'6.
If the inputs to the multiplier are:
\' = V sin{o) t +'" )
J I I \j.'/
l' =V sin(w t + '" )
U (I / 0/ U
The P.D output is:
,. =K ,. ,.
d /11 I V
where Km = multiplier constant resulting in:
Vd -'1>0
and
$, - =6.<jl
yielding
Vd = K",f"JJsin + sin(2o),t - 6.<jJ)]
(3.16)
(3.17)
(3.18)
(3.19)
(3.20)
Equation 3.19 shows the two sinusoidal components 0f the phase detector output. For
a constant the output of the P.D should be constant according to equation 3.20.
The second term however varies \,'ith a frequency 2(o)i as shown in equation 3.20.
fnitiallnvestigations 30
Because the second tenn is removed by the loop-filter, the average dc equivalent .of
the output is given by:
(3.2t)
where km is a constant associated with the multiplier. For small values of ti<j>, sin ti<j> '"
ti<j> and the P.D gain is given by:
I
k =-k VV
cl 2 fI[ ( v (3.22)
It is therefore evident that the P.D gain (k
d
) is a function of the input signal level.
Therefore if the input signal amplitude varies, k
d
and all loop parameters dependant on
loop gain will also vary16 As ti<j> increases with time, the average component of Vd
varies sinusoidally, resulting in the P.D characteristic shown in fig (3.14).
"dill
Ii -2-
./
Fi g.3.14: ut" anal0g ph:bl:
dcteetllr (4-Quadrant muhiplir.:r).
3.8.2.2 Switch tvpe phase-detectors
Also a common type of P.D consisting of a switch. The :;\\itch could be any1hing
from a transistor to a diode-quad or evcn an analog switch.
Signal Input
. Loop Filler
A
o Ulput

l) r 1\ C
J'lg 3 JS B':hh hpl' I'D ..:"nI I::'-U13IJ,.jJ
Initial Investigations
Common types of switch-type phase detectors are:
Gilbert Multiplier
Double-Balanced Multiplier
Half and full-wave transistor multiplier
31
These phase detectors also have a sinusoidal characteristic as shown in fig (3.14). Thc
switch is driven synchronously with the input signal and on alternate half-cycles it
allows the input either to pass or not to pass as shown in figure 3.16. Assuming the
input signal to be Es cos (It"t +0 ) and the switch changes at the zero crossings of sill
wt, the output will be Es cos (It"t +0 ) for 0 < wt < n and zero for n < It"t < 2n. The
average d-e output of the P.D is:
F f-
Ed =~ "eos(wt +<1> )d\\'t
2rc (,
E . ,h
= __5 SIn,+,
n
(3.23)
Figure 3.16 illustrates the operation of a half-wave detector. A full-wave detector can
also be used and the d-c output will be doubled, as well as the ripple frequency. This
is an advantage in wide-band loops as it climinates problems caused by low phasc
detector ripple getting to the VCO and causing phase jitter.
f
lnpu t
Sy, ilchlnc-
function -
U nfiltnt'd
au pUl
lnitiallnvestigatiolls
3.8.2.3 Triangular phase detectors
32
Unlike sinusoidal characteristic phase detectors, linearity in triangular P.D's are near
perfect for phase angles as large as 90. Figure 3.17 and 3.18 show a comparison
between sinusoidal and triangular P.D characteristics. A triangular characteristic is
realized by driving the inputs to the multiplier with
gives the P.D an exclusive-OR characteristic
27
\'d",
square waves. This operation
\U
A-
A
Liocar
region
-IT\. -IT 2
Vdm

. !
if
I
4/:
"'J I
IT2
n ... 80:
-II'
-:\
fig.3.l7: Sinusoidal charu(.tcristic of analog rha:,e
dctcctor shov.'ing: its limitu..l linear opIXating region
fig3 IK 1ii.-q,'UL.,. rXlN..d.rt.Ul.l dUU:.1t.ri...:ric
liffilr qu-Jiing rdl'p,:
Digital phase detectors are realized wheIT using aIT XOR gate or an edge-triggered R-S
flip-flop. These form part of the triangular family of P.D's but have a slightly
different output characteristic as shown in fig (3.19) below:
\' d
A
\' dd _.
11 2 n
Fig.3.19:XOR phase-detector
characteristic showing optimum 0peraril,':;
poinr at 90'
\'d
A
\" J d :;
n 'II
Fig.3.20: R-S lakh deleCtor
characll:ristic sho\\ ing optimum (lp,-'rallng
point at 1XU)
Initial Investigations
3.8.2.4 XOR Phase Detector
33
Operation from a single supply and a close examination of the XOR truth table yields
the digital P.D characteristic. It should be observed that preferred operation of this
device would be when the two input signals arc phase shifted by 90. This puts the
P.D in the center of its linear region and cnsures accurate lock operation over thc
range 0 < <P < 11.
The XOR gate being a digital device is relatively immune to switching and input
signal noise. The trade-off however, is that the input signal rangc is limited to a 50 %
duty cycle in order to ensure correct operation of this device.
3.8.2.5 R-S LATCH
The extended operating range of (0 < <P < 211) for the R-S latch makes it an attractive
option for a P.D. This device is not duty-cycle limited like the XOR but has its
disadvantages. Being an edge-triggered device makes it susceptihle to noise effects
and therefore the two input signals must be of a quality that will trigger the nip-nop
reliabli
6
Also the input signal-ta-noise ratio must be high and is of no value if a
signal must be recovered from a larger noise.
Other types of Triangular P.D's are:
2 and 3 state P.D
charge-pump P.D
sample and hold P.D
3.8.3 Loop tiller
The output of the phase detector is tiltered by the loop tilter. which provides a phase
error voltage to drive the VCO keeping the loop in lock. Since the P.D and the veo
designs are usually innexible, the design of the loop tilter provides more flexibility in
controlling the PLL characteristics"' 27. ". The desired PLL response will detem1ine
Initia/lnvestigations 34
the loop-order. The loop-order required therefore dictates the loop filter type. Loop
filters are generally of2-types namely, passive and active.
3.8.3.1 Passive loop filter
Passive loop filters are of the low pass type or of the phase-Iead-Iag type. For simple
phase-locked applications requiring low loop gain, marginal phase-accuracy and
transient loop stability, passive loop filters provide a quick and easy solution.
3.8.3.2 Active loop filter
For a passive loop filter the maximum dc ga,n achievable is I. An activc loop filter
provides dc loop gains that are essentially ,nfinite and provide bctter tracking
performance. Many types of active loop filter configurations (such as the integrator,
integrator and lead, lead-lag filter) are available in refcrences
l6
. 26. n The final loop
filter configuration used for this research will be discussed briefly.
3.8.3.2.1
Intcgrator and lead filter
R2 C1
L
I Rp' I
1 '/, :
,"r---'
R1 -Ga i
----fv'/'I ----1
,/
Fig.3.21: Simplified representation
of an active Integrator and Lead
loon filter
The integrator plus lead filter forms a basic PI controllcr as : ,!Own in tigure 3.21. Thc
prime purpose of introducing an integral tenn into the controller is to remove any
steady state phase error. At high freq L1ency the ac gain (proportional tcrm Kp) is
formed by R2!R I. The ac amplifier is actually used as an attenuator to the high
frequency ripple, providing a jitter free signal to the veo. The de gain of the filter is
usually infinite as mentioned before. In many applications howewr, involving high
order loops it is always desirable to control the dc loop gain to prcvent instability.
RpiR I controls the de gain component of the loop tilter and thcrdorc also indirectly
controls the entire loop gain.
Initial Investigations 35
Design of a PLL requires the ability to be able to control the natural loop frequency
(w
n
), damping factor ( ~ ) and de loop gain (K). Passive loop filters such as the single-
pole low-pass and the two-pole low pass filter, do not allow for the control of W
n
, ~
and K independently. The control of K ensures good tracking as mentioned before but
a high gain loop (large K) also comes with a wide bandwidth. Therefore narrow
bandwidth and good tracking are usually incompatible in first order loops. If it is
necessary to have large gain and small bandwidth, the loop will be badly under-
damped (low 1;;) and transient response will be poor (low w
n
). The active integrator-
plus lead filter having two independent time constants (,I and ,2), draws on the
concept of tachometer feedback which allows for the independent control of natural
frequency (transient response), damping factor (overshoot) as well as the de gain
(good tracking).
3.8.4 Voltage Controlled Oscillator (VCO)
The voltage-controlled oscillator provides an output frequency, which is controlled by
the filtered error voltage it receives from the loop filter. Since !requeney is the
derivative of phase, the VCO operation may be described as:
d<jJo =K V
dt U III
where Ko = VCO gain
I ~ " = VCO input voltage
d<jJo =VCO output phase
(3.24)
lt is therefore apparent that the phase of the VCO output will be proportional to the
integral of the input voltage Vino The vea should be operated within its linear range
to ensure a constant loop-gain parameter (Kveo). For the purposes of this research, a
linear relationship between input control voltage and output frequency is assumed and
is given by equation 3.25
k = Mo (3.25)
" L'.I
o
The veo's employed in the PLL system !()r this research were derived tram two
4046 PLL integrated circuits.
Implementation ofAutomatic Frequency Control 36
CHAPTER 4
IMPLEMENTATION OF AUTOMATIC FREQUENCY CONTROL
4.1 SYSTEM DESCRIPTION AND OPERATION
53
; Php-t..LLEL P.ES'J:Lt..l;r CCT
"':
---. __ .._._---
,I!..TO::::
; . Ill'JEP.TEP.
; ===== ;- _.. -
;
: :
:..--- _.-. _.- ~
;
;
;
;
....__. .__.__ ..;. __._. __ pi,:-:-r,IE; ; . ~ : : ,1:'7.. "
._._.- _._._'- -..- - -.- --_.- -- . ---
~ _. _._.- -- _._._..,- _.- - _...,._.- -_._.. -- _._._..-- _.- - --- _._,
, :.. , ~ _ . - ---- _.-:
: \ : ; ~ ~ I I i
AFC
Figure 4.1: Schematic layout of the induction fumace <lnd all it..; component..;
The induction furnace comprises the following components, with reference to figure
4.1:
A variable DC power source that is derived from rectified mams voltage. This
feeds a rectifier bridge from a variac. The isolation transfonner between the variac
and the mains voltage (50Hz), serves to provide isolation for test purposes. By
varying the DC bus voltage, the input power to the inverter is controlled thereby
controlling the input power to the load;
A filtering inductor or iron core reactor, which is situated in the positive DC bus
rail. The iron core reactor serves to feed a constant current to the inverting stage.
The iron core reactor also provides inherent short ci,.cuit protection because it
restricts the rate of rise of current i" the event of a short circuit occurring in the
induction-heating coil. Because of the slow rate of rise of current under fault
conditions in the iron core reactor, this topology is advantageous since it now
gives the necessary protection circuitry some time to sense and operate under fault
conditions. The result is that protection circuitry can be easily implemented to the
'" system .
Implementation ofAutonwtic FrequcnLy Control 37
A 100kHz full-bridge load-resonant MOSFET inverter which operates at
approximately IkW. The inverter switches are operated in alternate pairs to
generate the high frequency alternating current needed to produce strong eddy
currents in the heating coil. The inverter operates at the resonant frequcncy of the
load circuit thereby allowing zero voltage switching, hence no external high speed
diodes are needed across the MOSFET switches to carry reactive freewheeling
current. 16 The result being that the total switching losses in the inverter is grcatly
reduced, thereby increasing the inverter efficiency;
The gating and gate drive circuitry which are used to convcy thc switching signals
to the inverter switches;
The load which consists of a water cooled iuduction heating work-coil in which
the crucible and work-piece arc situated;
A water-cooled high frequency matching transformer which is used to step up the
current in the work-coil to a high value, which is nccessary for good induction
heating and also serves to provide electrical isolation;
A capacitor bank which is used to resonate with the reflected inductance of the
load and matching transformer at a frequency of approximately I(JOkHz.
To enable maximum power transfer to the load at all times the automatic
frequency control system is included which forms the basis for the current
research. This is given by the AFC block in figure 4.1.
4.2 LOADING EFFECT
The placing of metal in the heating coil tends to change the frequency characteristic of
the load circuit. This facilitates the need for frequency con,rol to ensure maximum
power transfer. Table 4.1 shows the resonant frequencies of the same load circuit with
different metals placed inside the coil.
Metal
DIAMETER MASS FREQUENCY
(mm) (g) 1kHz)
Copper 12.5 278 195.5
Gold 10 20 160.4
Steel 12 18.5 126.1
Nickel 9 10.5 134.5
Lead 10 12 156.4
Brass 12 243 183.3
TabJC'. 4.l: i{t:SPT)<lnf frc:qucnci..::-\ fur \ arlou:-> llK{;..i1-; at rnoll\
tclllp.....raturc, v. hen pbccll in rh..... ph1tntypc Induction fumacl.:
Implementation ofAutomatic Frequency Control 38
The resonant frequencies for different metals at room temperature were measured at
low power levels using a function generator and oscilloscope to determine the
frequencies at which zero phase shift between the driving voltage and current were
observed in the load circuit.
The inner diameter of the heating coil was approximately l4mm. The natural resonant
frequency of the tank circuit with the coil not loaded was 148kHz. It was observed
that when a high conductivity, closely coupled metal (copper) is inserted into the coil,
it causes the inductance of the tank circuit to decrease. This results in a shift in
resonance, which means that the tank circuit must now be resonated at a higher
frequency. When a steel work-piece is inserted into the coil its magnetic properties
(permeability) tends to increase the inductanc of the tank circuit, causing its resonant
frequency to decrease.
This dynamic behavior of the load circuit (induction-heating load) is of major interest
for the implementation of automatic frequency control. In a basic sense, automatic
frequency control is implemented to compensate for changes, which occur in the load
during the heating cycle. A basic understanding of the load bchavior under various
conditions is essential for the effective implementation of the RLL circuit.
4.3 LOAD CIRCUIT
The induction-heating load forms part of a parallcl resonant circuit, which is
continuously driven at its natural resonant frequency by the inverter. The idealised
equivalent circuit model for the induction-heating load is shown in figure 4.2.
1
~ Rp
l
[
~
-I
j
l
Lp
:::;:: Cp
FigA.2: Idealized equivalent circuit for induction heating load
The expression for the complex impedance of the parallel tuned circuit in figure 4.2 at
any frequency (t) is given by equation 4.1":
Z I I) = 11'-
( I
I'N!'!
I, /0
(4.1 )
lmplemefllation ofAutomatic Frequency Control 39
where:
Rp = Equivalent resistance of the tank circuit as seen by the source,
Qp = Quality factor of the tank circuit and is given by Qp = Rp / XLp,
fa =Natural resonant frequency of the tank circuit.
The equivalent circuit parameters were measured at low power with sinusoidal
excitation from a signal generator. These tests were conducted in order to determine
the load circuit parameters and calculations were performed where necessary. The
load circuit was then simulated on ORCAD 9.1 using the measured and calculated
values determined in the experiment. simulated load circuit parameters
transformed to the terminals of the source are c:scussed for three discrete conditions
namely:
4.3.1 Unloaded heating coil
Impedance Characteristic
'00
0----
60 80 lOO 120 140 160 180 200 220 240 260
FrequencJ<kHz)
VI 80
E

E- 60
o
u
40

o
20
E
R1
02458
L1
4 7SCJuH
C,
243n
r 11
'C.Y
Fig. 4.3: Equivaknt load cirl.:uit
par:.lnlcters of induction heating load
measured with an heating
c:oil
Fig 4.4: Impedance characteristic of unloaded induction-
coil. The circuit ha:; ? natural resonant frequency
of 1.f8 kHz and a Q of 18. The IOJd circuit has a
maximum impedance 11.
The frequency response of the unloadeJ induction-heating coil is shown in tigurc 4.4.
The resonant impedance is higher (79 Q) for unloaded copjitions, which improves the
systems no load performance
30
because of minimal current drawn from the supply
(higher impedance at no-load). When the coil is loaded the load impedance is reduced
and more current is drmm trom the DC supply. The resonant frequency is
approximately 148kHz with a Q of 18.
Implementation ofAutomatic Frequency Control
4.3.2 Copper work-piece

o
40
Impedance Olaracteristic
40

E30

. 25
820
fij 15
"C
10
E 5
- -----------
0-
60 &l 100 12) 140 160 180 2CO 220 240 330
Frequene,(kHz)
Fig. 4.5: Equivalent load circuit
parameters of induction heating load
measured with a copper work-piece
placed in the heating coil
Fig 4.6: Impedance characteristic of induction-
heating load with a copper work-piece placed in the
healic
s
coil. The circuit has a natural resonant
frequency of 195 kHz and a Q of 10. Thc load
circuit h.... s a maximum impedance of 33 Q.
The frequency response of the loaded induction-heating coil is shown in tlgurc 4.6.
The copper work-piece has the parameters as shown in table 4.1. The resonant
impedance is lower (33 D) for the loaded condition and more current is therefore
drawn from the supply. The inductance of the coil (L2) is decreased due to the
insertion of the copper work-piece resulting in an increase in the resonant frequency
of the load circuit to approximately 195 kHz "ilh a loaded Qof 10. The increase in
resonant frequency results in a reduction in skin depth thereby mcreaslllg the
equivalent resistance (R2) of the load circuit.
4.3.3 Steel work-piece
13
VY
C3
243n

L3
6.555uH
Impedance Characteristic
!
o
50 80 11)) 120 140 160 15G 2(J: 22C 240 2tiO
FrequenC)(kHz)
Fig. 4.7: EqUIvalent load circuit
parameters of induction heating load
measured with a steel work-piece
nhcerlm the heatinl! coil
Fig 4.8: charactt:flstic of mduc[iull-
hearmg load wilh J steel wDrk-piece placed in [he
hC'3tlOt! coil. The Circuit has a natural n:sonant
frC'quC'ncy of 120 kilz and a Q of 3.5. The load
Implementation ofAutomatic Frequency Control 41
The frequency response of the loaded induction-heating coil is shown in figure 4.8.
The steel work-piece has the parameters as shown in table 4.1. The resonant
impedance is the lowest (18 D) for this loaded condition and more current is drawn
from the supply. The inductance of the coil is increased due to the insertion of the
steel work-piece resulting in a decrease in the resonant frequency of the load circuit to
approximately 126 kHz with a loaded Q of3.5.
The Q acts as an impedance transformer in a parallel resonant circuit'. The lowering
of the circuit Q as a result of inserting a steel work-piece, results in the reduction of
the load circuit impedance. The steel work-piece is a better conductor of the magnetic
flux in the coil than air is, which tends to increase the inductance of the coil (L3) as
can be seen in figure 4.7. The equivalent resistal.ce of the work-piece is also incrcased
(R3) henee the power loss in the work-piece increases. This relationship is given by
equation 4.2 for a relative permeability of several hundred in steel at room
temperature
l8
4.2 CONCEPT OF RESONANCE LOCKING
Phase Characteristic
260 240
'.
Coppl.'r
200 220
__
\ -.
Stl.:l.:l \ Unloaded
.140 \ 160 180
'-. "
'-
'--'---------'-"-'
100
..
50

.- 60 80 100

50
"
a.
-100
Frequency(kHz)
Fig. 4.9: Phase relationship between driving voltage and driVing current to tanJ... cIrcuit
as a function of The cn;,cacteristic illustrates the response for the three
conditions dIscussed above. The respect(\'e resonant frequencies occur et the points of
zero-nhase disnlacement.
The analysis of the induction-heating load has shown that different resonant
characteristics exist for different loading of the heating coiL It is clearly apparent that
different loading changes all the parameters of the load circuit such as the natural
resonant frequency. resonant impedance and inductance of the coil as wdl a, the Q. It
is evident that at a trequency f = fa, the impedance of the tank circuit is a maximum.
At this frequency the phase displacement between the driving voltage and current to
Implementation ofAutomatic Frequency Control 42
the tank circuit is equal to zero. Figure 4.9 shows calculated phase characteristics for
the three load conditions presented.
For a load circuit Q of greater than 10, this implies that maximum real power transfer
is taking place at resonance as given by figure 4.9. This maximum operating point is
where the induction furnace should operate at all times.
Figure 4.10 shows the combined complex impedance magnitude versus frequency
plots for three conditions namely:
1. Coil unloaded (no work-piece)
2. Copper work-piece in coil
3. Steel work-piece in coil
,"
/.
: , :'.j ,
r r r
, r" r I "
,'

h <: ;I 1 III l! - <: U I I
, ,
1, '.'
, J Il P (, l I,
I ! J I)
I 2 U
p le c e
o "
1
S 1 <: I:" I
S 0

c:

" C-
N
Gl
S 0
0
l:
ell
'0
"0
Gl
C- H
E
....
' 0
l:
ell
l: , 0
0
I/)
Gl
I ,
0:::
0"
Operating Frequency
Fig. 4.10: Frequency response for th..: induction h..:ating !J.nk cir,,:ull. Th..: unlll:'ltkd coil h:J-; rdJti\dy high Q
(approximately IS). When the coil is loaded the Q tcnd..; to dccrc:J;;c (S.2-; fur Ctlppcr :Jnd 2.511 for ..;It.-'I,.'IL Thl;
rl..'Son3nce locked loop tracks the operating points 110 11 and l for difkn:nt ltldd condition:'> :.Hld lht.:rcfnrl; ll1:lintilin:-
maximum real power transkr to the load through.1U' the heatIng
Figure 4.10 shows the resonant frequencies, for an unloaded coil, ji for a copper
work-piece and 12 for a steel work-piece placed in the coil. The unloadcd coil
resonates at approximately 148kHz, and has a Q of approximately 18
When a stcel work-piece is inserted into the coil. the inductance of the coil increases,
changing the Q of the tank circuit as well as its resonant frequency. If the induction
Implementation ofAutomatic Frequency Control 43
furnace were to run in open loop, at frequency fo with a steel work-piece, the system
would be operating at point A on the steel work-piece curve. Operation at point A
results in a reduction of power transfer to the load since point A is relatively close to
the 3dB (1/2 power) point on this curve. When a copper work-piece is inserted into
coil, the system operates at point B on the copper work-piece curve. With no
frequency-tuning present, operation at point B would result in very little power
transfer to the copper work-piece. Another drawback of operating at points A (steel)
and B (copper) is that significant switching losses develop in the power source when
driving a load off resonance
4
,7,8 The resonance locked loop therelore tracks the
optimum operating pointsj("ji andJ2 for different loading in the coil.
4.3 RESONANCE LOCKING METHODOLOGY
The implementation of the resonance locked loop required the control of two distinct
variables whose phase relationship was a function of the applicd frequency of thc
power source. A simplified schematic of the current fed invertcr (power sourcc) is
shown in figure 4.11.
!\
" ,

,"
.......... ' \
"'/ ..
/
(\ "I.ud
. ,
! \
!
\ \I
1I
I
..
r
I
I
, I I
,
,
\
,
,
\
i
,
,

1\

f---
.
" G"lr
,
\
!
,
i
I; ,
,
--'0':'
Fig. 4.11: current-fed inverter
configuration employing po\',;er MOSFET's.
Gate driver Circuits ha\'e been JmHtcd for
Fig. .t.12: Ideal wavefomls of the dri\'ing
\'oltage and current to the I03d circuit. It is
apparent that the gate control sIgnal
(VGATE) IS an approxImate phase
The induction-heating load can be characteriscd by the equivalent circuit shown in
figure 4.1. The load circuit is currcnt supponi\'c and is modeled with an ideal currcnt
source which warrants the use if the iron-core rcactor in the invcner DC bus, The
Implementation ofAutomatic Frequencl' Control 44
switching elements in the inverter drive the load at a frequency determined by the
switching rate of the control signals fed to the gate of the power MOSFETs. Switches
SI, S2 and S3, S4 operate alternatively each to produce one half cycle of the RF
power presented to the load terminals. Simulation results of the equivalent load circuit
driven at resonance are shown in figure 4.12. V
Load
is the driving voltage across the
tank circuit and l
Load
is the driving current through the load produced by the closure of
switches SI, S2 and S3, S4 respectively.
Due to the principal of forced commutation
JO
it is evident in figure 4.12 that the
control voltage to the power MOSFET Vg"" is an actual phasc representative of thc
driving current through thc load. This concept is treated in the idcal sensc and omits
the propagation delay time taken to drive thc MOSFET into thc saturation mode of
operation. This delay time is typically in the order of 200 - 300ns and is affected hy
the following factors:
Rise and fall times of gate drive si!,'l1al
Value of gate resistor chosen for damping
Input capacitance of the power MOSFET
Stray inductance in thc gate drive loop
Characteristics of the load being switched by the power MOSFET (resistive or
reactive)
This propagation delay results in a small offset phase error within the resonance
locked loop. This phase error is encouraged as it has the effect of producing a nonzero
output from the phase detector, whieh is required to maintain thc control voltage at
thc veo input, holding the system in lock!6.
4.3.1 Signal Measurement
In summary the control strategy employed utilized the following concepts:
The inverter output voltage (V
Lmd
) was transformed to logic levels (900Vp-p to
25Vp-p) The voltage transformer was wound on an ETD29 ferrite core with a
turns ratio of 40:1. This transformer is gi\'en by T4 in schematic 1 of Appendix B.
Gate control signal fed to power MOSFET is used as a phase representative of the
driving current. This factor eliminates the need lc)r curn:nt measurement and
simplifies the layout of the im"Crter, making it compact, and provides for stahk
operation.
Implememation ofAutomatic Frequency Control 45
Control of the inverter is achieved by continuously locking the gating control
signal (Vgate) to the inverter output voltage (VLoad) over its entire operating range.
4.4 CONTROL CIRCUIT IMPLEMENTATION
Research into the development of an Automatic Frequency Control systcm resultcd in
two final implementations. The implementation of the gate voltage locking method
has eliminated the need for current measurement. Both systems were tcstcd on thc
prototype induction furnace at full power where various work-pieces wcre heatcd. The
systems (Revl and Rev2) proved to bc stablc over thc entire opcrating rangc at both
low and full power. A comparative discussion will be prescnted to summarizc the
individual system's perfonnances.
4.4.1 RLL revision I
Automatic frequency control of the invener was achievcd by means of resonant mode
locking. The control system, which is called a resonant locked loop (RLL) employed
essentially two second-order phase locked loops. The basic system is shown below in
figure 4.13.
I LOOP 11
CLlc
Fig. 4.13: Simplitied schematic represt:,;Lltion of thl: rcsonancl: comprising. two
phasc locked-loups (Loop 1 and Loop 2). Loop I an acti\c tilter ",hid, and is used to
gcneratc a 90" pna:;c-shifr In \\J\'donn B. LOdP :! C' ..... othcr acti\c filter and IS used
generate a 90" phasc-shift in A. Tht' AGe i-; used tu supply a tixcd amplitude signal
toPD2.
Phase detcctor I (PD I) is a type I, exclusivc-OR phase detector derived trom the
MC14046 PLL chip. Loop 1 operated as an active filter and was used to generate a
90" phase-shift in the current sample (wavefonn B). The 90" phaoe-shitl is
characteristic of the XOR gate PLL and was used to hold the phase detector in the
center of its linear range (chapter 3). The phase-shitied current-sample wavefonn was
Implementation ofAlltomatie Frequency Control 46
multiplied by the tank-circuit voltage (waveform A) in phase detector 2 (P02). Phase
detector 2 incorporated the A0734 4-quadrant analog multiplier. The analogue
multiplier was used so that the transformed sinusoidal tank circuit voltage (waveform
A) could be fed directly into the phase detector. PO 2 operated by locking the phase-
shifted current sample 90 out of phase with the voltage waveform A. The 90 phase
shift method was employed in order to ensure operation in the phase detector's (PO I
and PO 2) linear region
27
This operation locked waveforms A and C 180" out of
phase. The result was a relative zero phase shift (anti-phase) between wavefonns A
and C. Inverting one of the waveforms initially resulted in a near zero phase shift
when in locked operation. vca I and vca 2 were derived from two MCI4046 PLL
integrated circuit.
The automatic gain control stage (AGC) was used to convey a fixed amplitude signal
to PD 2. It operates by amplifying or attenuating an incoming signal in ordcr to
maintain a fixed amplitude output signal. Undcr different load conditions the Q of the
tank circuit changed, resulting in an amplitude change at a specific resonant frcqucncy
as shown in figure 4.10. Another reason for employing an AGe was to allow thc
induction furnace to operate at reduced power Icvels. It was f()und that by changing
the amplitude of waveform A, an offset phasc error was produced in phasc detector 2
(analog multiplier) due to signal amplitude bcing behl\\ the minimum input offset
voltage, which caused the loop to lock incorrcctly. The AGC which incorporatcd the
VCA610 was used to hold the amplitude of waveform A constant over the opcrating
range of the induction furnace, hence produced no offsct phase error in thc multiplier.
The following derivation has proven the necessity for an AGC implementation III
conjunction with an analog phase detector (POl) in the system implemented.
Assuming two uniformly time varying signals multiphed such that the multiplier
output M is:
M ~ Acos(t>l! +cjJ, )xBcos(w +9,)
AB (. .) AB ( ~ , )
~ - c o s 9 , - 9 , +-cos d"l+qJ, +cjJ,
:2 ':2 .
After low pass jillCi'illg Lcm'cs :
(4.3 )
AB ( )
= -cos t.cjJ
2
(4.4)
Implementation 0/Automatic Frequc/1(:r Control 47
From the final expressIOn of the output it can be seen that output phase of the
multiplier (cos L1</ is dependant on the amplitude of the input signals (AB/2). It is
therefore apparent that a fixed amplitude signal has to be fed to the multiplier in order
to eliminate the problem of phase errors being produced over the operating range of
the RLL. The actual circuit implementation of revision I is shown in appendix B2.
4.4.2 RLL revision 2
The cost and complexity of RLL revision I has led to the development of a simpler,
cheaper and more effective means of phase locking. Revision 2 introduced a similar
system to the previously presented model, except for a few changes as shown in
figure 4.13.
PARA". l
R""ONA""
LOAll
ZE'" ,
<...",-".. o.,N..-.
"L''''':'''-'
Fig:. 4.14: Block di:.J.gnln <:.:ontrol sy:-;to.::m
system comprises t\\o <:.:as<:.:aded y,J order PLL <:':lr<:.:uits, which lock at 90' pha... I.--
shift rdatl\C to its input. PO I and P02 compn-;L' XOK digital pha-.;e-dct\.'ctnfs.
The frequency control system also composes two 2
nd
order phase locked loops as
shown in figure 4.14 but does not employ an AGe or an analog phase detector.
The two loops operate as 90 phase shifters maintaining lock over the entire operating
range. Operation is also realised by comparison of the phase difference between the
load voltage (VLOAO) and the switch gate \oltage (VGATE>. This phase difference is
processed by loop 2 and a frequency change to the phase di fference is
generated by veo 2. This frequency difference is the clock signaL which is used to
either drive the inverter to the new load resonant frequency, or hold it at the current
resonant frequency.
Implementation ofAutomatic Freqll(!IKy Control 48
The automatic frequency control system employed Type I Exclusive-Or phase
detectors in both loops. Active 2
nd
order PI controllers where employed as the loop
filters in LPF I and LPF 2. The use of active loop filters provided the necessary high
gain to the loop and ensured good tracking performance with minimal static phase
error. The total loop can be modelled as a 4
th
order PLL system and was found to be
stable over the entire operating range. The actual circuit implementation is shown in
appendix B3.
4.4.3 Discussion
The following aspects were observed to be critical aspects in the design of the two
RLL circuit implementations:
Loop stability was !,'Teatly influenced by the bandwidth of the op-amps used in
the phase shifter 100p26.27. Op-amps with high gain bandwidth products were
used.
A second order PI controller was employed as part of the loop filter. Op-amps
with very low input bias currents were used to avoid the integrator from
charging in the wrong direction as well as drifting during nonnal operation] [.
Loop time constants were a critical factor in the design of a stable RLL
system. Stable operation of the loop was achieved by making the time constant
ofLPFI much fasterthan that ofLPF2 (at least 10 times).
No extra filtering circuitry was employed to condition signals before being fed
to the RLL system. This factor simplifies the design and allows effective
operation over a wide frequency range.
The implementation of the zero-crossing detector ir revision 2 was a major
contributing factor to the simplicity of the second design.
Slew-rate limiting in the analcg multiplier resulted in a phase error offset at
the loop output.
Employing active loop filters was a necessity because the low DC gam of
passive loop filters did not enable lock in operation when the system was
started up.
Limiting the RLL lock range gives the system the propenies of a highly
selective filter. This feature gave the system extremely good noise rejection
capability, which assisted in automatic start-up operation.
Implementation ofAutomatic Frequency Control 49
4.4.4 Anti-Lock protection circuitry
An electronic protection circuit was incorporated to monitor the RLL operation during
a heating cycle. The basic system is shown in figure 4.14.
ANT'LOcK
PROTECTION
CIRCUIT
CCK
PARALLEL
RESONANT
LOAD
INVERTER
ZERO-
CROSSING
DETJ"CTOR
ANALOG
SVVITCH
PO 2
TIMER
LPF 1
,-PF :z
VVINDoVV
COMP
vco'-'
LOOP 1
Fig. 4.15: Block diagram representation of the frequency (ontrol sysh:m showing the an\i-Io-:k
protection circuit. Operation ofloopl is monitored by a window comparator circuit. In thc c\"cnl OfLl
loss of lock, the triggered timer dcacti\'utt.'S the invcrter P\\'M and 0PCfLltcs thc analog switch <.:ircuit,
which simultaneously resets both \('.\)\1:;, pulling the system back into Il)l..:k operation.
The anti-lock or loss oflock protection circuit was developcd as part of the electronic
protection circuitry for to the induction furnace. The protection circuit section on
figure 4.14 monitors loopl status checking for an invalid operation. The input voltage
to vea I is fed to a window comparator circuit, which monitors the operating range
of the vea I. If loss of lock occurs, the vea driving voltage goes out of range and
triggers the window comparator circuit. This circuit then triggcrs a eMaS timer
configured as a monostable. Activation of thc monostable deactivates the PWM
signals to the inverter section and also activates analog switching circuit. The analog
switch circuit simultaneously resets the loop-filtcrs LPFI and LPF2 by shorting out
the integrating filter capacitor. This re<et action pulls the RLL circuit to its center
frequency, which is designed to be close to unloaded resonant Irequency of the
induction furnace. \Vhen the monostable has timed out thc loop is reacti\'ated and
returns to normal lock operation.
The complete implementation of the anti-lock protection circuit is shown in appcndix
B3.
Experimental Results
CHAPTERS
EXPERIMENTAL RESULTS
50
Two final circuit implementations resulted from the research into automatic frequency
control of the induction furnace. Both systems were individually tested on thc
induction furnace at full power and at low power levels.
The AFC system was tested on the induction furnace where 50g slugs of steel and
copper were heated respcctively. The load circuit comprised a multi-turn induction-
heating coil, which formed part of a high Q parallel resonant circuit. The system was
driven in open loop and the frequency was adjusted to the natural resonance of thc
unloaded tank circuit. When a steel work-piece is placed inside the coil the inductance
of the tank circuit increase. This effect makes the tank circuit capacitive! y reactive as
shown in figure 6.
1 5.00Y 2 5.00V
.. '
f1 RUN
-;-;:-; -,-,-;- -- i- -- -- -, ;:-; -.- -- -1-- -- --,--;- -;- <i
.::. l


I
,
FreqCl) 218.7kHz
Fig. 5.1: Capacitively tank circuit being dri\cn hy th... inv.:n... r. Tr.lce 1 shows the
s\\itching control signal fed to the \10SFET gate. TrJce 2 SIHl\\S the loss of zero \oltagl'
switching: across tht': MOSFETs. O\"L'r vob;e tum-on and tum-off srikcs ;m: a1:-;o pn.:senr.
which could lead to the destruction of thl' S\\ itcht.:s <:It highcr PU\\ er le\ds.
EJ.perimemal Results 51
The control-switching signal (VGATE) fed to the power MOSFET is shown in trace I
of figure 5.1. Trace 2 shows the drain-source voltage (VDS) being switched by a
MOSFET in the current-fed inverter. It is evident that the mismatch between the
natural resonance and the current driving frequency has resulted in a loss of zero
voltage switching as shown by trace 2. The loss of ZVS has also brought rise to over-
voltage transients at both turn-on and turn-off of the switch. These transients increase
dramatically in amplitude as the power is increased. This often results in the necessity
to use special snubber circuitry to prevent MOSFET destruction. Driving the load off
resonance also results in a reduction of load circuit impedance (as shown in figure
4.9) which resulting in excessive current being drawn from the DC supply.
5.1 REVISION 1
The AGC circuit employed in revision I performed well over the entire operating
range with no noticeable phase shift incurred by its operation. A high-speed (15Mhz)
4-quadrant analog multiplier (AD734) employed in PD2 was used to provide minimal
phase error introduced by the multiplier at the operating frequency range in question
(80kHz - 220kHz). A low speed (5Mhz) 4-quadrant multiplier (AD633) was initially
incorporated as PD2 but slew rate limiting in the multiplier eore produced ofbet
phase errors in LOOP2.
15.00'; 22.00'1
I
.I.
I
.-0.00:::' 2.00'g/
f1 STOP
.: . I ...
______._ 1 __:.... -- -- -- - _
..: ....1 _
FreqC Phase( '"
Fig. 5.2: Gate voltage (tract: I) and imr.:nt:r Illidpninr \tllw.gr.: (tra,.:e
locht:d 9U' out of pha;,r.: by !,wp I.
Experimental Results 52
Figure 5.2 shows the loop in lock at an operating frequency of approximately 150kHz
with a gold work-piece placed inside the crucible. The 90 phase shifted gate voltage
(trace I) and the transformed sinusoidal midpoint voltage (trace 2) are both fed to
PD2 which locks the two incoming signals by phase displacing them a further 90.
The output of PD2 is shown in trace 2 of figure 5.3 with a copper work-piece placed
inside the coil. Switching noise fed from the midpoint of the inverter to the RLL
circuit causes the noise on the rising slope of the multiplier output (trace 2). The fast
falling edge in the output of PD2 is the main factor which dictates the necessity for a
high-speed (15 MHz) analog multiplier. Trace I shows the zero-voltage switching
drain-source voltage (l50Vpeak) across a pl)wer MOSFET in the inverter-bridge and
is free of over-voltage transients.
vp p(l)-159.4 V Freq(l) 149.7kHz
Fig. 5.3: The inverler operating with RLi.. ro:\isioll I in pha:.<.:-luck. Trace I shu\\j
the zero-voltage switching drain-source voltage across a MOSFET in thl' bridge.
The 90 phase shifted gate vDltage and the transfonned sinusoi.u<1\ tank cin.:ult
voltage is multiplied together by the high.spi:i:d an3log plwsc detcl..":tor (AD7J-l)
P02. The output of PD2 is ShOV.ll in tracc 1. The fast falling cd:i:s in the output
wavefonn is the factor whil.':h dit.:tates the us.: of a hig.h skw-rati: .... ,la\og mu\iiplii:f
Figure 5.4 shows the system in lock w:th the coil unloaded. Trace 1 is the transformed
signal waveform A (figure 4.12) of a 400Vp-p voltage applied to the tank circuit at
resonance. Trace 2 represents the 90" phase shifted current sample of loop I, which is
180 out of phase with waveform A (figure 4.12) at 159kHz. \\nen different loading
occurs in the coil, the resonance locked-loop will change the driving frequency of the
power source to maintain lock between the current sample (waveform A) an" the tank
circuit voltage waveform B (figure 4.12) o\"l:r its full operating range (80kHz-
220kHz).
Experimental Results
1 5.00Y 2 5.00V
...... f,o!...._-;....:
.-0.005 f2 STOP
53
FreqC 152 .4k.Hz
I
J.
J
... 1..
I
;',
V
,
,
1
,
. ! i
',',! __.-.i-j ',1
,
1'1' !. I . I ....
0
Fig. 5.4: Gate voltage (trace 2) and transfonncd im'cTtcr midpoint voltage (tra{;t: I)
wavcfonlls lot:ked 180
0
our of phasc to hold tilC tank circuit at resonance v,hcn
operating thc prototype induction furnace.
5.2 REVISION 2
The following results are were taken from reVISIon 2 of the automatic frequcncy
control system implemented. This system was found to be the most feasible and cost
effective solution of the two investigated for this research.
The resonance locked loop was tested on the prototype induction furnacc. which was
used to melt 30g of copper and 30g of gold at IkW of DC input powcr with closed
loop frequency control using revision 2. It was found that the system held the load at
resonance throughout the heating cycle with no frequcncy drift or instability occurring
over the operating frequency range (85k -220kHz).
The PLL system employed acts as a highly selectiv'e filtcl. This feature gIves thc
system extremely good noise rejection cu?ability. which assists in automatic start-up.
With linle power applied to the inverter, the zero-crossing detector generates random
oscillations on its output. This acts as a noise input to the loop as shown in trace I of
figure 5.5. This noise injected into loop occurs at a frequency. which is outside of the
bandwidth of the AFC loop. The frequency control syslcm therefore locks to the
closest multiple of this noise within its bandwidth thereby holding the in lock
at start-up. Trace 2 of figure 5.5 shows onc half cyclc of the inverter output phase-
locked to Ihe 43'd harmonic of the noise injected into the loop.
Experimental Results 54
1 IO.OV 2 2.00V -0.005 2.00l/
1'2 RUN
-------
---""""
-_._----
.----.
, I
: ...1
: I
,I,
I
,T
I '
.... ,. ..1.
: I
-.=----,
f----I".. : . I.

2
FreqCl) =4 . 167MHz Freq(2)=95.51kHz Vp-pCZ)=3.313 V
Fig. 5.5: Trace I illustrates the zero crossing detector output as the automatic frcquem:y
control system acquires lock when thc powcr is applied. The circuit acts as a selcctivc tilter
extracting only the fundamental load resonant frequency component and rejects the high
frequency noise injected into the loop. The drain- source voltage across a lower MOSFET
in the bridge is given by trace 2.
Figure 5.6 shows the implementation of automatic frequency control to thc induction
furnace. It is evident that the ZVS is occurring in evcry cycle and no ovcr-voltage
transients are present as shown in trace 2. With thc AFC system in operation thc gatc
control signal (trace I) is always phase-locked to thc zero-crossing points of the tank
circuit voltage (trace 2).
1 IO.OV 2 50.0V -O.00s
f2 RUN
7\
, , -'-', , "
I
I I I I
I
I
I
I

FreqCl):::oIOO.9t:Hz FreqCZ)=101.QI-cHz Y!J-pCZ)=209.4 Y
Fig. 5.6: Tank circuit dri\cn at its n..Hural reson:mt frequency by the power source The
AFC system is controlling the in\crkr $witchlllg IhL'r::by holding the IOJd CIrcuit
in resonance at all times. Z\'S CJn bL' obscf\cJ in trac\,:"2 \\ith O\L'TyolragL' trJthiL'nh
across thc \10SFET s\\ir..:h.
E'(perimelllal Results 55
Figure 5.7 shows the heating cycle of a steel work-piece. At room temperature the
tank circuit resonates at 126kHz. As the work-piece is heated, its relative permeability
decreases and approaches unity. This causes a decrease in the resonant frequency of
the tank circuit. At the curie transition (",710C to ",nOC) in figure 5.7, the relative
permeability of the work-piece has fallen to unity and the steel loses its magnetic
properties [4]. This results in a decrease in inductance of the tank circuit, resulting in
a major shift in the resonant frequency (from 125k of the tank circuit. The
work-piece was heated to I 180C. After the transition through curie temperature, the
resonant frequency increases slightly due to the change in resistivity of the steel work-
piece. The temperature of the work-piece was measured by means of a radiation
pyrometer, which was immune to the magnetic fields produced in the heating coil.
Resonant Frequency vs Tern perature
''"
N
ne
I
"'-
'e e
>-
u
c
"e
w

IT
ne
w
u:
CO
"e
rn
c
"e 0
w
w
'"
"e
, e
0

, 00
Start of Curie
transition
'00
Temperature (QC)
Fig. 5.7: Heating cycle of a steel work-piece in the prototype Induction fUrtl3CI.... ing
the frequency change as the metal is heated through its cunt: poinl.
Cunclusions and Recommendations/or Future JVork
CHAPTER 6
CONCLUSIONS AND RECOMMENDAnONS FOR FUTURE
WORK
6.1 CONCLUSIONS
56
The automatic frequency control system has been successfully implemented by vinue
of "gate-voltage locking" and the induction-furnace has been tested on a number of
different metals. The rapid frequency changes that occurred when heating steel
through curie temperature (figure 5.7) has proven that the resonance locked-loop can
track changes and maintain lock at the natural resonant frequency of the tank circuit.
The implementation of the resonance locked-loop eliminates the need for manual
tuning and provides for a more accurate and effective means of closed loop frequency
control, providing maximum power transfer to the load at all times.
The system proved to have the following advantages:
1. The implementation of the actual circuit utilized fewer and less expensive
components than revision I and therefore provided a relatively cost effective
approach for frequency control.
2. The implementation of AFC eliminates the need for manual open loop frequency
control and has optimized the inverter performance.
3. The continuous ZVS achieved has eliminated the need for snubber circuitry and
also allows the MOSFET switches to be driven closer to their maximum voltage
ratings.
4. No current measunng circuitry was needed for the approximation of the load
current phase displacement. This technique of phase locking is simpler and only
utilizes the measurement of the load 'oltage and gate control ,"oltage to the
Invener.
Conclusions and Recommendations for Future Work 57
5. No special matched filtering circuitry was needed to filter the signals to be phase-
locked. The AFC system performed an inherent filtering function as mentioned in
chapter 5.
6. The high gain active loop filters employed, provided optimum tracking
performance with reduced steady-state phase error.
7. Automatic start-up operation was achieved by virtue of the implementation of
active loop filters. At startup, the smallest phase error signal fed to the loops from
the phase detectors (PO I and P02) are integrated to zero. This feature holds the
system in lock from the start, hence allowing automatic start-up operation of the
induction furnace.
8. The use of the XOR PO's provided good circuit immunity to the radiated EMI
radiated by the magnetic field insidc thc coil and powcr source during a typical
heating cycle
9. The system response to a stcp change in phase whcn a work-piecc was inscrtcd
into the coil proved to be satisfactory. Tracking the curic-point transition of a steel
work-piece during a typical heating cycle simulated thc system response to a
velocity change in phasc, which also provides satisfactory results.
10. The basic electronic loss of lock protection was provided for the AFC system. It
monitored the status of the control system and detected a loss oflock. The systcm
then performcd a corrcctivc action by simultaneously r:setting both loops and
providing a trip signal for future auxiliary protection.
The resonance locked-loop was therefore found to bc su;table for the application of
automatic frequency control of the prototype miniature induction furnace. The
successful implementation of AFC on this system has encouraged investigation into
the application of this control strategy to other resonant-mode power electronic
converters for induction heating. The concept of gate-voltage locking" has provided
a breakthrough for this research with regards to frequency control and possibilities of
other forms of frequency control using this technique can be im'Cstigated.
Conclusions and Recommendations/or Flllure Work
6.2 RECOMMENDATIONS FOR FUTURE WORK
58
Current research is underway to melt platinum slugs (20g), which would test the
system's stability at higher output power levels (2kW). Furthcr tests to investigate the
effect of the phase transformation of a solid work-piece to its molten liquid state are
to be conducted. These results will provide valuable infonnation regarding the
detection of the melting point of a metal by virtue of a frequency shift during the
heating cycle. This method could save major costs invested in radiation pyrometers
for temperature measurement.
A mathematical model of thc load and frequercy control circuit will aid thc designing
of effective frequency control systems. The two working systems (Revision I and
Revision 2) will provided the foundation on which the numerical model will bc bascd.
The aim of this study will be to provide a working model which can bc applicd to the
designing any frequency control system for powcr elcetronic converters.
The following improvements could be implemented to the existing frequency control
system:
High bandwidth optical isolation between the AFC system and the inverter drivers
could bc implemented. This procedurc would separate thc control circuit ground
from the invcrter power ground thus providing bettcr noise immunity to the
system.
PCS prototyping of Revision 2 is currcntly u n d e r w ~ y in prcparation for thc
melting of platinum. The current prototypes (Rcvision I and Rev'ision 2) were
constructcd on vcraboard for testing.
A theoretical model of the working systcms (Revision I and Revision 2) will
provide valuable information for the design proccdure of future AFC systems at
any operating frequency range f"r various induction heating applications.
Conclusions llnd Recommendations/or Future Work 59
A frequency control system incorporating the use 0 f the type 11 phase detector (in
place of the XOR) and active loop filters could be investigated for future research.
The noise immunity of the edge triggered PD (RS latch) in the new PLL system
would have to be investigated further. Special noise shielding techniques could be
employed to allow stable operation in this mode.
A simpler lock-detection circuit incorporating an R-S latch could also bc
investigated. This system would eliminate the use of thc window comparator
circuit thereby simplifying the overall design.
Application of "gate-voltagc locking" to other resonant-modc power electronic
converters for induction heating. A voltage-fed invcrter is to be developed for
induction heating and the control strategy employed in this research is to be
implemented on the inverter, as a means of automatic frequency control.
A self-oscillating resonant inverter incorporating "gate-voltage locking" is to bc
investigated. It is believed that the zero crossing points across the load circuit
voltage in a present cycle of operation could be used to generate the switching
transition signals for the next cycle of operation. This system could be
implemented, but requires some thought with regards to start-up operation.
Future projects on the development of the induction-furnace include:
Temperature control
The temperature of the work-piece has to be monitored thro'lghout the heating cycle
to ensure that the work-piece temperature never exceeds thE: maximum temperature of
the crucible. The work-piece is hened to its molten fonn, hence no contact
measurement can be allowed as contamination of preci,)us metal quickly occurs. A
radiation pyrometer could be cmployed to monitor the temperature of the work-piece
throughout the heating cycle. The output signal Irom a pyrometer can be used to feed
a translator circuit. which would either advance or dclay the tiring angle of the
controlled rectilier bridge. and accurate power control to the work-piece- can be
achieved. The implcmentation of temperature control would be advantageous because
it would extend thc applications of the induction furnace. The system could then be
Conclusions and Recommendations/or Future Work 60
used for special laboratory applications, which reqUIre preCISion heating of small
quantities of metal. Examples of applications are silicon crystal growing, tungsten
refining and special high-purity alloying with metals like titanium, ruthenium and
platinum.
Protection circuitry
Overload and short circuit protection needs to be implemented to the system. This
kind of protection could involve inserting a circuit breaker into the DC bus, which
would operate when a fault was being sensed. Due to the presence of the iron core
reactor in the DC bus, the protection circuitry will be given adequate time to respond
to a fault condition.
Cooling water monitoring
The most common type of failure present In induction furnaces is cooling water
failure. Dangerous consequences could result if no monitoring of the flow rate and
temperature of cooling water was present. A temperature sensor such as the LM35
could be employed to monitor the temperature of the water. Whcn the sct point
temperature of the water is reached, a signal could be fed to the cooling water pumps
to increase the flow rate of the water, hence lowering the tempcrature of the cooling
water. Differential pressure sensors could be employed to monitor the flow rate of the
water. When an undesirable condition is reached, a signal could be fed to the
protection circuitry to operate and trip the system.
Front end powcr factor correction
Investigations need to be conducted to detennine what kin" of harmonics the system
could be injecting back into the line frequency power source. If the need arises a
front-end power factor correction 'ystem could be implemented. which would
incorporate a DC-DC converter in place of the controlkd recti tier. If the system is to
be sold to foreign markets (e.g. Europe) it would ha\'e to comply with ccrtain
harmonic standards.
Conclusions and Recommendations for Future JVork 61
Microprocessor implementation
An embedded micro-controller could be implemented as the main unit which would
monitor and control all of the above mentioned processes. A simple PlC or DSP could
be employed for this application.
References
REFERENCES
62
[I] E Swift (verbal consultation), Platinum Perfect.
[2] Alvan Hirner (verbal consultation), Franz Hirner Jewellers.
[3] Prof. CLang (verbal consultation), Department of Material Science,
University of Cape Town.
[4] 1. Khan, HA Miniature High Frequency Induction Furnace," BTech. Thesis,
School of Electrical Engineering, Cape Technikon, November 1998.
[5] D. L. Loveless, "An Overview of Solid-State Power Supplies for Induction
Heating," Metal Productioll, vo\. 33, August 1995.
[6] 1. Khan, J. Tapson and 1. De Vries, "Automatic Frequcncy Control of an
Induction Furnace", Proc. IEEE COil!, Aji-icon '99, vo!.2, Scptcmber 1999, pp.
913-916.
[7] M. Kamli, S. Yamamoto, and M. Abe, "A 50-150 kHz Half-Bridge Invcrter
for Induction Heating Applications," IEEE hallsactiolls 011 Illdustrial
Electronics, vo\. 43, No. 1, February 1996, pp. 163-172.
[8] J. M. Ho and F. C. Juang, "A Practical PWM Inverter Control Circuitry t"r
Induction Heating and Studying of the Pcrfonnancc under Load Variations",
Proc. IEEE Con!, Intenwtiollal Snllposiwn all Industrial n('clrollics, \o\. 1.
July 1998, pp. 294-299.
[9] D. L. Loveless, R. L. Cook and V. 1. Rudnev, "Considering Nature and
Parameters of Power Supplies for Efficient Induction Heat Treating,"
Industrial Heating, June 1995.
[10] D. Tebb, L. Hobson and W. Wilkinson, "A Currcnt Fed MOSFET Inverter for
Induction Heating Applications," Proc. ]rjh L'llh. j'o\\D' Engineering Cant:,
April 1985, pp. 390-392.
[11] M. Bartolini, "An Induction Furnace Using a 100- I50 kHz Voltage-Fed Full-
Bridge Load Resonant Inverter," BTeeh. Thesis. School of Elcctrical
Engineering, Cape Technikon, Octobcr 1997.
[12] L. Bardenhorst, "High Frequency Induction Melting Furnace". BTech. Thesis.
School of Electrical Engineering, Cape Technikol1, October 1996.
[13] L. Hobson, and D.W. Tebb, "Transistorized pO\\'cr supplies for induction", 1nl.
J. Electronics, vo!. 59, No. 5, June 19X5, pp. 543-552.
References 63
[14] H. Akagi, T. Sawae and A. Nabae, "130kHz, 7.5kW Current Source Inverters
using Static Induction Transistors for Induction Heating Apllications," Proc.
IEEE PESC.. 1986, pp.395-400.
[15] A. Veldhuizen, "Investigation into High Power Ultrasound for Industrial
Applications," BTech. Thesis, School of Electrical Engineering, Cape
Technikon, November 1998.
[16] F.M Gardner, PhaseLock Techniques. John Wiley & Sons Inc., USA, 1967.
[17] K. Billings, Switchmode Power Supply Handbook 2
nd
edition, McGraw-Hill,
USA, 1999.
[18] E. J. Davies, Conduction and Induction Heating. Peter Perq,'linus Ltd., UK,
1990.
[19] E. J. Davies, and P.G. Simpson, Induction Heating Handbook. Maidenhead,
McGraw-Hill, 1979.
[20] C. A. Tudbury, Basics oJInduction Healing, vo1. I, New Rochelle, New York,
1960.
[21] H. Barber, Electroheat, London, Granada, 1983
[22] N. R. Stansel, Induction Heating, l;t Ed., McGraw-Hill Inc., USA, 1949.
[23] R. L. Boylestad, Introductol)' Circuit Anall'Sis, Macmillan, 1994.
[24] J. W. Nilsson, Electric Circuits. 4
th
Ed, Addison-Wesley Inc., USA, 1993.
[25] N. Mohan, T. M. Undeland and W. P. Robbins, POll'er Electronics:
Converters. Applications and Design. John Wile)' & Sons Inc., 1989.
[26] D. H. Wolaver, Phase-Locked Loop Circuit Design. New Jersey, Prentice
Hall, 1991.
[27] P. Young, Electronic Communication Techniq' cs. Englewood Cliffs.
Macmillan, 1994.
[28] W. Egan, Phase-Lock Basics, Julcn Wile)' & Sons Inc., 1998.
[29] I. Khan, J. Tapson and I de Vries, "An Inducli:m Furnace for the Jcwelrv
Manufacturing Industry", Proc. 2
nd
BTech. conr. October 1998, pp. 41-44.
[30] M. H. Rashid, POIl'er Electronics: Circuits, Dn'ices and Applications. 2"" Ed,
Prentice-Halllnc., 1993.
[31} S. Schrier, EE35911', E!cctronic Componcnts. Circuits and :\lodulcs.
University of Cape Tuwn, \998.
Appendices
APPENDICES
APPENDIX A: LOOP DESIGN EQUATIONS
B(s) .
l ~
+. ~
A
K ~
-I
~
+ A
Fig.l: Block diagram model of frequency control system.
The equivalent model for the frequency cOltrol circuit of rcvision 2 is given by tigure
L The system consists of two cascaded 2
nd
0: :fcr phase-lockcd loops which opcratc by
tracking changes in the rcsonant frequency of the load circuit.
LOOP COMPONENTS
Phase detector
The type 1 Phase detector (XOR) has a lincar op<:ratillg range of IXii degrecs as
shown in figure 3.19, The phase detector gain is thcrcfore:
K<jJ ~ Vdd/rr (V/rad)
Loop filter
An active loop filtcr was used to providc optimum tracking and minimal static phasc
error. The loop filtcr consists of an integrator plus lead fiJ'.er and its coniiguration is
shown in figure 2
R2 Cl
--.' ..~ - . - - ~
Rp
Rl
~ " ' - - - - ~
-Ga
Fig.": simplified representation of
an acti\ e Lcad-Lag loop tiltcr
Appendices
The loop transfer function K
LF
is represented in the frequency domain by:
F(s) ~ A(r,s+ I)
(".1'+1)
where:
A = Rp/RI,
'1 = (R2+Rp).C and
'2 = R2.C
VCO
The transfer function ofthc VCO in thc frcqucncy domain is givcn by:
Ko ~ K vis (rad/sIY)
where
Kv ~ 2IT (finax - finin)IYdd-3.6V (rad/slY)
Feedback
[I]
II
The fcedback loop usually contains a gain, Kn which rcprcsents a counter module of
valuc IIN where N is the didviding ratio of the counter.
TRANSFER FUNCTIOi'l
The open loop transfer function of a second order loop is given by:
K("s+l)
GH (s) ~ - - - - 7 ~ " - _ c _
s("s+l)
[21
where: K = A. K<jl.Kv.Kn
The open loop transfer function yields a typc I. second ordcr system which should
produce zero steady state phase error t,)r a stcp phase input.
The characteristic equation for the loop is givcn by:
c.1:" :
, (K,.+I) K
.1'" + --"---.- s + - ~ (j
1" I "( ,
[31
Appendices
This allows for the fonnulation of the expressions for W
n
and C; :
W = ~ "
"
"
and
C; = (K,] +1)
2(J)/I"( J
[4]
[5]
III
This allows for the design of a desired loop response. It is evidant that (J)n can be
controlled by adjusting the value ohl' It is also evident that the damping factor C; can
be controlled by adjusting '2.
APPENDIX B: SCHEMATICS
1-

"
I[
co
0,,'
1R2113
co
l)'O
"
't,
Ll
LJr '007
116 I

W
"
O-JOOVDC
.\:

"t'::"


'YI_'"
,',,, I
11

"IR'
" '
"

"l1:'
I
,,'-.
Uf I<JU' lJ
1 I
,,,'
LT
'-11 '1';1\
"

------ ,,-*
- .- "r tUn,' \Jf
- ",
-A,/V'-
'
____
',0
1R2113
'"
""',1
v:',:,
'.'
" I
H"
v['P
..)0.1 Ht"
CC
"
I.
-1_, .'
T'n"
......
- --]-
'IY
V)('Lld
\''1, iT "

('11\'," J 1
(: c ';' ;,
I. SCHEMATIC LAYOUT OF INDUCTION FURNACE
\;" 1-:'"
: L :
T "1""'['''''-
1- .\ .\, .. I .
.....L: . ::r:: ,".- "'I' ' 1,1 ----:: '''.,.... :::::r:: c"1( *;,'
T.\'c\. r ,,',').. ; ',")" [ J J 0,,'
1.
1
-
01
,r 0
,
-I
1
z z
~ ,
0
,
~
u
'"
;0-
>
0
0
-
c 11
M
-
,
-
0
I
I
'. ;:.
~ - -
Iol
U
-<
Z
~
~
""
z
o
-
t-
U
~
o
z
U
-
t-
-<
:;;:
~
-
U
z
Appl?l1di('('s v
2. AIITOMATIC FREQUENCY CONTROL SYSTEM REVISION 1
uw

C
"" '.V -J n --t--l
'.' ..C1,1: H:: rH'" '::" J:: 1
\
I"""''',] ..- In".. . -0 -c__ 0
HI!,
'"
Rn

_-\fV\,.1 r;- U11
'"
t.'v'h)fl'
o
[
o
."
:r---
-
uo
''';;,''' l"
V
01 rl:'I
. VV'v R:'O
cn won.
--j-
fl;',' C j
o o
INVERTER
'M
eLK 1 IVload I
I I l
;. ''''
f') '1 l' ""
'\1': I 0

t
'"
I . C: 11 ...
I
."" r z. IJ:? - 0
ln4;-=:::.J "V
I' ., -..:: --_. '. --. I
. ,
'.v
0

Vl. ()1I 1
,;
" I
,:"
,,,7,,,1
'"
V,""N
IN"
l Ill'
H'
H'
.'
'( ".
... ,.,,,
..' A,'"
I
"
0 0
.
V,"HI]
( ",]
1 ,'X
"''',l
\"
V' '''N
IN"
l ""
'"

Ho'
'j' .,'..
,. I-
f
0
" "
Hq
VoA
.\,'
1
f"I,.
""', .
A,1(l

1<'''-1

- \.
1
:"C
R"

r-:-...
....-
lr 347
C'
PLL revision 1
N,,"d'p' ---
1.'1" 1 ,,'
T"3
Appendices VI
3. AlJTOMATIC FREQUENCY CONTROL SYSTEM REVISION 2
I .
';IN

CIN
VC')Ul Pl
INVERTER
,,,
R'
y
. 'l
VVY
5l30R
1<;:>5
P1 1-
R'
veorN
Lr347
"M
U,
j f
c.
."
7-t",
X
1,1=]4
[
R'
----NV'- .. -
-- 33k L .
u"' - r- 33k
....-
1<,;'
I", .'fM

I r-rl
//11:147
--------
-------
[1.\

() \ I T "JI 0." -
h,,,,, PLL revision 2
IN"'"'' -----';,.;;-, . -I H.l' Oil
1 \' ... 1 I [)o<;ll"'''''' Numl, .. , I 5
o 'V'""'v-----"
o 'O"k ,",,, \'\i"'I" .. I., . 0 ,1"... , or
-,,-- --- ---- :lU=-------"
--- - CON
_.cv _ _ veOUT
1R" ," S<N
- '" Ire' ex
1; 000" 1- ex
"",. - I.'.' 'NH
" 't' _.. R1
-. '-Moo. j__ __ _ L
R
_'__
"\ --. - (."l""
0
"..1
'f
I
n;'<I
"',V
I

I

I.Mn:Hll
" I<l"
n:',.'
"

j
IHl
AI VVY - II JB
<I .'ro I
I (l
- -- VV\, -
IIlI\1
"VV'v HIl
". I .>
_ .. /'LI.l41
I ')
'.N

l
"
J
llX4(
J, (
?1:
1
"
'Ill:
I
V"UIN

;':,
',.,

,.

I

VVY
D
'l',V
I

INII
,<0
I{;'
I
, ,
Vload
<: ., .\

t HO
l
I
I k I "
!
.. 1k
D D
"
____ L _
CLKJ
J lh
_A
r
B
oX y.
I'" >.n"
1111

"
Appendices
APPENDIX C: TECHNICAL DATA
IRFF460 ENHANCEMENT-MODE POWER MOSFET
IR2113 HIGH AND LOW SIDE MOSFET DRIVER
AD734 HIGH-SPEED ANALOG MULTIPLIER
VCA 610 AUTOMATIC GAIN CONTROL IC
CD4046 CMOS PLL IC
DG301 ANALOG SWITCH
62
International
IJ:ORI Rectifier
PD-9.512B
IRFP460
HEXFE-r<" Power MOSFET
Dynamic dv/dt Rating
Repetitive Avalanche Rated
Isolated Central Mounting Hole
Fast Switching
Ease of Paralleling
Simple Drive Requirements
V
OSS
=500V
ROS(on) = O.27U
1
0
= 20A
Description
Third Generation HEXFETs from International Rccllflcr providc the designcr
with the best combination of fast switching. ruggcd1zed devIce design, low
on-resistance and cost-effectiveness,
The T0-247 package is preferred for commerciaHndustrial applications
where higher power levels preclude the use of 10-220 devices, 1he 10-247
is similar but superior to the earlier TD-218 pacl<age because of Its Isolaled
mounting hole. It also provides greater creepage distance bch'leen Dins 10
meel the requirements of mosl safety specifications
I
TO247AC
A
'C
___ J
Vir.s
- ,
VI
_. - -
vrc

mJi
A- ,
-55!0 .. 150
Absolute Maximum_R'-a'-Ii'-n:.:g,,5----=__,---
r----- -- --- Parameter
" .?5'C I Continuous Drain 10V- 20
1I0 @ Tc = 'GGOC IContinuous CUH'''. VG, '" 'A V r-===- 13_
r ----- IPulsed Drain Current (e 60
Po@le25'CJ.O.w.'_'D.'_ssipat1on__--=..--==_-_-___-:-
Factor 2 2
IVGS IGate-to-source-YOlfage - - __ _
f.EAs _ ._ 51n9[7 Pulse Energy '- 96C
.1 Avalanche Current 12:- _
Repetitive Avalanche Energy _ _ 28
I Peak DIOde Recovery __ ___ 3.5
'TJ i O?rating Junction and
Ts10 _. Storage
'- 1CS.C?n<lS I 300 (1.6n'TI fr.:lrT! cas-eJ
10. tbf
o
..r2J1 I
Junctl:X1-to-Arrbier. t
Thermal Resistance
Rt>Jc JU:lctlOn-tc-Case
I __.--f?se..to.'?ini\, Fla: SJ1ase
_

045

,
eN'1 I
1025
International
I\?R Rectifier
Data Sheet No. PD601471
IR2110/lR2113
HIGH AND LOW SIDE DRIVER
Features Product Summary
Floating channel designed for bootstrap operation
Fully operational to +500V or +600V
Tolerant to negative transient voltage
dV/dt immune
Gate drive supply range from 10 to 20V
Undervoltage lockout for both channels
Separate logic supply range from 5 to 20V
Logic and power ground 5V offset
CMOS Schmitt-triggered inputs with pUll-down
Cycle by cycle edge-triggered shutdown logic
Matched propagation delay for both channels
Outputs in phase with inputs
Description
The IR2110llR2113 are high voltage, high speed
power MOSFET and IGBT drivers with independent
high and low side referenced output channels. Pro-
prietary HVIC and latch immune CMOS technologies
enable ruggedized monolithic construction. Logic
inputs are compatible with standard CMOS or LSTIL
output. The output drivers feature a high pulse
current buffer stage designed for minimum driver
cross-eonduction. Propagation delays are matched
to simplify use in high frequency applications. The
floating channel can be used to drive an N-channel
power MOSFET or IGBT in the high side configura-
tion which operates up to 500 or 600 volts.
VOFFSET (IR2110)
(IR2113)
10+/-
VOUT
ton/off (typ.)
Delay Matching
Packages
16 Lead PD1P
'11,10 lead5 4 & 5
IR2110-21IR2113-2
500V max.
600V max.
2A/2A
10 - 20V
120 & 94 ns
10 ns

,
,.
-
III 14 Lead POIP
w:o Lp-Cl':! 4
IR2110-1IIR2113-1

%:: ,\'
".
16 le<:ld sOle
IR2110S/IR2113S
Typical Connection
U;:J If) 5001/ 0< 600/
-,
r-,
;[::- J..

_____ ..... .--:-J /
.. . '----i .
, I
;-0
_c-_
LOAO
- ----. - ..----'.' / /
-- - --+-'
-
HO I
V
v
'::

I
-
.,
HIN
Vs i
SO
-
..
L1N
Vc:

Car.l

Vs:..
I
".,
i
Voc:,
HIN
SO
UN
.... ANALOG
-"'OEVICES
10MHz, 4-Quadrant
MultiplierIDivider
AD734 I
CO;-';!'\ECTW:" DIAGRAM
Y2 7 a NEGATIVE
'----_---.r'
nIP
(Q Package and !"l Package)
'to OUTPUT
[INPUT
" ,
!l 01 VOl UGE
,",P POSITIVE SUPPI \
1) DD OE"IO"WiATOR DISABLE
'"
'"
AD734
ur ( TOPVIEW
UZ 5 ("101'0 $col.j ID
'Yl 5
xINPUT
Y INPUT
DENOMINATOR
I"lTERfACE
FEATU RES
High Accuracy
0.1% Typical Error
High Speed
10 MHz FullPower Bandwidth
450 V/fJ.s Slew Rate
200 os Settling to 0.1% at Full Power
Low Distortion
-80 dBc from AnV Input
Third-Order IMD Typically -75 dBc at 10 MHz
low Noise
94 dB SNR, 10 Hz to 20 kHz
10 dB SNR. 10 Hz to 10 MHz
Direct Division Mode
2 MHz SWat Gain of 100
APPLICATIONS
High Performance Replacement for AD534
Multiply, Divide. Square. Square Root
Modulator, Demodulator
Wideband Gain Control, RMS-DC Conversion
Voltage-Controlled Amplifiers. Oscillators, and Filters
Demodulator with 40 MHz Input Bandwidth
PRODUCT DESCRIPTlOX
The ADi34 is an accurate high speed, four-quadrant analog
muhiplier that is pin-compaTible with the Industry-standard
AD534 and provides the transfer functlon W = XYiC. The
AD734 provides a Iow-impedance voltage output with a fu:l-
power (20 V pk-pk) bandwidth of 10 Total statIC error
(scaling, offsets, and nonlineamies corn bmed) is 0.1 of fu!:
scale. Distortion IS typically less than -SO dBe an.:! g;Jaranteec.
The low capacitance X, Yand Z inputs are ful1}" differentia:. In
most applications, no external componer.ts are required to
define the function.
The internal scaling (denominator; voltage Lt is !0 r, derivec
from a buried-Zener voltage reference. A r.ew feature provides
the option of substituting an externa: denominator voltage,
allowing the use of the AD734 as a two-quadrant dIvider wllh a
1000:1 der:.omlOator range and a signa: bandWidth that
10 to a gain of 20 dB, 2 MHz at a gaIn of 40 dB a:1d
200 kHz at a gaIn of 60 dB, for J gam-bandwidth p:od"..1ct of
200MHz.
The advanced performance of the AD734
combination of new circult ted.nlq"Jes, the use of a hl;h spee':
complementary bipolar prOCess and a novel approach to laser-
ttlmming based un ac s:gnals rather than the customary dc
methods The ""ICe bandWidth (>.1(1 ;\1Hz; of the
input SI ages and the 200 .\\ Hz ga:n-ba:;...-iv. iClh ;:r<Jc.::r of
mulup!ier corc a;:ov. thc AD71.f!0 bc 1,.:"cC as a:,)'.', ':h",n:,_,r.
demodulatf)r WIth mrut fretj"Jencl":' a<, high a<; -i1J .\\Hz a<; lonl.'
J) thr deSired o:.Jl/l"Jt frq'Jer.cy I, ;e<;\ than ]1) ,\1 Ill.
The anc AD734BQ arc "fCClf::.-d for the mcu,trla:
temperature nf 1', ;'f\)C Crtme 1r. a I j';tJc
CerJmlC D1P. The J,Jl:ar.:e p,occ,"cc to
.\HL-STD-853B the rJi1I'C If ))'C t" 125 C, I"
a';allJb:c In a 14-lnd It''fa::lIC DH'
PRODCCT HIGHLIGHTS
Tht' AD73'; mere tha;. t;;(, r:f exreoence In
the a::d r.,an"Jfact..:re I}f ar.Jlol; r.:u:lli':,ers, [fl provice:
I. A ne';. O'ltp'Jt ar.Jp;lfler dn1gn ';'I:n mn"e than llmn
Ihe s:e .... -rate ef the ADS Ino \.... , 20 \'j.Js; for J
pov.er \. ba:ld... 0: 10 .\1Hz
2. \'e,y h... Cl"ort:OfJ, eve" a: L;:: r0',l,er. thro'Jgh the use of
CI,C''':1l a:lc I:I:r.r.,lng lechmq:.:es tr.at vlr;;,;al:y elimInate al: of
the Sj::"JriO"JS fJO:-:;lneanues fO"Jr.: Ir. ear;ler deSIgns.
3. DIrect cun,r,,;: 0: the de:-:Offilr.:3.lIlr. re)u:llng Ii': hlghe:
a:c"J .cy ar,c! a p Ir:-ba:: c';' Idth rroc:..iet a:
\3::.:n tha::' t:"l:a1::, tlr.:e'
tb: A[;53.; H1 c:',::er
.... \'er:, ::<:3:: reSt':;';;:' J:hle,;:c t:':, ;,;se of 3
r..).e: lr.r:..:: St3?:' ce,l';::: :;'-. -: \.\ :c:e-b:lc a:r.rlJ:ler.
.. h::h ;"'0 tr.3! ;'_'v. even at hIgh
fre.;:.:;:nc;c:>
r!'Jl,:' t':":Jrr.:a::.": c:'OJ::e of ':::'Vlce
a::: ... h!:h prO";l':e:!.
p3:l,,:-:"':: '-.:' .:8 L: '-::,;'3:-:'.:, ra::>:t If: J 20 kHl bancv,],jth
REV.C
tnformal,On lurnJshed by Analog Dev,ces 's Del,eved 10 tl'" EC"fill.: <l'1'j
rellabie no respanSlool,ly '5 a5s"mE<d by Ana;og DC!.c.es for It::.
use, nor for any ,nfrtngemenls of patenls or otner fights 0' I" r.j PJr!
wh'Ch m<lY result from 'Cs USe No I,cense .s gr<lnted Oy ,mpl'Cd('on or
otherwise under any patent or patent fights of Analog Dey,ces
One Technology Way. P 0 Box 9106. Norwood. MA 020029106, U.S A
rei 781/3294700 World WIde Web S'te http !/www.snalogcom
Fax 781/326-8703 10 Analog De.... ices.lnc. 1999
BURR - BROWN
1
1313
1

-
VCA610
WIDEBAND
VOLTAGE CONTROLLED AMPLIFIER
FEATURES
WIDE GAIN CONTROL RANGE: BOdB
SMALL PACKAGE: B-pin SOIC or DIP
WIDE BANDWIDTH: 30MHz
LOW VOLTAGE NOISE:
FAST GAIN SLEW RATE: 300dBIIlS
EASY TO USE
DESCRIPTION
The VCA610 is a widcband, continuously \ ari3blc,
voltage controlled gain amplifier. It lir.car
dB gain control with high impedance mputs. It b
designed to be used as a flexible gain control ch:ment
in a variety of clcctromc systems.
The VCA610 has a gain control range of 80d13I-40dll
to +40dB) providing both gain and for
maximum flexIbIlity in a small 8-lcad SO-S or plastic
dual-in-Iine package. The broad attenuation can
be used for gradual or controlled channel turn-on and
turn-off for applications in hich abrupt gain changes
can create artifacts or other errors. In addition. the
output can be disabled to -80dB of altcnu:.l-
tion. Group dcby \ariation with gain IS tYPICally less
than 2ns across a bandwidth of I to 1:5.\1Hz.
The VCA610 has a nOIse figure of3.5dB h\i::l an R:>
of 200Q) including the effects of hoth current and
voltage noise. lnstantancuus output dyn:1flllC range
70dS for gains of OdB to .,..40dB \\ ith 1.\IHL nOI:'C
bandwidth. The output capable of dflvlIlg 1002
The high speed. gain control slgn:.ll IS :.l
unipolar (0 to -2\') \o[tagL' that \aries th..: gain lIn-
early in dB \'.
APPLICATIONS
OPTICAL DISTANCE MEASUREMENT
AGC AMPLIFIER
ULTRASOUND
SONAR
ACTIVE FILTERS
LOG AMPLIFIER
IF CIRCUITS
CCD CAMERAS
Th..: \'( ':\{l If) IS deSIgned with a very b.'>t O';t;r!IJ:ld
n:con:r;. tlInc of onl;. 2(j(jns all,m,.... a Llrf'L'
slgrul tran<"lcn! tn O\crlO:.lJ the llulrUl at tl1;tll g:.lln,
WIthout oh"curmg km -Ievcl fq! Im\lflt' d'lsely
behind The excelient O\"Crlr):id tUlle and
dist0r110!1 spccific:ilions opwni/c tillS dc\ ice rilr km-
level dorrkr mcasurements
5'/ -5 :
61
!
" 7! 2 i --:--
,

_i,
,
/
.
1-, -
3 ,
'le.


VCA61Q
lnltrn4llonll .... "'31long PO BOI 1140C T.. Al.!57J.( 6!30 S T.usonBI.d. lel h. 9109521111
Inlernet http Ilwww burr bro",n com, FAXL'M BBRCORP. leiu 066-&491 FAX IS20J!!9lS10' Imr.td,jltProd"Cllnfo I!OOIS.(!61l2
F=AIRCHILD
October 1987
Re..ised January 1999
SEMICDNDUCTORTtl
CD4046BC
Micropower Phase-Locked Loop
General Description
The INHIBIT input, when high disables the vea ood
source follower 10 minimize s!:Jndby power consumption
The CD4046BC micropower phase-locked loop (PLL) con-
The zener diode IS provided for power supply regulation, If
Slsls of a low power, linear, voltage-controlled oscillator
necessary
(VeO), a source follower, a zener diode. and two phase
comparators. The two phase comparators have a common
Features
signal inp:.Jt and a common comparator input. The signal
input can be directly coupled for a large voltage Signal. or Wide supply voltage range: 30Vto18V
capacitJvely coupled to the self-biasmg amplifier allhe $19-
Low dynamiC power consumption: 70 (lyp ) at
naf input for a small voltage signal.
'0 kHz, V
oo
=:. 5V
Phase comparator I, an excfuslve OR gale, prOVIdes a dlgi
vea frequency 1.3 MHz (typ) at VDU = 10\1
taf error signal (phase comp lOut) and malnlaH1S 90'
phase shifts at the vea center frequency Betwcen Signal LON frequency drift o06%t'e at VO:J 10') w;lh tem
input and comparator input (both at 50% duly cycfe). it may pcratun;:
lock onto the Signal input frequencies thal are close to har
High vca I,nearlty 1% (typ)
monies of the VCO center frequency.
Phase comparator 11 is an edge-conlrolled d,gltal memory
Applications
network. It prOVIdes a digital error Signal (phase camp. fJ
FM demodulalor and modulator
Oul) and lock.jn signal (phase pulscs) to indicate a locked
conditIon and malntams a 0' phase shift between signal Frequency syntheSIS and mull,plic.allon
I
Input and comparator Inpul Frequen:y d,scr,mlf1at,on
The Ilf1ear vollage-conlrollcd OSCillator (VeO) produces an
Data Si'lshronl.1.aIIOn and conrii1Ion'f13
output signa! (Vea Out) whosc frequency IS determoned by
c.on/']rs,on
the voltage at the VC0
Ir
Input. and the capaCltor and resls
Tone d<::co1,n'1
tors connected to pin C1
A
. C1
8
. R1 and R?
FSK moeulJtlon
The source f,Jllower output of lhe (demodulJtor Out)
t.lot')f spee<j J)'llroi
I
IS used With an external of 10 k:':l Of more
- ---- - ----- - - -- - - - -
Ordering Code:
Order Number Package Number I Package DescriptIOn
CD4046BCM t.11cA I16-lead Sm<!I1 OuUme in\(;-:;rated CJeud (SOIC), JEDEC t,1'3-012, 015;)- tJ;:mo... 8-yj/
CD4(}46BCN N16E I15-Lead Plast,c Dualln-lrne Pao.a38 (PDIP). JEDEC MS-001, 0 300- V:,'je
Ir ",,,,e'
t-, ...
Connection Diagram
Pin Assignments for SOle and DIP
!
C)
o
""-
o
""-
en
OJ
C)
s:
n
o
"0
o
::
'"
,
"U
::T
'"
'J>
'!'
r
o
n
'"
'"
0-
r
o
o
"0
, .. I
Top View
intersil
DG300A, DG301A, DG303A
TTL-Compatible, CMOS Analog Switches
The DG300A through DG303A family of monolithiC CMOS
switches are truly compatible second source of the original
manufacturer. The switches are latch-proof and are
designed to block signals up to 30Vp_p when OFF. Featuring
low leakage and low power consumption, these sWitches are
ideally suited for precision application in instrumentation,
communication, data acqUisition and battery powered
applIcations. Other key features include Break-Before-Make
switching, TTL and CMOS compatibility, and low ON
resistance. Single supply operation (for positive switch
voltages) IS possible by connecting V to OV.
Features
Low Power Consumption
Break-Before-Make Switching
- tON
tOFF
TTL, CMOS Compatible
Low rDS(ON) (Max).
Smgle Supply OperatIon
True Second Source
.... 150ns
130ns
SOl
Ordering Information
TEMP.'----lPKG" -I
PART NUMBER I RANGE (0C) PACKAGE I NO. :
DG300A8K -25 la 85 J1J. Ld 3__ !
OG301ACJ 0\070 114 la POt? 14") J
lDG303AAK -55 to 125 :14 CERDIP f143 i
IOG303A8K -25 Ld 3 -:
iDG303ACJ 0 to 70
-----'-- --._- -----i ----
I __,_? 70 E:1
Functional Diagrams and Pinouts .... for a I0(;S T l:1p';:j
DG300A TRUTH TABLE
DG300A
(SPST)
5,
,
,
- ,
5,
0,
0,
LOGIC
o
"0' -0: 0 8'.' lcYJ c"1
SWITCH
Oi=F
O'J
4 G'I
OG300A (CEROl?)
TOD VIe-V;
DG301A (POIP)
TO;::>

GND 7
__-J
SWITCH 2
OFF
SWITCH 1
OG301A TRUTH TABLE
o
LOGIC
DGJ01A
ISPOT)
5, <>--+--0 02
IN
5, 0,
4-1 I
CAuT1Qr. ..C 10 ".::.r_d'::)" 10:c" le Hafld
J
.f1g
'I#/>-n "'le'S,! c.om o' .:J1- 721 9z'V Cu:>,""';'" ,; l"ler:.,1 Corporabor. 1':J9';