Characterization and Modeling of Brushless DC Motors

and Electronic Speed Controllers with a Dynamometer

Robert Brown Inderjit Chopra
Graduate Research Assistant Distinguished Professor

Alfred Gessow Rotorcraft Center, Department of Aerospace

Engineering
University of Maryland
College Park, MD, 20742

ABSTRACT
Proper characterization of brushless DC (BLDC) motors and electronic speed controllers (ESCs) is key to improving
the range, endurance, and payload capacities of electric aircraft. Many components for electric vertical take off and
landing (eVTOL) aircraft are sourced from commercial-off-the-shelf vendors, however, these vendors often do not
provide all the important characteristics that are essential for rigorous engineering application. To address this issue,
a custom dynamometer was developed to study the performance characteristics of different BLDC motors and ESCs,
and is capable of testing motors up to 200 W DC from 7.2 V to 15 V. Recording DC power, AC power, and mechanical
power at various operating conditions allows for the evaluation of the BLDC motor/ESC system. From the sets of
experimentally recorded data, a mathematical model is developed using first principles analysis. Unique to this work
is separate modeling of the BLDC motor and ESC, with a focus on the ESC, which is typically neglected in formal
analysis.

INTRODUCTION and shipping solutions to Uber and other customers. Most of

these companies aim for “on demand aviation”, where users
The world of Aerospace engineering is changing fast. Electri- summon aircraft directly to their location and will be flown
fication of aircraft systems has allowed for flying to become directly to their destination in a short 20 minutes. To meet the
an everyday occurrence, resulting in a dramatic increase in the rigorous safety requirements associated with manned aircraft
number of unmanned aircraft systems over the past decade. flight, new analysis methods are required to adequately and
Emergence of light weight composite structures, high energy robustly model the unique challenges associated with multi-
density lithium polymer batteries, and reliable low-cost sen- rotor all electric aircraft.
sors have allowed for an increased interest into electric avia-
tion. One of the major advancements in recent decades are
high power to weight ratio brushless DC (BLDC) motors.
Increased operating efficiency and peak output power allow
for small, unmanned autonomous aircraft to complete essen-
tial tasks such as surveillance, infrastructure inspections, and
package delivery. The promise of a new electric aircraft mar-
ket has excited investors, as the global drone market is ex- Fig. 1: Electric propulsion systems on today’s aircraft.
pected to grow from $4.9 billion in 2019 to $14.3 billion by
2028 (Ref. 1).
One such problem is electric aircraft propulsion sizing. As
Compact and efficient drones have caused a transportation with gas engines, designers must not over-size or under-size
revolution - urban air mobility. Recent developments in urban their motor and motor controller, otherwise this will have
air mobility have placed a great deal of importance on electric detrimental effects to the aircraft’s performance. For exam-
aircraft. Prominent companies such as Uber have advocated ple, over-sizing a motor results in a physically heavier motor,
rigorously for air taxis, and promise to deliver full scale 4,000 which adds weight to the airframe and reduces payload capac-
lbs autonomous electric aircraft by 2023 (Ref. 2). Numerous ity. Under-sizing the electric motor requires additional current
startups have emerged recently to provide low cost mobility to be supplied from the battery, resulting in reduced range and
Presented at the VFS International Powered Lift Conference 2020,
lifetime of the aircraft. Additionally, for electric aircrafts the
San Jose, CA, January 21–23, 2020. Copyright c 2020 by the Ver- battery consists of a significant fraction of the aircraft’s total
tical Flight Society. All rights reserved. mass, and thereby augmenting power requirements dramati-
1
cally increases or decreases the battery’s mass, voltage, and but require less power to operate. Exchanging mass and power
capacity (Ref. 3). With many tradeoffs possible, identifying can change the aircraft’s mission. An unaddressed question
an optimum design requires a fundamental understanding of in research is considering which motor controller to use, as
the propulsion system’s effect on aircraft design. again heavier controllers tend to be more efficient. This exam-
To increase the performance of electric aircraft, the rotor, mo- ple demonstrates a key point, that performance characteristics
tor, and motor controller should be paired in such a way that of commercial-off-the-shelf components must be available to
the rotor’s operating point is located at the optimum efficiency designers for use in detailed electric aircraft design.
point for both the motor and motor controller. A properly Despite the need for the motor and motor controller’s perfor-
paired propulsion configuration requires the least amount of mance curves for optimum rotor pairing, few commercial-off-
power to operate, translating into increased aircraft service the-shelf vendors release information related to their BLDC
life, endurance, range, and payload. Information on the rotor’s motors and ESCs. Without the knowledge of the motor’s
torque, RPM, and power are critical pieces of information, as performance, the optimum pairing problem cannot be solved.
different motors and motor controllers have different peak op- While data unavailability may be acceptable for hobby and re-
erating points. To accomplish this integrated objective, a ref- search grade aircraft, scaling up these designs requires a sys-
erence flight condition is selected, for example cruise, and the tematic understanding of the propulsion system. For future
aircraft is trimmed for this operating point. Once trimmed, designs, it is pertinent to experimentally determine the perfor-
the required rotor power can be calculated by integrating the mance curves for BLDC motors and ESCs, which is the focus
blade loads, and the process of selecting the proper motor and of this paper.
motor controller begins.
This work addresses the optimum pairing problem by devel-
For a properly designed aircraft, the rotor’s operating point
oping an experimental apparatus to study BLDC motor and
must be matched with a motor and motor controller that have
ESC performance. From these experimental datasets, a first
the highest efficiency at this operating point. For this task,
principles based analytical model for both the ESC and the
performance curves of electric motors are essential to design-
BLDC motor are created. An important point about this work
ers, as each motor and motor controller have their own unique
is that the BLDC motor and ESC are uniquely modeled un-
operating point. Identifying optimum motor and motor con-
der in a single analytical model. By modeling the ESCs and
troller operating point allows for an informed decision to be
the BLDC motors separately, a fundamental understanding of
made with regards to pairing the motor with a given rotor.
power loss in the system is gained, with the goal of identify-
To better understand the motor and rotor interaction, consider
ing unique parameters to describe different ESCs and BLDC
the following design example. Suppose a rotor operating at
motors. Characterizing the ESC and the BLDC motor sepa-
2500 RPM is selected to meet propulsion and lift require-
rately allows for a more programmatic and discipled approach
ments. To power the rotor, two motors are under consider-
to aircraft design. Unique to this work is separate modeling of
ation, with masses of 30 grams and 50 grams respectively.
both the BLDC motor and the ESC, and is accomplished by
Ideally, the least massive motor should be selected. However,
testing 6 BLDC motors (920 RPM/V to 2500 RPM/V) with 3
the efficiency curves of the motors must be considered, and
ESCs (18A to 40A) at 3 voltage levels (7.2V to 14.8V). The
are overlaid with the rotor’s requirement in Figure 2.
explored propulsion configurations were selected due to their
prevalence in commercial and research aircraft.

Fig. 3: Propulsion system power and efficiency breakdown

Fig. 2: Motor efficiency curves overlaid with rotor RPM for an ESC and BLDC motor.
requirement.
A typical electric propulsion system is shown in Figure 1, and
Once the motor’s performance is factored into the design, it consists of a DC power supply, an ESC, and a BLDC motor.
is clear that motor 2 is more efficient at the rotor’s operating Shown in Figure 3 is a simplified block diagram representing
RPM. Now a design tradeoff is carried out between motor ef- the different powers in the setup: DC power PDC , AC power
ficiency and motor mass, as more efficient motors are heavier PAC , and mechanical power Pm . The DC power supply emits
2
a steady voltage and current signal to the ESC, the ESC mod- Literary Review
ulates this power into the AC voltage and current signals for
the BLDC motor, and the BLDC motor converts AC power Despite the growing popularity of electric aircraft, few liter-
into mechanical power. The expressions for DC power and ary papers address BLDC motor and ESC performance. The
mechanical power are: most complete work was by researchers at the Army Research
Lab in 2013 (Ref. 4). Their work used a dynamometer to vary
PDC = VDC IDC (1) the torque applied to small (< 10 gram) BLDC motors, and
showed that BLDC motors have similar performance relation-
ships (in terms of torque-RPM, mechanical power, and effi-
Pm = Qω (2) ciency) to traditional brushed motors. Different BLDC motors
DC, AC, and mechanical power analyzers are used to isolate were characterized with the same Castle Creations Phoenix 6
the performance of the ESC and BLDC motor. As a result, ESC. An important conclusion from (Ref. 4) is that ESC ef-
there are three different efficiencies one may discuss for this ficiency decreased with increasing current for a given throttle
system: the ESC efficiency ηESC , the motor efficiency ηm , setting. Although it gave good insights into BLDC motor op-
and the overall system efficiency ηsys . These definitions are eration, their work did not rigorously study the effect of ESC
expanded below: performance, nor was any modeling attempted.
There are numerous electrical engineering and BLDC mo-
PAC PAC tor textbooks that describe 3 phase BLDC motor operation
ηESC = = (3)
PDC VDC IDC (Refs. 5–7). They outline the commutation timing of BLDC
motors, their construction, operation, and key performance
Pm Qω parameters. Other textbooks (Refs. 8, 9) focus on DC/DC or
ηm = = (4)
PAC PAC DC/AC inverter modeling, and are useful for providing back-
ground information on motor controller modeling. A focal
PAC Pm Qω
ηsys = ηESC ηm = = (5) point of the work in this paper is to combine motor and motor
PDC PAC VDC IDC controller sources together to capture both the ESC and the
where PDC , PAC , and Pm are the DC, AC, and mechanical pow- BLDC motor with a single, concise model. Analytically rep-
ers, whereas ηESC , ηm , and ηsys are the ESC, motor, and sys- resenting both the ESC and BLDC motor is critical for electric
tem efficiencies, and VDC , IDC , Q, and ω are the DC voltage, aviation design, as it allows designers to evaluate trade-offs in
DC current, motor output torque, and motor rotational speed, an informed manner.
respectively. Mills and Datta (Ref. 10) at the University of Maryland pro-
Each efficiency relates to a particular transformation of power vide a more complete model on brushed and BLDC motor us-
in the propulsion system. An ESC transforms DC power into age in electric aircraft design. Their work describes the pair-
AC power to run the motor, whereas the BLDC motor trans- ing of rotor torque and RPM requirements with electric motor
forms AC power into mechanical power. Increasing the ESC operation. Specifically, given a set of motor and rotor param-
efficiency allows for more power to be available to the motor, eters, they were able to predict the required motor power and
meaning that a different performance is expected for the same thus size the remainder of their aircraft. Both a brushed and
amount of DC input power. Typically, researchers use equa- brushless motor model are discussed in their work. Combin-
tion 5 when discussing propulsion system performance, only ing a rotor’s operating point with a properly selected electric
relating the DC input power to the mechanical output power. motor can have significant impacts on the final aircraft design.
This approximation does not distinguish between motor and In their case, they developed an 1800 lb twin-rotor helicopter
motor controller inefficiencies during the power conversion that was capable of hovering for 24 hours.
process. By properly recording the AC power of the system, Researchers at the University of Sydney (Ref. 11) in 2017
researchers can use this information to search for high effi- studied the ESC efficiency by measuring the input DC power
ciency ESCs and BLDC motors. and the output AC power of an ESC. Their setup consisted of
This paper describes the efforts to experimentally capture testing multiple ESCs with a single BLDC motor and rotor.
BLDC motor and ESC performance by varying the load ap- Critical to their work was an AC power analyzer, which was
plied to a motor and recording the power at different operating inserted in between the ESC and BLDC motor. By record-
points. Experimental data is then used to derive an analytical ing the DC power entering the ESC and the AC power exit-
model to explain the nature of the motor response. To accom- ing the ESC, the ESC’s performance can be evaluated. With
plish this, this paper describes the basic operation of motors, this, they showed ESC efficiency increased with higher throt-
the experimental setup, and the analytical model. The remain- tle settings and that ESC power loss increased with increasing
der of this section describes a review of available references DC voltage. A key result from their work was modeling the
on the brushed motor model and ESC operation. Next, the ESC as a single MOSFET transistor consisting of a conduc-
experimental setup to record the system’s performance is dis- tive and switching power loss. Similar work was carried out
cussed. Then, the experimental data is used to develop a first in (Refs. 12–14).
principles based model for ESCs and BLDC motors. Finally, Additional information on Metal Oxide Silicon Field Ef-
conclusions are drawn and future works are pointed out. fect Transistors (MOSFETs) and ESC modeling is given in
3
(Refs. 15–17). These works describe the conduction and nets. Based on Figure 4, the sum of voltages around the loop
switching power losses of a MOSFET. Switching the state is:
of the ESC (i.e., from non-conducting to conducting, or vice VDC = IDC Rm + KE ω (8)
versa) takes a small amount of time to achieve. During this
Output torque is related to the DC current, as in equation 7,
rapid transition, power loss occurs. In order to minimize the
but must be augmented with an additional torque KT Io to over-
power loss from the switching event, the number of switching
come the static friction of the motor:
events should be minimized, which corresponds to keeping
the MOSFET always conducting. This observation is consis- Q = KT (IDC − Io ) (9)
tent with the experimental results on ESC efficiency shown
from (Refs. 4, 11, 12). Adapting the information from MOS- Equations 8 and 9 are considered to be the fundamental equa-
FET power loss into ESC efficiency performance is not well tions of an electric motor. Equation 8 relates the motor’s ro-
documented, but is a core focus of the following sections. tational speed with the voltage of the motor, and equation 9
The work in this paper studies BLDC motors and ESCs. How- shows that the current flowing though the motor relates to the
ever, before describing an apparatus to study brushless mo- torque of the motor. Motor constants KT and KE are used to
tors, basic motor performance shall be discussed. relate the input electrical quantities of VDC , IDC , and PDC to
the output mechanical properties of Q, ω, and Pm . These two
Brushed Motor Model equations can be combined to determine the linear relation-
ship between RPM and torque:
Brushed motors have simple models that allow for powerful
KT KT KE
insights to be drawn into the basic motor performance. An Q= [VDC − Io Rm ] − ω (10)
understanding of the brushed motor model is necessary for un- Rm Rm
derstanding motor operation. The tools and concepts needed which is plotted in Figure 5. Equation 10 states that the Q vs
to analyze brushed motors will be used to analyze BLDC mo- ω relationship is linear in nature, with a slope related to the
tors. Motor theory relies on drawing an equivalent circuit di- parameters of the motor and a x-intercept related to the DC
agram, which represents the electromechanical properties of supply voltage VDC .
the motor controller and motor. Electromechanical properties
of motor operation can be summarized (Refs. 5, 6, 14):
VE = KE ω (6)

Q = KT IDC (7)
where ω is the motor’s rotational speed, VE is the induced
voltage generated in the motor’s coils from a rotating mag-
netic field, IDC is the DC current flowing through the motor’s
coils, and Q is the mechanical torque of the motor. Motor
constants KT and KE relate electric properties to mechanical
properties for a motor, and are elementary constants in mo- Fig. 5: Left: Torque vs RPM relation. Right: Torque vs
tor theory. They are functions of the magnetic field strength, current relation.
material design of the motor, number of turns in the induc-
tors, and overall electrical design of the motor. To demon-
strate the key concepts of motor performance, an equivalent There are two key points along the Q vs ω line: the stall torque
brushed motor circuit (Refs. 5, 6, 14) is shown in Figure 4. Qs and the no-load speed ωNL . Stall torque Qs is defined as the
torque generated from the motor when the shaft is no longer
rotating, and is the maximum torque the motor can produce.
Such a torque is useful for preventing mechanical rotation,
for example holding robot arms at constant angles. The other
point is the no-load speed ωNL , which is the RPM that the mo-
tor shaft spins when no load is applied to the motor. This is the
maximum RPM that the motor shaft spins. Both of these are
significant as they relate to the maximum performance of the
motor: either maximum rotational speed or maximum load.
Fig. 4: Simplified brushed motor electrical circuit, the arrows Power output from the motor is the product of the mechanical
indicate component voltage drop. output torque with the mechanical RPM, given by equation 2,
and can be reformulated, using equation 10, as:
A simplified steady-state circuit contains a motor resistance KT KT KE
Rm and an induced back voltage KE ω from the rotating mag- Pm = ( [VDC − Io Rm ] − ω)ω
Rm Rm
4
 KT KT KE 2
Pm = [VDC − Io Rm ]ω − ω (11)
Rm Rm
Equation 11 is plotted in Figure 6, revealing a concave down
parabola. With this, the maximum power rotational speed
ωmax P can be found by differentiating equation 11 with re-
spect to ω:
dPm KT KT KE
= [VDC − Io Rm ] − 2 ωmax P = 0
dω Rm Rm

1 VDC − Io Rm ωNL
ωmax P = ( )= (12)
2 KE 2 Fig. 7: System powers and efficiency vs RPM.
Maximum power lies at half of the no-load speed ωNL , when 75% for brushed motors. A typical plot of a motor’s efficiency
the distribution between torque and RPM is balanced at half of curve is shown in Figure 7. Motor performance is augmented
the maximum speed ωNL and at half of the maximum torque by changing the voltage of the motor, which is done using the
Qs . throttle setting.
The motor equations thus far have assumed that the motor
was operating at the full supply voltage VDC . A more real-
istic model for motor operation is shown in Figure 8, which
includes a voltage regulating device. The voltage regulator
scales the output voltage proportionally to some input signal,
called the throttle TR . With a lower throttle setting, the voltage
acting across the motor is reduced. For example, if the throttle
setting is 60% (TR = 0.6), 0.6VDC acts across the motor.
User throttle TR is a value between 0-1, and has been normal-
ized appropriately. The throttle is analogous to the gas pedal
on a car, depending on how far down the gas pedal is pressed,
more or less fuel enters the engine. With more fuel, the en-
gine produces more power, and the car responds appropriately.
Fig. 6: Motor output power vs RPM. For a BLDC motor, increasing the throttle changes the voltage
acting on the motor.
Note from Figure 6 that at both the no-load speed ωNL and at
stall torque Qs the mechanical output power is 0, again cor-
responding to the limits of the motor. Since the input and
output powers have been identified, the overall efficiency can
be evaluated.
The ratio of mechanical output power to DC input power is
the system efficiency:
Pm Qω
ηsys = =
PDC VDC IDC
Fig. 8: Simplified brushed motor diagram with a voltage
regulator.
KT ( VDCR−K
m
Eω
− Io )
ηsys = VDC −KE ω
(13)
VDC ( Rm ) Changing the voltage seen by the motor is accomplished by
augmenting the throttle TR into the voltage regulator. This de-
Maximum efficiency of the system occurs at ωopt , and can be
creases the voltage acting across the motor, and thereby mod-
found by differentiating equation 13 with respect to the RPM:
ifies the performance of the motor. The influence of throttle
dηsys KT (KE2 ω 2 − 2KE VDC ω + (VDC
2 −I R V )
o m DC
TR is apparent when an equivalent circuit, as shown in Figure
= =0 8. Applying Kirchhoff’s voltage law to Figure 8 yields:
dω VDC (VDC − Rm ω)
√ VDC TR = IDC Rm + KE ω (15)
VDC − Io RmVDC
ωopt = (14) the torque-current relation, equation 9, remains unchanged for
KE
a given VDC . However, the user throttle TR appears when eval-
The optimal RPM for system performance is typically be- uating the motor’s performance:
tween 75% to 95% of the motor’s no-load speed, and can KT KT KE
have a maximum efficiency of 95% for brushless motors and Q= [VDC TR − Io Rm ] − ω (16)
Rm Rm
5
 KT KT KE 2
Pm = [VDC TR − Io Rm ]ω − ω (17)
Rm Rm

KT ( VDC TRR −K
m
Eω
− Io )
ηsys = (18)
VDC ( VDC TRR −K
m
Eω
)
changing the input throttle to the motor shifts the plots in a
predictable manner. The change in performance for differ-
ent throttles is shown in Figures 9, with throttle TR increasing
from left to right. Fig. 10: Six MOSFET configuration for an ESC.

Six MOSFETs are used to commutate a BLDC motor, with 2

conducting at any time. If Q1 and Q5 are conducting, then cur-
rent forms the DC supply, through Q1 , into the motor by line
A, and out of the motor from line B, through Q5 , to ground.
During this time, iA > 0 and iB < 0, which changes the orien-
tation of the magnetic field inside the motor. One half of the
coils are energized with positive current that pulls on the rotor
via magnetic forces. The second half of the coils are ener-
gized with negative current, resulting in a magnetic repulsion
which pushes the rotor. During a commutation event, the rotor
reaches a point where the active MOSFETs must be changed.
Fig. 9: Torque-RPM and mechanical power-RPM curves For example, keeping Q1 active and switching the second ac-
with different throttle settings. tive MOSFET from Q5 to Q6 , current no longer exits via line
B but now exits though line C. Which MOSFETs are active
Throttle has a linear relationship in the torque-RPM curves, depends upon the rotor’s position, with a typical commutation
and can be thought of as shifting the plots left or right. In- sequence shown in Table 1.
creasing throttle allows for a higher no-load speed ωNL and a
higher stall torque Qs to be achieved, which consequently in- Table 1: Six step, 120o conduction commutation sequence
creases the maximum power and RPM. Maximum power still used for BLDC motors.
occurs at half of the no-load speed ωNL , and since the no-load
speed increases with voltage, so must the maximum power θe [deg] Active MOSFETs iA iB iC
point by equation 12. 0 − 60 Q1 , Q5 + − 0
The user, or closed loop controller, changes the throttle value 60 − 120 Q1 , Q6 + 0 −
to the voltage regulator. This regulator acts like a valve, and 120 − 180 Q2 , Q6 0 + −
changes the voltage available for the motor. Changing the 180 − 240 Q2 , Q4 − + 0
motor voltage allows for more or less power to be delivered to 240 − 300 Q3 , Q4 − 0 +
the motor, and has 2 effects: faster RPM for the same load, or 300 − 360 Q3 , Q5 0 − +
more current (and thus more torque) for the same RPM. For
a BLDC motor, changing the motor’s voltage is achieved via Two MOSFETs control a commutation event and by alternat-
the ESC. ing the two out of six MOSFETs, the rotor is kept spinning.
Most low cost commercial ESCs and BLDC motors use 120o ,
Electronic Speed Controller Operation 6-step commutation methods. The 120o refers to the conduc-
tion interval of the MOSFET, meaning each MOSFET is on
ESCs commutate, or spin, the BLDC motor by electronically for 1/3rd of an electrical revolution, with two active at any
regulating the direction of current into the motor. By alter- one time. Coils are connected together, in an either a Y or a
nating which phases of the BLDC motor are energized, the ∆ configuration, to allow the same current to pass through the
direction and strength of the magnetic field inside the BLDC both positively and negatively charged coils. Six step refers to
motor is controlled. This develops a push-pull action between six commutation intervals, required for one electrical revolu-
the motor’s permanent magnets and the energized electromag- tion, θe = 0o to 360o , with θe being the electrical angle. The
netic coils, causing the rotor to chase the rotating magnetic electrical angle θe is related to the mechanical shaft angle θm
field. MOSFETs are used to regulate the direction of cur- by the number of pole pairs NP of the motor:
rent into the BLDC motor. By changing which MOSFETs are
θe = NP θm (19)
conducting, the direction of current into the motor can be con-
trolled. A typical ESC and MOSFET configuration is shown The primary means of controlling induction motors are with
in Figure 10. MOSFETs, which act as electronic switches that open or
6
close rapidly based on a control signal. These MOSFETs
can be combined into useful arrangements, such as H-bridges
or ESCs, that allow for precise control of a motor. A MOS-
FET consists of three terminals: a drain, a gate, and a source,
shown in Figure 11.

Fig. 12: MOSFET switching waveforms for on and off

operation. The MOSFET is switched on between ti,on and tv, f ,
whereas the MOSFET is switched off between ti,o f f and ti, f .

Fig. 11: MOSFET model.

Switching, as the name implies, occurs when the MOSFET is
switched from on (i.e., non-conducting to conducting) and off
When a sufficient voltage is applied to the gate VGS , the MOS- (i.e., conducting to non-conducting). A detailed description of
FET will allow current Irms to flow from the drain to the the nature of MOSFET switching can be found in (Refs. 15,
source. Power loss modeling of a MOSFET is critical to un- 16). To calculate the switching power loss, both the energy
derstanding ESC operation. Both the act of switching the per switch and total number of switching events per second
MOSFET on and off, as well as the act of allowing current must be evaluated. From figure 12, the on time is defined as
to flow results in power losses. Power loss of a MOSFET ton = tv, f −ti,on . Energy consumed during the on transition can
contains two primary components (Refs. 11, 15–17): be calculated by examining the area under the VDS Irms curve
from Figure 12:
1. A linear Irms component - related to the rapid switch-
ing on and off of the MOSFET. Referred to as switching 1
Esw,on = VDS Irms (tv, f − ti,on )
losses Psw . 2
2 component - related to the conductive
2. A quadratic Irms 1
power loss of any physical component. Referred to as Esw,on = VDS Irmston (23)
2
conduction losses Pcond . Energy lost during the on transition per second Psw,on can be
found by:
These losses can be written as:
Psw,on = Esw,on × (number o f switches each second)
PMOSFET = Psw + Pcond (20)

1
1 Psw,on = VDS Irmston fsw (24)
Psw = VDC Irms (ton + to f f ) fsw (21) 2
2
where fsw is the switching frequency, and is set by either the
2
Pcond = Irms R (22) user or closed loop controller.
Similar logic is used to determine the switching off losses:
where ton ,to f f , and fsw are the time to switch on in seconds,
time to switch off in seconds, and switching frequency in Hz, 1
respectively. Switching losses originate with a changing volt- Psw,o f f = VDS Irmsto f f fsw (25)
2
age and current acting across the MOSFET. Resistive losses
are evident in all physical conducting elements. Diode losses Here, to f f is the time for the MOSFET to stop conducting, ie
and capacitive effects have also been included in some mod- between ti,o f f and ti, f . With the switching on and off power
els (Refs. 15, 16). Favorable qualities to have in a MOSFET modeled, the total switching loss is:
are fast switching times, low forward voltage drop, low on
resistance, and high breakdown voltage (Ref. 9). Psw = Psw,on + Psw,o f f

Transitioning from no current flowing to current flowing

across the MOSFET is not instantaneous. A switching event 1
Psw = VDS Irms (ton + to f f ) fsw (26)
takes some finite time and energy input to occur, which can be 2
found be examining the changing waveforms for a MOSFET, Conduction losses occur between tv, f to ti,o f f , that is: when
shown in Figure 12. the MOSFET is no longer switching and current is flowing.
7
 tains a large amount of chattering, resulting in a disorganized
appearing waveforms. On the other hand, the 100% throttle
waveforms look relatively smooth and continuous.
The difference in appearance of these signals relates to the
amount of time the MOSFETs are open. At a lower throttle
setting, more switching occurs and thus the signal appears to
be noisy in comparison to the 100% throttle setting. From this,
3 conclusions can be drawn: increasing throttle decreases the
number of switching events for the same amount of time, in-
creasing throttle increases effective motor voltage, and wave-
forms are ‘ideal-like’ at 100% throttle. These results will be
useful when determining the BLDC motor and ESC model.
Now that a basic understanding of motor operation and per-
formance has been described, the experimental apparatus to
record BLDC motor performance is discussed next.

EXPERIMENTAL SETUP
A dynamometer is an instrument used to study the perfor-
mance of a motor, which consists of a load varying device
and sensors to monitor the motor’s performance. The main

Fig. 13: Instantaneous voltage and current waveforms over 2 Fig. 14: Key components of a dynamometer.
cycles for a BLDC motor at VDC = 7.2V . Note the difference
in time between waveforms of different throttles. component of any dynamometer is the brake, which is respon-
sible for absorbing the motors mechanical output power and
Total power loss during MOSFET operation is the sum of the for changing the load on the motor. Recording the power at
switching and conductive losses, and is stated as: each operating point allows for performance plots of the mo-
1 2 tor to be experimentally obtained. As a result, a Magtrol mag-
PMOSFET = VDS Irms (ton + to f f ) fsw + Irms R (27) netic hysteresis brake (HB-10) was acquired to vary the load
2
on the motor. Construction of the brake consists of a rotating
A single MOSFET contains both a linear and quadratic com-
pole structure and a coil, shown in Figure 15. Due to the con-
ponent of power loss with the rms current. If there were
struction of the brake, (Ref. 18): mechanical torque is only a
only one MOSFET to block the entire supply voltage, then
function of the input current to the brake, Q = f (Ibrake ), and
VDS = VDC . For a given amount of Irms , it is evident from
is easily controlled. Unlike an electric motor, the mechani-
equation 27 that minimizing switching losses occurs when fsw
cal torque of the brake is independent of operational RPM,
= 0, indicating that the MOSFET only has conductive losses.
making a magnetic hysteresis brake an attractive options for
In an ESC, the switching frequency fsw shows a strong corre-
studying electric motor performance. The brake’s input cur-
lation to the input throttle TR . For an actual ESC, the effect of
rent to output torque relationship can be obtained, as shown
MOSFET switching on the trapezoidal voltage and rectangu-
in Figure 15 (Ref. 19).
lar current waveforms are shown in Figure 13.
A snapshot of the instantaneous waveforms over approxi- Augmenting the torque on the motor via the brake changes
mately 2 electrical cycles for throttle settings of 40%, 70%, the motor’s torque, RPM, and mechanical power. A change
and 100% demonstrates the effect MOSFET switching has on in mechanical power also alters the AC and DC power of the
the waveforms. When comparing individual waveforms, the system. Recording the mechanical, AC, and DC powers at
most obvious result is that there is less chattering in the wave- each operating point allows the performance of the system
forms as the throttle increases. For example, 40% throttle con- to be evaluated. This operation is then repeated for different
8
 Referencing a line voltage with respect to another line voltage
is referred to as a line-to-line voltage. Voltages in this system
can be referenced to 3 different points: the DC ground N, the
system’s neutral point N ∗ , or another line voltage. Line-to-
line voltages of vAC and vBC are combined with currents iA and
iB to find the instantaneous power pinst in the 3-phase system
via the 2-wattmeter method:

pinst = vAC iA + vBC iB = pA + pB (30)

Fig. 15: Exploded view of a magnetic hysteresis brake.
Image on the left is from (Ref. 19). Image taken
with pA and pB indicating the wattmeter readings for lines A
from (Ref. 20).
and B, respectively. It is important to emphasize that this is
throttle setting and DC voltage levels. In the next section, the not the power flowing through line A and line B, since they are
power analyzers used to determine the system’s performance referenced to line C. AC power can have inductive or capac-
are discussed. itive elements that result in an oscillatory power waveform,
meaning multiple types of power exist in an AC system. Ac-
Power Analyzers tive power is the average power transferred over one wave-
form cycle and is referred to as the real power in the AC sys-
Table 2: Dynamometer power analyzer’s sensors and ranges. tem (Refs. 8, 21, 22). Therefore, active power is the proper
AC power to use for efficiency analysis. Together the mea-
Power Analyzer Sensor Rating surements pA and pB allow for the active AC power PAC to be
DC Current, IDC + 30 A calculated as the average of the instantaneous power pinst :
DC, PDC
DC Voltage, VDC + 20 V Z T
1
AC Current, iA iB +/− 30 A PAC = pinst dt
AC, PAC T 0
AC Voltage, vAN vBN vCN + 20 V
RPM, ω 25,000 RPM
Mechanical, Pm
Torque, Q +/− 50 in-oz 1
Z T
PAC = (vAC (t)iA (t) + vBC (t)iB (t))dt (31)
T 0

DC, AC, and mechanical power analyzers were developed in- where T represents the period of the AC signal.
house using Labview DAQs. Two DAQs were required, as
the sampling frequency of the system operated at 250 kHz Another important concept when discussing AC circuits is the
to properly capture the AC signals of the BLDC motors. Of root-mean-square (rms) value of a signal. This allows for a
key importance was developing an AC power analyzer that single, steady-state, value to describe the intensity of a signal.
could adequately sample the motor’s trapezoidal voltage and For a line-to-line voltage signal, for example vAC (t), the rms
rectangular current waveforms. value of the signal is:
s
Z T
1
VAC = rms{vAC (t)} = v2AC (t)dt (32)
T 0

Although the AC signals are oscillating with different peak

values and with changing frequencies, the rms value of a
signal captures ‘the equivalent DC amplitude’ of a signal
(Ref. 22).
Fig. 16: Left: Generic 3-phase AC system. Right: Phasor
diagram for a 3 phase system. Each waveform is 120o
electrical degrees apart. Final Setup

The two-wattmeter method was used to experimentally record Multiple iterations of the experimental dynamometer were re-
the AC power entering the motor (Refs. 12, 21, 22), and re- quired before necessary sampling requirements were satisfied.
lies on recording the three phase voltages vAN (t), vBN (t), and The resulting dynamometer was a modular, reconfigurable ex-
vCN (t) along with two of the line currents iA (t) and iB (t). perimental apparatus allowed for testing different BLDC mo-
vAC (t) = vAN (t) − vCN (t) (28) tors and ESCs under a variety of operating conditions. Two
NI-USB 6251 DAQs and two SC-2345 signal conditioning
devices were required to accurately capture the motor’s AC
vBC (t) = vBN (t) − vCN (t) (29) waveforms that were previously shown in Figure 13.
9
 Both of these observations are consistent with the brushed mo-
tor model. An important distinction between the DC and AC
motor model for BLDC motors is the evaluation of AC power
in a 3-phase system. In the brushed motor model, power to
run the voltage regulator and brushed motor are both DC. In
the presented AC motor model, AC power runs the brushless
motor and DC power runs the ESC, shown in Figure 3. Pri-
mary objectives for the proposed model are:

1. Determination of an equivalent AC motor model, which

relates the inputs VLL,rms , Irms , and TR to the both the
ESC’s and BLDC motor’s performance.

2. Prediction of the torque-RPM curve and mechanical out-

put power.

3. Calculation of the required AC power to run the motor.

4. Generation of the ESC’s DC to AC power conversion

characteristics, resulting in the steady-state DC current
from the DC power supply.

Fig. 17: Final dynamometer configuration. This model adapts the concepts from the brushed motor model
for BLDC motors, by mapping DC values of VDC and IDC into
AC values of VLL,rms and Irms . Key points of the motor model
The motor was mounted directly to the torque sensors, which include predicting the torque-RPM relation, properly captur-
in turn was mounted to a physical structure, shown in Figure ing the torque-current relationship, and accurate prediction of
17. A central shaft connected the BLDC motor to the brake the AC power required to run the motor. A novel mapping be-
via bearings and flexible shaft connectors. Along this central tween VLL,rms , Irms , and PAC is found by examining the trape-
shaft was an optical tachometer that recorded the shaft’s rota- zoidal voltage and rectangular current waveforms of BLDC
tional speed. This setup was used to experimentally record motor.
BLDC motor and ESC performance before modeling work
was attempted. Initial testing of BLDC motors revealed ex-
Waveform Evaluation
perimental performance plots similar to brushed motor per-
formance plots, indicating that a similar model can be used
Waveforms entering a BLDC motor are time varying in na-
for both brushed and brushless motors.
ture, and are shown in Figure 18. The derivation of the
AC power entering a BLDC is built upon the following as-
SYSTEM MODELING sumptions: The voltage waveform is trapezoidal (Refs. 5–7),
the current waveforms are rectangular (Refs. 5–7), and that
Once sufficient experimental data was recorded with the dy- power, for any waveforms, can be calculated using the 2-
namometer, an analytical model of the system was created wattmeter method (Refs. 8, 21, 22). By using the exact wave-
using first principles based engineering. This model relies form shapes of BLDC motors, the AC power can be calculated
on evaluating the root-mean-square (rms) values of voltage from recorded rms values for VLL,rms and Irms . Note that ex-
and current waveforms, adapting the bushed motor model for perimentally the 2-wattmeter method is used to determine the
BLDC motors, and developing equations that capture the DC AC power, but this section describes a method to model the
to AC power transformation across an ESC. AC power using rms values VLL,rms and Irms .
This section describe model of an electric propulsion systems
by using a novel equivalent circuit approach. To accomplish
this, during motor testing the active power PAC , as well as
other key parameters related to AC power, are recorded:
VAB +VBC +VCA
VLL,rms = (33)
3
Fig. 18: Expected voltage and current waveforms entering a
IA + IB BLDC motor.
Irms = (34)
2
It will be shown that VLL,rms maps to the motor’s mechanical From Figure 18, it can be seen that the voltage and current
rotation ω, whereas Irms relates to the motor’s output torque. waveforms oscillate between peak voltage Vpk and peak cur-
10
rent I pk values. To determine the AC power in these wave- Now that PAC is written in terms of peak voltage and current
forms, the waveforms for vAN ∗ , vBN ∗ , and vCN ∗ are mathe- values, equations 36 and 38 can be used to relate these quan-
matically defined and then subtracted from one another to tities to VLL,rms and Irms values.
find vAC and vBC . Once piece-wise functions for vAC and r
vBC exist, the rms values for any piece-wise waveform can 27
PAC = VLL,rms Irms ≈ 1.64VLL,rms Irms (41)
be found (Ref. 8). The relation between the line-to-line rms 10
voltage and peak voltage is:
Equation 41 is similar in nature to DC power PDC = VDC IDC .
For AC power, voltage and current must be evaluated as rms
r
20 p
VAC = Vpk (35) quantities, and the factor of 27/10 can be thought of as an
9
exchange rate between rms values and active power for 3-
Due to the balanced loading condition between vAB , vBC , and phase AC circuits. Values for VLL,rms and Irms are not only
vCA , the rms values for all line-to-line voltages are equal to important for the power content of the AC signal, but are used
one another: to model the motor’s mechanical performance as well.
r
20 Brushless DC Motors
VAB = VBC = VCA = VLL,rms = Vpk (36)
9
An equivalent motor circuit for BLDC motors and the ESCs
Equation 36 can be extended to include any phase voltage sub-
can be found by modifying the existing DC motor model, Fig-
tracted from one another, the resulting rms value is the same
ure 8, to include AC rms quantities. This augmented model
as VLL,rms . This is most notable for:
serves to predict the performance metrics of the motor and the
VLL,rms = VAC = VBC (37) required AC and DC powers of the system. In the next section,
this motor model is expanded to include the ESC.
and describes the AC line-to-line voltage acting across the An actual circuit for a BLDC motor requires 3-phases to com-
motor. The rms value of iA can be computed in a similar man- mutate the motor. However, it has been shown that the AC
ner to vAC : r power to run the motor is only related to VLL,rms and Irms ,
2 meaning these values can be used to make a simplified ‘quasi-
IA = I pk (38)
3 single phase’ circuit, like a DC circuit. With this, the pre-
Again, these are equal to the line rms current Irms due to the sented motor model initially used the circuit diagram in Figure
balanced loading condition: 20 to capture the electromagnetic properties of the brushless
motor and ESC system.
Irms = IA = IB (39)

Once the waveforms have been evaluated, an expression for

the AC power using VLL,rms and Irms can be determined. Using
the 2-wattmeter method, the instantaneous power of the sys-
tem is given by equation 31. Combining the piecewise wave-
forms for the brushless motor’s voltage and current signals are
shown in Figure 19.

Fig. 20: Initial ESC and BLDC circuit.

The equivalent model has the following features added: a mo-

tor resistance Rm , a motor back voltage constant KE , and a
motor torque constant KT . By reducing the three phase sys-
tem into an equivalent DC type circuit, the motor’s complex-
ity is reduced, and evaluating its performance can be refer-
enced against the brushed motor model. A discussion on how
AC electrical parameters influence the motor torque-RPM line
starts by referring to the equivalent DC motor model that was
Fig. 19: Total instantaneous power entering a BLDC motor. initially presented:
VDC TR = IDC Rm + KE ω
From Figure 19, the total instantaneous 3-phase power enter-
Q = KT (IDC − IDC,o )
ing the BLDC motor from the ESC is:
DC values model brushed motors, however AC quantities are
PAC = 2Vpk I pk (40) needed to work with brushless motors. In the proposed model,
11
DC voltage VDC is replaced with the AC voltage Vrms and the From Figure 22, it can be seen that for a given throttle setting
DC current IDC is usurped with the AC current Irms . With this, TR , all voltages following the same linear trend with Irms :
the DC model is augmented to: 
VLL,rms = VLL,rms  − RESC Irms (44)

Irms =0
VLL,rms = Irms Rm + KE ω (42)
where RESC is the equivalent ESC resistance. Phase voltages
are measured directly at the output of the ESC, indicating that
some voltage drop occurs across the ESC’s internal resistance.
Q = KT (Irms − Io ) (43)
These qualitative relations between VLL,rms and Irms are sum-
where VLL,rms is the line-to-line rms voltage entering the mo- marized as:
tor, Irms is the rms input current to the motor, and Io is the
1. Motor rms voltage VLL,rms is a function of throttle TR , rms
required rms current to overcome the static friction of the mo-
current Irms , and the DC voltage VDC .
tor, all AC quantities. Equations 42 and 43 relate motor volt- 
age to mechanical rotation, and rms current to output torque. 2. Initial offset rms voltage VLL,rms 

is only a function
Equation 43 is shown in Figure 21 using predicted and exper- Irms =0
of the DC Voltage VDC and the input throttle TR , and has
imental data.
a value less than VDC .
3. Equivalent ESC resistance RESC is the same for all throt-
tle settings.
It was found (Refs. 5, 8) that the maximum theoretical rms
output voltage depends on the connection and commutation
setup. The ESCs studied in this work are of six step, 120o con-
duction commutation type, via Table 1. Based on this setup,
the maximum line-to-line rms voltage VLL,rms from a fixed DC
supply is (Refs. 5, 8):
3
VLL,rms = √ VDC ≈ 0.675VDC (45)
2π
Fig. 21: Q vs Irms for an EMAX 1900KV motor and Similar to the voltage regulator from the DC model, the ef-
MultiStar 30A ESC at VDC = 7.2 V. Note the relation is fective output voltage of the ESC is augmented based on the
analogous to the brushed motor model prediction, Figure 5. throttle command:
 3
VLL,rms  = √ VDC TR (46)

Irms =0 2π
Rms values for voltages and currents can be easily measured
with this, the rms motor voltage can be stated:
in an experiment, but how do these values relate to the DC
voltage and the input throttle? One expects a higher VLL,rms 3
VLL,rms = √ VDC TR − RESC Irms (47)
to be associated with a higher DC voltage and a higher input 2π
throttle, but by how much? To determine the link between rms which satisfies the original stipulations for rms motor voltage.
voltage and input DC voltage and throttle, consider the follow- Overlaying the predicted VLL,rms values from equation 47 with
ing experimentally retrieved data on motor voltages, shown in the original experimentally obtained VLL,rms values from Fig-
Figure 22. ure 22 results in the plot shown in Figure 23.

Fig. 22: VLL,rms vs Irms for an EMAX 1900KV motor and Fig. 23: Predicted and experimentally obtained VLL,rms vs
MultiStar 30A ESC at VDC = 7.2 V. Note that all voltages are Irms for an EMAX 1900KV motor and MultiStar 30A ESC at
below VDC . VDC = 7.2 V.

12
Predicted values for VLL,rms agree well with experiment up to which is similar to equation 10 for the equivalent brushed mo-
90% throttle. Somewhere between 90% and 100% throttle, tor model. From equation 49, the BLDC motor’s output power
the ESC begins to saturate and can no longer provide the re- can be found:
quired 0.675VDC TR voltage. Not all DC voltage is converted Pm = Qω
into AC voltage due to physical losses and thus equation 45
over-predicts the actual rms voltage value at higher throttle
settings. Saturation is a phenomenon that occurs at high con- KT ω 3 KT KE
Pm = [ √ VDC TR −Io (Rm +RESC )]− ω2
trol inputs, and takes place in this system between 90% and Rm + RESC 2π Rm + RESC
100% throttle. Physically, this means that some DC voltage (50)
cannot be converted into AC voltage due to inherit hardware Equations 49 and 50 are plotted against experimentally ob-
limitations in the ESC. Saturation is ESC dependent, as it was tained data in Figure 25. Careful examination of the torque
found that different ESCs had different maximum throttle set- vs RPM relation for a BLDC motor further demonstrates ESC
tings before saturation occurred. saturation at high throttle settings. Note that the horizontal
spacing between throttle settings is a consistent 1230 RPM,
As a result of this upper saturation limit, the model presented whereas between 90% and 100% throttle there is only a 750
here is only valid up to 90% throttle where no saturation RPM displacement. The ESC saturates and cannot produce
occurs and equation 46 is valid. At throttle setting >95% the required rms voltage, and thus the motor has a slower ro-
a nonlinearity occurs, which means that equation 46 over- tational speed from equations 10 and 49.
predicts the amount of voltage available to the motor. An
From equation 50 and 25, the motor’s output power is a con-
area of future work is to further explore the region between
cave down parabola and has a peak power point at half the no-
90% and 100% throttle, because at 100% the absolute lim-
load speed of the motor. With the output mechanical power of
its of the system’s performance can be evaluated. However,
the motor defined, an expression for the required AC power to
within the non-saturated region, the torque-RPM relationship
drive the motor can be determined.
is predicted well. To determine the relationship between volt-
age, throttle, and RPM, equation 42 can be combined with
equation 47:

3
( √ VDC TR − RESC Irms ) = Irms Rm + KE ω
2π

3
√ VDC TR = (RESC + Rm )Irms + KE ω (48)
2π
which yields a more useful form of the AC motor equation.
Here the influence of user throttle and DC voltage is linked
to the physical parameters of the motor, and is similar to the
brushed motor model, equation 15. This modified equivalent Fig. 25: Predicted and experimentally obtained Q vs RPM
motor model is shown in Figure 24, and has been updated to and Pm vs RPM curves for an EMAX 1900KV motor and
include the equivalent RESC . MultiStar 30A ESC at VDC = 7.2 V.

The strength of this equivalent motor model is that it combines

well with the rms AC power relations, equation 41. Given a
load Q and a rotational speed ω, the required voltage VLL,rms
and current Irms can be found, and are related to the active
power PAC in the circuit via their peak values. With the insight
of Figure 24 and equation 47, a modification to the AC power
entering the motor is required to account for the impact of
RESC . Power output from the ESC has an Irms RESC voltage
drop associated with the ESC, and reduces the effective motor
Fig. 24: Final equivalent BLDC motor and ESC circuit,
rms voltage. Therefore, power entering the BLDC motor has
updated to include RESC . 2 R
a Irms ESC power loss. This power loss can be modeled by
combining the rms-AC power relation with the modeled ESC
With the voltage-RPM relation and the torque-current relation voltage drop, equations 41 and 47, to yield:
defined, an expression relating torque to RPM can be found by r
combining equations 43 and 48: 27 3
PAC = ( √ VDC TR − RESC Irms )Irms (51)
10 2π
KT 3 KT KE
Q= [ √ VDC TR −Io (Rm +RESC )]− ω r r
Rm + RESC 2π Rm + RESC 3 27 27 2
(49) PAC = VDC Irms TR − I RESC
π 20 10 rms
13
AC power for a BLDC motor contains 2 components: a posi-
2 R
tive VLL,rms Irms and a negative Irms ESC component. In reality,
2
the Irms RESC is a conductive power loss to the ESC, which will
be explained further in the next section. As the rms current

Fig. 27: Simplified ESC transformer diagram.

Converting DC power PDC into AC power PAC requires some

power PESC , which is lost to the environment as heat. A power
balance relates these quantities:

Fig. 26: Predicted and experimentally obtained PAC vs RPM PDC = PESC + PAC (52)
and Irms vs RPM curves for an EMAX 1900KV motor and
MultiStar 30A ESC at VDC = 7.2 V. which states that the DC power is either lost to the ESC as heat
or is converted into AC power to run the motor. Using equa-
tions 3 and 52, the efficiency of the ESC can be reformulated
increases Irms , less power is available to the motor. Predicted
as:
and experimental results for the rms current and AC power are PAC PESC
shown in Figure 26. ηESC = = 1− (53)
PDC PDC
Representing the BLDC motor as an equivalent AC circuit Increasing ESC efficiency indicates that power lost to the ESC
predicts the motor’s performance well. The next section seeks PESC should be minimized. However, what if the power lost
to model the ESC. to the ESC PESC was exactly 0? Under these conditions, the
ESC is operating with an efficiency of 100%. An ideal ESC
Electronic Speed Controller implies that DC and AC powers are equal to one another:

An ESC is responsible for properly modulating the DC power PDC = PAC for an ideal ESC (54)
into the proper AC waveforms for a brushless motor. A series
of MOSFETs control the commutation interval, based on the
To explore how DC current is converted into AC current, first
rotor’s position. The length of time that the MOSFETs are
an ideal power transformer will be examined. Next, this work
conducting controls the rms voltage to the motor. Objectives
is extended to include power losses associated with a non-
for ESC modeling are to determine: the DC current IDC draw
ideal transformer. A key concept in the transmission of elec-
out of the DC power supply, physical characteristics of the
trical power, or any power, is that the input and output power
ESC that are invariant of the motor, and to accurately model
must remain accounted for:
the power loss of the ESC PESC . Traditionally, the ESC is
neglected in formal analysis. However, this work seeks to Pin = Pout
determine a physics based model of the ESC, and to identify
physical parameters that uniquely describe ESC operation. Vin Iin = Vout Iout
It was shown in a previous section that MOSFETs contain
some power loss related to switching and conduction. These For ideal electrical systems (Ref. 9), a reduction in voltage
MOSFET power losses must be included in the subsequent requires an increase in current such that if Vout < Vin then it
ESC power loss model. From the brushless motor model sec- is guaranteed that Iout > Iin . Exchanging voltage for current
tion, equation 47 has already identified a conductive power stems from the fact that power must remained balanced, and
loss of an ESC. To describe the ESC switching power losses, the electromagnetic system will work to achieve this. Power
first ideal ESC behavior will be examined. Next, actual ESC cannot be created or destroyed, but merely changes form. This
performance will be examined and the difference between the logic can be extended for an ideal ESC, by using the relations
ideal and actual relations represents the total power loss of the for DC and AC power from equations 1 and 51:
ESC. r
Figure 27 shows a diagram that relates the DC input power to VDC IDC =
27 3
( √ VDC TR −  RESC*I 0 )I
 rms rms
the AC output power of an ESC. Here VDC and IDC are aug- 10 2π
mented into VLL,rms and Irms . Based on the equivalent motor
r
circuit: VLL,rms , Irms , and VDC are known, whereas the DC cur- 3 27
rent draw IDC is unknown. VDC IDC = VDC TR Irms (55)
π 20
14
For an ideal ESC, the equivalent resistance RESC = 0, meaning
that no voltage is dropped across the inverter. Equation 55 can
be further simplified:
r
3 27
IDC = TR Irms
π 20

r
IDC 3 27 10
= TR ≈ TR (56)
Irms π 20 9

For a given amount AC current Irms , the required DC current

draw IDC out of the power supply is dependent upon the throt-
tle alone and is independent of the DC voltage. It is also worth
noting that at 90% throttle, the ratio IDC /Irms = 1. Equation
56, referred to as the ideal transformer relation, is plotted in
Figure 28 for all throttles.

Fig. 29: IDC vs Irms for a MultiStar 30A ESC and EMAX 935
KV motor.

The goal now is to determine an expression for the current

ratio, and then deduce an expression for ESC power. By plot-
ting the ratio of IDC /Irms vs throttle at different voltages for
this motor/ESC combination in Figure 30, it is apparent that
these points lie above the ideal transformer relation.
Fig. 28: Ideal IDC and Irms relationship for an ESC.

Note that this IDC /Irms relationship is a straight line, with re-
gion above the line representing a nonideal transformer with
ηESC < 1. Above the ideal line means more DC current
(power) is required to generate the same AC power when com-
pared to a point closer to the ideal line. By examining actual
ESC behavior, an expression for the ESC power loss can be
found.

The concept of an ideal transformer must be extended to in-

clude losses. Experimentally obtained plots for rms current vs
DC current are shown in Figure 29, and reveals plots similar Fig. 30: ESC transformer diagram for the MultiStar 30A
to the ideal transformer relation shown in Figure 28. In both ESC.
Figures, for a given throttle setting, all DC currents lie on the
same line originating from 0. When the throttle changes, the
For all voltages, the grouping of the points is consistent. To
slope of the line changes and all IDC follow the same trend.
describe the actual ratio of points, a simple linear expression
The higher the throttle setting, the more the ratio of DC cur-
is required:
rent to rms current approaches 1. IDC
= C1 TR +C0 (57)
Figure 29 implies that the ratio of DC current to rms current Irms
does not depend on voltage, which was first shown in the
IDC = (C1 TR +C0 )Irms (58)
ideal transformer relation, by equation 56. Experimentally
observed ratio of IDC /Irms is the same regardless of the op- where C1 and C0 are coefficients that describe the transfor-
erating voltage and is located along the same line for a given mation from DC to AC current. Physically, it will be shown
throttle. that C1 and C0 are related to the MOSFET switching operation
15
and timing, first referenced in Figure 11. Note that C1 = 10/9 r r
and C0 = 0 for an ideal transformer. This implies that a more 3 27 27 2
efficient ESC is expected to have a smaller value of C0 . Com- PESC = VDC Irms ((C1 − )TR +C0 ) + RESC Irms
π 20 10
bining the BLDC motor model with equation 57 is shown in | {z } | {z }
switching losses conduction losses
Figure 31. (61)
Equation 61 contains both switching losses and conduction
losses, mimicking the power loss of a single MOSFET from
equation 27. The result of this analysis is that each ESC can be
represented as a single equivalent MOSFET, with some modi-
fications. Equating the switching losses for a single MOSFET
with the switching losses expression for a single ESC reveals:

Psw,MOSFET = Psw,ESC

r
3 27
(ton + to f f ) fsw = (C1 − )TR +C0 (62)
π 20
Fig. 31: Predicted and experimentally obtained IDC and Irms
Recall that for a MOSFET, ton and to f f are set by the physi-
vs RPM curves for an EMAX 1900KV motor and MultiStar
cal characteristics of the device, whereas fsw is controlled via
30A ESC at VDC = 7.2 V.
the duty cycle to the gate of the MOSFET. It was first shown
in Figure 13 that the MOSFET switching frequency fsw de-
creases with increasing throttle setting TR .
Combining the ESC transformer relations with the equivalent
brushless motor model allows one to determine the required Equation 62 reveals that the coefficients C1 and C0 describe
DC current and power. Using the rms current expression, the on and off times of the specific MOSFETs used by the
throttle setting, and equation 57, the DC power can be evalu- ESC. Different ESCs use different MOSFETs, and hence can
ated: be uniquely described by their MOSFET timing coefficients
C1 and C0 . The microcontroller of the ESC regulates the
PDC = VDC (C1 TR +C0 )Irms (59) amount of switching to keep the motor commutating based
of the phase voltages and the user throttle.
which shows that the DC power is correlated to both the throt-
tle and the rms current. Recall that throttle sets the speed of
the motor and the rms current sets the torque of the motor.
Therefore, it becomes pertinent that both variables appear in
the final DC power equation. With AC and DC power deter-
mined, an expression for ESC power can be derived.
An ESC is a non ideal component, indicating that power loss
occurs across the device. The ESC operation section con-
tained a literature review of MOSFET power losses, revealing
that MOSFETs contain both conductive and switching power
losses. The next focus is to arrive at an ESC power loss equa-
tion that contains both a switching loss and a conduction loss,
similar to equation 27. From the BLDC motor modeling sec- Fig. 32: Left: ESC transformer diagram for a MultiStar 30A
tion, it was shown that the motor model has already identified ESC. Right: Predicted and experimentally obtained PESC vs
2 R
an Irms Irms curve for an EMAX 1900KV motor and MultiStar 30A
ESC conductive power loss across the ESC. The re-
maining item to quantify is the ESC’s switching losses. ESC at VDC = 7.2 V.

Considering again the diagram in Figure 24, a power balanced

is used to relate the DC and AC power to the power lost across Plotting equation 62 vs throttle shows how the C1 and C0
the ESC PESC : describe a decreasing switching frequency with an increased
throttle. Similarly, the conductionp
losses are modeled as hav-
PESC = PDC − PAC (60)
ing
p an equivalent resistance of 27/10RESC . The factor
Expressions for AC power and DC power, equations 51 and 27/10 is consistent with having defined the RESC in the pre-
59, are combined with equation 60 to evaluate the ESC power: vious motor modeling section and needing to convert to active
AC power.
r
27 3 A linear transformation relation, specifically equation 57,
PESC = VDC (C1 TR +C0 )Irms − ( √ VDC TR −Irms RESC )Irms seems well suited to describe the conversion between DC and
10 2π
16
AC current and is critical to model correctly. With the powers 2. Simplifying the 3 phase AC circuit into a single phase
of the system modeled, the efficiencies can be calculated us- equivalent circuit was shown to be an effective modeling
ing equations 3 to 5 and compared with experimental results. tool. Physical parameters related to the motor’s perfor-
mance can be applied to this model, allowing for mo-
tors and ESCs to be used in a rigorous design manner.
Four parameters are required to describe the BLDC mo-
tor: KT , KE , Io , and Rm . Unique to this work is the de-
scription of the three ESC parameters: C0 , C1 , and RESC .
3. ESC operation is a function of the user throttle. It was
found that both output rms voltage and rms current are
functions of the user throttle. This allowed for the devel-
opment of an analytical ESC transformer model, which
relates DC power to AC power using the input throt-
tle and motor rms current. Saturation occurs in ESCs
between 90% to 100% throttle, meaning that the motor
Fig. 33: System efficiency vs load, image shows never experiences the full 0.675VDC rms voltage, reduc-
experimental results from (Ref. 4). Image on the right shows ing the maximum ωNL of the motor by 5%.
experimental and predicted results from this work. Note the
4. ESC’s have higher efficiencies at full throttle, as the
linear decrease in ESC efficiency.
switching losses are minimized. Switching losses stem
from the rapid turning on and off of the MOSFET to aug-
A final note is that the model agrees with other experimental ment the output rms voltage and rms current. At 100%
results in this area. Figure 33 shows on the left a plot of ef- throttle, the MOSFETs are placed in an always conduct-
ficiency vs torque from (Ref. 4), and on the right information ing mode, indicating that only switches related to com-
collected and predicted on ESC, motor, and system perfor- mutating the motor occur. Additionally, switching losses
mance from this work. In addition to validating other’s work, increase with increasing DC voltages.
this paper contains an explanation as to ESC power loss and
operation in a single, concise model. 5. Maximum efficiency points for brushless DC motors oc-
cur at high RPM and low torque. ESC efficiency gen-
erally decreases linearly with rms current, making the
CONCLUSIONS
overall maximum efficiency for the system occur when
An custom dynamometer was developed in house to charac- the motor is operating at maximum efficiency. Selection
terize the performance of BLDC motors and ESCs. By record- of ESC changes the motor rms voltage, with higher rated
ing the DC, AC, and mechanical power through the system, ESCs having a lower RESC , associated with more expen-
efficiency metrics can be obtained for the ESC, BLDC motor, sive MOSFETs.
and overall system. 6 motors, with RPM/V ratings between
920 to 2500, were operated with 3 ESCs at DC voltages be- Areas of future work are to create a catalog containing BLDC
tween 7.2 V to 14.8 V. Key features of the dynamometer are motor and ESC parameters and to develop a formulation to
use of a 2 DAQ setup to record the AC power entering the match rotor’s operating conditions with high efficiency motor
BLDC motor using the 2-wattmeter method. and motor controllers. A sizing methodology can be devel-
A total of 34 combinations of ESCs, BLDC motors, and VDC oped similar to (Ref. 23) for optimal pairing between motors,
were tested using the custom dynamometer. From the results motor controllers, and rotors for a given reference flight con-
of these experiments, an analytical model was developed to dition.
predict motor performance with different ESCs and BLDC With an improved understanding of electric motors and mo-
motors. Significant work to model the ESC, which is typi- tor controllers, the original question of aircraft sizing can be
cally neglected in research, shows that 2 primary power losses addressed. From this research, a catalog of BLDC motor and
occur: a switching loss and a conduction loss. ESC performance parameters exist to aid designers in select-
From the sets of experimental data and modeling, the follow- ing the proper motor for their application. Informed decisions
ing conclusions can be drawn: can be made concerning battery voltage, rotor selection, and
motor performance and their effects on aircraft operation. As
1. A first principles analytical model was developed to un- a result, it is envisage that the future electric aircraft designs
derstand the type of power losses that occur within an will be able to fly longer and further than current designs.
ESC and BLDC motor system. A model of brushed DC
motors was modified to include AC voltage and current
quantities that better describe the inputs to the motor.
This model was fitted to the experimental data and shows
good agreement for Q, Irms , VLL,rms , PAC , and IDC .
17
 NOTATION 6 Kenjō,
T., Nagamori, S., Permanent-Magnet and Brush-
less DC Motors, Oxford University Press, NY, USA, 1985.
ηESC ESC efficiency, PPDC
AC

ηm Motor efficiency, PPAC

m 7 Hanselman,D. C., Brushless Permanent Magnet Motor
ηsys Overall system efficiency, PPDC
m Design, The Writers’ Collective, RI, USA, 2003.
fsw MOSFET switching frequency 8 Mohan,
i(t) Instantaneous current N., Power Electronics: Converters, Applications,
v(t) Instantaneous voltage and Designs, John Wiley & Sons INC., NY, USA, 2007.
IDC DC Current 9 Erickson,R., and Maksimovic, D., Fundamentals of
Io Profile current Power Electronics, 2nd Edition, Springer Science & Business
Irms Root mean square value of current, rms{i(t)} Media, NY, USA, 2007.
KE Motor back voltage constant, VE = KE ω
KT Motor torque constant, Q = KT I 10 Mills,
B. and Datta, A., “Analysis of a Permanent Magnet
N DC Ground Synchronous Motor Coupled to a Flexible Rotor for Electric
N∗ Neutral point of AC system VTOL”, AHS International 74th Annual Forum & Technology
NP Number of pole pairs Display, Phoenix, Arizona, USA, May 14-17, 2018.
ω Motor RPM
ωNL Motor no-load speed 11 Gong, A., and Vestraete, D., “Experimental Testing of
Active AC Power, T1 0T v(t)i(t)dt
R
PAC Electronic Speed Controllers for UAVs”, Paper AIAA 2017-
Pcond Conduction power loss, I 2 R 4955, 53rd AIAA/SAE/ASEE Joint Propulsion Conference,
PDC DC Power, VDC IDC Atlanta, GA, 10-12 July, 2017.
PESC ESC Power, PDC − PAC 12 Green, C. R. and McDonald, R. A., “Modeling and Test
pinst Instantaneous power, v(t)i(t)
Pm Motor output power, Qω of the Efficiency of Electronic Speed Controllers for Brush-
Psw Switching power loss less DC Motors”, Paper AIAA 2015-3919, 15th AIAA ATIO
Q Mechanical torque Conference, 22-26 June, Dallas, TX, 2015.
RESC ESC equivalent resistance 13 Gabriel,
D. L., Meyer J., Plessis, F., “Brushless DC Motor
Rm Motor resistance
Characterisation and Selection for a Fixed Wing UAV”, IEEE
TR Throttle input to ESC, 0 ≤ TR ≤ 1
Africon ’11, Livingstone, Zambia, September 13-15, 2011.
vAN (t) Instantaneous phase voltage
VDC DC Voltage 14 Bronz, M., Moschetta, J. M., Hattenberger, G., Multi-
VLL,rms Line-to-line rms voltage, ex. VAC , VBC , etc. point Optimisation of a Propulsion Set as Applied to a Multi-
Tasking MAV, IMAV 2012, International Micro Aerial Vehi-
BLDC Motor Brushless DC motor cle Conference and Competition, Braunschweig, Germany, 3
ESC Electronic Speed Controller - 6 July, 2012.
MOSFET Metal Oxide Field Effect Transistor
rms Root mean square 15 Xiong, Y., Sun, S., Jia, H., Shea, P., and Shen, Z. J., “New
physical insights on power MOSFET switching losses”, IEEE
Transactions on Power Electronics, Vol. 24 (2), Feb. 2009, pp.
REFERENCES 525–531.
1 Teal Group Corporation, “2019 World Civil Unmanned 16 Glitz, E. S., and Ordonez, M, “MOSFET Power Loss Esti-
Aerial Systems Market Profile & Forecast”, 2019. mation in LLC Resonant Converters: Time Interval Analysis”,
2 Swartz, Kenneth, “Charging Forward New eVTOL Con- IEEE Transactions on Power Electronics, Vol. 34 (12), Dec.
2019, pp. 11964-11980.
cepts Advance”, Verti-flite, American Helicopter Society In-
ternational, June/July 2017, pp. 24-29. 17 Yao, Y. Lu, D.C., and Verstraete, D., “Power Loss Mod-
3 Reddy, T. B., Linden’s Handbook of Batteries, 4th Edi- elling of MOSFET Inverter for Low-Power Permanent Mag-
tion, McGraw-Hill New York, NY, USA, 2011. net Synchronous Motor Drive”, 2013 1st International Fu-
ture Energy Electronics Conference (IFEEC), Tainan, Tainan,
4 Harrington, A. and Kroninger C., “Characterization of November 3-6, 2013.
Small DC Brushed and Brushless Motors”, Report No. ARL-
18 Magtrol, Hysteresis Devices for Tension and
TR-6389, Army Research Lab Vehicle Technology Direc-
torate, Aberdeen Proving Ground, MD, 2013. Torque Control, 2019. https://www.magtrol.com/wp-
content/uploads/brakes-clutches.pdf
5 Hendershot,
J. R. and Miller, T. J. E., Design of Brushless
Permanent-Magnet Machines, Motor Design Books LLC, FL, 19 Magtrol, HB/HC User’s Manual, 2019.
USA, 2010. https://www.magtrol.com/wp-content/uploads/hbmanual.pdf
18
 20 Magtrol, HB-10-2 Nominal Performance
Characteristic Curve Torque vs. Current, 2019.
https://www.magtrol.com/wp-content/uploads/hb-10-
2curve.pdf
21 Sarma, M. S., Introduction to Electrical Engineering, Ox-
ford University Press, NY, USA, 2001.
22 Irwin,J. D., Basic Engineering Circuit Analysis, Macmil-
lan Publishing Company, NY, USA, 1987.
23 Winslow, J., Hirshikeshavan, V., Chopra, I., ”Design
Methodology for Small Scale Unmanned Quadrotors”, Paper
AIAA 2017-0014, 55th AIAA Aerospace Sciences Meeting,
9-13 January, Grapevine, TX, 2017.

19