Content uploaded by Ehsan Abdi
Author content
All content in this area was uploaded by Ehsan Abdi on Mar 24, 2021
Content may be subject to copyright.
IET Power Electronics
Special Section: Selected papers from the 8th International
Conference on Power Electronics, Machines and Drives (PEMD 2016)
Energy harvesting and wireless data
transmission system for rotor instrumentation
in electrical machines
ISSN 1755-4535
Received on 1st November 2016
Revised 17th March 2017
Accepted on 24th May 2017
E-First on 6th July 2017
doi: 10.1049/iet-pel.2016.0890
www.ietdl.org
Danilo X. Llano1,2 , Salman Abdi1,2, Mark Tatlow1,3, Ehsan Abdi4, Richard A. McMahon2
1Electrical Engineering Division, University of Cambridge, 9 JJ Thomson Avenue, Cambridge CB3 0FA, UK
2WMG, University of Warwick, Coventry CV4 7AL, Coventry, UK
3McLaren Automotive Ltd, Surrey, UK
4Wind Technologies Limited, St Johns Innovation Park, Cambridge CB4 0WS, UK
E-mail: dxl20@cantab.net
Abstract: It is desirable to measure rotor quantities such as currents and temperatures in an electrical machine for design
verification and condition monitoring purposes. A Bluetooth module which sends data from the rotor was previously reported in
literature, but this module was battery powered, and therefore the duration of the tests was limited. This study presents a
solution to this problem by developing a rotor-mounted power supply system which can harvest energy from the magnetic field
inside the machine, by fixing an external loop to the rotor and making use of the induced voltage in the loop. A full-bridge
rectifier, boost converter and battery charging module were developed to supply sufficient power to a bespoke Bluetooth
transmission system and associated sensor circuitry.
1 Introduction
The ever increasing use of variable speed drives (VSD) in both
traditional industries and in new areas such as the automotive
sector, robotics and more electric ships and aircraft has led to a
need for electrical machines with high performance, reliability,
robustness, efficiency and power density. Achieving successful
machine designs requires a good knowledge of machine
parameters, especially thermal parameters, for the prediction of
performance but their determination and verification are difficult,
especially for rotor-related quantities.
Unfortunately, rotor measurements cannot be wired out to a
data acquisition system without great difficulty, thus any signal
must be transmitted wirelessly. A battery powered system to
measure rotor currents and transmit the data using Bluetooth has
been reported [1, 2]. In that approach the batteries required regular
recharging and the system had to be partially dismounted from the
rotor to be charged, which limits the practicality of the
instrumentation system and the tests than can be carried out. On the
other hand, when the drive is in operation, there is a potential
source of energy in the magnetic field of the machine that could be
used to feed low-power instrumentation, but achieving stable DC
power from this field is challenging due to its variable frequency
and amplitude.
This paper describes an energy harvesting module which draws
power from the magnetic field in an electrical machine and
transmits rotor data wirelessly via Bluetooth to a data acquisition
tool based on LabVIEW. The system has been implemented in a
brushless doubly fed machine (BDFM), where the presence of two
separate stator windings with different pole numbers and supply
frequencies, and a special rotor winding design, creates an
advantageous magnetic field distribution for energy harvesting [3].
Also, the BDFM used in this work has been extensively studied in
[4] so the data acquired with the rotor instrumentation system can
be confirmed against previous simulation and experimental results.
Although the system has been demonstrated in a BDFM, most of
the electronic designs and control algorithms are applicable or
easily extendable to other machines. The approach also opens
possibilities for the long-term condition monitoring of machines.
The current prototype has been designed to be as small as
possible but its physical size is still rather big for small machines
or machinery with high power density. Bearing this in mind, there
are two main target applications: using this instrumentation in large
size machines with enough room to accommodate the system or
looking at design verification of prototypes where it is practical to
extend the shaft to accommodate the module, noting that this may
alter the thermal behaviour of the machine.
Instrumentation to measure rotor temperatures in electrical
machines has been reported in the literature as part of thermal
modelling studies or the verification of sensorless temperature
estimators and observers [5–8]. In all cases, those systems were
battery powered and transmitted data wirelessly using infrared
communication [5, 6]. Kovačić et al. [9] used an isolated DC/DC
converter integrated circuit (IC) connected to the synchronous
machine excitation to feed temperature sensors and a Bluetooth
data transmission system. Another energy harvesting technique
using a small DC generator and an attached eccentric mass was
proposed in [10, 11]. The armature of the DC generator is attached
to the prime mover rotor and field magnets are held in a frame free
to rotate relative to the rotor. A small mass is also attached to the
frame so that it remains stationary as the rotor/armature spins but
the mass will be displaced so that a gravitational torque equal to
the DC generator torque (in steady-state conditions) is established.
The reported arrangement uses a maximum power point tracking
(MPPT) algorithm to maximise the power extracted from the DC
generator, i.e. when the load and armature resistances are equal.
The system has an energy conversion stage composed of a boost
converter to perform the MPPT algorithm and a buck converter to
regulate the DC link at 3.3 V. The system was characterised on a
test bench and generated about 60 mW at 2000 rpm [10]. This
solution has limited power and requires a DC generator, energy
conversion and energy storage circuitry leading to a complex
system with large volume.
Battery powered solutions are relatively attractive if only
temperature measurements are required since low sampling rates
are acceptable and the Bluetooth and analogue circuitry can operate
in energy saving mode. On the other hand, batteries are not
particularly suitable if rotor currents are also measured because
higher sampling and data transmission rates are required to record
waveforms (0–200 Hz) with a reasonable resolution and the
circuitry will be operating in an active mode most of the time,
demanding more power. The energy harvesting and rotor
instrumentation reported in this paper are in the form of modules
that can be easily adapted to a final user's instrumentation
IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
1259
requirements. For instance, the size of the energy harvesting
module could be reduced if only temperature sensors are used since
low power and hence less energy storage are demanded.
This paper is organised as follows: first at all, a suitable method
for extracting energy from the BDFM rotor is investigated and
calculations are carried out to determine the expected range of
voltages and frequencies that this would yield. Other practical
issues such as construction, ease of retro-fitting to the rotor and
mechanical reliability are also considered. Then, an energy
conversion system which converts low AC input voltages from the
machine rotor into a stable DC power supply suitable for feeding
electronic systems is presented. A metal–oxide–semiconductor
field-effect transistor (MOSFET)-based active rectifier is proposed
for AC/DC conversion and voltage boosting. Later on, a current
and temperature sensing, signal conditioning, Bluetooth data
transmission system and its associated software are described. The
system includes Rogowski coils for current measurements, PT100
probes for temperature sensing, amplifiers and analogue filters for
signal conditioning and a Bluetooth module and microcontroller
for wireless communication. Finally, a data logger based on
LabVIEW is presented for data management.
2 Overview of the BDFM and the machine
measuring system
2.1 BDFM machine
The BDFM is a variable speed electrical machine which shows
particular promise in applications such as wind energy generation
or as a VSD, but still needs design and thermal modelling
verification [12]. The BDFM has two separate stator windings,
which must be of different pole numbers to avoid direct coupling
between them. The machine of interest in this paper has a four-pole
winding connected directly to the 50 Hz three-phase grid, and an
eight-pole winding supplied via a power converter. The basic
outline of the BDFM system is shown in Fig. 1 (left). The rotor,
however, must couple to the two air-gap fields associated with the
two stator windings. The machine used in this paper has a nested-
loop rotor; this is formed from sets of concentric copper bars.
Fig. 1 (right) shows a view of the nested-loop rotor with six ‘nests’,
and three loops per nest.
An experimental BDFM has been constructed in a frame size
180; its key specifications are given in Table 1. A rotor-mounted
wireless system for measurement of the rotor bar currents and
temperatures was initially designed by Abdi [13]. It used Rogowski
coils [14] to measure the currents, and incorporated signal
conditioning, analogue-to-digital (ADC) conversion and a
Bluetooth module to transmit the measured data to a computer. The
system was battery powered, but the life of the batteries was
insufficient to perform tests of extended duration.
2.2 Rotor measuring system
This electronic system requires a stable 6 V DC link to match the
minimum input voltage necessary for a 5 V linear regulator. The
latter is used to achieve a good quality DC link for the electronic
components. Subsequently, two different voltage levels are derived,
5 V for the energy harvesting module and gate drivers and 3.3 V
for the Bluetooth module and analogue signal conditioning board.
The current requirement was conservatively estimated based on the
analysis of the system. Table 2 lists the main elements of the
system and their predicted contributions to the current draw, based
on the maximum power consumption figures in the manufacturers’
datasheets. Overall the energy harvesting system must be able to
supply at least 200 mA at 6 V to power the measuring system and
electronics in the energy module itself. The 6 V DC link was
selected based on the fact that the MOSFETs used in the active
rectifier are optimised for 5 V applications and the system was
developed in modules so that the power supply and the
instrumentation/Bluetooth boards can be used independently if
necessary. The latter requires having a voltage regulator on each
board. However, this design can be optimised according to the user
specifications, for example, if only temperature measurements are
required, the size, number of components and power consumption
will be reduced significantly.
Fig. 1 Block diagram of the BDFM (left), end view of D180 nested-loop BDFM rotor (right)
Table 1 Specification of the 7.8 kW D180 BDFM
Characteristic General Grid tied
winding
Converter fed
winding
frame size 180
poles 4 8
voltage rating 240 V at 50
Hz (delta)
172 V at 25 Hz
(delta)
current rating 9.5 A (line) 6.8 A (line)
speed range 750 rpm ± 33%
rated torque 100 N m
rated power 7.8 kW at 750
rpm
stack length 0.19 m
efficiency (at full load) ∼92%
Table 2 Power specification for the energy harvesting
system
Circuit element Current, mA
Bluetooth module 38
microcontroller 20
Rogowski coil circuitry 24
temperature sensor circuitry 25
energy harvesting microcontroller 24
power distribution system 19
total (including 33% margin) 200
1260 IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
3 Energy harvesting techniques
The energy harvesting system was designed to meet the
specifications outlined in Table 3. The output current of 200 mA is
the estimated maximum current consumption of the rotor
instrumentation operating at 6 V (later reduced to 5 or 3.3 V with a
linear regulator).
Different methods of energy harvesting from rotor magnetic
fields have been assessed based on analytical calculations to
determine the expected voltages and frequencies that each option
would yield. Other practical issues such as construction, ease of
retro-fitting to an existing rotor and mechanical robustness were
also considered. This analysis has been reported in [15] and it was
concluded that a dedicated rotor loop, as described below, is the
best option.
3.1 Energy harvesting rotor loop
The energy harvesting loop is realised by making connections to
two rotor bars which are pitched at the desired angle and shorted at
one end through the end ring. The pitch angle can be chosen so that
the energy harvesting loop couples with the magnetic field of either
or both stator windings. A pitch angle of 90° is attractive because
this will couple fully to the grid tied winding (see Fig. 2).
Applying Faraday's law to the machine, the voltage Erms
induced in a single loop that fully couples to the air-gap field Brms
is given by
Erms =lωd
pBrms
(1)
where l is the stack length of the machine, d is the air-gap diameter,
p is the number of pole pairs for the relevant winding and ω is the
frequency in the rotor reference frame. In [16], the measured air-
gap field of the four-pole winding (grid tied with rated supply) was
0.30Trms at 230 V. The frequency f in the rotor reference frame is
given by
f=fs−p2πN
60
(2)
where fs is the electrical frequency applied to the relevant stator
windings, p is the number of pole pairs for the relevant winding
and N is the mechanical speed of the rotor. Using (1), the voltage
and frequencies expected in the external rotor loop are given in
Table 4. This method has the advantage of being physically robust,
in contrast to the end winding transformer, provided the wire taps
are securely fastened to the rotor bars. However, the induced
voltage is low and there is no possibility of using additional turns
to increase this voltage, though a step-up converter can be used to
raise the voltage to a more useful level. Ultimately, this method
was chosen due to its mechanical simplicity, despite its low output
voltage levels.
3.2 Finite-element (FE) analysis
The induced voltage in a rotor loop can be also determined by FE
analysis. The D180 BDFM machine is modelled in synchronous
mode at the full load specification listed in Table 1. Full load
conditions were chosen for the FE analysis because they can be
compared with the analytical calculations to validate the
estimations done for sizing the energy harvesting module. Fig. 3
(left) shows the magnetic flux in the iron circuit with four/eight-
pole stator winding configuration. A single-turn coil with a pitch
angle of 90° is wound around the rotor slots through the machine
stack length in the FE mesh to simulate the external loop formed
from taps to the rotor bars, so that coupling with the eight-pole air-
gap field is mainly eliminated. Fig. 3 (right) shows the induced
voltage in the external rotor coil at different mechanical speeds.
The root mean square (RMS) value of the induced voltages
obtained by FE is shown in Table 4. The voltages estimated
analytically by Faraday's law and FE analysis are similar.
4 Design of the energy conversion module
The system is designed to meet a list of specifications set out in
Table 3. The energy harvesting module must operate in the
challenging environment inside an electrical machine, where it will
be subjected to high temperatures, strong magnetic fields, vibration
and centripetal acceleration. These specifications include a
moderate margin to allow for uncertainty over the precise
characteristics of the machine. The energy conversion system must
be designed to:
• Rectify and step-up the low AC input voltage from the rotor
bars.
• Provide a small amount of energy storage to keep the system
operating for short periods when the machine is idle.
Fig. 4 (top) shows the block diagram of the proposed solution.
4.1 Rectifier
A passive diode rectifier is not suitable due to the low input voltage
as the voltage drop across the diodes would use up almost all of the
available voltage, even if Schottky barrier diodes with low forward
voltage were used. Therefore, a MOSFET-based active rectifier (H
Bridge) with additional line inductance was chosen for rectification
and stepping up the DC voltage at the same time. The continuous
output power requirement was conservatively estimated up to 1.5
W. It can be shown that if the effects of dead time, MOSFET Ron
resistance and other parasitic effects associated with pulse-width
modulation (PWM) switching are neglected, the magnitudes of the
rectifier input voltage and the DC link voltage are related as
α=Vin
VDC
(3)
Table 3 Specification for the rotor energy harvesting power
supply
Characteristic Value
input voltage range 0.75–1.5 Vrms
input frequency range 20–50 Hz
output voltage 6 Vdc
output current 200 mA
run time with no input power 10 min
temperature rating 85°C
Fig. 2 Energy harvesting loop comprising two rotor bars pitched 90°
apart and shorted at one end through an end ring. The rotor has 36 slots
Table 4 Voltage and frequencies induced in a single rotor
loop of 90° pitch
Rotor
speed, rpm
Electrical
frequency, Hz
Voltage, Vrms –
calculation
Voltage, Vrms
– FE
200 43.3 1.39 1.49
500 33.3 1.07 1.12
800 23.3 0.75 0.79
IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
1261
where α is a single parameter such that α< 1. A closed-loop
controller is necessary to deal with the variable input voltage and
compensate for the elements that were neglected in (3).
A detailed circuit diagram of the proposed solution is shown in
Fig. 5.
4.2 Energy storage
Supplying 200 mA at 6 V for 10 min would require a total energy
storage capacity of 720 J. However, it would be desirable to allow
some additional margin in the design subject to space and weight
constraints. This energy storage could be implemented either with a
small rechargeable battery or using double-layer electrolytic
capacitors. However, both of these technologies rely on chemical
processes and are therefore temperature sensitive. Different battery
options were considered and among them a nickel-metal hydride
rechargeable battery rated at 510 mAh and 3.6 V, about 1100 J in
10 min, was chosen to power the system when the rotor is idle. The
battery can withstand a continuous trickle charge, so accurate
charge management is not required. The battery is rated at 85°C,
Fig. 3 FE analysis – distribution of flux density in the D180 BDFM iron circuit in synchronous mode of operation (left), induced voltages in the additional
loop mounted in rotor slots with a pitch angle of 90° (right)
Fig. 4 Energy harvesting scheme (top), control scheme (bottom)
Fig. 5 Circuit diagram
1262 IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
which is the maximum operating temperature recommended by the
manufacturer. Unfortunately, batteries with a higher temperature
rating were not available from manufacturers in a small quantity.
4.3 Output stage
A DC/DC converter (LM2731) was selected to regulate the DC
link at 6 V. This IC has embedded the main active elements for a
standard boost converter (power MOSFET, gate driver, PWM
controller, feedback amplifier and protection circuits) and is used
with an external inductor, diode, capacitors and resistors, the latter
for setting the output voltage. The switching frequency is 600 kHz.
4.4 Shut down circuitry
For practical use inside the machine it was required that the power
supply could be shut down wirelessly, to switch off the measuring
circuit and avoid draining the battery when the machine was not
running. This was especially important since charge in the battery
is needed in order for the power supply to start up. This cannot be
done via the wireless system since the radio is powered down in
the off state. A centripetal switch was considered, but the
acceleration at minimum speed (200 rpm) was only about 2g,
making it difficult to distinguish from gravity depending on the
orientation of the switch. It was therefore decided to detect the
presence of voltage at the input terminals and use this to trigger the
power supply to start. The circuit in Fig. 6 was devised to sense
this voltage and switch on the power supply. When VAB ≥ 0.6 V,
the transistor Q1 turns on, pulling the gate of the P-MOSFET
towards ground and allowing it to connect the battery to the output
stage. The 10 kΩ resistor and 22 μF capacitor allow the gate
voltage to remain low even during negative half-cycles of the input
voltage, when the transistor is off. Furthermore, to allow operation
to continue for periods while the machine is idle, a second
transistor Q2 was added, with its base current controlled by the
microcontroller. During normal operation, the microcontroller
keeps this transistor turned on so the circuit will remain active.
However, the microcontroller can be instructed to pull this pin low,
turning off the transistor. If this occurs while the machine is
inactive, both transistors will then be off, and the P type MOSFET
(PMOS)'s gate voltage will rise due to the 10 kΩ resistor. This will
turn off the power to the output stage, shutting down the system.
4.5 Control system
4.5.1 Control structure: A closed-loop controller is needed to
regulate the power flow through the rectifier as the relationship
between α in (3) and the output current is not straightforward. A
control scheme with two nested proportional–integral (PI)
controllers, an inner loop to regulate the input current within limits,
and an outer loop to regulate the battery current, is used. The block
diagram of the proposed controller is given in Fig. 4 (bottom). RC
low-pass filters were added to the sensor circuits to remove
switching-frequency noise.
This strategy is similar to the control structure used for single
phase active rectifiers with regulated DC link voltage [17, 18].
However, the presence of the battery introduces the need for using
its charging current in the outer control loop. First at all, charging
currents, rather than voltage figures, are commonly given in battery
specifications. Second, it is highly desirable to control the charging
current of the battery since this is the only way of ensuring that the
battery will be sufficiently charged to power the system when the
machine is not rotating. The system also measures the battery
voltage, which in turn is used to identify any overcharging
condition. The charging current would be set to zero in this latter
case.
The power balance equation is written as follows:
PAC =PDC ⇒VinIin =VDC IBAT +ICAP +ILOAD
(4)
Considering ICAP =C(dVDC/ dt) and VDC =VBAT =IBATZBAT,
where ZBAT is the battery impedance, and (3) it is possible to write
αIin =IBAT +Cd
dtIBATZBAT −ILOAD
(5)
Equation (5) shows that the battery current can be regulated by
controlling the AC input current and ILOAD can be treat as a
disturbance. ILOAD was varied for characterising of the module
(efficiency measurements in Section 5), but it was fairly constant
during actual operation. Also, when the machine is running the
load and any circuitry will be powered by the AC side and the
current to the battery will be specifically used to charge the battery
so that its stored energy can be used when the machine is idle, but
the instrumentation is still operating, or to wake up the system. The
DC/DC converter is represented by the ILOAD term in this analysis.
Unfortunately, the manufacturer does not provide enough details to
model this component. However, no stability issues were identified
during characterisation with loads that varied significantly more
than the values expected for normal operation. In addition, the
specifications for the DC/DC converter are given for an input of
2.7–14 V and maximum load current of 1.8 A, which are within the
operation range of the module.
4.5.2 AC control (feed-forward loop): It is well known that PI
controllers cannot track sinusoidal references (as the input current)
without a steady-state error [19], but this limitation can be
overcome by adding a feed-forward loop using the measured input
voltage. In addition, to prevent instability when the peak of the
output current is reached, a limit was placed on the maximum input
current that could be requested by the outer loop. By applying this
limit to the output of the outer PI controller, operation in the
unstable regime beyond the point of maximum power transfer is
avoided.
Dynamic response: Finally, some consideration must be given
to the transient response of the control system. A formal
mathematical analysis could be conducted, using techniques such
as state-space averaging to determine the transfer functions of the
rectifier [20], but this is beyond the scope of this paper. Instead, the
following general points were used to estimate appropriate values
for the filter cut-offs and integrator time constants:
• The inner loop should have a phase lag as close as possible to
zero over the whole AC input frequency range. This was
necessary to maintain the power factor near unity.
• The outer loop should have low bandwidth to strongly attenuate
the ripple in the battery current at twice the fundamental input
frequency.
• The low-pass filter on the input current reading must pass the
full range of input frequencies, but should strongly attenuate the
switching frequency and its harmonics. The cut-off point must
not be set too low to avoid limiting the response rate of the inner
loop. The gains of the loops were adjusted by trial and error to
obtain the desired response.
Fig. 6 Shut down circuitry
IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
1263
5 Construction and testing of the energy
harvesting system
5.1 Construction
Results for the first energy harvesting prototype were reported in
[15]. A second version of the system has been built and tested. The
original prototype was built using three single printed circuit
boards (PCBs) (80 mm × 49 mm each one), while the new design
has only two boards of the same size. This was achieved by using
components with the smallest package, suitable for hand assembly,
available in the market. Additionally, nexFET MOSFETs
(CSD16321Q5) from Texas Instruments were used in an effort to
minimise Ron resistance (2.1 mΩ at VGS = 4.5 V from datasheet
specifications), track resistance and parasitic inductances as the
devices are packaged in SON (small-outline no leads) 5 mm × 6
mm packages. Having fewer boards also reduces losses due to
connectors and wires between PCBs. The components are rated up
to 125°C and automotive qualified if available. As already stated,
the limiting component is the battery. Heavier components such as
inductors and the battery can be mechanically secured to the
mounting bracket, but this was not done in the current prototype.
The hexagonal bracket shown in Fig. 7 (left) has been designed to
allow the energy harvesting boards, Bluetooth module and three
other circuit boards (if required) to be secured to the rotor shaft.
The two halves separate to allow fitting the whole unit onto the
rotor shaft.
5.2 General characteristics
A low-cost AVR 8-bit microcontroller from Atmel was used to run
the control algorithms detailed in Section 4.5. Unipolar PWM at
25 kHz and 125 ns dead time was used to drive the active rectifier.
The MOSFETs and gate drivers were selected to operate at VGS =
5 V. The current consumption of the microcontroller was 39 mA,
while the gate drivers, sensing amplifiers and voltage reference ICs
require about 32 mA. These values are comparable with the
estimates listed in Table 2. The charging current reference can be
wirelessly set from the graphical interface described in Section 6.3.
The battery manufacturer recommends a trickle current between 5
and 15 mA for long-term operation. A 25 mA reference value was
used during characterisation as this is the minimum current
recommended for normal charging.
5.3 Maximum load current and efficiency tests
The second prototype of the energy harvesting module was
characterised on the laboratory bench with a low-voltage, variable-
frequency AC power supply as was done for the first system
reported in [15]. The power source was composed of a signal
generator and amplifier to achieve the power level required by the
circuit. Later, in order to match the real rotor as closely as possible,
it was necessary that the power source should have as low output
impedance as possible. For this, a commercially available toroidal
transformer with a third winding wound by hand through the toroid
using thick wire is used. The measured output impedance of the
circuit in this configuration is 83 mΩ over the full range of
frequencies.
The battery current was set to 25 mA and the load was a
variable resistor. The results presented here correspond to the first
(in blue) and second prototype (in red) to show the improvement
not only in size but also in performance. The maximum available
load current is plotted in Fig. 8 (left). The 200 mA target was
achieved at an input voltage of 0.74 and 0.67 V for the first and
second prototypes, respectively, which meets the specifications
outlined in Table 3. Varying the input frequency from 20 to 50 Hz
was found to have no measurable effect on performance. The
efficiency of the modules is shown in Fig. 8 (right). The efficiency
is calculated as
η=Pload +PAR +Pbat
PAC
(6)
where PAR is the power required to feed the active rectifier
circuitry (gate drivers, voltage references etc.), Pload is the power
consumed by the load (variable resistor in these tests, but the
instrumentation and Bluetooth boards in the final implementation),
Pbat is the power required to charge the battery at 25 mA and PAC is
the input power.
Fig. 7 Hexagonal bracket for mounting the circuit boards to the rotor shaft (left), power electronics, energy storage and control PCB (right)
Fig. 8 Maximum load current delivered from the converter module over an input voltage range with a battery trickle charge of 25 mA (left). Efficiency of the
converter module with a battery trickle charge of 25 mA (right)
1264 IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
6 Bluetooth data transmission system for rotor
instrumentation
Fig. 9 shows the block diagram of the rotor instrumentation and
Bluetooth data transmission system.
6.1 Bluetooth module
A WT12 Bluetooth 2.1 chip from Bluegiga Technologies is used
for wireless data transmission. It has an integrated chip antenna,
UART communication, 30 m range in line-of-sight and can be
easily integrated with a microcontroller for a fully functional
control and data transmission system. The microcontroller used for
this task was a SAM4S from Atmel. This microcontroller manages
the communication with the Bluetooth module, ADC sampling,
temperature multiplexing, serial communication with the energy
harvesting module to get data about the state of the battery (voltage
and current), input voltage and input current. This system is built
independently from the energy harvesting module to have
modularity so that different transmission systems (e.g. with
different sensors) can be integrated to the same power source.
6.2 Temperature, current sensors and signal conditioning
6.2.1 Temperature sensors: PT100 temperature sensors rated at
−50 to 500°C were selected. The system has eight temperature
channels multiplexed with an 8 : 1 digitally controlled analogue
switch CD74HC4051M96 from Texas Instruments. The sampling
process is done as follows: the microcontroller selects the
temperature channel to be connected to the output of the
multiplexer and allows 250 ms to let the signal to settle. The
resistance of the selected PT100 probe is sensed using a
Wheatstone bridge configuration. Afterwards, this measurement is
amplified (gain = 100) with an instrumentation amplifier
(AD627BRZ from Analog Devices) and low-pass filtered with a
second-order Sallen–Key filter with unity gain, Q= 0.5 and cut-off
frequency 3.5 Hz. The same temperature channel is then sampled
50 times at 2 kHz and finally the average value is transmitted to the
Bluetooth module. This process is repeated for all the channels
under control by the LabVIEW user interface.
6.2.2 Current sensors: The Rogowski coils were manufactured
as described in [1, 13] to measure the rotor currents. The voltage
induced at the coil terminals is proportional to di/ dt rather than i t ,
therefore an integrator and low-pass filter are required for signal
conditioning, in this case an analogue implementation. The rotor
currents are expected to be within 20–1000 Arms and 0–200 Hz.
Assuming a sinusoidal current i=Isin ωt, then the output voltage
at the coil terminals will be
Vcoil =γdi
dt=γωIcos ωt
(7)
where γ is the Rogowski coil constant and depends on its physical
construction [1, 13].
A practical integrator circuit has a transfer function given by
Vout
Vin
= − kG
j(ω/ωn)+1
(8)
where kG is the integrator gain and ωn is its natural frequency.
Assuming ω≫ωn, then the magnitude of (8) can be approximated
as
Vout
Vin
=kG
ωn
ω
(9)
If (7) is applied to (9), then the frequency term ω is effectively
cancelled out, leading to a constant gain kGωn regardless of the
frequency. The integrator was designed with ωn= 0.378 rad/ s
(0.06 Hz) and kG= 1833. The gain and natural frequency could be
modified independently if required. The low-pass filter is a Sallen–
Key filter with gain G= 1.6, Q= 0.707 and cut-off frequency 5.3
kHz to maintain constant gain and almost zero phase shift in the
frequency range 5 Hz–1 kHz as there is an interest in measuring
the third and fifth harmonic components in the current waveform.
The system has four current channels. The microcontroller
samples and stores 100 points for each channel. The sampling
frequency is adapted in each data acquisition run to maintain good
resolution and 400 data points are transmitted via Bluetooth. The
LabVIEW interface calculates the frequency of the measured
waveforms and sets the sampling frequency accordingly for the
next iteration (at least 20 points per cycle).
6.3 LabVIEW data acquisition system
A data logger based on LabVIEW was developed for data
acquisition, processing and recording. This interface communicates
wirelessly with the Bluetooth module using a serial protocol at
115.2 kbps. The program acquires temperature, rotor current and
data from the energy harvesting module at different rates. The
program also manages the communication, does all the numerical
transformations (sampled signals from the ADCs at 12 bits to
values with a physical meaning), displays the measurements in real
time, calculates characteristics such as the DC offset (for self-
calibration), the RMS value and frequency of the rotor current and
records data for further analysis if required. The user interface is
shown in Fig. 10 (top). It has all the communication settings on the
left, information about the energy harvesting module in the centre,
and rotor current and temperature measurements on the right. A
second tab displays time trends for temperature and energy
harvesting values.
Fig. 9 Bluetooth system scheme
IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
1265
6.4 Bench calibration
Fig. 10 (bottom) shows the sensor and signal conditioning board on
the left and the Bluetooth module and microcontroller on the right.
The current consumption of the Bluetooth module and sensor board
at 6 V were 35 and 45 mA, respectively. The temperature sensors
were calibrated by comparing the PT100 resistance measured with
a four-wire precision multimeter Fluke 8846A and the
corresponding value calculated with the LabVIEW interface at
different temperatures. The Rogowski coils were calibrated at 50
Hz as follows:
• A variac was connected to a resistive load to set different current
values.
• The Rogowski coils were mounted around 36 turns of
conducting wire to simulate higher currents.
• The RMS value of the current measured with a precision
multimeter multiplied by the number of turns was compared
with the RMS value obtained with the LabVIEW program.
The system was designed for 1.5 W power consumption, with
an actual measured power of about 0.9 W. Both values are
negligible compared with the power rating of the machine 7.8 kW.
On top of this, the LabVIEW interface samples the input voltage
and current, battery voltage and battery current in the energy
harvesting module, therefore these values can be used to
compensate any efficiency measurement if required.
7 Conclusions
The aim of the paper was to demonstrate an energy harvesting
system and Bluetooth module which can be mounted on the rotor
of an electrical machine, in this case a BDFM, to measure and
transmit rotor current (four channels) and temperature (eight
channels). Two techniques were considered. The best solution was
to insert physical taps between loops in the rotor such that small
electromotive force voltages are induced in an external circuit. An
active rectifier and DC/DC converter were designed to convert the
low AC input voltage into a stable DC voltage for battery charging
and powering the energy harvesting microcontroller, sensor
circuitry and Bluetooth module. The system was successfully
characterised (efficiency measurement) and calibrated (sensors and
signal conditioning) on the laboratory bench to check it met all the
design requirements. The current prototype might experience space
limitations especially in compact machines with high power
density, but the system could still be further miniaturised. As an
example, if only temperature measurements are required the
Bluetooth module and SAM4S microcontroller can be replaced
with a single Bluetooth 4.0 IC featuring the same tasks but at lower
sampling rates. This leads to lower power consumption, smaller
electronic components, less energy storage requirement and a
smaller footprint. Also, the size of the current prototype is limited
by the use of components suitable for hand assembly and the
presence of testing points/connectors that can be removed in
further versions. Once the energy harvesting is proved to work, this
current design can be optimised depending on the user
specifications to reduce size and power consumption, for example
if only temperature sensing is required.
8 Acknowledgment
The research leading to these results has received funding from the
Innovate UK technology programme (Project reference number:
102155).
9 References
[1] Abdi-Jalebi, E., McMahon, R.: ‘High-performance low-cost Rogowski
transducers and accompanying circuitry’, IEEE Trans. Instrum. Meas., 2007,
56, (3), pp. 753–759
[2] Roberts, P.C., Jalebi, E.A., McMahon, R.A., et al.: ‘Real-time rotor bar
current measurements using Bluetooth technology for a brushless doubly fed
machine (BDFM)’. Second Int. Conf. Power Electronics, Machines and
Drives, 2004 (PEMD 2004) (Conf. Publ. No. 498), March 2004, vol. 1, pp.
120–125
[3] Abdi, S., Abdi, E., Oraee, A., et al.: ‘Equivalent circuit parameters for large
brushless doubly fed machines (BDFMs)’, IEEE Trans. Energy Convers.,
2014, 29, (3), pp. 706–715
[4] Abdi, S., Abdi, E., Oraee, A., et al.: ‘Optimization of magnetic circuit for
brushless doubly fed machines’, IEEE Trans. Energy Convers., 2015, 30, (4),
pp. 1611–1620
[5] Ganchev, M., Kubicek, B., Kappeler, H.: ‘Rotor temperature monitoring
system’. XIX Int. Conf. Electrical Machines (ICEM), 2010, September 2010,
pp. 1–5
[6] Ganchev, M., Kral, C., Oberguggenberger, H., et al.: ‘Sensorless rotor
temperature estimation of permanent magnet synchronous motor’. IECON
Fig. 10 LabVIEW interface (top left), system installation (top right), signal conditioning board (bottom left), Bluetooth module (bottom right)
1266 IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
2011 – 37th Annual Conf. IEEE Industrial Electronics Society, November
2011, pp. 2018–2023
[7] Xue, X., Sundararajan, V., Brithinee, W.P.: ‘The application of wireless sensor
networks for condition monitoring in three-phase induction motors’. 2007
Electrical Insulation Conf. and Electrical Manufacturing Expo, October 2007,
pp. 445–448
[8] Jianzhong, G., Hui, G., Zhe, H.: ‘Rotor temperature monitoring technology of
direct-drive permanent magnet wind turbine’. Int. Conf. on Electrical
Machines and Systems, 2009, ICEMS 2009, November 2009, pp. 1–4
[9] Kovačić, M., Vražić, M., Gašparac, I.: ‘Bluetooth wireless communication
and 1-wire digital temperature sensors in synchronous machine rotor
temperature measurement’. 14th Int. Power Electronics and Motion Control
Conf. (EPE/PEMC), 2010, September 2010, pp. T7–25–T7–28
[10] Toh, T.T., Mitcheson, P.D., Yeatman, E.M.: ‘Continuously rotating energy
harvester with improved power density’. PowerMEMS, Sendai, Japan, 2008
[11] Toh, T.T., Mitcheson, P.D., Holmes, A.S., et al.: ‘A continuously rotating
energy harvester with maximum power point tracking’, J. Micromech.
Microeng., 2008, 18, (10), p. 104008
[12] Abdi, E., Mcmahon, R., Malliband, P., et al.: ‘Performance analysis and
testing of a 250 kW medium-speed brushless doubly-fed induction generator ’,
IET Renew. Power Gener., 2013, 7, (6), pp. 631–638
[13] Abdi, E.: ‘Modelling and instrumentation of brushless doubly-fed (induction)
machines’. PhD dissertation, Engineering Department, University of
Cambridge, 2006
[14] Ward, D.A.: ‘Measurement of current using Rogowski coils’. IEE Colloquium
on Instrumentation in the Electrical Supply Industry, June 1993, pp. 1/1–1/3
[15] Tatlow, M., McMahon, R., Abdi, S.: ‘Wireless energy harvesting for rotor
instrumentation in electrical machines’. 8th IET Int. Conf. on Power
Electronics, Machines and Drives (PEMD 2016), April 2016
[16] Oraee, A., Abdi, E., Abdi, S., et al.: ‘A study of converter rating for brushless
DFIG wind turbines’. 2nd IET Renewable Power Generation Conf. (RPG
2013), September 2013, pp. 1–4
[17] Blaabjerg, F., Teodorescu, R., Liserre, M., et al.: ‘Overview of control and
grid synchronization for distributed power generation systems’, IEEE Trans.
Ind. Electron., 2006, 53, (5), pp. 1398–1409
[18] Rodriguez, J.R., Dixon, J.W., Espinoza, J.R., et al.: ‘PWM regenerative
rectifiers: state of the art’, IEEE Trans. Ind. Electron., 2005, 52, (1), pp. 5–22
[19] Ogata, K.: ‘Modern control engineering’ (Prentice-Hall, 2010)
[20] Middlebrook, R.D., Cuk, S.: ‘A general unified approach to modelling
switching-converter power stages’. IEEE Power Electronics Specialists Conf.,
1976, June 1976, pp. 18–34
IET Power Electron., 2017, Vol. 10 Iss. 11, pp. 1259-1267
© The Institution of Engineering and Technology 2017
1267