See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332844136

Energy-saving opportunities of direct-DC loads in buildings

Article  in  Applied Energy · August 2019

DOI: 10.1016/j.apenergy.2019.04.089

CITATIONS READS

26 10,032

3 authors, including:

Daniel Lin Gerber Richard E. Brown

Lawrence Berkeley National Laboratory Lawrence Berkeley National Laboratory
25 PUBLICATIONS   255 CITATIONS    70 PUBLICATIONS   2,453 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Metacapacitors: Printed Thin Film, Flexible Capacitors for Power Conversion Applications View project

Energy Star program support View project

All content following this page was uploaded by Daniel Lin Gerber on 18 July 2019.

The user has requested enhancement of the downloaded file.

 Energy-Saving Opportunities of Direct-DC Loads in Buildings

Daniel L. Gerbera,∗, Richard Lioua , Richard Browna

a
Lawrence Berkeley National Laboratory, Building 90, 1 Cyclotron Rd, Berkeley, CA 94720

Abstract

Despite the recent interest in direct current (DC) power distribution in buildings, the market for
DC-ready loads remains small. The existing DC loads in various products or research test beds are
not always designed to efficiently leverage the benefits of DC. This work addresses a pressing need
for a study into the development of efficient DC loads. In particular, it focuses on documenting
and demonstrating how to best leverage a DC input to eliminate or improve conversion stages in
a load’s power converter. This work identifies how typical building loads can benefit from DC
input, including bath fans, refrigerators, wall adapters, task lights, and zone lighting. It then
details the development of several prototypes that demonstrate efficiency savings with DC. The
most efficient direct-DC loads are explicitly designed for DC from the ground up, rather than from
an AC modification.
Keywords: DC distribution, plug loads, direct DC, buildings, variable frequency drive, LED
lighting

1. Background and Motivation

1.1. Motivation for DC Distribution in Buildings

Direct current (DC) power distribution systems have become a recent topic in building energy
research as a means of reducing power consumption. In the past, alternating current (AC) was well
suited to power the nineteenth and twentieth century loads such as incandescent lamps, resistive
heating, and fixed speed motors. However, with the proliferation of electronics, light emitting
diode (LED) lighting, and variable frequency drives (VFDs) for motor-driven loads, an increasing
fraction of todays electric loads operate internally on DC [1]. In addition, the recent growth in

∗
Corresponding author
Email addresses: [email protected] (Daniel L. Gerber), [email protected] (Richard Liou), [email protected]
(Richard Brown)

Preprint submitted to Elsevier April 8, 2019

on-site photovoltaic (PV) generation [2] and battery storage [3] adds even more DC electronics to
the building network. DC power distribution for on-site generation, storage, and loads can reduce
losses in converting between AC and DC, leading to electricity savings across the board [4].
The potential electricity savings from DC distribution systems in buildings have been addressed
in numerous studies [5–27]. For the commercial building sector, the reported savings vary widely
from 2% [6] to as much as 19% [7]. Gerber et al. [21] performed a series of highly-detailed
parametric simulations that varied solar and storage capacity in several equivalent AC and DC
buildings. The results suggested that the highest savings occur in buildings with large PV and
storage capacities, such as zero net energy buildings. An additional loss analysis revealed that
low-power load rectifiers contributed the most loss in the AC buildings. In general, the reported
savings are highly dependent on the system topology, converter efficiencies, distribution voltage
levels, and the coincidence of generation and load.
Systems with 380 V DC distribution have been proposed and successfully implemented in data
centers. Fig. 1 shows a typical high-level DC architecture. The estimated savings between 7% and
28% are a result of the load being predominantly electronics [5]. Commercial buildings have seen
several instances of early adoption for DC distribution systems, primarily in lighting applications
[27–29]. Several other experiments developed DC loads and test beds to evaluate the savings and
power quality with DC [25–27, 30, 31]. Weiss et al. [25] used measured efficiency data in a variety
of loads to estimate the DC savings at 5%. Fregosi et al. [27] performed a similar analysis on
lighting test beds with the Bosch DC microgrid, resulting in savings of 6% to 8%. Finally, Boeke
and Wendt [26] reported 2% measured and 5% potential electricity savings from an office LED
lighting test bed at the Philips High Tech Campus in Eindhoven, Netherlands. Many of these
experimental DC loads and test beds exhibit significantly lower savings than those predicted by
past analyses and simulations. This work suggests ways in which experimental loads in DC test
beds can be improved.

1.2. Previous Works in DC Modification

Currently, there is a small market for DC plug loads for RV and boating applications [32].
These loads are often compact and designed for 12 V or 24 V DC from the vehicle’s battery. In
the next few decades, this market will expand to low-voltage DC solar home systems in developing
countries that currently lack universal electrification.

2
 1

380 V

120/208 V 380 V 380 V ≈ 380 V

3 2

380 V 380 V
High
Voltage
DC Loads

4
Direct-DC Loads
Low
48 V Voltage
DC Loads

Figure 1: Example DC building topology with distribution at 48 and 380 V DC. Converters: 1. MPPT power
optimizer, 2. integrated battery charge controller, 3. grid tie inverter, 4. PoE switch. Certain loads such as LEDs
may require an additional DC-DC converter at the load (not shown).

Several previous works have prototyped DC plug loads for newly constructed DC buildings in
the residential and commercial sectors. Many of these efforts create prototypes for a variety of
DC appliances as a means of performing system or microgrid tests [33–37]. Recent research has
produced DC prototypes in many end-use categories such as lighting [25, 26, 34–38], motor loads
[34, 36, 37, 39–42], electronics [34–37, 43], and induction cooking [35, 44].
Despite the wealth of research on developing DC loads, there are still several important gaps.
First, many of the previous works compare AC and DC loads of different technologies. It has been
common, for example, to compare a DC variable frequency drive motor appliance to an outdated
AC synchronous fixed-speed motor. Second, many of the previous prototypes are not designed to
operate at the DC distribution voltages and require DC/DC converters at the input. Such practice
neglects the many design optimizations allowed by DC that can benefit efficiency and cost. Finally,
most of the previous studies use prototype DC plug loads, but do so as a means of showcasing
another research topic such as a DC microgrid demonstration or a new control strategy.
This work aims to improve on previous works by focusing on the optimal development of DC
loads. It introduces several new concepts and analyses, including:

3
 • A formal categorization of building end-use loads by how they benefit from DC distribution

• An analysis that shows how standard variable speed motors can be improved and optimally
designed for DC input

• Two completely new system topologies for efficiently powering LEDs from DC distribution

The main purpose of this work is to suggest ideas and practices that can and should be implemented
in future products. It gives detailed explanations of the benefits of DC from the load perspective.

2. DC Load Modifications
2.1. Existing DC Voltage Standards
There are several considerations in selecting the DC distribution voltage. First and foremost
is electrical safety, and OSHA standard 1910.303(g)(2)(i) [45] considers voltages below 50 V AC
or DC to be touch-safe (the UK IET BS 7671:2008 [46] specifies it at 50 V AC and 120 V DC).
Second, higher voltages require more insulation to protect against dielectric breakdown. Finally,
low-voltage systems can experience high wire loss and voltage drop. For a device with constant
power requirements PL , the current can be expressed as I = PL /V . Since the wire loss Pw = I 2 Rw
decreases quadratically with the wire’s voltage, high-power devices are often wired on high-voltage
lines.
This work investigates several existing or emerging standards for DC plug voltages:

• USB Type C: 5-20 V, 100 W

• PoE: 48-56 V, 100 W

• EMerge Alliance Data/Telecom Center Standard: 380 V

These standards are further described in Appendix A.

2.2. Devices that Benefit from DC

Internally DC loads are classified as being direct-DC or native-DC, depending on whether the
building distribution is DC or AC, respectively [6]. Direct-DC loads connect to the DC network
directly or through a DC/DC converter. Native-DC loads always require a rectifier to interface
with the AC distribution.
This work further classifies loads based on how they can benefit from an interface to a DC
distribution network. In Fig. 2, common plug loads are classified as being:
4
 • DC-connected: The internal DC stage of these loads can be connected or hardwired directly
to the DC network. A direct-DC input would bypass the input voltage conversion (and/or
rectification) stage, thus allowing for savings in efficiency and cost. DC benefits these loads
the most.

• DC-converted: The internal DC stage of these loads requires a DC/DC converter in order to
connect to the DC network. However for most loads, the direct-DC version outperforms its

Incandescent

Lighting Fluorescent

LED

Computer
Electronics
Wireless
Electric Loads Charger

Heating
Water Heater
Element

Fixed Speed Cheap Low

Motor Power Motor

HVAC Unit
Motor Loads
Compressor

DC-connected
VFD Motor Refrigerator
DC-converted

DC-indifferent Water Heater

Figure 2: DC-connected loads have an internal DC stage that can be connected or hardwired to DC distribution.
DC-converted loads require an input DC/DC converter, but still benefit from a DC input. DC-indifferent loads do
not benefit from a DC input.

5
 native-DC counterpart in efficiency, cost, and size. DC offers some benefits for these loads.

• DC-indifferent: The load can be hardwired to an AC or DC distribution with equivalent cost

and efficiency. DC does not directly benefit these loads.

2.2.1. DC-connected
Includes loads with a fixed internal DC bus, many of which require AC power at a non-mains
frequency.
VFD motor loads contain a brushless DC (BLDC) inverter-driven synchronous or asynchronous
motor. Due to their efficiency benefits, VFDs will eventually replace fixed-speed motors in air
conditioners, refrigerators, heat-pump air and water heaters, and ventilation fans [47–49]. As
shown in Fig. 3a, native-DC VFD motors require a rectification stage, followed by a DC capacitor
bus. The inverter supplies the stator coils with a variable-frequency AC drive. In well-designed
direct-DC VFDs, the DC capacitor bus could be connected directly to the DC distribution, as
shown in Fig. 3b. A direct connection bypasses the rectification stage, thus allowing for savings in
efficiency and cost. In addition, the DC bus capacitors can be reduced, as described in Appendix
AC AC
120 VRMS DC PWM Amplitude
60 Hz Variable Frequency

AC
Rectifier Inverter
Input
CDC

(a)

AC
120 VRMS AC
60 Hz PWM Amplitude
DC Variable Frequency

AC
Input Rectifier
Inverter
CDC

DC
Input

(b)
Figure 3: Block schematic of a VFD motor, as present in the refrigerator and bathroom fan. The inverter is powered
from an internal DC stage (blue), and outputs AC at a variable frequency (orange). (a) A rectifier is required to
convert 60 Hz AC (red) to DC. (b) A proposed modification, with multiplexed AC and DC inputs.

6
C. Although they are still required for electromagnetic compatibility, they do not need to filter
AC mains frequency power ripple.
Wireless transmitters power an antenna with high-frequency AC. In this way, wireless transmit-
ters are structurally similar to VFDs: they contain a rectifier, a DC stage, and a power amplifier,
which outputs AC at the antenna’s resonant frequency. Transmitters are present in any wirelessly
connected plug-loads, including routers, computers, and Internet of Things devices. Other appli-
cations, such as wireless chargers and induction stoves employ near-field wireless power transfer.

2.2.2. DC-converted
Includes current-controlled loads and loads with multiple internal voltage rails.
LEDs are a current-controlled load since their luminosity is nearly proportional to their current.
DC LED drivers are often 95-98% efficient, whereas AC LED drivers commonly have 86-93%
efficiency [21, 27]. In addition, AC LED drivers are more expensive because they have to rectify
the AC input, apply power factor correction (PFC), and cancel the 120 Hz AC power ripple with
a large electrolytic capacitor.
Laptops require several DC/DC converters to power the internal components at their respective
voltages. Nonetheless, an AC/DC wall adapter is also required to supply the laptop’s primary power
rail. DC input to the primary rail could obviate the need for a wall adapter, allowing for benefits
similar to DC-connected devices.

2.2.3. DC-indifferent
Includes heating elements and fixed-speed motors. Resistive heating elements are often found
in water heaters, ovens, and incandescent bulbs. These devices can be connected to AC or DC
distributions with equal efficiency. Fixed-speed motors are often found in old or inexpensive motor
loads. Due to their low efficiency in many applications, they will eventually be replaced by VFDs.

2.3. Experimental Method

This work examines a bathroom fan, refrigerator, wall adapter, LED task lamp, and LED zone
lighting rig. These loads are selected to represent the end-use categories of ventilation, refriger-
ation, electronics, and lighting in residential or small commercial buildings. The prototypes and
modifications are intended to demonstrate savings by leveraging a DC input to reduce the number
of conversion stages. The method of reporting savings varies by load, and includes measuring the

7
input power consumption, nominal operating efficiency, and efficiency curves. Efficiency measure-
ments use η = Pout /Pin , where Pout is the power measured at the device’s internal DC stage, and
Pin is measured at the AC or DC input.

3. Bath Fan

3.1. Modifications for DC Input

As described in Sec. 2.2, variable frequency drive (VFD) loads have an internal DC capacitor
that buffers the inverter. There are two proposed modifications to the Delta GBR80 bath fan.
The 48 V DC-converted modification in Fig. 4b is intended for PoE, and the 12 V DC-connected
modification in Fig. 4c is for USB-C. Although PoE requires a 48/12 V DC/DC conversion, it is
practical due to the distance limitations of USB-C. These modifications both maintain the DC bus
at its nominal 12 V, since increasing or decreasing the bus voltage may damage the inverter or
stator coils, respectively. As such, this experiment does not affect the motor’s speed and torque.
AC DC DC
120 Vrms ~170 V 12 V

Bridge Flyback DC
Wall
Step-Down Cap
Mains Rectifier Converter Bus

(a)

DC DC
48 V 12 V

Murata DC
PoE MYBSP0122BABF Cap
DC/DC Converter Bus

(b)

DC
12 V

DC
12 V
Cap
Input
Bus

(c)
Figure 4: Methods of providing power to the bath fan’s DC capacitor bus and inverter stages. (a) The fan’s original
AC/DC rectifier board, shown in Fig. 6a. (b) The DC-converted modification for 48 V PoE input, using a DC/DC
converter, shown in Fig. 6b. (c) A DC-connected modification for a hypothetical 12 V DC input. This configuration
represents how DC input would be beneficial if the motor and inverter were designed for 48 V DC.

8
Figure 5: Prototype of the Delta GBR80 bath fan, modified to accept a PoE input. The blue box contains the boards
in Fig. 6, with a switch to toggle between AC or DC input.

3.2. Experimental Prototype and Results

As shown in Fig. 5, the bath fan prototype has a multiplexed AC or DC input. The original
rectifier board in Fig. 6a uses an isolated flyback topology. The DC-converted input uses the
Murata MYBSP0122BABF isolated 48/12 V DC/DC converter in Fig. 6b.
The results suggest that direct-DC can improve the fan’s power consumption. The efficiency
curves in Fig. 7 show that the DC/DC converter is up to 6% more efficient than the original rectifier
board. Tab. 1 shows a decrease in the fan’s input power with DC input. In addition, the fan draws
even less power with a 12 V input, which represents the savings potential of a DC-connected design.
In this experiment, the exhaust valve is held open or shut to simulate a lightly or heavily loaded
fan, respectively.

3.3. Discussion

The 48 V DC-converted modification improves upon the efficiency of the rectifier board. Not
only does it remove the AC/DC rectification stage, but it also allows for a smaller DC/DC con-
version ratio Vout /Vin . Large conversion ratios require the switching elements to frequently block
power flow to the load, yet the converter still suffers parasitic loss regardless. The 120 VRMS AC
9
 12 V DC Capacitor Bus DC-DC Flyback Switch

DC
Terminal

Bridge
Rectifier

AC
Terminal

DC-DC
Flyback
Transformer
DC High-side EMI Filter
Filter Capacitor Components

(a) (b)
Figure 6: Input power conversion boards for the Delta bath fan. (a) The original rectifier board that converts
120 VRMS AC to 12 V DC. The board dimensions are 8 x 6 cm. (b) A Murata MYBSP0122BABF 48/12 V DC/DC
converter for interfacing the DC capacitor stage with PoE. The board dimensions are 3.5 x 2 cm.

Bath Fan Board Efficiency

90.0
87.5
85.0
Efficiency (%)

82.5
80.0
77.5
75.0
AC/DC
72.5 DC/DC
2 4 6 8 10 12
Output Power (W)
Figure 7: Efficiency curves for the prototype bath fan’s AC/DC and DC/DC boards.

converter board mitigates some of this loss by instead converting through the flyback transformer’s
turns ratio. However, inclusion of a flyback transformer adds cost, size, core loss, and winding loss.
Although the 12 V DC-connected modification has the lowest cost and consumption, it is
impractical for a bath fan due to low-voltage wire loss. Nonetheless, the 12 V modification is
explicitly intended to show the benefits of a DC-connected design. Ideally, a PoE bath fan would
use a 48 V inverter and motor. Appendix B shows how a BLDC motor can be redesigned to operate
with a different internal DC voltage.

10
 Table 1: Fan prototype power consumption

Input Power Input Power

Input Type Conversion AC Power Factor
Light Load Heavy Load

Original 120 VRMS AC AC/DC 7.75 W 9.01 W 0.64

DC-converted 48 V DC DC/DC 7.43 W 8.65 W N/A

DC-connected 12 V DC None 6.6 W 7.72 W N/A

AC DC
120 Vrms 340 V

Delon DC
Wall
Doubler Cap
Mains
Rectifier Bus

(a)

DC
340 V

DC
340 V
Cap
Input
Bus

(b)
Figure 8: Methods of providing power to the refrigerator’s DC capacitor bus and inverter stages. (a) The refriger-
ator’s original AC/DC Delon doubler rectifier circuit. (b) The DC-connected modification for 340 V input, which
approximately represents 380 V DC.

4. Refrigerator
4.1. Modifications for DC Input

Like the bath fan, the DC input modification of the 11 ft3 LG LTNC11121V inverter-based
refrigerator removes the VFD rectification stage. The main difference is that the refrigerator’s
rectifier is a Delon (full wave) doubler, whose unloaded DC output is 340 V DC. As such, the
refrigerator is modified to accept a 340 V DC input, which although not a standard DC voltage,
approximates the efficiency and power draw at 380 V DC.

4.2. Experimental Prototype and Results

The refrigerator prototype, shown in Fig. 9, adds a 340 V DC-connected input to the DC
capacitor stage. As shown in Fig. 10 the inverter board’s original AC input contains an input EMI
filter and protection components. The DC prototype instead contains a thermistor that protects
11
Figure 9: Prototype for the DC input modification of a refrigerator. The blue box on the left side allows for a
340 V DC input to the inverter board’s DC capacitor bus. The compressor is the BMA069LAMV model.

the DC capacitors from over-current at start-up. In this experiment, the refrigerator’s defrost coils
have been deactivated, and power is drawn solely by the compressor and electronics.
The efficiency of the rectification stage in Fig. 8a is 99%, averaged over fifty measurements with
power ranging from 50 W to 80 W. There is little to no correlation between power and efficiency
in this range. The AC power factor is 0.65, and is caused entirely by harmonic distortion.

4.3. Discussion

The refrigerator barely benefits from a high-voltage DC input, since bypassing the rectification
stage only saves 1% of the input power. 340 V operation can be very efficient, but requires the
large inverter and capacitors shown in Fig. 10. In contrast, many low-voltage DC refrigerators can
operate with smaller electronics, such as the 24 V Danfoss 101N0212 in Fig. 11. Compared to the
LTNC11121V, its compressor has fewer and thicker windings, as discussed in Appendix B.
It is important to note that the LNTC11121V has a poor power factor of 0.65 due to its lack of
power factor correction (PFC). Although inverter-based refrigerators are generally subject to the
stringent IEC 61000-3-2 Class D harmonic specifications, these requirements only apply for devices
that draw more than 75 W of power [50, 51]. While the LG refrigerator sometimes draws up to
80 W, it usually operates below 75 W (the defrost coils draw up to 175 W, but they are a resistive

12
 DC-DC Flyback DC-DC Flyback Switch Control
340 V DC Capacitor Bus
Transformer Electronics
340 V DC
Inverter

Bridge
AC Input EMI Filter and Defrost Coil Terminal
Rectifier
Protection Components and Relays

Figure 10: Inverter board for the LG LTNC11121V refrigerator. The inverter is a high-voltage 3-phase SIM6822M
chip. The board dimensions are 23 x 16 cm.

Figure 11: Danfoss/Secop 101N0212 inverter board, for a Danfoss 12/24 V DC refrigerator. The board dimensions
are 10 x 6.5 cm. This inverter board can provide up to 113 W when connected to a BD50F compressor at 3500 RPM
and 30◦ C ambient temperature.

13
 AC DC
120 Vrms 5-20 V

Elec-
Wall Wall
tronics
Mains Adapter
Load

(a)

DC
5-20 V

Elec-
DC
tronics
Mains
Load

(b)
Figure 12: Two different methods for powering the internal DC bus in electronics. (a) Native-DC electronics require
a wall adapter, which often rectifies 120 V AC to 170 V DC and steps-down to the 5-20 V DC appropriate for
electronics. (b) This work considers direct-DC electronics to be DC-connected.

load with unity power factor). Standard-sized refrigerators would require a PFC rectifier, whose
losses and cost are much greater than the Delon rectifier. For these refrigerators, a DC-connected
input would be considerably more beneficial.
DC input allows for a reduction in the size of the DC capacitor bus. Not only are the high-
voltage high-capacitance electrolytics in Fig. 10 bulky and expensive, but they are also prone to
failure and may limit the life span of the entire board [52, 53]. The DC capacitors are sized to filter
120 Hz AC power ripple, provide an energy buffer for rapid changes in motor torque, and filter
the high-frequency inverter PWM ripple. Appendix C describes how the giant DC capacitors in
Fig. 10 are mainly required for filtering the 120 Hz AC power ripple, and can be greatly reduced
with direct-DC.

5. Wall Adapters
5.1. Modifications for DC Input
Wall adapters provide low-voltage DC to many household plug-load electronics. Future trends
in electronics and the Internet of Things will see an increase in distributed AC/DC wall adapters.
In general, low-power units with high voltage-conversion ratios are often inefficient. The US only
requires wall adapters to be 74% to 88% efficient, depending on power level [54, 55]. DC distribution
is attractive because it can eliminate the wall adapter, improving efficiency, overall cost, and
convenience.
14
 This work studies the losses in native-DC electronics by measuring the efficiency of several
USB-C wall adapters. As shown in Fig. 12, this work considers direct-DC electronics as being DC-
connected. In this sense, the savings with DC is equal to the wall adapter’s measured loss. Other
more realistic scenarios may require a DC wall adapter, which is further discussed in Section 5.3.

5.2. Experimental Prototype and Results

The five USB-C wall adapters, shown in Table 2, vary in power capacity and quality. These
adapters are loaded with a 5 V Anker Powerpack battery and a Macbook that attempts to request
up to 20 V. The efficiency is calculated from ten-minute measurements of the input and output
energy.

Table 2: Wall adapter efficiency

Average Measurement
Adapter Output Voltage (V) Efficiency (%)
Output Power (W)
Choetech 15 W 5 84.73 15.7
5 87.18 11.0
Pixel 18 W
9 88.24 17.6
5 90.82 10.7
Anker 30 W
20 91.22 29.2
5 87.15 10.4
Choetech 39 W
15 91.96 37.7
5 89.92 10.6
Macbook 87 W
20 94.91 59.8

5.3. Discussion

The results in this study show that wall adapters contribute significantly to the AC system
loss. However, direct-DC electronics would more realistically be DC-converted, since building
distribution voltage must be greater than 5-20 V to avoid wire loss. As such, direct-DC electronics
will require a DC/DC wall adapter with a potentially large step-down ratio. Nonetheless, DC
distribution allows for highly-efficient centralized conversions. For example, a single centralized
direct-DC charging station could power all the electronics in a quad of cubicles. High-power
converters such as the 800 W Vicor BCM6123xD1E1368yzz perform a 380/12 V DC conversion
with 97% efficiency.

15
 DC DC
380 V 5V
Desktop
Wall USB-A
Mains Charging Ballast Resistor
Station

(a)

DC DC DC
380 V 20 V ~15-20 V
Desktop
Wall USB-C LED
Mains Charging Driver
Station

(b)

DC DC
380 V ~15-20 V
Desktop
Wall USB-C Current
Control
Mains Charging Signal Current
Station Sense

(c)
Figure 13: Methods of powering a task lamp with DC input from a desktop charging station. Note that the wall
mains could also be 120 VRMS AC or 48 V DC. (a) Use a ballast resistor for current control. (b) Connect the LED
driver in the lamp to the charging station’s output. (c) If the charging station has PPS capability, it can act as a
constant current LED driver.

6. Task Light

6.1. Modifications for DC Input

Task lights in a workspace can benefit from localized low-voltage DC distribution. Many USB
task lights in today’s market are powered by a 5 V USB-A port and use a ballast resistor for current
control, as shown in Fig. 13a. The ballast resistor makes the lamp inefficient, susceptible to input
voltage swing, and difficult to dim. Advanced task lights overcome these difficulties with an LED
driver. Future lamps may be powered by a USB-C charging station, as shown in Fig. 13b.
A further improvement is possible, shown in Fig. 13c, in which the charging station becomes
the LED driver. This solution requires a charging station that supports USB protocols with a
programmable power supply (PPS) capability (USB-PD 3.0 or QC 3.0). These lamps need only
to sense the LED current and program the supply voltage accordingly. This method effectively
removes the LED driver conversion stage, while still offering all its benefits.

16
 Figure 14: Prototype of the driverless USB task light.

6.2. Experimental Prototype and Results

The task light prototype, shown in Fig. 14, demonstrates the driverless topology in Fig. 13c. It
connects to an Anker PowerPort charger with QC 3.0 (USB-PD 3.0 was not available during this
experiment). The Anker PowerPort provides up to 12 V to three 1.5 A LEDs (Cree MTG7-001I-
XTE00-NW-0GE3). At the brightest operating point, the LEDs draw 1.5 A at 14.3 W (9.815 V).
An Arduino microcontroller generates the brightness control signals, which are sent via the USB
data lines as per the QC protocol [56]. Although this proof-of-concept prototype uses voltage
control, current control is generally recommended to account for temperature effects and such.

6.3. Discussion

The prototype effectively uses the charging station as an LED driver, thus reducing the number
of power conversions. This method is possible in any point-to-point DC topology that supports a
protocol with PPS capability.

7. Zone Lighting

7.1. Modifications for DC Input

Zone lighting systems simultaneously operate many light fixtures, and are often found in office
and retail buildings. Traditional LED bulbs have internal drivers, and are designed to plug into
17
 Dimmer
Hot
Neutral

+V + VFIX
_ FIX _

Fixture 1 Fixture N

(a)

Neutral
Positive

Hot
Negative

+ VFIX _ + VFIX _
Fixture 1 Fixture N
Driver
(b)

380 V
_ _ GND
+ VFIX + VFIX
Fixture 1 Fixture N _ Driver
+ VOUT
(c)
Figure 15: Methods of wiring lighting fixtures. (a) The conventional design with distributed drivers and a dedicated
0-10 V analog dimming line. (b) Most remote driver topologies use an array of parallel current-controlled lines.
(c) Series-remote driver topology with a buck-based driver. Although depicted with a DC input, series drivers
are currently designed for an AC input. In this topology the number of fixtures N is limited to N*VFIX <380 V.
Nonetheless, a buck-boost LED driver can allow for more fixtures to be attached.

existing incandescent or fluorescent fixtures. As shown in Fig. 15a, the LED drivers are distributed
between the fixtures and may require a separate signaling line for dimming. Although internal LED
drivers are standard in AC buildings, they are also present in many of the experimental 380 V DC
test beds [26, 27].
External or remote LED drivers are physically separate from the lamp’s LEDs, and have ap-
plications in outdoors lighting systems such as streetlights and lit signs [57]. Several companies
have also started promoting their use in building lighting systems. First, remote drivers reduce
component cost since each driver can power several fixtures, as shown in Fig. 15b. Second, easily
accessible and replaceable remote drivers reduce maintenance costs, since the driver often limits
the lifespan of the entire lamp. Finally, dimming and ancillary functions are more cost-effective
with a centralized driver for multiple fixtures.

18
 Figure 16: Prototype of the series-remote driver zone light.

Most commercial remote LED drivers use a parallel design, shown in Fig. 15b, which is ideal
for applications that require individual fixture control. Another topology, shown in Fig. 15c, uses a
single line to power the LED fixtures in series. Series drivers are uncommon, and the few companies
that develop them seldom intend them for an application in indoor zone lighting.
This work aims to demonstrate that series-remote drivers are an excellent design for zone
lighting in 380 V DC buildings due to their low overall cost and high efficiency. The cost benefit
is a result of using simple low-power hardware to drive many fixtures, as shown in Fig. 15c. In
addition, the driver’s total switch stress does not increase with the number of fixtures, implying a
constant hardware cost.
Series-remote drivers can also be significantly more efficient than their parallel counterparts.
In general, buck converters are most efficient at a high MOSFET conduction duty cycle, which
happens when the output voltage is closely matched to the input. The drivers in Figs. 15a and 15b
undergo a significant step-down from the mains to the fixture voltage VFIX . In contrast, the series-
remote driver in Fig. 15c performs a much lighter step-down when several fixtures are attached.

7.2. Experimental Prototype and Results

A 380 V DC series-remote driver was prototyped using the buck-based topology in Fig. 15c.
The prototype, shown in Fig.16, uses an MP4000 controller, a 600 V IGBT, and a 10 mH inductor.
The experimental fixtures are a set of 75 V 18 W four-foot T8 LED tubes, which come from
removing the driver of a PT-T84FP18W. As shown in Fig. 17, the driver’s efficiency increases with
the number of tubes. For reference, the original tube-integrated AC LED driver, shown in Fig. 18,
operates at 92% efficiency and 0.97 power factor.

19
 100 Efficiency Curves
98

Efficiency (%)
96

94

92 1 tubes
2 tubes
3 tubes
4 tubes
90
0.00 0.05 0.10 0.15 0.20 0.25
Output Current (A)
Figure 17: Efficiency curves for the 380 V DC series-remote driver with 1-4 LED tubes attached in series. The
driver’s efficiency increases with the number of attached tubes.

Figure 18: Original AC LED driver packaged in a PT-T84FP18W four-foot LED tube. The board dimensions are
3.5 x 2 cm.

7.3. Discussion

Remote drivers have many advantages over conventional internal drivers, such as cost, main-
tenance, and control functionality. However, the main drawback is a loss of the plug-and-play
simplicity of traditional light bulbs. In addition, remote systems may experience voltage drop
over long wire runs. Series-remote drivers add the potential to further improve on hardware costs
and efficiency. Nonetheless, their drawbacks include non-trivial wiring and the inability to dim
individual fixtures. As such, series-remote drivers are best suited for commercial zone lighting.
Overall, further study is required to fully determine the value of series-remote drivers in industry
applications.
One potential concern with series fixtures is that a single LED failure would disable the entire
multi-fixture string. This drawback can be overcome by equipping each fixture with a bypass

20
 + VFIX _

Figure 19: Bypass circuit for the LED fixtures connected to a series-remote driver. This circuit will allow other
fixtures to remain powered despite a failure in the series string.

circuit. An example bypass circuit, shown in Fig. 19, is based on a design for bypassing individual
LEDs [58]. In this case, the SCR and Zener diode would be sized such that their breakdown voltage
is greater than the nominal LED fixture voltage.
Although the prototype only allows five 18 W 75 V tubes to be stacked in series, newer high-
brightness LEDs provide more power with less voltage drop [59, 60]. For example, an 18 W bulb
with three modern 6 V 1 A LEDs would only drop 18 V. Twenty-one such bulbs would stack to
380 V, which is enough to cover a moderately sized room. A buck-boost series-remote driver allows
the flexibility of even more bulbs at the incremental cost of efficiency and safety.

8. Conclusion

Buildings with DC power have taken the recent spotlight in research, but the development
of highly efficient DC-ready loads has lagged. This work categorizes the types of loads whose
efficiency directly benefits from a DC input. Several types of loads are studied, and some of
them are modified or prototyped as direct-DC. This work focuses on low-power residential and
small commercial loads, which include a bath fan, refrigerator, wall adapter, task lamp, and zone
lighting system. The main focus of each design is to leverage DC input to eliminate or improve
the conversion stages.
This project obtained new insights specific to the categories of motor loads, electronics, and
lighting. In motor loads, the most efficient type of BLDC motor is designed such that its internal
DC capacitor bus is coupled to the DC input. Although there is very little loss across a diode bridge
rectifier, a PFC boost rectifier is notably less efficient. In electronics, DC distribution can allow
21
for downsizing or eliminating wall adapters. The efficiency benefits generally favor DC, though
the comparisons must be made with careful attention to the distribution voltage and conversion
process. In lighting, DC task lamps can leverage a USB-PD charging station as an LED driver.
Zone lighting can benefit greatly from series-remote drivers, but further research must validate
their feasibility. In all loads, DC input will improve power quality and reduce the size of the DC
capacitors. Further study is required to determine the full value of these secondary effects.

Appendix A. DC Voltage Standards

USB-C is emerging as the single standard and connector for consumer electronics. Its point-
to-point power delivery architecture allows for simultaneous power and communications. Power
over USB-C strictly adheres to the USB power delivery (USB-PD) protocol [61]. Past USB-PD
standards set the programmable voltage to 5 V, 9 V, 15 V, or 20 V. However, as of USB-PD 3.0
(2018), the port can now function as a programmable power supply (PPS) and outputs voltages
between 3 V and 21 V with 20 mV increments. USB-C connectors contain twenty-four pins of
which eight are for power, two are for dedicated USB-PD communications, and the rest are for
data. Due to voltage drop and wire loss inherent with low-voltage distribution, USB is best suited
for short-range power distribution such as a cubicle.
Another competing USB protocol is Qualcomm Quick Charge (QC) [62]. QC works with
traditional USB-A or B ports, but requires the data lines for power control signaling. It also has
a PPS capability with 200 mV increments (QC 3.0), allowing it to charge batteries quickly and
efficiently. Although fast charging made QC popular, USB-PD 3.0 now has much of the same
functionality. Recent versions of QC also support USB-PD, and these standards may eventually
merge.
Although Power over Ethernet (PoE) nominally operates at 48 V, it allows up to 56 V to com-
pensate for wire drop [63–65]. The 802.3bt Type 4 standard allows PoE to transfer 100 W of power
over Category 5 Ethernet cable using 4-pair power transfer mode. PoE can be found in phones,
cameras, routers, LED luminaires, and other electronics that require power and communications.
Since many buildings are already wired with Ethernet cables, PoE represents one of the easiest
ways for DC power distribution to enter the market. In addition, PoE operates near the OSHA
touch-safe limit (50 V), thus optimally minimizing wire loss while ensuring safety. While USB is
best suited to power a work space, PoE can extend to multiple rooms or even an entire residence.
22
 The European Standardization Institute (ETSI) and the EMerge Alliance have established DC
standards with specifications for powering datacenters at 380 V DC [66–69]. This standard allows
for easy retrofitting since 380 V DC has similar insulation requirements with 277 V AC. Besides
datacenters, 380 V DC holds much promise for powering other high-power loads, and allows for
efficient coupling with solar and storage [21].

Appendix B. DC Capacitor Voltage in BLDC Motors

(a) (b)
Figure B.20: Winding area diagram for two motors with equal input power and magnetic flux. The winding area in
both motors is roughly equal. In this example, K = 4. (a) A high-voltage motor that requires 12 windings, but they
can be relatively thin. (b) An equivalent low-voltage motor only requires 3 windings, but they must be thick enough
for the relatively high winding current.

Brushless DC (BLDC) motors are designed with a specific AC voltage across the stator wind-
ings. The winding voltage dictates the inverter’s DC capacitor bus voltage, which can be DC-
connected if it equals the DC distribution voltage. This section shows how a motor’s winding
voltage can be designed independently of its power rating, and can ultimately be designed as a
DC-connected device.
For a given application, assume the motor’s electrical input power Pconst is specified and con-
stant. Also assume the motor’s magnetic flux φconst is constant and is directly related to the
mechanical output power. The relations between the electrical stator windings and the magnetic
core are:

Pconst = V I (B.1)
NI
φconst = (B.2)
Rtot
dφconst
E=N , (B.3)
dt
23
where V and I respectively are the winding’s input voltage and current, N is the number of
windings, Rtot is the total reluctance of the magnetic core, and E is the back EMF generated in
the windings.
Consider a comparison between motors 1 and 2, which both have the same input power P1 = P2 ,
and magnetic flux φ1 = φ2 . However, the main difference is that the winding voltage of motor 1
is designed to be K times greater than that of motor 2. Since V1 = KV2 , Eq. B.1 assures that
1
I1 = K I2 . If the two motors are specified with the same magnetic flux and core parameters, then
Eq. B.2 requires that N1 = KN2 . In other words, the motor 1 requires K times as many windings,
1
but the windings only pass K times as much current. Finally, Eq. B.3 follows that E1 = KE2 , but
since V1 = KV2 , then each motor has proportionally as much voltage headroom, and each motor
can attain the same maximum speed.
Overall, high-voltage motors use less current and have more stator windings than low-voltage
motors of the same input power and flux. The low-voltage motor requires thick high-current
windings, but the overall winding packing window is still comparable between the two motors, as
shown in Fig. B.20.
Finally, the losses in the high and low-voltage motors are also equivalent. The core loss is
equivalent because the flux φ is the same between the motors. For winding loss, the factor K falls
out of the equations:

Pw = I 2 R w (B.4)
L
Rw = ρ . (B.5)
A

The wire loss power Pw and resistance Rw are dependent on the wire length L, cross sectional area
A, and resistivity ρ. As shown in Fig. B.20, the number of turns N1 = KN2 requires motor 1 to
1
have a smaller area A1 = K A2 , and greater length L1 = KL2 . As such, motor 1 has a greater wire
1
resistance R1 = K 2 R2 . Nonetheless, the winding current I1 = K I2 , and so the overall winding loss
power P1 = I12 R1 = ( K1 I2 )2 (K 2 R2 ) = P2 .

Appendix C. Sizing the DC Capacitor Bus

Most types of native-DC loads contain an internal DC capacitor bus, whose size can often be
reduced in converting to a direct-DC input. These capacitors are sized such that the peak-to-peak

24
 Vbus
Pin Pou t LL
PC
PFC
C VEMF

Inverter
Figure C.21: A model of the energy flow in a VFD motor with PFC. The DC capacitor is responsible for absorbing
the AC component of Pin , allowing for a purely DC Pout . In this model, the inverter is a half-bridge and the motor
is represented by its winding inductance LL and a constant back EMF.

bus voltage ripple, ∆VC , is within specification. In native-DC loads, the DC capacitor bus is
required for:

• Filtering the 120 Hz power ripple from the AC input

• Filtering the inverter PWM current

• Providing an energy buffer for a change in load

In direct-DC loads, the DC capacitor bus is only responsible for the latter two. This section
explains how to analytically determine the requisite capacitance for VFD motor loads. It follows
with a design example that shows how direct-DC can reduce the bus capacitors.

Appendix C.1. 120 Hz Power Ripple from the AC Input

Many devices require power factor correction (PFC), including electronics over 75 W and all
types of lighting. PFC rectifiers control the input current to be sinusoidal and in-phase with
the 60 Hz input voltage. As such, the input power is a raised sinusoid at 120 Hz. As shown in
Figure C.21, the DC capacitor bus is responsible for absorbing the AC component of the input
power Pin . To allow for a DC output power Pout , the power into the capacitor PC must be:

PC,AC = Pin,AC+DC − Pout,DC = −Pin cos(2ω0 t) (C.1)

From PC , the capacitor bus voltage ripple ∆VC can be solved by an energy analysis [70]:

Pin,avg
∆VC = , (C.2)
VDC ω0 C

where VDC is the DC bus voltage.

25
Appendix C.2. Bus voltage ripple from the Inverter

In this analysis, the inverter is a half-bridge circuit with PWM switching frequency f . As shown
in Fig. C.21, the half-bridge switches the output between the positive and negative terminals of
the DC bus. This switching leads to a swing in output current ∆IL , which the DC capacitor must
ultimately absorb. Salcone and Bond [71] derive the bus voltage ripple as a function of the DC
capacitor:

D(1 − D)VDC
∆IL = iC = (C.3)
f LL
VDC
∆VC = , (C.4)
32LL Cf 2

where LL is the load or winding inductance, and a duty cycle of D = 0.5 results in the largest
ripple current.

Appendix C.3. Energy buffer for a change in load

In some applications, the DC capacitor is responsible for maintaining the bus voltage over
loading transients. For example, transients in VFD motors may arrise from braking, sudden
changes in torque, and start-up. This work develops a simple model to determine the effect of load
transients on the bus voltage. As shown in Fig. C.22, the model accounts for the line inductance
LS , which is assumed to be 19.41 μH for 20 m of 10 AWG copper wire. The load transient IL (t)
can be a step M u(t) or a ramp N tu(t). The system in Fig. C.22 is represented by the differential
equation:

d2 vC dIL (t)
LS C + vC = VS − LS (C.5)
dt2 dt

Although realistic systems have a damping component, this model is sufficient to find the worst-case
ripple on vC . The solutions to Eq. C.5 are:
s
LS t
IL (t) = M u(t) ⇒ vC = VS − M sin( √ )
C LS C
s
LS
∆VC = 2M (C.6)
C
t
IL (t) = N tu(t) ⇒ vC = VS − N LS + N LS cos( √ )
LS C
∆VC = 2N LS . (C.7)

26
 VS LS IL(t)
C

Figure C.22: Simple model of a DC load connected to DC distribution with line inductance LS and supply voltage
VS . This model examines the input capacitor voltage ripple subject to a current step or ramp.

Table C.3: DC bus voltage ripple due to each effect

C (uF) AC Ripple (V) PWM Ripple (V) Transient Ripple (V)

10.0 78.02 3.32 0.82
100.0 7.8 0.33 0.26
1000.0 0.78 0.03 0.08
As shown in Eq. C.7, the voltage ripple ∆VC of the ramp response is actually independant of C.
As such, Appendix C.4 only accounts for the step response.
The step response magnitude varies with application and technology. Many of the loads in
Fig. 2 have nearly constant power draw, and transients only arise at start-up. VFD inverters can
soft-start the motor with a low-voltage AC output, allowing for a start-up current that is always
lower than the full-load current (M = Iout,max ) [72].

Appendix C.4. Comparison and Discussion

This section uses the LG inverter refrigerator as a design example for comparing the three
metrics for capacitor sizing. As discussed in Section 4, the refrigerator draws as much as Pin = 80 W
from a standard 60 Hz AC distribution. The inverter operates at f = 20 kHz, and is powered from
a DC bus at VDC = 340 V. Finally, the compressor’s load inductance is dominated by a load-side
reactor with LL = 0.8 mH.
As shown in Table C.3, the DC bus voltage ripple ∆V caused by the AC input is considerably
greater than the other two sources. With direct-DC input, the DC capacitors can be reduced
a factor of ten to twenty. Applications with high input power Pin or low bus voltage VDC are
especially favorable for direct-DC.

27
Acknowledgements

This work was supported by the California Energy Commissions EPIC Project EPC-15-024.
The authors would like to thank Alan Meier, Leo Rainer, Bruce Nordman, Wei Feng, Chris Marnay,
Vagelis Vossos, and Seth Sanders for their technical advice and support. Special thanks to the
reviewers and editors.

References
[1] K. Garbesi, V. Vossos, and H. Shen, “Catalog of DC Appliances and Power Systems,” Tech. Rep. LBNL-5364E,
Lawrence Berkeley National Laboratory, Berkeley, CA, 2011.
[2] A. Perea, C. Honeyman, S. Kann, A. Mond, M. Shiao, S. Rumery, and A. Holm, “U.S. Solar Market Insight:
2016 Year in Review - Executive Summary,” tech. rep., GTM Research & Solar Energy Industries Association,
Mar. 2017.
[3] G. Research, “U.S. Energy Storage Monitor: Q4 2016 Executive Summary,” tech. rep., GTM Research, Dec.
2016.
[4] K. George, “DC Power Production, Delivery and Utilization: An EPRI White Paper,” 2006.
[5] G. AlLee and W. Tschudi, “Edison Redux: 380 Vdc Brings Reliability and Efficiency to Sustainable Data
Centers,” IEEE Power and Energy Magazine, vol. 10, pp. 50–59, Nov. 2012.
[6] S. Backhaus, G. W. Swift, S. Chatzivasileiadis, W. Tschudi, S. Glover, M. Starke, J. Wang, M. Yue, and
D. Hammerstrom, “DC Microgrids Scoping Study Estimate of Technical and Economic Benefits,” Tech. Rep.
LAUR1522097, Los Alamos National Laboratory, Mar. 2015.
[7] P. Savage, R. R. Nordhaus, and S. P. Jamieson, “From Silos to Systems: Issues in Clean Energy and Climate
Change: DC microgrids: benefits and barriers,” tech. rep., Yale School of Forestry & Environmental Sciences,
2010.
[8] D. Denkenberger, D. Driscoll, E. Lighthiser, P. May-Ostendorp, B. Trimboli, and P. Walters, “DC Distribution
Market, Benefits, and Opportunities in Residential and Commercial Buildings,” tech. rep., Pacific Gas & Electric
Company, Oct. 2012.
[9] A. Sannino, G. Postiglione, and M. Bollen, “Feasibility of a DC network for commercial facilities,” IEEE
Transactions on Industry Applications, vol. 39, pp. 1499–1507, Sept. 2003.
[10] B. A. Thomas, I. L. Azevedo, and G. Morgan, “Edison Revisited: Should we use DC circuits for lighting in
commercial buildings?,” Energy Policy, vol. 45, pp. 399–411, June 2012.
[11] V. Vossos, K. Garbesi, and H. Shen, “Energy savings from direct-DC in U.S. residential buildings,” Energy and
Buildings, vol. 68, Part A, pp. 223–231, Jan. 2014.
[12] K. Engelen, E. Leung Shun, P. Vermeyen, I. Pardon, R. D’hulst, J. Driesen, and R. Belmans, “The Feasibility
of Small-Scale Residential DC Distribution Systems,” in IECON 2006 - 32nd Annual Conference on IEEE
Industrial Electronics, pp. 2618–2623, Nov. 2006.
[13] B. Glasgo, I. L. Azevedo, and C. Hendrickson, “How much electricity can we save by using direct current circuits
in homes? Understanding the potential for electricity savings and assessing feasibility of a transition towards
DC powered buildings,” Applied Energy, vol. 180, pp. 66–75, Oct. 2016.
[14] D. Hammerstrom, “AC Versus DC Distribution Systems. Did We Get it Right?,” in IEEE Power Engineering
Society General Meeting, 2007, pp. 1–5, June 2007.
[15] Z. Liu and M. Li, “Research on Energy Efficiency of DC Distribution System,” AASRI Procedia, vol. 7, pp. 68–
74, 2014.
[16] M. Noritake, H. Hoshi, K. Hirose, H. Kita, R. Hara, and M. Yagami, “Operation algorithm of DC micro-
grid for achieving local production for local consumption of renewable energy,” in Telecommunications Energy
Conference ’Smart Power and Efficiency’ (INEC), Proceedings of 2013 35th International, pp. 1–6, Oct. 2013.
[17] M. Noritake, K. Yuasa, T. Takeda, H. Hoshi, and K. Hirose, “Demonstrative research on DC microgrids for
office buildings,” in Telecommunications Energy Conference (INEC), 2014 IEEE 36th International, pp. 1–5,
Sept. 2014.
[18] P. Paajanen, T. Kaipia, and J. Partanen, “DC supply of low-voltage electricity appliances in residential build-
ings,” in CIRED 2009 - 20th International Conference and Exhibition on Electricity Distribution - Part 1,
pp. 1–4, June 2009.

28
[19] M. Starke, L. Tolbert, and B. Ozpineci, “AC vs. DC distribution: A loss comparison,” in Transmission and
Distribution Conference and Exposition, 2008. T #x00026;D. IEEE/PES, pp. 1–7, Apr. 2008.
[20] S. Willems and W. Aerts, Study and Simulation Of A DC Micro Grid With Focus on Efficiency, Use of Materials
and Economic Constraints. PhD thesis, University of Leuven, Leuven, Belgium, 2014.
[21] D. L. Gerber, V. Vossos, W. Feng, C. Marnay, B. Nordman, and R. Brown, “A simulation-based efficiency
comparison of ac and dc power distribution networks in commercial buildings,” Applied Energy, vol. 210, pp. 1167
– 1187, 2018.
[22] D. L. Gerber, V. Vossos, W. Feng, A. Khandekar, C. Marnay, and B. Nordman, “A simulation based comparison
of ac and dc power distribution networks in buildings,” in 2017 IEEE Second International Conference on DC
Microgrids (ICDCM), pp. 588–595, June 2017.
[23] V. Vossos, D. Gerber, Y. Bennani, R. Brown, and C. Marnay, “Techno-economic analysis of dc power distribution
in commercial buildings,” Applied Energy, vol. 230, pp. 663 – 678, 2018.
[24] S. M. Frank and S. Rebennack, “Optimal design of mixed ac-dc distribution systems for commercial buildings:
A nonconvex generalized benders decomposition approach,” European Journal of Operational Research, vol. 242,
no. 3, pp. 710 – 729, 2015.
[25] R. Weiss, L. Ott, and U. Boeke, “Energy efficient low-voltage DC-grids for commercial buildings,” in 2015 IEEE
First International Conference on DC Microgrids (ICDCM), pp. 154–158, June 2015.
[26] U. Boeke and M. Wendt, “DC power grids for buildings,” in 2015 IEEE First International Conference on DC
Microgrids (ICDCM), pp. 210–214, June 2015.
[27] D. Fregosi, S. Ravula, D. Brhlik, J. Saussele, S. Frank, E. Bonnema, J. Scheib, and E. Wilson, “A comparative
study of DC and AC microgrids in commercial buildings across different climates and operating profiles,” in
2015 IEEE First International Conference on DC Microgrids (ICDCM), pp. 159–164, June 2015.
[28] N. P. Systems, “Fort Belvoir Direct Coupling DC Microgrid.”
[29] M. Wright, “Eaton demonstrates distributed DC power for LED lighting at LFI,” LEDs Magazine, May 2016.
[30] H. Kakigano, Y. Miura, and T. Ise, “Low-voltage bipolar-type dc microgrid for super high quality distribution,”
IEEE Transactions on Power Electronics, vol. 25, pp. 3066–3075, Dec 2010.
[31] Y. Ito, Y. Zhongqing, and H. Akagi, “Dc microgrid based distribution power generation system,” in The 4th
International Power Electronics and Motion Control Conference, 2004. IPEMC 2004., vol. 3, pp. 1740–1745
Vol.3, Aug 2004.
[32] V. Vossos, S. Pantano, R. Heard, and R. Brown, “Dc appliances and dc power distribution,” 2017.
[33] A. Mishra, K. Rajeev, and V. Garg, “Assessment of 48 volts dc for homes,” in 2018 IEEMA Engineer Infinite
Conference (eTechNxT), pp. 1–6, IEEE, 2018.
[34] M.-H. Ryu, H.-S. Kim, J.-W. Baek, H.-G. Kim, and J.-H. Jung, “Effective test bed of 380-v dc distribution
system using isolated power converters,” IEEE transactions on industrial electronics, vol. 62, no. 7, pp. 4525–
4536, 2015.
[35] G. Makarabbi, V. Gavade, R. Panguloori, and P. Mishra, “Compatibility and performance study of home
appliances in a dc home distribution system,” in Power Electronics, Drives and Energy Systems (PEDES),
2014 IEEE International Conference on, pp. 1–6, IEEE, 2014.
[36] M. Noritake, K. Yuasa, T. Takeda, K. Shimomachi, R. Hara, H. Kita, and T. Matsumura, “Experimental study
of a 400 v class dc microgrid for commercial buildings,” in Power Electronics and ECCE Asia (ICPE-ECCE
Asia), 2015 9th International Conference on, pp. 1730–1735, IEEE, 2015.
[37] A. Stippich, A. Sewergin, G. Engelmann, J. Gottschlich, M. Neubert, C. van der Broeck, P. Schuelting, R. Gold-
beck, and R. De Doncker, “From ac to dc: Benefits in household appliances,” in International ETG Congress
2017; Proceedings of, pp. 1–6, VDE, 2017.
[38] A. Jhunjhunwala, K. Vasudevan, P. Kaur, B. Ramamurthi, S. Bitra, K. Uppal, et al., “Energy efficiency in light-
ing: Ac vs dc led lights,” in Sustainable Green Buildings and Communities (SGBC), International Conference
on, pp. 1–4, IEEE, 2016.
[39] M. G. Jahromi, G. Mirzaeva, S. D. Mitchell, and D. Gay, “Dc power vs ac power for mobile mining equipment,”
in Industry Applications Society Annual Meeting, 2014 IEEE, pp. 1–8, IEEE, 2014.
[40] D. Das, N. Kumaresan, V. Nayanar, K. N. Sam, and N. A. Gounden, “Development of bldc motor-based elevator
system suitable for dc microgrid,” IEEE/ASME Transactions on Mechatronics, vol. 21, no. 3, pp. 1552–1560,
2016.
[41] R. K. Chauhan, C. Phurailatpam, B. Rajpurohit, F. Gonzalez-Longatt, and S. Singh, “Demand-side man-
agement system for autonomous dc microgrid for building,” Technology and Economics of Smart Grids and
Sustainable Energy, vol. 2, no. 1, p. 4, 2017.
[42] K. Yukita, T. Hosoe, S. Horie, T. Matsumura, M. Hamanaka, K. Hirose, and M. Noritake, “Air conditioning

29
 equipment using dc power supply system,” in Telecommunications Energy Conference (INTELEC), 2017 IEEE
International, pp. 229–232, IEEE, 2017.
[43] P. Rani, D. Vignesh, S. Gunaki, and A. Jhunjhunwala, “Design of converters for leveraging 48v dc power line at
homes/offices,” in Sustainable Green Buildings and Communities (SGBC), International Conference on, pp. 1–5,
IEEE, 2016.
[44] O. Lucia, I. Cvetkovic, H. Sarnago, D. Boroyevich, P. Mattavelli, and F. C. Lee, “Design of home appliances for
a dc-based nanogrid system: An induction range study case,” IEEE Journal of Emerging and Selected Topics
in Power Electronics, vol. 1, no. 4, pp. 315–326, 2013.
[45] OSHA 1910.303(g)(2)(i), “Guarding requirements for 50 volts or more DC,” standard, Occupational Safety and
Health Administration, Washington, D.C., 2015.
[46] BS 7671:2008, “Requirements for Electrical Installations. IET Wiring Regulations,” standard, Institution of
Engineering and Technology, Stevenage, England, 2015.
[47] R. Saidur, S. Mekhilef, M. B. Ali, A. Safari, and H. Mohammed, “Applications of variable speed drive (vsd) in
electrical motors energy savings,” Renewable and Sustainable Energy Reviews, vol. 16, no. 1, pp. 543–550, 2012.
[48] J. A. Rooks and A. K. Wallace, “Energy efficiency of vsds,” IEEE Industry Applications Magazine, vol. 10,
no. 3, pp. 57–61, 2004.
[49] M. Didden, J.-M. De Hoe, J. Callebaut, and E. De Jaeger, “Installing variable speed drives: Energy efficiency
profits versus power quality losses,” in Electricity Distribution, 2005. CIRED 2005. 18th International Confer-
ence and Exhibition on, pp. 1–5, IET, 2005.
[50] H. C. Emissions, “Guidelines to the standard en 61000-3-2,” European Power Supply Manufacturers Association,
2010.
[51] I. IEC, “61000-3-2: 2014 electromagnetic compatibility (emc)-part3-2: Limits-limits for harmonic current emis-
sions (equipment input current 16a per phase),” British Standards Institute, 2014.
[52] L. Gu, X. Ruan, M. Xu, and K. Yao, “Means of eliminating electrolytic capacitor in ac/dc power supplies for
led lightings,” IEEE Transactions on Power Electronics, vol. 24, no. 5, pp. 1399–1408, 2009.
[53] W. Chen and S. R. Hui, “Elimination of an electrolytic capacitor in ac/dc light-emitting diode (led) driver with
high input power factor and constant output current,” IEEE Transactions on Power Electronics, vol. 27, no. 3,
pp. 1598–1607, 2012.
[54] CUI Inc, “Efficiency standards for external power supplies.”
[55] SL Power Electronics, “Energy efficiency requirements by levels.” Application Note: AN-G-001/15.
[56] V. Deconinck, “Turning a quick charge 3.0 charger into a variable voltage power supply.”
[57] 1000bulbs.com, “Understanding led drivers,” May 2014.
[58] K. Bollmann and T. C. Penick, “Led lighting system with bypass circuit for failed led,” Apr. 2 2013. US Patent
8,410,705.
[59] S. Keeping, “High current, high brightness leds simplify power supply solutions,” February 2012.
[60] D. Solly, “Driving high intensity led strings in dc to dc applications,” 2007.
[61] “Universal Serial Bus Power Delivery Specification, Version 1.2, Revision 3.0,” standard, June 2018.
[62] Qualcomm, “Quick Charge 3.0 specs,” standard.
[63] F. G. Osorio, M. Xinran, Y. Liu, P. Lusina, and E. Cretu, “Sensor network using power-over-ethernet,” in
Computing and Communication (IEMCON), 2015 International Conference and Workshop on, pp. 1–7, IEEE,
2015.
[64] J. Petroski, “Power over ethernet thermal analysis with an engineering mechanics approach,” in Thermal Mea-
surement, Modeling & Management Symposium (SEMI-THERM), 2016 32nd, pp. 50–56, IEEE, 2016.
[65] J. Johnston, J. Counsell, G. Banks, and M. J. Stewart, “Beyond power over ethernet: The development of
digital energy networks for buildings,” in CIBSE Technical Symposium 2012-Buildings Systems and Services for
the 21st Century, pp. Session–5, 2012.
[66] E. Alliance, “380 Vdc Architectures for the Modern Data Center,” EMerge Alliance, San Ramon, CA, USA,
2013.
[67] D. E. Geary, D. P. Mohr, D. Owen, M. Salato, and B. Sonnenberg, “380V DC eco-system development:
present status and future challenges,” in Telecommunications Energy Conference’Smart Power and Effi-
ciency’(INTELEC), Proceedings of 2013 35th International, pp. 1–6, VDE, 2013.
[68] D. J. Becker and B. Sonnenberg, “DC microgrids in buildings and data centers,” in Telecommunications Energy
Conference (INTELEC), 2011 IEEE 33rd International, pp. 1–7, IEEE, 2011.
[69] EN ETSI, “Environmental Engineering (EE); Power supply interface at the input to telecommunications and
datacom (ICT) equipment; Part 3: Operated by rectified current source, alternating current source or direct
current source up to 400 V; Sub-part 1: Direct current source up to 400 V,” European standard EN 300 132-3-1

30
 V2.1.1, pp. 2012–02.
[70] D. L. Gerber, C. Le, M. Kline, S. Sanders, and P. Kinget, “An integrated multilevel converter with sigma-delta
control for led lighting,” IEEE Transactions on Power Electronics, pp. 1–1, 2018.
[71] M. Salcone and J. Bond, “Selecting film bus link capacitors for high performance inverter applications,” in
Electric Machines and Drives Conference, 2009. IEMDC’09. IEEE International, pp. 1692–1699, IEEE, 2009.
[72] W. Lukitsch, “Soft start vs ac drives-understand the differences,” in Textile, Fiber and Film Industry Technical
Conference, 1999 IEEE Annual, pp. 5–pp, IEEE, 1999.

31

View publication stats