Simscape

HELMHOLTZ-INSTITUTE FOR BIOMEDICAL ENGINEERING
RWTH AACHEN
CHAIR FOR MEDICAL INFORMATION TECHNOLOGY

Univ.-Prof. Dr.-Ing. Dr. med. Steffen Leonhardt
Prototyping a closed loop control

system for a neonatal incubator
Carlos Paternain Soler
Supervisor: Abbas Khudair Abbas
Version: 2009-11-28
Pauwelsstraße 20
D-52074 Aachen
Telefon: +49 (0)241 80 -23211

Fax: +49 (0)241 80 -82442
E-Mail: [email protected]
Declaration
I declare to the best of my knowledge and belief that, this assignment is my own work, all
sources have been properly acknowledged, and the assignment contains no plagiarism. This
material has not been submitted, either in whole or part at this or any other university.
Place, Date Signature

Acknowledgement
First of all, I would like to express my sincere appreciation to the MedIT chair with its
director Professor Dr.-Ing. Dr.med. Steffen Leonhardt for giving me permission to conduct
this student research project (master thesis), to do the necessary investigation and to use
the institute’s facilities. Special thanks to Abbas Kader Abbas, M.Sc., who accepted to be
my supervisor, for his commitment to this work, reading, correction and friendly supervision
throughout all of my work. My thanks also go to Mr. Walter Wrobel, for giving me his
advice when I needed it.
My sincere gratitude also goes to all of the people who have motivated and helped me in
many aspects during this work, in the good and the bad moments. The task of naming all
of them would be unfair for the ones forgotten. And, of course, a special thank to my family
for giving me the opportunity to arrive were I am nowadays and for their loving care and
support.
Abstract
Premature born neonates often need special cares because some vital organs, biochemical
or enzyme systems may not have developed sufficiently, with the results that the infant is
unlikely to survive undamaged without special protection. Depending on culture and on
geographic location, either thermal contact with the mother or thermal insulation has been
a major form of protection for many centuries. The purpose of this study has been to design
and implement a closed loop control system to regulate the temperature inside a neonatal
incubator. This closed loop control system has been designed using the software in the loop
technique, which minimizes the hardware related work and concentrates all the decision-
making in the software. The closed loop implemented is a combination of sensing devices
and actuators that operate synchronously to provide a stable thermal environment inside the
incubator. The sensing devices are five resistance temperature detectors placed following the
IEC-60601 standards, two current sensors, which give feedback information about the power
consumption of the actuators, and two disturbance sensors that provide information about
the opening of the incubator clappers. The actuators, which produce the stable environment
inside the incubator, are the heater and the motor that propels the fan. Results confirm that
the closed loop control system has been successfully implemented and the software designed
is able to maintain a stable temperature inside the incubator. Moreover, the prototype
implemented allows continuing the task started with this work, with further optimization and
improvement of the controller.
Contents
Declaration iii
Abstract v
1. Introduction 1
1.1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2. Aim of the project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3. Organization of the document . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Theoretical bases and background 3

2.1. Brief historical review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2. Incubator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3. Thermoregulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4. Thermoregulation in neonates . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.5. Heat transfer mechanisms in neonates . . . . . . . . . . . . . . . . . . . . . 16
2.5.1. Convection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5.2. Conduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.3. Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5.4. Evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.6. Control system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.6.1. Closed loop control systems . . . . . . . . . . . . . . . . . . . . . . 21
3. Control System Design and analysis 25

3.1. Specification parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.1. NI 9217 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.1.2. NI 9205 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.1.3. NI 9263 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.1.4. RTD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.1.5. Current sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.1.6. Potentiometer transducer . . . . . . . . . . . . . . . . . . . . . . . 31
3.1.7. Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.1.8. Heater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2. Incubator setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3. Simulation of the incubator dynamics using Simulink . . . . . . . . . . . . . 35
3.3.1. Simulation objectives . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3.2. Thermodynamics of the incubator model . . . . . . . . . . . . . . . 38
3.3.3. Results and conclusions of the simulation . . . . . . . . . . . . . . . 40
3.4. Sensor interfacing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4.1. Temperature sensors . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4.2. Disturbance sensors . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.4.3. Current sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4.4. Signal acquisition with LabVIEW . . . . . . . . . . . . . . . . . . . 50
3.5. Controller hardware design and analysis . . . . . . . . . . . . . . . . . . . . 52
viii Contents
3.5.1. Heater driver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

3.5.2. Fan driver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.5.3. Control signals generation . . . . . . . . . . . . . . . . . . . . . . . 57
3.6. Controller software-based design and analysis . . . . . . . . . . . . . . . . . 59
3.6.1. Decoupled controller . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4. Results and discussion 63

4.1. Optimizing the output ranges of the PID controllers . . . . . . . . . . . . . 64
4.2. Plant response dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.3. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5. Conclusions and future work 71

5.1. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2. Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
A. List of Abbreviations 73
B. List of Symbols 75
C. LabVIEW software and figures 79

C.1. System Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
C.2. Control Signals Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
C.3. Decoupled PID Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
C.3.1. Heater PID Controller . . . . . . . . . . . . . . . . . . . . . . . . . 81
C.3.2. Motor PID Controller . . . . . . . . . . . . . . . . . . . . . . . . . 81
C.4. Decoupled Auto-tuning PID Controller . . . . . . . . . . . . . . . . . . . . 82
C.4.1. Heater Auto-tuning PID Controller . . . . . . . . . . . . . . . . . . 82
C.4.2. Motor Auto-tuning PID Controller . . . . . . . . . . . . . . . . . . . 83
D. EAGLE software and circuit layouts 84

D.1. RTD Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
D.2. Heater driver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
D.3. Fan driver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
References 87
1
1. Introduction
One of the most important elements in a newborn’s survival is the infant’s temperature reg-
ulation. Mammals have the advantage of being homeothermic, meaning that they are able
to produce heat enabling constant body temperature to be maintained. However a preterm
newborn infant needs special care because some vital organs and/or biochemical/enzyme
systems may not have developed sufficiently, or because the growth of the fetus may have
been disturbed, with the result that the infant is unlikely to survive undamaged without
special protection. An infant is called preterm if it is born following a gestation period of less
than 37 weeks. The preterm infant has several disadvantages in terms of thermal regulation.
An infant has a relatively large surface area, poor thermal insulation, and a small amount of
mass to act as a heat sink. The infant has little ability to conserve heat by changing posture
and no ability to adjust clothing requirements in a response to thermal stress. Responses
may also be hindered by illness or adverse conditions such as hypoxia (below normal levels of
oxygen). Heat exchange between the environment and the infant is like any physical object
and its environment. Heat is exchanged by conduction, convection, evaporation and radia-
tion. Heat exchange by conduction is relatively small. Conduction depends on the thermal
conductivity of the substance in contact with the body. Since babies are usually laid on a
mattress, usually composed of material with low thermal conductivity, the heat loss from the
baby to the mattress is relatively small. Heat loss from the infant by convection is depen-
dent upon air speed and air temperature. Evaporative loss depends upon air speed and the
absolute humidity of the air. Radiation heat transfer is produced by electromagnetic infrared
waves between solid surfaces that are not in contact[MIH08].
Neonatal incubators, or isolettes (also known as incubators), are some of the oldest medical
devices used to maintain constant body temperature for preterm infants. There are a variety
of incubators in neonatal intensive care units (NICUs), but they all do basically the same
job i.e. keeping the neonate warm[Tro08]. A neonatal incubator is, usually, a small (approx-
imately: 0.5 x 0.5 x 1 m3 ) cabinet with transparent walls so that the infant can be easily
observed. An artificial climate is maintained inside the incubator, and which usually differs
from the local environment with respect to temperature, humidity and/or oxygen concen-
tration. The device may include an AC-powered heater, a fan to circulate the warmed air,
a container for water to add humidity, a control valve through which oxygen may be added,
and access ports for nursing care. With the technology available currently, incubators use
microprocessor-based control systems to create and maintain the ideal microclimate for the
preterm neonate[Boy96][Zha05]. The newest incubators, such as the Dräger Isolette C2000,
use a proportional integral differential (PID) control algorithm to drive the servo control
system[Sys04].
From an engineering point of view, an incubator is one subsystem among other subsystems
such as, for instance, the feeding system or the diagnostic systems within the more general
systems of medical care for the newborn. The designs of this subsystem and the evaluation of
the modifications occasionally proposed require an appropriate physical analysis of the process
to be controlled, and of the disturbances whose influence has to be minimized[Dan87].
2 1. Introduction
1.1. Motivation
The progressive increase in servo-controlled incubators has created an ever increasing need
for detailed assessment of the ability of these incubators to achieve the desired thermo-
regulation for neonates. Thermo-regulation in the incubator depends on infant- as well
as incubator-related factors. Infant-related parameters that influence thermo-regulation in-
clude size, physical constitution, surface-area-to-volume ratio, skin thickness, metabolic rate
and maturity level. Incubator-related factors include air flow rates, humidity levels, tem-
perature, structural factors (wall thickness, material type, incubator geometry) and control
mechanisms[Sim94][AG75].
Many studies have analyzed the incubator related factors and how they affect thermo-
regulation inside the incubator. System identification of neonatal incubator based on adap-
tive autoregressive technique investigates the thermal disturbance from opening of incubator
clapper[AKA08]. Disturbance sensing is implemented by using potentiometers hinged with
the incubator clappers. This disturbance information can be used to improve the temperature
regulation inside the incubator.
1.2. Aim of the project
The purpose of this study is to design and implement a closed loop control system to regulate
the temperature inside a neonatal incubator. The closed loop control system is a combination
of sensors and actuators that operates synchronously to provide a stable thermal environment
inside the incubator. Using the Software In the Loop (SIL) technique this study wants to
implement the prototype for this closed loop control system. SIL technique is a flexible and
powerful design strategy, which minimizes the hardware related work and gathers together
the entire decision making in the software design, thus facilitating any possible changes.
1.3. Organization of the document
After the introduction in chapter one, the theoretical basis behind this work are presented in
chapter two. A brief historical review and the state of art of incubators are explained. Ther-
moregulation functions in neonates and heat transfer mechanism in incubators are discussed.
Also a brief introduction to control engineering is presented in this chapter. Chapter three
develops the design stages to implement the incubator closed loop. A simulation of incubator
heat transfer dynamics is discussed in this chapter. The power electronics and software-based
controller design are explained. Chapter four presents, analyses and discusses the results. In
chapter six, the conclusions of the work are presented. Appendixes are included at the end
of the document. A list of abbreviations used in this work, arranged in order of appearance
are shown in appendix A. List of symbols, also in order of appearance are shown in appendix
B. LabVIEW block diagrams of the software designed are shown in appendix C. Appendix D
shows the EAGLE designs for the electronic circuits.
3
2. Theoretical bases and background
2.1. Brief historical review
Culture and geography have been major influences on thermal insulation over the centuries
with incubators coming into use about 170 years ago.
Marx (1968) records the historical development of the incubator. Perhaps motivated by
the severe Russian winter, the first known incubator was developed at the imperial foundling
hospital in St. Petersburg in 1835 - the days of Pushkin and Gogol - by an unknown craftsman
at the suggestion of Von Ruehl, physician-in-ordinary to Czarina Feodorovna, wife of Czar
Paul I. This incubator consisted of a double-walled zinc tub with an open top: the space
between the walls was filled with warm water, so that an infant placed in the tub could
be kept warm[Ant09]. By 1850, some 40 of these incubators were in use in the Moscow
foundling hospital and by the end of the 19th Century various modified version were applied
in Germany, France and England[Dan87].
After an initial period during which incubators were applied on an intuitive basis, accumulated
experience lead to systematic investigations into how the temperature inside the incubator
affected the survival of the infant. Around 1900, Pierre Budin, who had studied under
Stephane Tarnier the inventor of the first warm air incubator, found that the mortality
of infants with a bodyweight of less than 2 kg decreased from 98% to 23% if the rectal
temperature is maintained above 35◦ C instead of below 32◦ C. The following quote belongs
to the study that Budin published in 1907 called The Nursling “Here is an instance which
occurred yesterday. A woman in the wards was delivered at 6.50 P.M. of an infant weighing
only 950 grams. As the vaginal temperature of the mother at the moment of delivery was
37.3◦ C it is likely that the temperature of the infant in the uterine cavity was not less. But
ten minutes after birth, at 7 o’clock, the thermometer placed in the rectum of this weakling,
did not register more than 35.1◦ C; the infant was immediately put in an incubator at 32◦ C
which, during the night, was maintained uniform.
Time Temperature (◦ C)
9 p.m. 35.6
11 p.m. 34.2
1 a.m. 34.0
3 a.m. 35.2
5 a.m. 35.9
7 a.m. 36.9
Table 2.1: Budin temperature measurements results
Here, then, is a weakling whose temperature fell to 34◦ C in spite of its having been
placed in an incubator at 32◦ C. How far would it have fallen if this measure had not been
taken?“[Bud07]
As Hey and O’Connell (1970) report: “He kept all his infants clothed and covered with
4 2. Theoretical bases and background
light blankets, but came to the conclusion that this was not in itself enough for the smaller
infants. He therefore extended and popularized the idea of nursing the more vulnerable
infants fully clothed in specially constructed incubators. Clinical experience led him to
recommend an air temperature of 30◦ C in these incubators for the smallest (1kg) infants
and a temperature of 25◦ C for most of the other small infants at risk“.
Figure 2.1: Tarniers Incubator used by Budin[Har]
Since Budin’s pioneering work, many other investigations have been conducted into the
clinical and physiological effects of micro-climate control and, as a result, several authors
have published guidelines for the practical use of incubators including temperature setting
for a particular infant given its weight, age and gestational age.
It was not until after the Second World War that special care baby units (SCBUs) also called
NICU were established in many hospitals. In Britain, early SCBUs opened in Birmingham
and Bristol. At Southmead Hospital (Bristol), initial opposition from obstetricians lessened
after quadruplets born there in 1948 were successfully cared-for in the new unit. Most early
units had little equipment and relied on careful nursing and observation. Incubators were
expensive so the whole room often was kept warm instead. Cross-infection between babies
was greatly feared. Strict nursing routines involved staff wearing gowns and masks, constant
hand washing and minimal handling of babies. Parents were sometimes allowed to watch
through the windows of the unit.
NICUs were developed in the 1950s and 1960s by pediatricians to provide better temperature
support, isolation from infection risk, specialized feeding, and access to specialized equipment
and resources. By the 1970s NICUs were an established part of hospitals in the developed
world. By the 1980s, over 90% of births were scheduled to take place in hospital and the
emergency dash from home to SCBU with baby in a transport incubator had become a thing
of the past; albeit transport incubators were still needed in some cases. Specialist equipment
and expertise were not available at every hospital, and strong arguments were made for large,
centralized NICUs.
On the downside was the long traveling time for frail babies and for parents. A 1979 study
showed that 20% of babies in NICUs for up to a week were never visited by either parent.
Centralized or not, by the 1980s few questioned the role of NICUs in saving babies. Hey
in “The care of babies in incubators“ (1971) thoroughly discusses the design, construction
5
and application of incubators from a medical point of view. His extensive review deals with
climate control, optimum warmth, oxygen supply, humidity, cleanliness, noise, light and cost.
Although it is his personal belief that most healthy infants can be nursed quite adequately
in an open cot because that is simpler and less expensive, incubators provide a convenient
means of controlling environmental warmth, humidity and oxygen, and of tailoring these
needs to the child’s requirements.
In the 1980s, 80% of babies born weighing less than 1.5kg survived compared to around 40%
in the 1960s. From 1982, in Britain, pediatricians could train and qualify in the sub-specialty
of neonatal medicine. Not only careful nursing, but also new techniques and instruments
now played a major role. As in adult intensive care units, the use of monitoring and life
support systems became routine. These needed special modification for small babies whose
bodies were tiny and often immature. Furthermore, by 1980, over 18% of newborn babies
in Britain were being admitted to SCBUs. The fact that these babies missed early close
contact with their mothers was a growing concern. As in other areas of medicine, the
1980s saw new questions being raised regarding the human, and the economic, costs of too
much technology and admission policies began to change. In addition, treating low birth
weight infants is expensive, especially when there are much cheaper ways of ensuring healthy
babies. The key is prevention with money being more effectively spent on programs educating
mothers on staying healthy during their pregnancy.
Currently, SCBUs concentrate on treating very small, premature, or otherwise sick babies.
Some of these babies are from higher-order multiple births, but most are still single babies
born too early. Preventing premature labor remains a perplexing problem for doctors. Even
though medical advancements enable doctors to save low-birth-weight babies, clearly it would
be better to prevent such births from happening in the first place. Over the last 10 years
or so, SCBUs have become much more ’parent friendly’, encouraging maximum involvement
with the babies. Routine wearing of gowns and masks has been abandoned and parents are
encouraged to help with infant care as much as possible. Cuddling, and skin-to-skin contact,
also known as Kangaroo care, is seen as beneficial for all but the frailest (very tiny babies, or
larger critically ill infants, who are exhausted by the stimulus of being frequently handled).
Neonatology and NICUs have greatly increased the survival of very-low-birth-weight and
extremely premature infants. In the era before NICUs, infants of birth weight less than 1400
grams (usually about 30 weeks gestation) rarely survived. Today, infants of 500 grams at 26
weeks have a fair chance of survival[Wik07].
2.2. Incubator
A neonatal incubator is a device consisting of a rigid box-like enclosure in which an infant

may be kept in a controlled environment for medical care. The device would include an AC-
powered heater, a fan to circulate the warmed air, a container for water to add humidity, a
control valve through which oxygen may be added, and access ports for nursing care. It may
also contain a servo control to help regulate incubator air temperature. The servo control
uses a temperature sensing thermistor, which is taped to the child’s abdomen.
Premature babies are not always placed in incubators. If a baby is in danger of going into
respiratory arrest or other significant problems, it is placed in an overhead radiant cradle so
that they are easily accessible to nurses and doctors. The radiation from overhead puts the
heat back into the baby while the baby is losing heat by other means.
Therefore the first distinction needs to be made between the two kinds of incubators available.
The radiant heat warmers are open incubators and use heat transfer by radiation to the
neonate. Closed incubators, also called isolette, use heat transfer by convection between the
neonate and the warm air[Mer79]. But both pursue the same objective i.e. to equalize the
thermal state of the neonate by transferring the same amount of heat lost by the neonate.
Heat losses and gains are difficult to monitor. Overhead radiant heaters can accommodate
for the heat lost by several routes but cannot account for water loss via the skin: this aspect
being crucial in preventing dehydration and in maintaining the infant for first 7 to 10 days
after birth[Che88][Mar87].
The current work was to design and to set-up a modified Dräger Isolette C500 (see Figure
2.2).
Figure 2.2: Incubator Isolette C500 from Dräger Medical

7
Some of the commercial models of incubators currently available are:
Ohmeda Giraffe OmniBed

The Ohmeda Giraffe OmniBed is a novel device designed to operate as both an incubator
and a radiant warmer, thus reducing the need to transfer the patient. This mobile
device incorporates all the usual features of modern incubators and radiant warmers. In
incubator mode, it has air temperature and baby-skin temperature control and integrated
humidification. In radiant warmer mode it has baby-skin temperature control and manual
heater control, where the user selects the percentage of power output from the radiant
heater[(GE02].
Figure 2.3: Ohmeda Giraffe Omnibed incubator
GE Lullaby Incubator XP
Lullaby Incubator XP has double wall clear acrylic hood that reduces radiant heat loss while
enabling unobstructed view of the infant. It also uses a microprocessor-based control system
and has a temperature variability of less than 0.5◦ C.
Figure 2.4: GE Lullaby Incubator XP

Medix Natal Care ST

The system includes a micro-processor control module and a graphic screen with large digital
displays for air and skin temperature monitoring. Also, two skin temperature channels are
available that allow peripheral as well as core temperature measurement.
Figure 2.5: Medix Natal Care ST incubator
Caleo Drger Medical incubator

The Caleo incubator allows the bonding benefit of kangaroo care combined with continuous
temperature monitoring outside the incubator. It is designed to facilitate physical access to
the neonate and sets the standard for a new generation of incubators.
Figure 2.6: Caleo Dräger Medical incubator

9
Finally, while surfing the web for information about incubators, the following interesting
project came to light. The project is being promoted by the Global Health Initiative at
the Center for Integration of Medicine and Innovative Technology (Cimit). It is a non-
profit consortium of Boston teaching hospitals and engineering schools which has designed
a neonatal incubator from automobile parts. The following is extracted from the article
published by Madeline Drexler in November 2008: “The heat source is a pair of headlights.
A car door alarm signals emergencies. An auto air filter and fan provide climate control. But
this contraption has nothing to do with transportation. It is a sturdy, low-cost incubator,
designed to keep vulnerable newborns warm during the first fragile days of life. Unlike the
notoriously high-maintenance incubators found in neonatal intensive care units in the United
States, it is easily repaired, because all of its operational parts come from cars. And while
incubators can cost $40,000 or more, this one can be built for less than $1,000. It could
prevent millions of newborn deaths in the developing world.“ More information at: [Tim08].
Figure 2.7: Design of the incubator promoted by Cimit made of car parts
2.3. Thermoregulation
Thermoregulation is the capacity to maintain normal body temperature under varying

environmental conditions and temperatures. Body temperature is a physiological variable
that is precisely controlled by complex metabolic processes. Just as an adequate supply
of oxygen is essential for cellular activities, an appropriate temperature control is critical
for the enzyme systems that regulate cellular functions. Normal body temperature of
37◦ C is the optimal thermal condition needed to support the organism’s internal functions.
Thermoregulatory responses balance heat production and heat loss to maintain this normal
body temperature. Body heat produced through metabolism is precisely balanced against
heat lost through convection, conduction, radiation and evaporation[KT]. The maintenance
of this “central set-point“ temperature allows the organism to maximize metabolic and
enzyme efficiency while minimizing oxygen consumption and caloric expenditure.
As in any other control system, two major elements can be recognized here: the controlled
part i.e. the body itself and the controller part which is largely the nervous system. The
controller aspect can be divided into three subsystems: a sensory, a processing, and an
actuator system.
The sensory system consists of a large number of temperature-sensitive neurons, some acting
as cold and others as warm receptors. They are located mainly in the brain (hypothalamus),
the spinal cord, the skin, and upper part of the respiratory tract. There are various
types of receptors. The firing rate of a central thermo-receptor responds to a constant
temperature level (relative to a certain threshold). Receptors in the skin also respond to
temperature variation with time (proportional plus rate sensitivity). Skin receptors are
unevenly distributed over the body-surface with facial receptors having a relatively high
density of distribution. In adults, the average depth of thermo-receptors in the skin is 0.15
-0.17 mm for cold receptors and 0.3-0.6 mm for warm receptors. It is in the hypothalamus
that sensory information describing thermal status throughout the body is processed and
compared to the body temperature set-point. The processing system has a static non-linear
characteristic without integral or derivative control action.
The actuator system consists of a metabolic heat production modulated by variation in
peripheral blood flow and resultant heat loss via evaporation. All these processes are involved
in other regulatory systems including respiration and voluntary muscle control systems. With
respect to actuator mechanisms, sweating is unique in that it is completely dedicated to the
thermo-regulatory system and receives only minor signals from other systems such as, for
instance, the salt balance.
Allan Hemingway (1963) reviewed shivering and non-shivering thermo-genesis. Shivering
is controlled by the spinal and supra-spinal motor system, whereas the other actuator
mechanisms are mediated by the sympathetic nervous system. Cooling of the spinal
cord increases the excitability of motor neurons thereby causes shivering. Non-shivering
thermo-genesis is a very effective way to produce heat through the oxidation of brown
adipose fatty tissues. Normally it occurs in newborn infants as a substitute for shivering,
and adults can use it as an additional source of heat production[Hul88].
Regional differences exist in the vasomotor control of peripheral blood flow. At least three
different zones can be distinguished: the extremities (hand, feet, lips and nose), the trunk
11
plus proximal limbs, and the head. In the first of these zones, vasoconstriction is the result
of increased sympathetic tone while vasodilatation is of decreased tone. In the second zone,
vasodilatation is the result of an active process, acting via the sweat glands. The third zone
shows a negligible constrictive response to cold, but a dilatation coupled to sweating when
heated. Passive heating of the body produces a blood-flow pattern different from heating
by active work; the main increase in blood flow occurring in zone 1 in the first case, and
in zones 2 and 3 in the second situation. Passive heating also causes sweating, initially in
the lower extremities if the whole body is heated. A rising temperature due to active work
causes sweating in zones 2 and 3[Dan87]. Local heating or cooling of a particular body part
results in a local response next to the central one. Local control systems are coupled to the
central system in a hierarchical structure.
Figure 2.8: Body temperature regulation

2.4. Thermoregulation in neonates
Newborn infants have only a modest capacity for sweating and heat production, and this
depends on gestational age and degree of maturation. Considerable variations do occur in
their ability to stabilize body temperature; some infants may even temporarily lose their
thermoregulatory capacity.
Unless immediate attention is given to heat loss, the neonate’s temperature can drop
approximately 4.5◦ C during the first minute after birth. Because the infant is dependent
on environmental temperature, provision of thermal support is a primary nursing objective.
Neonates have higher metabolic rates than do children or adults. Metabolism reflects
the overall energy needs supporting maintenance, repair and growth i.e. higher metabolic
rate is due not only to energy demands related to growth but also to the increased
thermoregulation requirements related to the neonate’s large body surface area and a large
surface-to-mass ratio. Body heat generated by body mass (weight of metabolizing cells
producing heat) is lost over surface area. Hence, the smaller the neonate, the greater the
imbalance between heat-producing ability (mass) and heat-losing capacity (surface area).
Sauer, Dane and Visser (1984) stated that of the calories consumed by a neonate 42% go
towards maintenance and thermoregulation. Posture significantly affects surface area and
heat loss. For example, the extended or spread-eagle posture increases heat loss by 35%
when compared to a curled-inward fetal position[KT][Hul88][Mot73].
The relationship between the body temperature and the oxygen consumption has been
widely studied. Adamsons, Gandy and James (1965) concluded that “the rate of oxygen
consumption of the newborn infant is predominantly a function of the temperature gradient
between body-surface and the environment, rather than absolute values of either deep-body
or surface temperature“.
The neonate’s lack of ability to regulate body temperature can be observed in Figures 2.10
and 2.11 in which the thermo-neutral range (stable body temperature and minimum oxygen
consumption) and the normo-temperature range (stable body temperature) for both adults
and neonates are depicted. The yellow chart represents the central temperature of the
body and the green chart represents the percentage oxygen consumption. Premature babies
have a much thinner range for both thermo-neutral and normo-temperature. Maintaining
neonates within this thermo-neutral range of temperatures is the purpose of micro-climate
control in incubators; the thermo-neutral environment makes minimal demands on the
subject’s energy reserves. Therefore, incubators must prevent neonates from cold stress
(hypothermia) or overheating (hyperthermia) by controlling the micro-climate to provide a
thermo-neutral environment for neonates.
13
Figure 2.9: Thermo-neutral neutral range of an infant[AL04]
Figure 2.10: Thermo-neutral range of a premature newborn[AL04]
Many studies have been conducted to ascertain the optimal environment for very low birth-
weight infants (VLBW) in incubators.
Hey and Katz (1970), sought to define what constitutes a neutral thermal environment for a
naked baby in a commercial incubator. They performed a study with newborns between 0.96
kg and 4.76 kg body-weight and measured the average heat production within a control range
of temperatures. The findings were used to establish estimates of what constitutes a neutral
thermal environment[HG70]. Yashiro, Forrest and Adams (1973), in a preliminary study on
the thermal environment of VLBW, observed that the optimal skin temperature in an open-
ended servo-controlled radiant heater was approximately 36.7◦ C. However, they recommend
more extensive observations to validate these findings[KYMRM73]. Scopes (1976), discussed
the control strategies of servo-controlled incubators for VLBW infants and concluded that
“servo-controlled incubators are as good as the person who uses it“[Sco77].
Ducker, Lyon, Rusell, Bass and McIntosh (1985), studied temperature stability in the first
four days of life in 22 infants of birth weight less than 1500 grams. Infants were nursed
in incubators using either air mode control or skin temperature servo-control. Data were
collected continuously using a computer linked monitoring system. The study showed that
changes in abdominal skin temperature occurred considerably later than the temperature
changes in a more peripheral skin measuring site and, therefore, the incubator heater responds
only after some delay when appreciable cooling had already occurred. Delay in switching off
the heater resulted in an overshoot of the infant’s temperature above the set value. Servo-
control therefore effectively maintains the reference temperature within a narrow range but
did so by exposing the infant to both heat and cold stress in an unstable thermal environment.
The infants in incubators controlled in air mode had a more stable thermal environment and
showed much less variation in the abdominal and toe temperature and, as well, in the central-
peripheral temperature difference. The study showed that the use of servo-control with the
abdominal skin temperature as the reference seemed to create further problems for the VLBW
infant. Using a more peripheral temperature as a reference allows a more rapid response of
the incubator heater[DADM96].
Telliez and Bach (1997), design a servo-controlled skin temperature derivative (SCS) heating
device to control the thermal environment in closed incubators without the necessity of
setting an air or skin reference temperature. The thermal environment obtained with the
SCS program is controlled by the neonate’s skin temperature changes. The results suggest
that the skin temperature derivative heating program takes into account both the ambient
and the physiological factors affecting body temperature regulation. Conversely, in current
commercialized closed incubators, the control of the thermal environment depends on air
temperature or skin temperature, which is set at a reference value. It does not take into
account the environmental and individual factors which can modify heat exchanges between
the neonate and the environment[FT97].
15
2.5. Heat transfer mechanisms in neonates
Heat is gained or lost by convection, conduction, radiation and evaporation. An understand-

ing of these avenues for heat exchange provides the basis for modification of the environment.
In assessing the thermal environment, it is important to account for all sources of heat gain
and loss. Since heat transfer is always down a heat gradient (i.e. from a higher to a lower
temperature), the larger the gradient, the greater the heat flow.
When attempting to modify the thermal environment it is important to differentiate between
application of heat (the heat source is warmer than the infant’s skin temperature) and re-
duction of heat loss (the heat source is lower than skin temperature). The infant is thermally
insensitive to the particular route of heat gain or loss. In other words, the thermal responses
to cooling are the same regardless of how the heat was lost. Figure 2.12 depicts the heat
loss mechanisms from a neonate.
Figure 2.11: Heat loss routes from neonates
2.5.1. Convection
Convective heat transfer can be described by the following equation:
Q = (T surf − T env) ∗ hconv ∗ Asurf (2.1)
Where Q represents the heat flow, hconv the convective heat transfer coefficient and Asurf
the surface area with convective heat transfer.
Convection is the transfer of heat between a solid surface (the infant) and either air or liquid
(as in bathing). Factors determining convection losses that are relevant to nursing care
include:
1. The infant’s large surface area. A large surface-area-to-body-mass ratio results in
increased loss by convection. Reducing exposed surface area is an effective means of
minimizing convection loss.
2. Air-flow velocity and turbulence. These are directly related to convection heat loss.
Air-flow is a concern when infants are exposed to drafts created by doors, ventilation
systems, and traffic flow around the infant’s bed. The faster the air-flow velocity and
the greater the turbulences in the air-flow the greater the convection heat loss.
3. Temperature gradient between infant skin and air or liquid. The greater the gradient
between the infant’s skin temperature and the ambient temperature, the greater is
the convection heat loss, or gain.
4. Diameter of infant’s body limbs. The smaller the diameter of infant’s limbs, the less
external insulation and, thus, the greater is the potential for heat loss via convection.
Incubators operate by convection rather than “warm“ infants. Incubators reduce convection
heat loss by decreasing the temperature gradient between air and skin. The exception is when
incubator air temperature is higher than skin temperature. Although convection heat loss is
modified by the incubator, radiation and evaporation remain avenues for heat loss. Air-flow
velocity in a closed incubator is fairly consistent, but opening portholes or side panels alters
air-flow and creates turbulence. Use of clothing and blankets in incubators is an effective
means of reducing the infant’s exposed surface area and providing external insulation[KT].
2.5.2. Conduction
Conductive heat transfer can be described by the following equation:

T1 − T2
Q= L
(2.2)
K∗A
Where Q is the heat flow, L the width of the conductive material, k the conductive heat
transfer coefficient and A the area which is conducting the heat.
Conduction is the transfer of heat between two solid objects that are in contact. Conduction
refers to heat flux between the infant’s body surface and other solid surfaces. The following
factors influence conduction:
1. The solid surface conductivity coefficient. This coefficient quantifies the surface’s
ability to transfer or conduct heat. Metals are highly conductive; plastics and wood
are less so. The greater the conductivity, the greater is the heat flux between the
infant and the solid surface.
2. The size of the surface area of contact between the infant and the solid surface. The
larger the surface in contact with the object, the greater is the heat flux. In the
supine position, an infant has approximately 10% of its surface are in contact with
the mattress. In a pre-warmed incubator, or under a radiant warmer, conduction heat
loss is not substantial.
17
3. The temperature gradient between surfaces. The infant is not heated by conduction
unless the contacting solid surface is warmer than skin temperature. Warming pads
and similar devices typically reduce heat loss by providing a smaller gradient between
the infant and the solid surface. In clinical practice, measurements of the infant’s skin
temperature and the temperature of surfaces in contact with the infant are means of
assessing conduction heat loss.
Measures to prevent conduction heat loss include warming solid surfaces before they come
in contact with an infant, and providing insulation between the infant and the solid surface.
Examples include wrapping x-ray plates with warmed blankets and placing them between the
infant and the surface of the scales[KT].
2.5.3. Radiation
Radiative heat transfer can be described by the following equation:
Q = (T surf − T surr) ∗ hr ∗ Asurf (2.3)
Where Q is the heat flow again, hr is the radiative heat transfer coefficient and Asurf is the
surface which is radiating.
Radiation is the transfer of heat between solid surfaces that are not in contact. Heat energy
is transferred by electromagnetic infrared waves. Since radiant heat flux does not involve
physical contact and is not measured by ambient air thermometers, this source of heat loss
is often misunderstood, or ignored. Feeling cold while sitting beside a cold exterior window,
despite a comfortable room air temperature, is an illustration of radiant heat loss. Likewise,
sunshine heating a car’s interior, even on a cold day, is an example of radiant heat gain. The
following factors affect radiant heat flux:
1. The emissivity of radiating surfaces. Emissivity, the power to emit or give-off heat by
radiation, is a comparison of the radiant capability of the surface in question to that of
a standardized black sphere at a similar temperature. The human body is similar to a
black object in that it reflects little infrared heat. The emissivity of a neonate’s skin is
relatively constant. However, clothing and blankets reduce emissivity as well as provide
insulation. Since radiation involves heat exchange between solid surfaces, the surface
temperature and emissivity of solid objects surrounding the infant should be considered.
2. The temperature gradient between solid surfaces. Radiant heat transfer is driven by
the temperature gradient between solid surfaces. The infant’s skin temperature is
typically warmer than other surrounding surfaces. Hence, the direction of radiant heat
transfer is usually from the infant’s surface to surrounding solid surfaces. Heat gain
by the infant via radiation occurs when a radiant heating source is warmer than the
infant’s skin temperature.
3. The surface area of solid surfaces. The infant’s large body surface-area-to body-mass
ratio potentiates radiant heat loss. Compared to the infant’s surface area, surrounding
solid surfaces are enormous and so contribute to even greater radiant heat loss. For
the supine infant, the largest surface areas from which radiant heat is lost are directly
above (30%) and to the side (17%) of the infant. Thus, the mattress surface adjacent
to the infant can be an important avenue for radiant heat exchange.
4. The distance between solid surfaces. The closer the two solid surfaces, the greater
is the radiant heat loss. This is of importance when considering radiant heat gain as
well as loss.
Ambient air temperature does not reflect the radiant temperature. Radiant temperature
is measured using a black globe thermometer; typically a black metal sphere surrounding
a thermistor, or other temperature measuring device. Unless a radiant heat source is
present in a nursery, radiant temperature is typically lower than the air temperature.
Sunlight or other radiant heat sources can cause a radiant temperature greater than the
air temperature; the result is the “greenhouse effect“ observed in incubators. When an
incubator is exposed to sunlight, the Plexiglas material of its construction transmits nearly
100% of the sun’s short-wave radiation, which heats the infant. However, virtually none of
the infant’s long-wave radiation is transmitted through the Plexiglas, so the infant easily
becomes overheated. Therefore, incubators should be placed away from direct sunlight.
Radiant heating devices such as heat lamps or phototherapy units used with incubators can
also result in overheating. Since air temperature is not a measure of radiant temperature,
the incubator temperature will not accurately represent the degree of infant heating[KT].
2.5.4. Evaporation
Evaporation produces heat loss through the energy used in the conversion of water to its
gaseous state. Evaporation of one liter of water utilizes 600 Kcal of heat, termed “latent
heat“ of vaporization. Evaporation loss may be insensible (from skin and respiration) or
sensible (from sweating). Trans-epidermal water loss (insensible loss from the skin) is
correlated with gestational age and the degree of keratinization of the epidermal stratum
corneum. Mature keratin, the nonliving layer of tough fibrous protein protecting underlying
epithelium, is relatively water impermeable. Keratin formation is directly related to
gestational age and, as such, premature infants have increased evaporation losses as the
body water diffuses across the permeable skin barrier and evaporates. In the VLBW infant,
evaporation loss alone is greater than the infant’s heat-producing capabilities. Influenced by
the extra uterine environment, keratinization increases over the first three to four weeks of
postnatal life and contributes to a reduction in losses through evaporation[KT][Abd98].
Skin damage that strips the keratin layer (and possibly affecting the underlying epithelial
layer), promotes skin evaporation loss. Such damage can be caused by repeated application
and removal of tape or other adhesive products. Factors affecting evaporative losses include
19
the following:
1. The infant’s surface area. The greater the surface area available for evaporation, the
greater is the heat loss.
2. Vapor pressure, which is governed by air pressure, temperature and humidity. The
greater the ambient air pressure, the less is the evaporation loss. Although air pressure
is relatively stable, it decreases with increasing altitudes such as during infant transport
in non-pressurized aircraft. Since evaporation and temperature are directly related,
increases in temperature would decrease vapor pressure and increase evaporation.
Therefore, warm air temperatures required by the infant are associated with high evapo-
rative loss. However, humidity increases vapor pressure and decreases evaporation loss.
3. Air velocity. Evaporation heat loss, like convection heat loss, is potentiated by the
increased speed and turbulence in air-flow.
2.6. Control system
Fundamental to any control system is the ability to measure the output of the system, and
to take corrective action if the value deviates significantly from some desired value. This in
turn necessitates a sensing device. The human organism has a number of ’in-built’ senses
which have been used for self-monitoring and control of action, or to control the action of
others and, more recently, to control the actions of machines. The aim of our study is to
control the temperature inside the pre-term infant incubator. The sensing devices used are 5
RTD sensors, the potentiometers hinged in the door clappers and the current sensors in the
actuators. To reach the desired temperature, the necessary actions to the actuators will be
applied, and which include a motor that propels the fan and a heater resistance that produces
heat.
Before discussing the structure of a control system it is necessary to define what is meant
by a system. Systems mean different things to different people. However, all systems have
certain things in common. For example, they all require inputs and outputs to be specified.
In our case, as mentioned above, the inputs are the control signals to drive the power for
the heater and the fan, the two mounted potentiometers for the frontal clappers used for
triggering the event of closing and opening. The outputs are two current sensors providing
feedback information about the actuator status and the temperature inside the incubator
which is monitored with 5 RTD sensors.
In control engineering, the way in which the system outputs respond to changes in the
system inputs (system response) is critical. Knowledge of the system inputs, together with
the mathematical model, will enable the system outputs to be calculated. It is conventional
to refer to the system being controlled as “the plant“ and this, as with other elements, is
represented by a block diagram. The engineer will have direct control over some inputs
which can be used to control the plant outputs. These are termed control inputs. There are
other inputs over which the engineer has no control, and these will tend to deflect the plant
outputs from their pre-set (desired) value. These are termed disturbance inputs.
In our study (depicted in Figure 2.13), the heater and fan control signals are our control
inputs. The clapper position inputs are disturbance inputs since they introduce errors in the
outputs, the temperature (termed controlled variables) sensed with 5 RTD sensors. The
fan current sensor and the heater current sensor provide information on the state of the
actuators[Hei05].
Figure 2.12: The incubator as a dynamic system

21
The following essential generic elements are required to design and implement a control
system:
- Knowledge of the desired value: It is necessary to know what is being controlled, to

what accuracy, and over what range of values. This must be expressed in the form of
a performance specification.
- Knowledge of the output or actual value: This must be measured by a feedback sensor,
again in a form suitable for the controller to understand. In addition, the sensor must
have the necessary resolution and dynamic response so that the measured value has
the accuracy required from the performance specification.
- Knowledge of the controlling device: The controller must be able to accept measure-
ments of desired as well as actual values and to compute a control signal in a suitable
form to drive an actuating element.
- Knowledge of the actuating device: This unit amplifies the control signal and provides
the “effort“ to move the output of the plant towards its desired value.
- Knowledge of the plant: Most control strategies require some knowledge of the static
and dynamic characteristics of the plant. These can be obtained from measurements
or from the application of fundamental physical laws, or a combination of both.
With all of this knowledge and information available to the control system designer, all that
remains is to design the system. The first problem to be encountered and resolved is that
the knowledge of the system will be uncertain and incomplete. In particular, the dynamic
characteristics of the system may change with time, and so a fixed control strategy will not
work.
Measurements of the controlled variables will be contaminated with electrical noise and
disturbance effects. Some sensors will provide accurate and reliable data while others,
because of the difficulties in measuring the output variable, may produce highly random,
and almost irrelevant, information. The design of a control system is a mixture of technique
and experience[Hei05].
2.6.1. Closed loop control systems
A closed-loop or feedback control system is any system in which the output quantity is
monitored and compared with the input and any difference is used to actuate the system
until the output equals the input. In our case we compare the incubator temperature to
our desired temperature value and, depending on the difference encountered, we will apply
different actions to the actuators. A generalized closed-loop control system is shown in
Figure 2.14. The control problem can be stated as: The control action u(t) will be such
that the controlled output c(t) will be equal to the reference input r1(t) for all values of
time, irrespective of the value of the disturbance input r2(t). In practice, there will always
be transient variations, but the transient period should be kept as small as possible. Usually,
it is possible to design the controller so that steady-state errors are minimized, or ideally,
eliminated[Hei05].
Figure 2.13: Generalized closed-loop control system
This generalized diagram does not reflect exactly the behavior of our system, although any
kind of closed-loop system can be expressed this way.
The following are types of controllers for closed-loop systems.
Proportional control
In this case the control action or signal is proportional to the error, as can be seen in
Equation 2.1 where K1 is the proportional gain constant.
u(t) = K1 ∗ e(t) (2.4)
For a first order plant, proportional control will always produce steady-state errors.
Proportional plus Integral (PI) control

Including a term that is a function of the integral of the error can eliminate steady-state
errors. Now the control signal is:
Z
u(t) = K1 ∗ e(t) + K2 e ∗ dt (2.5)
The variable Ti is termed the integral action time, and is formally defined as: The time
interval in which the part of the control signal due to integral action increases by an
amount equal to the part of the control signal due to proportional action when the error is
unchanging.
Ti is given by Equation 2.3:
K1
Ti = (2.6)
K2
For a first-order plant, PI control will produce a second-order response. There will be
zero steady-state errors if the reference and disturbance inputs r1(t) and r2(t) are either
unchanging or have step changes.
23
Proportional plus Derivative (PD) control

PD control action is expressed in Equation 2.4:
de
u(t) = K1 ∗ e(t) + K3 ∗ (2.7)
dt
In PD control, another variable, Td , is termed the derivative action time, and is formally
defined as: The time interval in which the part of the control signal due to proportional
action increases by an amount equal to the part of the control signal due to derivative action
when the error is changing at a constant rate.
Td is given by Equation 2.5:
K3
Td = (2.8)
K1
The inclusion of a derivative term in the controller generally gives improved damping and
stability.
Proportional plus Integral plus Derivative (PID) control

Most commercial controllers provide full PID (also termed “three-term“) control action.
Including a term that is a function of the derivative of the error can, with high-order plants,
provide a stable control solution. PID control action is expressed as:
Z
de
u(t) = K1 ∗ e(t) + K2 e ∗ dt + K3 ∗ (2.9)
dt
The proportional component depends only on the difference between the set-point and
the process variable. This difference is referred as the error term. The proportional gain
K1 determines the ratio of output response to the error signal. In general, increasing the
proportional gain will increase the speed of the control system response. However, if the
proportional gain is too large, the process variable will begin to oscillate. If K1 is increased
further, the oscillations will become larger and the system will become unstable and may
even oscillate out of control.
The integral component sums the error term over time. The result is that even a small
error term will cause the integral component to increase slowly. The integral response
will continually increase over time unless the error is zero, so the effect is to drive
the steady-state error to zero. Steady-state error is the final difference between the
process variable and set point. A phenomenon called integral windup results when in-
tegral action saturates a controller without the controller driving the error signal toward zero.
The derivative component causes the output to decrease if the process variable is increasing
rapidly. The derivative response is proportional to the rate of change of the process variable.
Increasing the derivative time Td parameter will cause the control system to react more
strongly to changes in the error term and will increase the speed of the overall control
systems response. Most practical control systems use very small Td , because the derivative
response is highly sensitive to noise in the process variable signal. If the sensor feedback
signal is noisy or if the control loop rate is too slow, the derivative response can make the
control system unstable.
The selection of PID control parameters K1 , Ti , and Td can be obtained using control system
design techniques. However, in the 1940s, when such tools were just being developed,
Ziegler and Nichols (1942) devised two empirical methods for obtaining the controller
parameters. The Process Reaction Method and The Continuous Cycling Method, both
of this PID tuning methods provide a useful ’rule of thumb’ empirical approach. Modern
control system design techniques will generally yield better design solutions. Of the two
techniques, the Process Reaction Method is the easiest and least disruptive to implement.
The Continuous Cycling Method, although more disruptive, has the potential to give better
results.
Another method to obtain the gain of a PID controller is the trial and error method. In this
method, the Ti and Td terms are set to zero first and the proportional gain is increased until
the output of the loop oscillates. As one increases the K1 , the system becomes faster, but
care must be taken in order to not make the system unstable. Once K1 has been set to
obtain a desired fast response, the integral term (K2 ) is increased to stop oscillations. K2
reduces the steady-state error, but increases overshoot. Some amount of overshoot is always
necessary for a fast system so that it could respond to changes immediately. K2 is tweaked to
achieve a minimal steady-state error. Once K1 and K2 have been set to get the desired fast
control system with minimal steady-state error, the derivative term (K3 ) is increased until
the loop is acceptable quick to its set point. Increasing derivative term decreases overshoot
and yields higher gain with stability but would cause the system to be highly sensitive to
noise. Often times, a tradeoff of one characteristic of a control system for another must be
done to meet the requirements.
The Ziegler-Nichols method is very similar to the trial and error method wherein Ti and Td
are set to zero and K1 is increased until the loop starts to oscillate. Once oscillation starts,
the critical gain Kc and the period of oscillations Pc are noted. K1 , Ti , and Td are then
adjusted as per Table 2.2[Ins06].
Control K1 Ti Td
P 0.5Kc - -
PI 0.45Kc Pc /1.2 -
PID 0.65Kc Pc /2 Pc /8
Table 2.2: Ziegler-Nichols tuning, using the oscillation method

25
3. Design and analysis

Throughout this chapter the requirements of the temperature controller inside the incu-
bator will be analyzed and, subsequently, designed and implemented. As stated earlier,
the purpose of this study is to design and implement a closed loop control system to
regulate the temperature inside a neonatal incubator. The use of software in the loop
technology enables hardware-related work to be minimized and, hence, increasing the design
flexibility since software changes are easier to introduce than hardware modifications. A sim-
ulation of the incubator thermodynamics to be implemented using Simulink will be presented.
A modified plant of a neonatal incubator C500 Isolette (from Dräger Medical) is used in
this work. Real time temperature data are collected inside the incubator from five points,
according to the IEC-60601 standards, using five different Resistance Temperature Detectors
(RTD) from Omega with a 0.082◦ C/50◦ C thermal resolution. Two mounted potentiometers
for frontal clappers are used for triggering the closing and opening events and two current
sensors provide feedback information on the state of the actuators. The sensors and control
signals for the actuators are interfaced using Modular Compact DAQ 9172 (National Instru-
ments) acquisition system with a PC-Windows XP SP2. LabVIEW software is used to create
the software-based controller.
3.1. Specification parameters
Defining the specifications of the devices that are to be used in the prototype is a task that
needs to be done as a first step of the design. This enables successful design and subsequent
implementation based on the limitations and requirements identified in this stage. Compact
Data Acquisition Modules, RTD sensors, current sensors, and potentiometers hinged in incu-
bator clappers, motor and heater will be defined and any limitation or need will be highlighted.
The main limitations in the design derived from the Compact Data Acquisition Modules
(DAQ Modules) installed in the compact DAQ 9172. These modules are used to obtain and
generate the system signals. For the purpose of this work, there are three DAQ modules
available.
26 3. Design and analysis
3.1.1. NI 9217
4 channel, 100Ω RTD, 24-Bit Analog Input Module.

- Works at a sample rate of 400S/s.
- Gives a temperature measurement.
- 24 bits resolution with 50/60 Hz noise rejection.
- Temperature accuracy 0.15◦ C (typical) and 0.35◦ C (maximum).
Figure 3.1: DAQ Module NI 9217 terminal assignments
More information available at:[Ins07a].
3.1.2. NI 9205
32 channel ±200mV to ±10V, 16 Bit, 250kS/s Analog Input Module.
- 32 single ended or 16 differential analog inputs.

- Gives a voltage measurement.
- ±200mV, ±1V, ±5V, ±10V programmable input ranges.
Table 3.1 summarizes the accuracy of the measuring ranges.
Range (V) Range accuracy (µV)

10 6230
5 3230
1 690
0.2 174
Table 3.1: Ranges and accuracies of NI 9205

27
More information available at: [Ins07b].
3.1.3. NI 9263
4 channel, 100kS/s, 16 Bits, 10V Analog Output Module.

- 4 simultaneously updated analog outputs, 100kS/s, 16 Bits resolution
- Voltage output
- 10V range, 0.11V accuracy
- 1mA/channel current drive
More information available at: [Ins07c].

It is immediately clear that only four of the five RTDs can be connected to the
DAQ Module (see Figure 3.1). Therefore an interface must be designed and introduced to
convert the signal from the RTD to an analog voltage that can be an input to the DAQ
Module of Figure 3.2.
Also, only four outputs can be generated, with a maximum output voltage of 10V and
a maximum current of 1mA/channel, from the DAQ Module (in Figure 3.3). The DAQ
Module (in Figure 3.2) also introduces a limitation in any signal used as an input to the
software-based controller; the maximum input voltage being 10V.
All these modules are controlled through LabVIEW software with the DAQmx Virtual
Instruments (VIs) and its functions.
3.1.4. RTD
The RTD sensors used are from OMEGA, four SA2C and one SA2F models. They have a
0.01◦ C temperature resolution, but the DAQ Module (Figure 3.1) limits the resolution to
0.15◦ C. These sensors are placed in the incubator following IEC-60601 standards.
Figure 3.4: RTD sensor from OMEGA

29
3.1.5. Current sensor
Current sense transformers have been the standard for precise current measurement in
instrumentation, and other high reliability equipment applications. They are accurate, easy
to implement, and reliable under harsh environmental and thermal conditions. In electronic
systems applications such as switch-mode power supplies, current transformers are generally
used for control, circuit-protection, and monitoring features.
In this work, current sense transformers will be used to monitor the power that is being
delivered to both actuators; the heater and the motor for the fan. This will be a feedback
input for our software-based controller.
The selection of a current transformer begins with the definition and verification of
certain factors such as size, frequency, function and range of current being sampled; the
accuracy and effectiveness will be dependent on these parameters. Typically, selecting a
current-transformer that is rated approximately 30% above the expected maximum of the
“sample current“ is a prudent starting point.
As can be seen in Figure 3.5, a Burden Resistor must be placed at one end of the
current sense transformer to create a voltage drop that is proportional to the current Io. In
this way, and through the DAQ Module 9205, real-time monitoring of the power consumed
by the load is possible[AG].
Figure 3.5: Current sense circuit

Equation 3.1 and Equation 3.2 show the theoretical behavior of the current sense transform-
ers. Nprim:Nsec, being the primary:secondary turns ratio.
VO
RO = (3.1)
IS
IO
IS = (3.2)
Nprim
For the purpose of the present work, TZ77V current sense transformer from Taehwatrans,
has been chosen. Table 3.2 shows the TZ77V electrical performance:
Model Current ratio (Nprim :Nsec ) Ratio error Burden at 20Ω Current rating
TZ77V 2500:1 ±0.2% 0.0135VA 1A,6A
Table 3.2: Current sense transformer characteristics[tC05]
Since neither the motor, nor the heater, consumes more than 6A, this current sense trans-
former is appropriate for the purpose of this work. Figure 3.6 demonstrates the TZ77V
current sense transformer:
Figure 3.6: TZ77V current sense transformer
Care needs to be taken to ensure that the maximum input voltage for the DAQ Module NI
9205 is ±10V and, hence, in a future work, an appropriate Burden Resistor must be placed
in order to not exceed that maximum voltage input.
31
3.1.6. Potentiometer transducer
The existence of thermal disturbance, such as when opening the incubator clapper, was
considered as the principal cause in significant incubator heat loss. The heat losses during
routine clinical use (when nursing staff provide medication, nutrition or visual inspection of
newborn infant), may cause a considerable drop in temperature surrounding the incubator.
To provide information regarding the position of the incubator clappers, two potentiometers
have been mounted on the frontal clappers and these used to trigger the closing and opening
events.
Figure 3.7: Mounted potentiometers in door clappers
Figure 3.7 depicts the two mounted potentiometers (circled in red). These potentiometers
give a voltage that is proportional to the opening angle of the clappers. The given voltage
must not exceed the maximum input analog voltage of the DAQ Module NI 9205 which is
±10V.
3.1.7. Motor
The motor is used to propel the fan. The same type of motor that is used on the original
incubator has been chosen for the purpose of this present work. That the same motor has
not been used is due to the lack of information related to the Isolette C500 Incubator power
electronics. The company has ceased to exist, and the actuators used are old. Therefore,
it has been decided that a new motor should be purchased to ensure correct control. The
motor VD-1-54.14 from Ebmpapst depict in Figure 3.8 has been purchased for the purpose
of fan propulsion.
Figure 3.8: Ebmpapst Variodrive Motor VD-1-54.14
Table 3.3 shows the nominal data relating to this motor:
Type VD-1-54.14
Nominal voltage (VDC ) 24 (1228)
Nominal speed (rpm) 4000
Nominal torque (mNm ) 62
Nominal current (A) 1.75
Nominal output power (W) 26
Maximum starting current (A) 2.7
Rotor moment of inertia (kg2 ∗ 10−6 ) 200
Motor mass (kg) 0.52
Table 3.3: Nominal data of motor
This model allows varying the speed via connection voltage. This is done through two single
leads, AWG 24, and has clockwise rotation. More information about the motor is available
from the manufacturer’s datasheet[eG06].
3.1.8. Heater
The heater is used to warm-up the air that is propelled by the fan. One of the requirements
of the heater is that it should have the same heating surface area. With present technology,
heaters have reduced in volume and, subsequently, their surface as well but still producing
the same amount of heat. A search has been performed to identify a model of heater that
satisfies the requirements of the present work, but no appropriate model has been found
that fulfills the demands. Hence, the same heater that is used in the original Isolette C500
Incubator will continue to be sued. This heater is depicted in Figure 3.9.
33
Figure 3.9: Vulcan x 94 heater used in incubator
The heater has an AC single phase power supply of 220-240V at 50Hz and a power
consumption of 377W. It is connected to the power supply through two electrodes. For the
purpose of the present work, the heater has been removed from the incubator cage and will
be installed in our prototype.
As can be seen, both actuators need external power supply; their power ratings are
too high for the DAQ Modules. Therefore, the design and implementation of electronic
drivers for the motor as well as the heater will be needed. As commented-upon earlier, the
purpose of the present work is to design a prototype closed-loop control system by applying
software in the loop methodology. This means that the signals generated by the software
must control the actuators. To conclude: the power electronics drivers for the motor and
the heater have to be controlled by the signals generated through the DAQ Module NI 9263
(depicted in Figure 3.3). Also, an exact copy of the incubator power electronics cage is
needed.
3.2. Incubator setup
A modified plant of a neonatal incubator C500 Isolette from Drger Medical is used in this
study. In order to create the closed loop control system, information is gathered from the
incubator by different means. As has been mentioned earlier, real time temperature data are
collected inside the incubator from five points, according to the IEC-60601 standards, using
RTD sensors. Also, the current sensors provide information on the power consumption by
the actuators. Finally, the two hinged potentiometers in the incubator clappers are used
for triggering the event of closing and opening, and enables an estimation to be made of
the amount of heat loss in the incubator. All these signals are inputs to the software-based
controller, interfaced through the DAQ input Modules. Subsequently, and taking into
account the temperature set-point, the controller generates the appropriate control signals
to the motor and heater power electronics, which are interfaced through the DAQ output
Module, in order the achieve the desired temperature inside the incubator.
Figure 3.10: Incubator closed loop setup
As depicted in Figure 3.10, the current sensors, the potentiometers and one RTD sensor, are
interfaced through the DAQ Module 9205 which supports analog voltage input data. The
remaining four RTD sensors are interfaced through the DAQ Module 9217, which supports
temperature measurement data. The input signals to the controller are processed in the
software-based controller designed with LabVIEW. Depending on the temperature set-point,
the controller must generate the corresponding control signals interfaced through the DAQ
Module 9263 which can generate analog output voltage. The control signals are used to
drive the motor that propels the fan and the heater.
35
3.3. Simulation of the incubator dynamics using Simulink
A simulation of the incubator thermodynamics is presented in this section. This simulation

has been performed with the SimscapeT M toolbox in Simulink. Simulink is an environment
for multi-domain simulation and model-based design for dynamic and embedded systems
integrated in Matlab. It provides an interactive graphical environment and a customizable
set of block libraries that enable the design, simulation, implementation and testing of a
variety of time-varying systems. SimscapeT M extends Simulink with tools for modeling
systems spanning mechanical, electrical, thermal and other physical domains such as physical
networks. It provides fundamental building blocks from these domains to enable models of
custom components to be created.
For the purpose of the present study, a prior simulation of the heat transfer dynamics inside
the incubator can provide useful information for the implementation of the controller. The
Simscape toolbox provides thermal building blocks for modeling and simulating thermal
effects in the modeled system. Conductive, convective, radiation heat transfers, together
with the thermal mass of elements, are represented by fundamental blocks.
As depicted in Figure 3.11, SimscapeT M toolkit provides fundamental blocks that enable
modeling and simulation of any kind of physical domain. The variety of thermal domain
fundamental blocks facilitates the modeling task. A brief explanation of the fundamental
thermal blocks is detailed, below.
Figure 3.11: Simscape libraries regarding thermal domain
The thermal elements library provides blocks that represent conductive, convective and
radiation heat transfer. Connections A and B are thermal conserving ports associated with
material layers. The block positive direction is from port A to port B i.e. the heat flow is
positive if it is from A to B. The thermal mass block represents the ability of a material
to store internal energy. The heat flow is positive if it flows into the block. The thermal
reference block, represents a zero or constant temperature point to which temperatures in
the system are related.
Thermal sensors library provides ideal heat flow and temperature sensors. Connections A
and B are conserving thermal ports. The sensor is oriented in both blocks from A to B. The
remaining port in both sensors outputs the heat flow value or the temperature value sensed.
Thermal sources library provides ideal heat flow and temperature sources. The ideal heat
flow source maintains a specified heat flow at its outlet, regardless of the temperature
difference across the source. The heat flow through the source is directly proportional to
the control signal. The ideal temperature source maintains a specified temperature at its
outlet, regardless of the heat flow consumed by the system. The temperature differential
across the source is directly proportional to the control signal.
3.3.1. Simulation objectives
The main purpose of this simulation is to obtain data regarding the thermodynamic behavior
inside the incubator. The model must represent the heat transfer dynamics inside the
incubator, the power electronics for the fan and the heater, as well as the air flow propelled
by the fan. Therefore, the model needs to be segregated into the three parts, as stated
above.
Even though simulations always represent ideal situations and conditions, the parameters of
the model components need to be adjusted to the maximum of possible real values in order
to obtain simulation of reality occurrence.
The model must enable the monitoring of the heater and motor status in order to obtain
useful information for future design and implementation of the power electronics that must
drive both actuators. Also, simulations for a wide range of temperature set-points for the
incubator can provide interesting information regarding the incubator dynamics in different
situations.
Some considerations have been taken into account in order to simplify the model to be
implemented. The geometry of the incubator will be simplified in order to calculate the air
thermal mass; the only assumption is the air in the neonate cage. Finally, and the most
importantly, a neonate model will not be included. The vast complexity and non-linearity
of thermoregulation functions of a neonate, and the difficulty in representing the neonate
thermal interaction with the surrounding environment, implies a work-load that is beyond
the scope of the present study.
To implement this model several physical properties must be known, as well as the incubator
dimensions. Air density, specific heat, thermal conductivity and heat transfer coefficient
are functions of temperature and, therefore, to simplify the model these properties will be
applied for a constant temperature of 26◦ C. Also Plexiglass (which is the material of the
incubators cage) density, specific heat, thermal conductivity, heat transfer coefficient and
radiation heat transfer coefficient must be known. Motor and heater characteristics will be
applied, as described in the parameters specifications.
Table 3.4 and Table 3.5 summarize the air and Plexiglass properties applied in the model.
37
Air properties (at 26◦ C)

Density (kg/m3 ) 1.17
Specific heat (J/kgK) 1005
Thermal conductivity (W/mK) 7.29*10−4
Heat transfer coefficient (W/m2 K) 35.42
Radiation heat transfer coefficient (W/(m2 ∗ K 4 )) 2*10−7
Table 3.4: Air properties at 26◦ C
Plexiglass properties
Density (kg/m3 ) 1190
Specific heat (J/kgK) 5.4744
Thermal conductivity (W/mK) 0.2
Heat transfer coefficient (W/m2 K) 9
Radiation heat transfer coefficient (W/(m2 ∗ K 4 )) 4*10−7
Table 3.5: Plexiglass properties
To implement a simplified geometry for the model , only the main Plexiglass cage of the
incubator has been taken into account. The dimensions are depicted in Figure 3.12 (all
measurements are in centimeters).
Figure 3.12: Incubator dimensions used in Simulink model

3.3.2. Thermodynamics of the incubator model
Figure 3.13 depicts the implemented model of the incubator with thermodynamics designed
with Simulink.
For clarification purposes, it is possible to divide this model in two different components
parts. These are clearly differentiated by the colors used in the blocks.
The blue and grey blocks represent the power electronics related to the motor that propels
the fan. A thermistor, which is constantly sensing the temperature variations in the incubator
air, has a resistance that is a function of temperature. These changes in the thermistor
resistance affect in the voltage drop and, subsequently, in the duty cycle of the Pulse Width
Modulation (PWM) signal that is used to drive the model of the DC motor. The parameters
used in this DC motor are as presented in the section on parameter specification for the
motor (see 3.1.7). The grey blocks sense the speed at which the motor is running. This
speed is subsequently used to quantify the amount of air that the fan propels.
The red block represents the heat transfer dynamics inside the incubator. The inputs in this
block are the fan rate (ω) and the temperature set-point. As outputs, the block delivers
a thermistor sense port, the incubator temperature, the power consumed by the heater,
the Plexiglass temperature and the temperature difference between the set point and the
incubator air temperature.
Figure 3.13: Incubator model implemented in Simulink

39
Figure 3.14 depicts the heat transfer dynamics block.
Figure 3.14: Heat transfer dynamics block used in Incubator model
In this block the heater power is a function of the difference between the temperature set
point and the air temperature inside the incubator. A look-up table has been used for this
purpose. When the temperature difference is 11◦ C or greater the heater produces 377W of
power. If the temperature difference is between 10◦ C and 3◦ C the heat power produced by
the heater ranges between 370W to 300W in steps of 10W. Finally when the temperature
difference is between 3◦ C and 0◦ C the heat power ranges from 300W until 0W in steps of
100W. Therefore, for 0◦ C temperature difference the heater produces 0W of heat power
which, in reality, is the equivalent of switching the power source to the heater. The look-up
table uses a linear interpolation to calculate any other temperature differences that are not
stated above. The initial temperature of the air is 26◦ C. This temperature is also used to
calculate the amount of heat power produced by the fan. The heat power produced by the
fan is affected by convective heat transfer with the air in the incubator.The heat transfer
process inside the incubator can be modeled by the following. The air heats up through
convective and conductive heat transfer and looses heat through radiative, convective and
conductive heat transfer with the Plexiglas hood of the incubator. To simplify, the variation
of heat transfer heats up the air thermal mass of the incubator. The equation represented
next describes mathematically this process.
dT emp
Qconv + Qcond − Qprad − Qpconv − Qpcond = c ∗ m ∗ (3.3)
dt
Where c represents the specific heat of the air and m the air mass.
The yellow box represents the heat transfer mechanisms inside the incubator and is depicted
in Figure 3.15.
Figure 3.15: Conductive, convective and radiative heat loss inside the incubator model
The heat power provided by the heater is affected by convective and conductive heat transfer
between the heater and the air. Radiative, convective and conductive heat loss is introduced
later to represent the interaction between the air and the incubator cage, made of Plexiglass.
Thermal mass blocks, representing the air inside the incubator and the Plexiglass incubator
cage, are used to measure the temperature.
3.3.3. Results and conclusions of the simulation
Simscape toolbox allows modeling and simulating of a wide variety of physical domains, and
interactions between them. It has not been possible to model the interaction between the
air flow proceeding from the fan and the heat produced by the heater. Simscape does not
have available the necessary tools to represent the temperature gradient inside the incubator
due to the air flow produced by the fan. Hence, the heat transfer mechanisms between the
heater and the air flow proceeding from the fan are not represented in the model. Moreover,
Simscape only facilitates the physical type of signal in feedback nodes. The switching device
to control the heater cannot be implemented with the physical type of blocks. Thus, the
look-up table has been used for this purpose, as has been commented-upon earlier.
As highlighted earlier, the objective of this simulation is to represent the heat transfer dynam-
ics inside the incubator, while monitoring the motor and heater status. For this purpose, the
physical signals derived from the Simscape model must be converted to Simulink signals in
order for them to be analyzed. The simulation is carried-out over a period of 2500 seconds.
Two different temperature set-points have been used in order to observe any differences in
the behavior of the model. Air temperature inside the incubator, power consumed by the
heater, motor revolutions, and current applied to the motor are the variables represented in
the following figures. The results will be analyzed and discussed later.
41
Figure 3.16: Simulation results for a temperature set point of 33C
Figure 3.17: Simulation results for a temperature set point of 35C

As depicted in Figure 3.16 and Figure 3.17, in the air temperature plot, the temperature
of the air inside the incubator does not reach the desired temperature set-point. In both
cases the steady-state temperature reached is below the desired temperature set-point. For
the 33◦ set point, the steady-state temperature reached is 31.83◦ and for the 35◦ set point
the steady-state temperature reached is 33.15◦ . This is due to the power transferred from
the heater to the incubator. As highlighted above; it is not possible to simulate a switch to
control the heater power. Therefore, a look-up table has been used for this purpose. This
table is constructed as a function of the temperature difference between the temperature
set point and the incubator air temperature. Any value that is not described in the look-up
table is linearly extrapolated. Thus, the heater is driven with not enough power to reach
the desired temperature set-point and stabilizes at the temperature mentioned above. This
effect can also be seen in the same Figures where the heater power is plotted. For the 33◦
set-point the steady-state heat power response reached is 116.8W and for the 35◦ set-point
the steady-state heat power response reached is 185.1W.
With respect to the temporal analysis, the results are within the order of magnitude expected.
The steady-state response is reached at about 25 minutes. The duration of the transient
response can vary depending on the dynamic margins for the signal. Due to the lack of tools
of Simscape, the influence of the air propelled by the fan in the incubator thermodynamics
is not properly modeled and this has an effect on the transient response of the system. One
hypothesis is that if the influence of the air propelled by the fan is correctly modeled, the
duration of the transient response should decrease. The air flow produced by the fan can
regulate the air temperature inside the incubator more rapidly. The inappropriate modeling
of the fan interaction with the thermodynamics inside the incubator can be observed in the
plots representing the motor revolutions and the current consumed by the motor. The values
obtained are far from real and reinforce the need to improve this model to include the air
flow produced by the fan in the heat transfer dynamics inside the incubator. This aspect
needs to be taken into account in future works.
43
3.4. Sensor interfacing
In this section, the inputs of the controller and how they are connected to the DAQ Modules
will be explained in detail. Also the signal acquisition software designed in LabVIEW will
be presented. As has been already established, the inputs to the software-based controller
are five RTD sensors which continuously gather the temperature inside the incubator; two
potentiometers hinged in the incubator clappers to trigger the event of an open door, and
two currents sensor to monitor the state of the actuators. Figure 3.18 depicts these inputs
in the incubator closed loop setup.
Figure 3.18: Inputs of the software based controller highlighted
There are two NI DAQ Modules that collect the data coming from the incubator plant.
Therefore, it is possible to make an early distinction between the inputs depending on the
NI DAQ Module used for their interface. DAQ 9205 supports analog input voltage measures
up to ±10V and DAQ 9217 supports temperature measurement. As described earlier, DAQ
9217 supports only four RTD sensors. As such, there is a need to design and implement an
RTD interface that produces a voltage proportional to the temperature sensed. This RTD
interface is sensed through DAQ 9205 as an analog input voltage.
In Figure 3.19, the sensor placement in the incubator is depicted from the top and the
side respectively. RTD sensor placement follows the IEC-60601 standard for temperature
measurement in medical devices. RTD1, RTD2, RTD3 and RTD4 are interfaced through
the NI DAQ 9217 Module. Therefore RTD5 must be connected to NI DAQ 9205 Module as
an analog voltage through the RTD interface design: This is explained later in this section.
The potentiometers and the current sensors are also connected to the NI DAQ 9205 Module
as analog voltage inputs.
Figure 3.19: Incubator sensor placement: top and side view

45
3.4.1. Temperature sensors
RTD
An RTD is a temperature sensing device with resistance that increases with tempera-
ture. Signal conditioning is generally required to measure temperature using an RTD.
Because an RTD is a passive resistive device, a current needs to be passed through the
device to produce a measurable voltage. Providing current to take a resistance measurement
is a form of signal conditioning, termed current excitation. In addition to producing current
excitation for the RTD, signal conditioning amplifies the output voltage signal, and filters
the signal to remove unwanted noise. All this signal conditioning and processing is performed
by the LabVIEW software.
As highlighted earlier, RTD1, RTD2, RTD3, and RTD4 will be interfaced through the
DAQ 9217 Module, which measures temperature. The distribution in the incubator of
these sensors is depicted in Figure 3.19. In the present study, a four wire configuration is
being used to make the connections, enabling the excitation current nodes to be separated
from the measuring nodes and, therefore, minimizing the error contributed by the signal
conditioning elements.
Connection diagrams are depicted in Figure 3.20:
Figure 3.20: Terminal assignment of the RTD interfaced by DAQ 9217 Module
RTD interface
IEC-60601 standard for temperature measurement in medical devices states that tem-
perature inside an incubator must be measured from five different points. Therefore,
since the DAQ 9217 Module only supports four RTD measurements using the four wire
configuration mode, this implies that one RTD (RTD5 in the present case), must be
interfaced through DAQ 9205 Module which supports analog voltage input values. Hence,
there is a requirement to create an RTD interface to obtain a measurable voltage with DAQ
9205 Module.
As explained earlier, to design an accurate interface to measure the voltage across the RTD
using a four wire configuration it is necessary to produce a constant current excitation
between the nodes 1 and 4 of the RTD and amplify, without creating disturbance in the
signal, the measured voltage across nodes 2 and 3 of the RTD. The design is depicted in
Figure 3.21.
Figure 3.21: RTD interface circuit design
REF5025, from Texas Instruments, is a precision voltage reference that is supplied with +5V
and provides 2.5V with very low drift and low noise[Ti04]. This 2.5V is used throughout the
circuit as a stable reference. LMP2011, from National Semiconductor, is a high precision
operational amplifier. Its V + input is supplied with 5V, and its V − input is grounded[TI].
This integrated circuit is used to provide constant current excitation to the RTD. INA118,
from Burr-Brown, is a low power instrumentation amplifier. The V + input is supplied with
5V and V − input is grounded[TId]. This integrated circuit is used to amplify and condition
the measured signal in the RTD nodes 2 and 3. The RTD has a resistance of 100Ω at 0◦ C,
and it is situated between the RTD 4 nodes which are the resistance between nodes 2 and
3. An RTD transfer function of resistance versus temperature, for temperatures greater
than 0◦ C, is approximated by Equation 3.3:
47
RRT D = 100 + 0.39083 · T − 0.00005775 · T 2 (3.4)
RTD is a passive resistive device; current must flow through the device to produce a mea-
surable voltage. Providing current to take a resistive measurement is a form of signal condi-
tioning, termed current excitation. This is done with the LMP2011, REF5025 and R8 .
The constant-current value is expressed in Equation 3.4:
VREF 2.5
IRT D = = = 9.259 · 10−4 A = 0.93mA (3.5)
R8 2700
In addition to producing current excitation for the RTD, signal conditioning amplifies the
output voltage signal, and filters the signal to remove unwanted noise. This is done in
the instrumentation amplifier INA118. To calculate the gain (G) of the instrumentation
amplifier, it is first necessary to calculate the sensory voltage margins. In the current case,
the temperature range of interest is 10◦ C to 40◦ C, which translates into a resistance of
103.903Ω to 115.541Ω. With the excitation current of the RTD, voltage on the sensory
leads will range from 0.0966V to 0.107452944V. The voltage reference of 2.5V is also
used as a reference for the instrumentation amplifier. Therefore, the required gain of the
instrumentation amplifier is:
VREF
G=1+ = 232.66 (3.6)
0.107452944
The gain of the instrumentation amplifier INA118 is given by Equation 3.6:
50kΩ
G=1+ = 232.66 (3.7)
RG
Therefore RG is shown in Equation 3.7:
50kΩ
RG = = 215.83Ω (3.8)
231.66
The layout has been designed using EAGLE version 5.6.0 for Windows a CadSoft tool. It is
depicted in Appendix D.
The analog voltage signal is connected to the DAQ 9205 Module using differential mea-
surement configuration at channel 0. Therefore, VOU T is connected to AI0, and GND is
connected to AI18. This connection is represented in Figure 3.22.
Figure 3.22: Connection to DAQ NI 9205 Module using differential measurement

configuration
LabVIEW function RTD-Measure conversion enables the conversion of the voltage measure-
ment from the RTD interface into a temperature; the excitation current must be given to
this block; in the present case this is 0.93mA.
This RTD interface design enables future studies to implement the closed loop control system
with only two DAQ Modules. All of the RTDs could, in the future, be interfaced through
DAQ 9205 Module as analog voltage inputs.
3.4.2. Disturbance sensors
Disturbance is sensed through a potentiometer transducer that gives a voltage proportional

to the opening angle of the incubator clappers. Due to the mechanism of the incubator
clappers, they are either open or closed. The analog voltage signals from the potentiometers
are connected to DAQ 9205 Module using differential measurement configuration. The
right door potentiometer is connected to channel 1 i.e. to terminals AI1 and AI9, and the
left door potentiometer is connected to channel 2 i.e. terminals AI2 and AI10. The door
nomenclature is depicted in Figure 3.19 and in Figure 3.23.
Depending on the position of the sliding contact the potentiometer creates an adjustable
voltage divider. Table 3.6 shows the voltage values measured for open and closed door events.
Right door Left door

OPEN 1V 1V
CLOSED 4V 4V
Table 3.6: Voltages representing the door status (open or closed)

49
Figure 3.23: Right door and left door terminal assignments
Theoretically, the voltage should range between 5V and 0V but, in both cases, the sliding
contact of the potentiometer does not reach the complete circular resistance since this is
caused by the particular movement of the incubator doors. The voltages represented in Table
3.6 will be used to define the two possible states of the incubator doors: open or closed.
3.4.3. Current sensors
The standard for precise current measurement in instrumentation and other high reliability
equipment application are current sensor transformer. They are accurate, easy to implement
and reliable under harsh environmental and thermal conditions. They are used, for the
purpose of this work, to monitor the power consumption from the actuators.
As has been described earlier in section 3.1.5, a burden resistor is used to create a voltage
drop that is a function of the current flowing through the current sensor. Combinations of
Equation 3.1 and Equation 3.2 determine the current flow.
VO · Nprim
IO = (3.9)
RBU RDEN
The sensing limit imposed by DAQ 9205 Module is ±10V. Therefore the voltage drop (VO )
in the burden resistor must not exceed these limits. Knowing the ranges of sensor current
enables calculating the RBU RDEN maximum for both actuators to be calculated. From
3.1.7. The maximum current consumption by the motor is 2.7A. The maximum current
consumption by the heater is 3.2A, known from 3.1.8. To simplify, and to introduce security
margins for possible anomalies in the signal, a current of 5A (IO ) can be used to determine
the RBU RDEN limits.
IO · RBU RDEN
VO = < 10V (3.10)
Nprim
10V · Nprim
RBU RDEN < = 5000Ω (3.11)
IO
Using RBU RDEN as equal or lower to 5kΩ ensures measurement inside the limits imposed
by DAQ 9205 Module. The value chosen for RBU RDEN in both cases is 2.7kΩ.
Figure 3.24 depicts the terminal assignment and connection configuration of the current
sensors.
Figure 3.24: Terminal assignment and connection configuration for the current sensors
As shown in Figure 3.24, both signals are connected to DAQ 9205 Module using differential
measurement configuration. The signal coming from the motor current sensor is connected
to channel 3 i.e. to terminals AI3 and AI11, and the signal coming from the heater current
sensor is connected to channel 4 i.e. to terminals AI4 and AI12.
3.4.4. Signal acquisition with LabVIEW
Once all the sensors are connected correctly to the DAQ Modules, it is time to create
the software to acquire the data from the sensors. LabVIEW DAQ toolkit enables the
configuration and control of the channels from the DAQ Modules. The purpose of the
software in designing this section should be to collect the data from all the different
sensor inputs used in the system. Therefore, data should be collected from the four RTD
channels in DAQ 9217 Module and from the five analog input channels in DAQ 9205 Module.
Table 3.7 summarizes the channel distribution of the input signals to our system.
51
Signal DAQ/Channel (terminal assignment) Measurement configuration

RTD1 NI 9217/Ch1 (1,2,3,4) Four wires
RTD5 NI 9205/Ch1 (0,8) Differential
Disturbance right door NI 9205/Ch2 (1,9) Differential
Disturbance left door NI 9205/Ch3 (2,10) Differential
Current sensor motor NI 9205/Ch4 (3,11) Differential
Current sensor heater NI 9205/Ch5 (4,12) Differential
Table 3.7: Signal to channel distribution in system inputs
Figure 3.25 depict the Virtual Instrument (VI) created to collect all the data derived from
the input sensors.
Figure 3.25: System inputs VI
The block diagram and front panel of this VI is depicted in Appendix C. The only input of
this VI is the Error signal. The remaining signals are all output of this VI, and will be used
to design the controller.
3.5. Controller hardware design and analysis
In this section the design of the power electronics to drive the motor and the heater is will
be described. Also the software, designed with LabVIEW, to generate the control signals
will be presented. Figure 3.26 highlights the part of the controller designed and described
in this section.
Figure 3.26: Outputs and power electronics of the closed loop highlighted
Only one DAQ NI Module is available to generate the control signals. As indicated earlier,
DAQ NI 9263 Module can generate up to ±10V of analog output in four different channels
with 1mA of current drive per channel. This must be taken into account for the design of
the power electronics. Only 4 signals can drive the motor and the heater, and these control
signals cannot exceed ±10V and 1mA of current drive. Of necessity, the actuators drivers
will be designed independently.
53
3.5.1. Heater driver
As indicated in 3.1.8, the heater uses an AC single phase power supply of 230V at 50Hz,
and has a power consumption of 377W. Thus the design must enable a high power load to
be driven with a low power control signal. A a Solid State Relay (SSR) is perfectly suitable
device for this purpose. SSR is a switching device that requires a low energy control circuit
to switch the output from OFF to ON and vice versa. This output can drive high power
loads; the heater in the case of the present study, and which is electrically isolated from the
controlling circuit.
For an SSR to operate, it is necessary to drive a suitable current through its coil. When the
coil draws the right current the output switch turns ON. It is a simple mechanism.
The FINDER 65.31.9.012.0300 SSR has been used in designing the heater driver[ele01].
This SSR allows driving AC loads with DC control signals. It is not possible to activate the
coil with the ratings provided by DAQ NI 9263 Module and,hence, a coil driver circuit needs
to be designed. Figure 3.27 depicts the circuit designed for this purpose.
Figure 3.27: Heater driver circuit design
When the coil is activated the output of the SSR is turned ON and the AC power is delivered
to the heater. The coil driver circuit is quite simple and consists of a NPN transistor, BC337
from Fairchild Semiconductor[ele98] which is used to control the coil from the SSR which,
in turn, is connected to a 12V power supply. Resistor R1 is used to set the base current
for the transistor so that the transistor is driven into saturation when the relay coil is to be
energized. In this way, the transistor has minimal voltage drop and, hence, dissipates very
little power while delivering most of the 12V to the relay coil. Assuring that the BC337
transistor remains saturated for the collector current level needed to activate the coil (this
is 109mA) the value of R1 can be obtained. To ensure saturation of the transistor, the base
current must be greater than the collector current divided by the transistors minimum DC
current gain hF E : this is 250. Therefore:
109mA
IB ≥ = 0.436mA (3.12)
250
Base current must be at least 0.436mA. In practice, and to introduce safety margins, a base
current of about 0.8mA is enough to ensure saturation. If VIN is supplied with 9V, the
value for R1 is 11.25kΩ. Due to availability, the value of R1 finally chosen is 11.5kΩ. The
diode is used for safety purposes.
A simulation using PSpice 9.1 has been developed to strengthen the theoretical analysis of
the design. The DC part and AC parts have been simulated independently. Figure 3.28
depicts the scheme for the coil driver (DC part) and AC part simulation. The coil is modeled
as an 110Ω resistor and the control signal is modeled using a pulse that every two second
switches from 9V to 0V. Regarding the AC part of the heater driver simulation, the heater
has been modeled as a resistance of 70.159Ω. This resistance is calculated in Equation
3.12. From 3.1.8, it is known that the power consumption by the heater is 377W and it is
supplied with AC power, which is 230V at 50Hz. Therefore:
Vac 2 1
Rheater = ( √ ) · = 70.159Ω (3.13)
2 P
The switch models the activation of the relay coil which is closed after the first second. V3
models the AC power supply.
Figure 3.28: PSpice scheme for the heater driver circuit
Theoretically, when VCON T ROL is 9V, the voltage drop in RCOIL should be 12V and the
current through R1 must not exceed 1mA.
Simulation results are depicted in Figure 3.29. The circuit behaves as expected and the
current through R1 is 0.716mA. Moreover, the circuit is able to activate the relay coil with
a control signal that the DAQ NI 9263 Module can generate.
55
Figure 3.29: Simulation results for the heater driver circuit
With respect to the AC part simulation results, the maximum current that flows through the
heater is 3.2A. The SSR from FINDER used in the design supports up to 30A. Therefore,
the design is suitable.
Once the design has been simulated with satisfactory results it is time to create the layout
with EAGLE. For more information on the layout see Appendix D.
To summarize the behavior of this circuit, when the control signal Vin is 9V, the relay coil is
activated, the switch is turned ON and the AC power is conducted through the heater. This
causes heat generation and the control signal changes to 0V, the relay coil is deactivated,
the switch turns OFF, and AC power is not conducted to the heater anymore. This control
signal must be generated with LabVIEW and interfaced to the circuit through DAQ NI 9263
Module. The software generating this signal is presented later in this section. The connection
diagram between the circuit and DAQ NI 9263 Module is depicted in Figure 3.30. The control
signal is generated by channel 0 of the DAQ NI 9263 Module, which are terminals AO0 and
COM.
Figure 3.30: Terminal assignment for the heater control signal

3.5.2. Fan driver
As shown in 3.1.7, the DC motor used to propel the fan is from ebmpapst, specifically
VD-1-54.14 with clockwise rotation. This motor requires 24V and 1.75A of nominal current
to operate. Further, the motor can operate with voltages between 12 and 28 Volts and has
a maximum starting current of 2.7A. Therefore, it is clear that the motor must be supplied
externally and the control signals need to control the power delivered to the motor.
An integrated circuit, L298 from ST Microelectronics, is perfectly suited to this purpose. It
is a dual full bridge driver and is capable of driving DC currents up to 4A and operating with
supply voltage up to 46V. An additional supply input is provided so that the logic works at
a lower voltage, and accepts TTL logic levels[Mic00].
Each bridge is able to drive DC currents of 2A. A parallel configuration has been cho-
sen due to the starting peak current of the motor of 2.7A. Both bridges are connected
in parallel enabling the bridge to drive current up to 4A. Figure 3.31 depicts the circuit design.
Figure 3.31: Fan driver circuit design
The DC motor is connected between OUT1 and OUT2. VSS is the logic supply voltage and
VS is the motor supply voltage. Three control inputs are needed to drive the motor at the
desired speed . These inputs accept TTL logic levels, V H ranges between 2.3V and VSS
and VL ranges between -0.3V and 1.5V. The maximum current drive of the control inputs
is 100µA. These can be generated with the DAQ NI 9263 Module. Control input enables or
disables the bridge with a High state or Low state respectively, independently of the other
control inputs. IN1 and IN2 control inputs are used to control the motor speed. Table 3.8
shows the motor status with respect to the logic states of IN1 and IN2.
The DC motor used in this project only has clockwise rotation. Hence IN2 is permanently
57
Input status IN1=High IN1=Low

IN2=High Brake Counterclockwise rotation
IN2=Low Clockwise rotation Brake
Table 3.8: Motor status with respect to the logic states of inputs IN1 and IN2
driven to the Low state. IN1 will be used as a PWM signal. This allows controlling the
speed of the motor by varying the duty cycle of this signal.
EAGLE design and layout is depicted in Appendix D.
Figure 3.32 depicts the connections and terminal assignment from the DAQ NI 9263 Module
to the circuit. IN1, which is the PWM signal, is generated by channel 1 (terminals AO1 and
COM). IN2 is generated by channel 2, which is terminal AO2 and COM. Finally Enable is
generated by channel 3, which is terminal AO3 and COM.
Figure 3.32: Terminal assignment for the fan control signals
3.5.3. Control signals generation
When the drivers are implemented, properly installed in the power electronics cage, and the
control signals are correctly connected to the DAQ NI 9263 Module, the LabVIEW software
to generate the control signals needs to be designed and installed. The designed software
must generate four control signals. Relay control signal is generated by channel0, and has
to allow the control of the activation time of the relay coil. Motor control signals, IN1, IN2,
and Enable, are generated by channels1, 2, and 3 respectively. As stated earlier, IN1 must
facilitate controlling the motor speed by varying the duty cycle of the PWM signal.
Table 3.9 summarizes the channel distribution of the input control signals generated to drive
the power electronics.
Figure 3.33 depicts the virtual instrument (VI) created to generate the control signals to
drive the actuators.
Signal DAQ/Channel (terminal assignment) Units

Relay coil control NI 9263/Ch0 (AO0,COM) Volts
IN1 NI 9263/Ch1 (AO1,COM) Volts
IN2 NI 9263/Ch2 (AO2,COM) Volts
Enable NI 9263/Ch3 (AO3,COM) Volts
Table 3.9: Signal to channel distribution in control outputs
Figure 3.33: Control signals generation VI
The block diagram and front panel of this VI is depicted in Appendix C.

Relay duty cycle is an input to this VI and facilitate controlling the amount of time the
coil is activated and, hence, the heater is supplied with AC power. This is via a PWM
signal with a frequency of 0.2Hz, which is a period of 5 seconds and, therefore, a duty
cycle of 50% switches the relay ON and OFF every 2.5 seconds. Motor ON/OFF and
Motor duty cycle (%) are inputs to control the speed of the motor. Varying the motor
duty cycle controls the average amount of power delivered to the motor. This signal runs
at a frequency of 35kHz. Motor ON/OFF signal activates or deactivates the motor driver.
The remaining signals are output of this VI, and are used to monitor the control signal status.
59
3.6. Controller software-based design and analysis
At this point of the present work, the only task left is to design the software-based controller.
This controller must gather the data coming from the input sensors, process these data,
and finally generate the adequate control signals to drive the power electronics. Figure 3.34
highlights the software-based controller in the incubator closed loop.
Figure 3.34: Software-based controller of the closed loop highlighted
Any system in which the output quantity is monitored and compared with the input, any
difference being used to actuate the system until the output equals the input is called a
closed-loop or feedback control system. For the purpose of this work, PID control is going
to be used. This control is explained in 2.6.1. PID control is the most common control
algorithm used in industry and has been universally accepted in industrial control. The
popularity of PID controllers can be attributed partly to their robust performance in a wide
range of operating conditions and partly to their functional simplicity.
The basic idea behind a PID controller is to read a sensor, then compute the desired
actuator output by calculating proportional, integral, and derivative responses and summing
those three components to compute the output. In the case of this study, a Multiple Input
- Multiple Output (MIMO) control system must be designed. The process variable is the
system parameter that needs to be controlled, temperature in this case. RTD sensors are
used to measure the process variable and provide feedback to the control system. The
set-point is the desired temperature value for the process variable. At any given moment,
the difference between the process variable and the set-point is used by the control system
algorithm, which is PID, to determine the desired actuator output to drive the system
(plant). In this case, the actuator output is not the only signal that has an effect on
the system. Opening the incubator doors creates a source of cool air that disturbs the
temperature inside the incubator. This term is referred as disturbance. The design of the
control system must minimize the effect of disturbances on the process variable.
LabVIEW PID toolkit provides a wide array of VIs that help in the design of a PID based
controller. Following, the control strategy applied is presented[Ins06].
3.6.1. Decoupled controller
The design of this controller derives from the air circulation inside the incubator. As
depicted in Figure 3.35 the air is propelled by the fan then heated up by the heater, after
that it circulates through the incubator and finally is propelled by the fan again. Therefore,
the heated air is firstly sensed by RTD2 and RTD3, then by RTD5 and finally by RTD4 and
RTD1, the placement can be observed in Figure 3.19.
Figure 3.35: Air flow inside the incubator
In this controller it is assumed that the heater behavior has an impact on RTD2 and RTD3,
and the fan behavior is reflected in RTD5, RTD4 and RTD1. This assumption is done
taking into account the air circulation and the actuators action in it. Due to the proximity
to the air output of RTD2 and RTD3, hence from the heater, these two RTDs are going to
be the process variables to generate the heater control signals. In the other hand RTD5,
RTD4 and RTD1 are going to be the process variables for the motor control, the fact that
61
these RTDs are in the rear end of the air circulation implies that the fan speed has a greater
impact on its measurements.
From the assumptions made above, the heater and the motor will be controlled independently
using the PID technique. Figure 3.36 depicts the decoupled controller block diagram.
Figure 3.36: Block diagram of the decoupled controller
An independent PID controller must be designed for the motor and for the heater. The
motor PID controller has as process variables RTD1, RTD4, and RTD5, disturbance coming
from the right door is going to be included in this controller and also the current sensed
from the motor. The heater PID controller has as process variables RTD2 and RTD3,
disturbance coming from the left door is included in this controller and also the current
sensed from the heater. Both PID controllers generate an output which is going to drive
the actuators, in this case this output controls the duty cycle of the motor, for the motor
PID controller, and the duty cycle of the heater, for the heater PID controller.
LabVIEW PID toolkit provides two fundamental VIs to create a PID controller, the PID
algorithm VI and the PID Auto-tuning VI, this last VI includes an auto-tuning wizard in
addition to the basic PID algorithm which allows to recalculate the PID gains for the system.
Therefore a decoupled controller has been created for both PID VIs.
Figure 3.37 and Figure 3.38 depict the Decoupled PID and Auto-tuning PID controllers
respectively. The only difference between them is two control inputs to activate the
auto-tuning wizard and reinitialize the PID controller. Besides those two inputs, both
controllers have the same configuration and design. HEATER ON/OFF and Motor ON/OFF
are two control inputs added for security purposes, this way motor and heater actions
can be turned off or on independently from the rest of the software. The outputs allow
monitoring the evolution of the system inputs and the state of the actuators as well as the
PID controller output.
The output range from the PID controllers set the maximum and minimum duty cycle
at which both actuators are going to work. Therefore, these ranges must be tested and
discussed in order to find equilibrium between heating up the incubator and cooling down.
Figure 3.37: Decoupled PID controller VIs for the motor and the heater
Figure 3.38: Decoupled Auto-tuning PID controller VIs for the motor and the heater
More detailed information about the controllers block diagram and front panel is available
in Appendix C.
63
4. Results and discussion
In this chapter, the closed loop control system prototype design is described. The behavior
of closed system of a neonatal incubator that was developed and implemented will be
analyzed and discussed. The final implementation of the power electronics is presented, as
well as the protocol used to test the controller that was developed. The results obtained
from the pilot-study will be analyzed and discussed.
The previous chapters presented the main aspects of the design concepts and the prototyping
strategy of plant system developed in this project. With respect to the final prototyping step
which will be presented in the following sections, the controller, actuator driver and power
electronics were driven by the software-in-the-loop (SIL) controlling kernel which consisted
of software-PID controller, sensor interface and actuator operation system.
The implemented pilot-study system with the power-electronics, driver-units and the
actuators (heater and fan) system, is presented in Figure 4.1 where the final overall
implementation of the power electronics-driver is shown in the incubator front panel cage.
Every element is highlighted in the figure below.
Figure 4.1: The final implementation of the experimental controller hardware and associated
power electronics shown in the incubator cage
The PID controller gains used are shown in Table 4.1. These PID gains have been calculated
using the auto-tuning PID gains wizard and will be used in the tests conducted.
Parameter Value
K1 9449.404171
Ti 0.032262
Td 0.006452
Table 4.1: PID gains used in the testing of the proposed system
64 4. Results and discussion
4.1. Optimizing the output ranges of the PID controllers
The first step in analyzing the system is to select appropriate output ranges for the PID
controller of the motor and the heater. The output ranges are going to control the duty
cycle at which the actuators work. The purpose of this experiment is to obtain the optimal
output ranges for the PID controllers. This involves using the output ranges that allow a
faster heating and cooling of the incubator air temperature. Therefore, tests will need to be
conducted with different output ranges for both PID controllers.
The test protocol will be a room temperature of 25◦ C starting from an incubator air
temperature of 31◦ C with steps of 1◦ C being applied to the temperature set-point, until
34◦ C is reached. Every set-point step will be applied when the incubator air temperature
has reached the previous set-point. Once the air temperature has reached 34◦ C, decreasing
steps of 1◦ C will be applied until 31◦ C. As before, every set-point is applied when the
incubator air temperature has reached the previous set-point.
This protocol will be tested for three different output ranges, as shown in Table 4.2:
Running Heater PID output ranges (min-max) Motor PID output ranges (min -max)
Test 1 15-85 0-85
Test 2 25-75 30-65
Test 3 0-50 15-100
Table 4.2: Output ranges tested
The hypothesis is that the wider ranges of the PID output will increase the heating
and cooling speed of the system. The faster cooling will enable obtaining the optimum
minimum output values to be obtained and the faster heating will enable the optimum
maximum output values to be obtained. Results are processed to assess the amount
of time taken by the system to reach the maximum temperature (which is the heating
time) and to return to 31◦ C (which is the cooling time). Figure 4.2 depict the results
obtained for the three tests run. The x-axis represents the time in seconds and the
y axis represents the temperature in ◦ C. This temperature is a result of averaging the
five RTDs. The blue-legend line represents the Tsetpoint curve, while the temperature
response of the incubator is plotted in red. The time spent by the system to heat up
and cool down is calculated. Table 4.3 shows the response time for each set of output ranges.
65
Figure 4.2: Temperature response of the incubator to step set-points

Time heating up (s) Time cooling down (s)

Test 1 703.61 610.57
Test 2 1136.9 1136.9
Test 3 1444 589.88
Table 4.3: Time results for each set of output ranges
Time for heating up the air inside incubator is the time spent by the system between the
start of the test until the maximum temperature is reached, while time cooling down is
calculated from the maximum temperature reached until the end of the test.
In Table 4.2, the output range parameters of the PID controllers used in Test 3 are translated
into a really slow heating-up speed of the incubator air up to the point that the down
step in the set-point is produced before the temperature reaches 34◦ C. The output range
parameters of the PID controller for Test 1 and Test 2 are able to drive the air temperature
inside the incubator to the desired set-point values introduced at the onset of the test.
However, the parameters used in Test 1 produce a much faster response in heating-up the
incubator air than any of the others. Hence, the maximum output ranges for the PID
controllers will be 85 for the heater as well as the motor.
With respect to the cooling-down process, Test 2 parameters cause the system response to
stabilize at about 31.2◦ C and, as such, not reaching the 31◦ C set-point. Further, the time
spent cooling-down the incubator with these parameters is much higher compared with the
parameters used in Test 1 and Test 3. The cooling-down responses from Test 1 and Test
3 are similar. The parameters for the minimum outputs of the PID controllers used in Test
1 are inverted in Test 3. This means that both actuators have a similar impact, in relation
to temperature, on the incubator system. Also, if both actuators are turned off completely,
which translates to 0 in the minimum output of the PID controller, the system cools down
at a faster speed.
To summarize the results of this experiment in relation to formulated hypothesis, the wider
the output range of the PID controllers, the faster the response; both in the heating-up
stage and in the cooling-down stage. The heater PID controller output range must be 0
to 85. This translates on a time scale to a maximum of 4.25 seconds on every 5 seconds
and, as such, when the controller output is 0 the heater is never turned on. The motor PID
controller output range must be 0 to 85. This is equal to a maximum voltage delivered to the
motor of 20.4V and, when the controller output is 0, the voltage delivered to the motor is 0V.
67
4.2. Plant response dynamics
To analyze the behavior of the plant system which, in the present case, is the incubator-
air-circulation and heat transfer dynamics, and once all the controller parameters have been
set, the following testing protocol is conducted.
Starting from an incubator air temperature of 29◦ C, the temperature set-point profile
depicted in Figure 4.3 is tested. Steps of 1◦ C in the temperature set-point from 30◦ C to
37◦ C and then from 37◦ C to 30◦ C again, are applied. In each set-point the process vari-
able, which is the temperature, must reach the steady state, then the next step is introduced.
Figure 4.3: Temperature set-point profile used in the system response dynamics test
From this measurement, several system parameters are extracted and discussed. The reverse
test, this is the down steps for the set-point, is performed to observe symmetry in the
behavior in the system dynamics. The rise time and peak time and percentage overshoot of
the plant response, are calculated for each set-point step.
Figure 4.4 represents the air temperature response from the incubator to the step tempera-
ture set-point profile input. Plotted in blue is the air temperature response. Plotted in red
is the temperature set-point profile input. The x axis represents time in seconds and the y
axis represents temperature in ◦ C. The system reacts as a second order response oscillating
around the temperature set-point. Data processed to obtain the rise and fall time between
every set-point step and the percent overshoot are presented in Table 4.4 and 4.5.
Figure 4.4: System response to temperature set-point profile input
Temperature set-point step (◦ C) Percent maximum overshoot (Mp %)

30 0.76
31 0.67
32 0.51
33 0.55
34 0.37
35 0.25
36 0.14
37 0.023
Table 4.4: Percent overshoot for every temperature set-point

69
Temperature set-point step (◦ C) Rise time/Fall time (s)

29-30 65.16
30-31 47.39
31-32 53.31
32-33 47.39
33-34 47.39
34-35 59.23
35-36 65.16
36-37 77.00
37-36 71.08
36-35 106.62
35-34 142.16
34-33 142.16
33-32 201.39
32-31 248.78
31-30 402.79
Table 4.5: Rise and fall times for every temperature set-point step
Another test needs to be performed to observe the steady state response of the system.
For this purpose a step set-point from 28◦ C to 34◦ C is applied. Data is collected over a
protracted period so as to draw conclusions regarding the steady state response of the
system.
Figure 4.5 depicts the steady state response to a step set-point input. The air temperature
in blue is plotted with the set-point step in red. The x axis represents time in seconds and
the y axis represents temperature in ◦ C. In this case the system reaches a marginal steady
state response which oscillates around the temperature set-point. The marginal steady state
error calculated from the data is 0.25◦ C.
Figure 4.5: Steady state response to step temperature set-point input

4.3. Discussion
The percentage overshoot decreases while temperature set-point increases (Table 4.4). The
reason is that the overshoot response of the controller is equal, no matter what temperature
set-point is reached. Hence, the percentage overshoot decreases if the temperature set-point
increases. The oscillations around the set-point are independent of the set-point reached.
The controller only reacts to the difference between the air temperature of the incubator
and the temperature set-point.
As observed in Table 4.5, the system does not present symmetry with respect to the rise
and fall times. Fall times increase steadily while the temperature is decreased. When
the temperature needs to be decreased, the controller shuts down both actuators, and
the incubator air cools down by loss of heat to the exterior via convection, conduction,
radiation and evaporation. Conversely, the incubator air heats-up through the action of the
actuators, which are driven by the controller and, therefore, is a faster process. On average,
the incubator air heats up 1◦ C every 60 seconds; this time increasing as the temperature
set-point is increased.
Figure 4.5 shows steady state response of the system. The response oscillates around the
temperature set-point due to the nature of the controller. As described earlier, the controller
implemented is a decoupled controller which features an independent control of the motor
that propels the fan and the heater. The process variables that control the motor have a
different behavior than the process variables that control the heater. A temperature gradient
is created inside the incubator while it is being heated-up. Therefore, the actions of the
heater and the motor are not really synchronized to obtain a steady state response which
does not oscillate. The marginal steady state response that results from this controller is due
to the temperature gradient created inside the incubator. When the process variables that
control the motor reach the set-point temperature, the motor is turned off. Meanwhile the
process variables that control the heater have not as yet reached the temperature set-point
and, without the action of the fan, the temperature inside the incubator decreases. Hence,
only when the process variables that control the motor fall below the temperature set-point
the motor is reactivated and is able to increase the temperature inside the incubator. This
creates the marginal steady state oscillation around the temperature set-point; for the
present controller, this marginal steady state error is 0.25◦ C i.e. a reasonable temperature
gradient inside the incubator.
71
5. Conclusions and future work
5.1. Conclusions
The following points are the conclusions from this project:
- A prototype of a closed loop control system for a neonatal incubator using the software
in the loop technique has been designed and implemented successfully. The software
designed together with the power electronics is able to control the temperature inside
the incubator with satisfactory results.
- An RTD interface has been designed and implemented with success. This enables
the implementation, in the future, of the closed loop control system with only two NI
DAQ modules, specifically the NI 9205 Module and the NI 9263 Module. The RTD
interface can substitute for the RTD NI 9217 Module because is able to provide an
output voltage proportional to the temperature sensed.
- A model, using SimScape toolbox in Simulink representing the incubator thermody-
namics, has been implemented in MATLAB platform. The simulations carried out with
this model represent the temperature evolution inside the incubator. The lack of the
appropriate tools in SimScape to represent and model the physical behavior of the fan
within the thermal state of the incubator has limited the results of this model.
- The controller design strategy followed in this project enables a marginal steady state
to be achieved, with an error of 0.25◦ C. A pure steady state is not reached, although
the marginal steady state error obtained is reasonably good. The decoupled controller
is able to drive the incubator temperature to the desired set-point values, which is the
main objective of this project.
- The software in the loop is a flexible and powerful design technique. The hardware has
been minimized and all the signal processing is concentrated in the software controller.
Changes in hardware are far more difficult to perform than changes in the software.
Hence, bringing together the entire signal processing in the software facilitates modi-
fication of any parameters of the system with the minimum effort.
72 5. Conclusions and future work
5.2. Future work
- Improve the behavior of the motor. The ramping produced in the motor is due to
the limitation in the PWM frequency signal generated from the NI 9263 Module. The
maximum frequency that this module is able to generate is 50 kHz. Taking into account
that the H-bridge network used has a fall and rise time of the order of microseconds
(µs), this reflects on the PWM frequency generated and which is not sufficient to
preempt the instantaneous ramping-behavior in the motor.
- The model created in Simulink is not complete. As highlighted earlier, the lack of tools
in SimScape for representing the interaction between the incubator thermal state and
the fan action, does not allow for a satisfactory model of the incubator thermodynamics
to be designed; albeit the model designed is fairly acceptable. Also, a model of a
neonate needs to be included so as to improve the results of the simulations. The
thermal interactions between the neonate and the environment complicate this task.
- The decoupled controller appears to have been a good control strategy, but it is not
the only one possible. Other controllers can provide a better performance. The task
of optimizing the controller is not part of this project. Future projects would need to
focus on either optimizing the decoupled controller or creating new control strategies
to obtain a better steady-state performance.
73
A. List of Abbreviations
NICU: Neonatal Intensive Care Unit
AC : Alternating Current
PID :Proportional Integral Differential
SIL : Software In the Loop
EAGLE : Easily Applicable Graphical Layout Editor
SCBU: Special Care Baby Unit
Cimit : Center for Integration of Medicine and Innovative Technology
VLBW : Very Low Birth Weight
SCS: Servo Controlled Skin
RTD: Resistance Temperature Detector
PI: Proportional plus Integral
PD: Proportional plus Derivative
IEC: International Electrotechnical Comission
DAQ: Data AcQuisition
PC: Personal Computer
SP2: Service Pack 2
NI: National Instruments
DAQmx: multifunction Data AcQuisition
VI: Virtual Instrument
SA2C: RTD adapted to curve surfaces, dimensions (15x50x6.4mm)
SA2F: RTD adapted to flat surfaces, dimensions (35x12x6.4mm)

74 A. List of Abbreviations
TZ77V: current sensor model from Taehwatrans
VD: Variable Drive
AWG: American Wire Gauge
DC: Direct Current
PWM: Pulse Width Modulation
REF5025: low drif precision voltage reference from Texas Instruments
LMP2011: high precision operational amplifier from National Semiconductors
INA118: low power instrumentational amplifier from Burr Brown
SSR: Solid State Relay
FINDER 65.31.9.012.0300: SSR used in the heater driver
NPN: type of bipolar transistor
BC337: NPN transistor from Fairchild Semiconductor used in the heater driver
Pspice: Personal Simulation Program with Integrated Circuit Emphasis
L298: dual full bridge driver from ST microelectronics
TTL: Transistor Transistor Logic
MIMO: Multiple Input Multiple Output

75
B. List of Symbols
m3 : cubic meters
Kg: kilogram
◦
C: degrees centigrades
mm : milimeters
Kcal: Kilo calories
u(t): control action
c(t): controlled output
r1 (t): reference input
r2 (t): disturbance input
K1 : proportional gain constant
Ti : integral action time
Td : derivative action time
K2 : integral term
K3 : derivative term
Kc : critical gain
Pc : period of oscillations with the critical gain
Ω: ohms
S/s: Samples per second
Hz: Hertz
mV: miliVolts
V: Volts
76 B. List of Symbols
kS/s: kiloSamples per second
µV: microVolts
mA: miliAmper
IO : load current to measure
Nprim :Nsec : primary to secondary turns ratio of ther current transformer
RO : Burden resistor
VA: voltamper
rpm: revolutions per minute
mNm: nominal moment
W: watts
kg 2 *10−6 : moment of inertia
Kg/m3 : density
J/kgK: specific heat
W/mK: thermal conductivity
W/m2 K: heat transfer coefficient
W/m2 K 4 : radiation heat transfer coefficient
V + : positive supply input of the LMP2011
V − : negative supply input of the LMP2011
RRT D : resistance from the RTD
IRT D : current through RTD
VREF : reference voltage from REF5025
R8 : resistance used to set the current through RTD
T: temperature in degrees centigrades

77
G: instrumentational amplifier Gain
RG : resistance used to set the gain of the operational amplifier
VOU T : voltage from the output of RTD interface
AI0: terminal 0 in the NI 9205
GND: Ground
AI2 terminal 2 in the NI 9205
R1 : resistance used to set the base current of the transistor used in the heater driver
hF E : gain of the transistor
IB : base current of the transistor
Rheater : equivalent resistance of the heater
Vac : AC voltage
V3 : AC power supply modelled in PSpice
VCON T ROL : control signal for the heater driver modelled in PSpice
RCOIL : resistance model of the coil of the SSR in PSpice

78 B. List of Symbols
VIN : control signal for the heater driver
AO0: terminal 0 in the NI 9263
COM: terminal Common in the NI 9263
OUT1: output 1 from L298
OUT2: output 21 from L298
VSS : logic supply voltage for L298
VS : supply voltage fro L298
V H : TTL high level
VL : TTL low level
Enable: input to L298
IN1: Input to L298
IN2: Input to L298

79
C. LabVIEW software and figures
C.1. System Inputs
Figure C.1: System Inputs VI

80 C. LabVIEW software and figures
C.2. Control Signals Generation
Figure C.2: Control Signals Generation VI
C.3. Decoupled PID Controller
Figure C.3: Decoupled PID Controller VI

81
C.3.1. Heater PID Controller
Figure C.4: Heater PID Controller VI
C.3.2. Motor PID Controller
Figure C.5: Motor PID Controller VI

82 C. LabVIEW software and figures
C.4. Decoupled Auto-tuning PID Controller
Figure C.6: Decoupled Auto-tuning PID Controller VI
C.4.1. Heater Auto-tuning PID Controller
Figure C.7: Heater Auto-tuning PID Controller VI

83
C.4.2. Motor Auto-tuning PID Controller
Figure C.8: Motor Auto-tuning PID Controller VI

84 D. EAGLE software and circuit layouts
D. EAGLE software and circuit layouts
D.1. RTD Interface
Figure D.1: EAGLE schematic and board files for RTD interface
Table D.1 lists the components used in this circuit:
Component Package Quantity

INA118U (Burr-Brown) SO-8 Surface Mount 1
LMP2011 (National Semiconductor) 5-Pin SOT23 1
REF5025 (Texas Instruments) SOIC-8 1
Molex connector 6410-02 2
WAGO connector 233-204 1
Capacitor (100nf) 0805 4
Resistor (2.7kΩ and RG ) 0805 2
Table D.1: Components used in the RTD interface circuit

85
D.2. Heater driver
Table 3D.2 presents the SSR properties.
Model FINDER 65.31.9.012.0300

Rated current 30A
Rated voltage 250V
Coil nominal voltage 12V
Coil resistance 110Ω
Coil consumption 109mA
Table D.2: SSR properties
Figure D.2: EAGLE board file for the heater driver

BC377 (Fairchild Semiconductor) TO-92 1
D1N4004 (Diode) DO-41 1
Fuse 6.3A (FLINK) 5x20mm 1
SSR - 65.31.9.012.0300 (FINDER) Flange mount/Faston 250 1
Resistor 11.5kΩ 1
Table D.3: Components used in the heater driver circuit

86 D. EAGLE software and circuit layouts
D.3. Fan driver
Figure depicts schematic and board layout designed in EAGLE for the Fan driver circuit.
Shottky diodes have been added for security purposes.
Figure D.3: EAGLE board and schematic file for the fan driver

L298 (ST Microelectronics) Multiwatt15- Vertical 1
BYV27-200-Shotky Diode (Philips) SOD-57 2
Capacitor 100nf 2
Resistor 1.5Ω 2
Table D.4: Components used in the fan driver circuit

References
[Abd98] G.; Delanaud S.; Bach V.; Villon P. & Libert J. P. Abdiche, M.; Farges. Humid-
ity control tool for neonatal incubator. Med Biol Eng Comput, 36(2):241–245,
1998.
[AG] General Applications Engineer Ariel G.E. Current-Sense Transformer
Application Design Guidelines. Datatronic Distribution Inc.
[AG75] N. R. & Rolfe P. Aynsley-Green, A.; Roberton. Air temperature recordings in
infant incubators. Arch Dis Child, 50(3):215–219, 1975.
[AKA08] Steffen Leonhardt Abbas K. Abbas. System identification of neonatal incubator
based on adaptive armax technique. In 4th European Conference on Medical
and Biological Engineering, volume 23, pages 231–236. IFBME, DGBMT, IF-
BME, Novemeber 2008.
[AL04] Drger Medical Systems Andrew Lyon, Peter Pschner. Thermo monitoring, drger
medical systems. Drger Medical Systems, August 2004. Thermoregulation and
thermal monitoring booklet for the neonatal intensive care unit.
[Ant09] Annalisa & Fanos Vassilios Antonucci, Roberto; Porcella. The infant incubator
in the neonatal intensive care unit: unresolved issues and future developments.
J Perinat Med, 37(6):587–598, 2009.
[Boy96] P. Boyd, H. & Lenhart. Temperature control: servo versus nonservo–which is
best? Neonatal Netw, 15(2):75–76, 1996.
[Bud07] Pierre Budin. The nursling, medical guideline for the health care system per-
sonnel. Research and clinical report, University of Paris, 1907.
[Che88] S. & Vaucher J. Chessex, P.; Blouet. Environmental temperature control in
very low birth weight infants (less than 1000 grams) cared for in double-walled
incubators. J Pediatr, 113(2):373–380, 1988.
[DADM96] R Ross Russell C A Bass D A Ducker, A J Lyon and N McIntosh. Incubator
temperature control: effects on the very low birthweight infant. Journal of
pediatrics medicine, 1996.
[Dan87] H. J. Dane. Climate control of incubators related to growth and
thermoregulation of newborn infants. TU Delft, Nederland, 1987.
[eG06] ebmpapst GmbH. 24V DC motor VD-1-54.14 datasheet. ebmpapst GmbH,
1st edition, 2006.
[ele98] Fairchild electronics. BC 337 BJN transistor datasheet, 1998.
[ele01] FINDER electronics. FINDER 65.31.9.012.0300 datasheet. FINDER electron-
ics, 2001.
88 Bibliography
[FT97] S.Delanaud B.Bouferrache G.Krim J.P Libert F Telliez, V.Bach. Skin derivative
control of thermal environment in a closed incubator. Medical & Biological
Engineering & Computing 1997, 26:82–97, September, 1997.
[(GE02] Ohmeda Medical Systems (GE). Ohmeda giraffe omnibed booklet, infant re-
diant warmer system., June 2002.
[Har] Biomedical engineering journal Harry, Fienberg. Tarniers incubator. Technical
report, Biomedical engineering journal.
[Hei05] Roland S.Burns. Butterworth Heinemann. Advanced Control Engineering, vol-
ume 1. CRC Press, 1 edition, 2005.
[HG70] E.N. Hey and G.Katz. The optimum Thermal Environment for Naked Babies.
CRC Press. NY, USA, 1970.
[Hul88] D. Hull. Thermal control in very immature infants. Br Med Bull, 44(4):971–
983, 1988.
[Ins06] National Instruments. PID Theory Explained. National Instruments.2006. Na-
tional Instruments, 2nd edition, April 2006.
[Ins07a] National Instruments. NI 9217 National Instruments cDAQ Data
AcquisitionSystem. National Instruments, device specification edition, July
2007.
[Ins07b] National Instruments. NI9205 National Instruments Analioge inputs module to
the cDAQ units. National Instruments, device specification edition, July 2007.
[Ins07c] National Instruments. NI9263 Analog output modules for cDAQ units. National
Instruments, 1st edition, July 2007. Device specification.
[KT] Ph.D. Karen Thomas. Thermoregulation in neonates. RN. Neonatal Network,
13:253–268.
[KYMRM73] Forrest H. Adams M.D. George C. Emmanouilidies M.D. Kimio Yashiro, M.D
and Ph.D. M. Ray Mickey. Preliminary studies on the thermal environment of
low-birth-weight infants. The Journal of Pediatricsm, Los Angeles, California,
37(991):78–92, June, 1973.
[Mar87] K. H. Marks. Incubators. Med Instrum, 21(1):29–32, 1987.
[Mer79] D. F.; Brown G. L. & Weisman L. Merenstein, G. B.; Koziol. Radiant warmers
vs incubators for neonatal care. Am J Dis Child, 133(8):857–858, 1979.
[Mic00] ST Microelectronics. L298 dual full bridge driver datasheet, 2000.
[MIH08] Eng MIIE MIHEEM. Baby incubation, volume 2. HEEM, 1st edition, 2008.
[Mot73] M. G. Motil, K. J. & Blackburn. Temperature regulation in the neonate. a
survey of the pathophysiology of thermal dynamics and of the principles of
environmental control. Clin Pediatr (Phila), 12(11):634–639, 1973.
Bibliography 89
[Sco77] J W Scopes. A new look at thermoregulation in the newborn. Technical

Report Volume 70-62, St Thomas Hospital, London, Section of pediatrics,
March 1977.
[Sim94] N. P. & Kantak A. Simon, B. N.; Reddy. A theoretical model of infant incubator
dynamics. J Biomech Eng, 116(3):263–269, 1994.
[Sys04] Drger Medical Systems. Draeger isolette c2000 infant incubator brochure, April
2004.
[tC05] taehwatrans Company. Voltage regulator TZ77V IC datasheet. taehwatrans
Company, 2005.
[TI] Analoge section Texas Instruments. LMP 2011 datasheet , operational amplifier
IC unit. Texas Instruments, Analoge section, 1st edition.
[Ti04] Analoge section Texas instruments. Voltage reference REF 5025 IC datasheet,
TI. Texas instruments, Analoge section, 1st edition, 2004.
[TId] Amplifier Texas Instruments and driver. Instrumentation amplifier IC INA 118
datasheet. Texas Instruments, Amplifier and driver.
[Tim08] New York Times. Neonatal incubator made of the car parts. News article and
technical hot spot, New York Times, 2008.
[Tro08] Joan Trombetti. Neonatal incubators technical report and illustrative intro-
duction. Technical report, DOTmed Industry Sector Report, June, 2008.
[Wik07] Wikipedia. Neonatal intensive care units (nicu). Electronic, March 2007.
[Zha05] Runjing Zhang, Lina & Zhou. Design of an incubator for premature infant
based on labview. Conf Proc IEEE Eng Med Biol Soc, 4:4079–4082, 2005.

Simscape

Uploaded by

Copyright:

Available Formats

Simscape

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Simscape

Uploaded by

Copyright:

Available Formats

HELMHOLTZ-INSTITUTE FOR BIOMEDICAL ENGINEERING

CHAIR FOR MEDICAL INFORMATION TECHNOLOGY

Prototyping a closed loop control

Telefon: +49 (0)241 80 -23211

Place, Date Signature

2. Theoretical bases and background 3

3. Control System Design and analysis 25

3.5.1. Heater driver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

4. Results and discussion 63

5. Conclusions and future work 71

C. LabVIEW software and figures 79

D. EAGLE software and circuit layouts 84

1.2. Aim of the project

1.3. Organization of the document

2. Theoretical bases and background

2.1. Brief historical review

Table 2.1: Budin temperature measurements results

Figure 2.1: Tarniers Incubator used by Budin[Har]

A neonatal incubator is a device consisting of a rigid box-like enclosure in which an infant

Figure 2.2: Incubator Isolette C500 from Dräger Medical

Some of the commercial models of incubators currently available are:

Ohmeda Giraffe OmniBed

Figure 2.3: Ohmeda Giraffe Omnibed incubator

Figure 2.4: GE Lullaby Incubator XP

Medix Natal Care ST

Figure 2.5: Medix Natal Care ST incubator

Caleo Drger Medical incubator

Figure 2.6: Caleo Dräger Medical incubator

Thermoregulation is the capacity to maintain normal body temperature under varying

Figure 2.8: Body temperature regulation

2.4. Thermoregulation in neonates

Figure 2.9: Thermo-neutral neutral range of an infant[AL04]

Figure 2.10: Thermo-neutral range of a premature newborn[AL04]

2.5. Heat transfer mechanisms in neonates

Heat is gained or lost by convection, conduction, radiation and evaporation. An understand-

Figure 2.11: Heat loss routes from neonates

Convective heat transfer can be described by the following equation:

Q = (T surf − T env) ∗ hconv ∗ Asurf (2.1)

Conductive heat transfer can be described by the following equation:

Radiative heat transfer can be described by the following equation:

Q = (T surf − T surr) ∗ hr ∗ Asurf (2.3)

2.6. Control system

Figure 2.12: The incubator as a dynamic system

- Knowledge of the desired value: It is necessary to know what is being controlled, to

2.6.1. Closed loop control systems

Figure 2.13: Generalized closed-loop control system

The following are types of controllers for closed-loop systems.

u(t) = K1 ∗ e(t) (2.4)

Proportional plus Integral (PI) control

Proportional plus Derivative (PD) control

Proportional plus Integral plus Derivative (PID) control

Table 2.2: Ziegler-Nichols tuning, using the oscillation method

3. Design and analysis

3.1. Specification parameters

4 channel, 100Ω RTD, 24-Bit Analog Input Module.

Figure 3.1: DAQ Module NI 9217 terminal assignments

More information available at:[Ins07a].

32 channel ±200mV to ±10V, 16 Bit, 250kS/s Analog Input Module.

- 32 single ended or 16 differential analog inputs.

Table 3.1 summarizes the accuracy of the measuring ranges.