2.8 MB PowerPoint

Download Report

Transcript 2.8 MB PowerPoint

Critical Design Review
December 11, 2012
Christopher Corey, Josh Crowley, John Fischer,
Tim Myers, Neil Severson, Kristine Thompson

Design and implement smart microgrid
energy delivery system

Combine multiple/varied energy sources in
most efficient use of resources possible

Design system to be as grid-independent as
possible

Detect real time power availability and load
demand

Convert sources to single DC bus and deliver
required energy to loads

Store energy in battery system for use when
resources are unavailable

Monitor load usage and display to user through
web interface





Predictive load profiling
Weather solar resource prediction
System mode control by the user
Load prioritization and control
Add scalability
 Allow for multiple source possibilities
 System architecture may be followed for higher
power applications

Power electronics
 Buck DC-DC converter
▪ Gate Driver
▪ Current Control
 Full-wave rectifier

Current and voltage sensing
 AC
 DC


Battery Charging/Monitoring
Interface with control architecture


Step down PV/rectified grid voltage to DC
bus efficiently
Design elements to minimize losses
 Conduction
 Switching
 Size for power level used

Control current draw and power point on PV
panel using feedback loop
 References provided by central controller
 fs

= 100kHz
∆iL = 0.3*IL
 Keep out of DCM
𝑉𝑖𝑛 −𝑉𝑜𝑢𝑡 ∗𝑇𝑠

L=

C = 3.3 mF
∆iL
= 180μF
 Cutoff above switching
frequency
Functional
Decomposition
Level 2
Main Controller
Solar DC-DC
Converter
Current Control Input
Current/Voltage
Current
CurrentProgrammed
Controller
Current
Converter Control Signal
Solar Panel
Variable DC
Buck
Converter
Battery Voltage



In contrast with a voltage comparator
To set duty cycle
Ideal for implementing a charge controller
 Necessary for OPPT
 Since current is being measured and compared,
most accurate

Not ordinarily stable at duty cycles > 50%
 Therefore add a slope compensator
Buffer creates a signal
appropriate for the MOSFET
Clock sets the output high
at the beginning of the period
Artificial ramp
stabilizes circuit
When comparator is triggered output
drops low, setting the duty cycle
Clock sets the
output high
Comparator condition
met, output set low
Slope compensation and
reference current
Inductor current

Choosing a chip to match our requirements:
 Large duty cycles
 100kHz frequency
 12V operation
 Current sense/ mode control
 Good documentation
Slope compensator
for stability
Current Transformer for
isolation and efficiency
Oscillator
Input filter


However the duty cycle output is 5V
Not sufficient to drive the MOSFET
 Vgs = Gate, Source Voltage
Minimum
12V


The gate driver takes the 5V duty cycle and
converts it to a signal for the MOSFET
Represented by the buffer on the output
1
2
3
4
12V
0V
Voltage
Limit
Step down
transformer
Vcc
Dampening
Resistor
C?
R?
Mosfet Gate
a
1
Duty Cycle
2
3
From CPC
4
NC
NC
IN A OUT_A_N
GND
VDD
IN B
OUT_B
8
C?
7
6
12V
T?
Cap Semi
1uF
Res3
1
Cap Semi
1uF
D?
1N4148
D?
1N4148
5
R?
Res3
10K
D?
12V Zener
TC4428
2:1
MOSFET Driver
12V
-12V
Mosfet Source
Charge
Capacitor
Bleed
Resistor
Functional
Decomposition
Level 1
SEND
Web Interface
Load Data
Display
Control Signal
Main Controller
User
Current / Voltage
Mode Command
Current
Voltage
Solar Panel
SCR Control
Solar DC-DC
Converter
Variable DC
Inverter
Current
SCR
120V AC
Load
Light bulb
10W
SCR
120V AC
Load
Laptop 20W
SCR
5V DC
Load
Phone Charger
5W
120V AC
Mode Command
Battery Voltage
AC Line
120V AC
Grid AC-DC
Rectifier
Buck
Converter
Legend
Measurement Signal
Control Signal
Power
Energy Storage
5V DC
Full bridge rectifier
T?
D?
Bridge1
P?
1
2
Grid Input
Grid Connection
C?
Smoothing Cap
Trans Ideal
GND
Smoothing capacitor
for DC voltage



Necessary to match the changing battery
voltage
Steps down input voltage
Requires current programmed controller and
gate driver
 Signal from main controller

3.3V UART
 Beaglebone

SPI
 AC Current/Voltage Sensing

ADC
 DC Current/Voltage Sensing

PWM
 Current Reference



ADE7753 5V TTL Logic
MSP430 3.3v
ADE7753 connected to
high voltages, such as
120V RMS on the grid
connection.


To minimize code refactoring, we isolated
hardware dependent code in driver software
modules.
Minimized changes when transitioning from
MSP430f636 to MSP430f6333





Software state determined by battery state
of charge.
State 0 – Initialization
State 1 – Low Battery
State 2 – Sufficient Battery
State 3 – Maximum Battery

Python
 Serial Interface
 Weather Forecasting
 Database Connection (Write)

Mysql
 Single Database Multiple Tables

Lighttpd
 Single site send.int.colorado.edu

PHP
 Database Connection (Read)

Software Drivers developed and tested on
the MSP430 F6736 series
 Serial
 Analog to Digital Conversion

Load Monitoring Prototype
 Open-Loop Toroid

User Interface
 Minimizing use of the BeagleBone


Testing and Prototyping done on F6736
series
Sampling Times
 ¼ second per 1000 samples

Considering a move to F433x series given an
increase in hardware ADCs
 F433x series provides 12 ADCs at up to 12bit
precision

Drivers would need to be ported
Current Sense Accuracy
1
0.9
0.8
0.7
0.6
Current Measured
0.5
[Amps]
0.4
Kill A Watt
Current Transformer
0.3
0.2
0.1
0
1
2
3
4
5
6
7
8
Sample Number
 Resistive loads should be more accurate
indicating incorrect calibration constant
 Non-Linear Differences when
adding/removing loads
 Mitigate using Energy Sense IC with
<0.1% error
Sample Number
1
2
3
4
5
6
7
8
Load Type
Complex Small Laptop
Resistive Fan Low
Complex Small Laptop +Fan(Low)
Resistive Fan High
Complex Single Laptop +Fan(High)
Complex Two Laptops
Complex Large Laptop +Fan Low
Complex Large Laptop +Fan High

Energy Data
 Real-time updating graphs of load usage

Weather Prediction
 Solar radiance prediction using cloud cover data
from weatherunderground.com

Javascript + Highcharts
 Fast rendering

PHP Development
 Future of the User Interface

A perf-board prototype was created to test
the buck converter design

Components sized to possible power output
of solar panel

Tested with power supply at a range of
voltages
Efficiency at 30, 50, 70% Duty Cycle
1
0.95
0.9
Efficiency
0.85
0.8
0.75
30%
0.7
50%
0.65
70%
0.6
0.55
0.5
0
10
20
V in
30
40

A Simulink model was also created for the
buck converter using the Simscape (circuit
elements) library

Useful for higher system level modeling

Matlab Simulink model created to test peak
power tracking algorithm

Changes value of a resistor connected to a
solar panel to draw max power

Have received a solar panel to use from the
department

Rated for 80 W

Thin film chemistry creates
slightly different IV curve

Battery usage

Measurement accuracy

Microcontroller usage

System integration

Accurate estimation of SOC
 Errors reduce life cycle of battery
 Adding temperature measurement

Overcharging protection
 Overcharging harmful to AGM batteries
 Conservative calculations

AC current measurement
 Using energy sense IC provides optimal accuracy

DC current measurement
 Current transformer
▪ Available for controller from hardware in converter and
rectifier
 Sense resistor amplification
▪ Battery current measurement

PV power measurements
 .2% error on PV Watt output with 10-bit ADCs
 Determined to be acceptable for PPT algorithms

Battery SOC calculations




Need accurate voltage set-points
10-bit ADC produces ~12mV step size
12-bit ADC produces ~3mV step size
12-bit preferable for charge control algorithm

Controller I/O
 MSP430 model has required ADC, UART, and SPI
channels

Computation timing
 Algorithms: PPT, Charge control, SOC

System failure
 Controlled boot-cycle reduces hardware fail-safe
usage

Power electronics
 Loading, noise, harmonics, interference
 Up to three board revisions planned and budgeted

Boot sequence
 Power controller regardless of battery SOC
 Hardware will be connected directly to battery

Failsafe mechanisms
 Overcurrent Protection on each board

Enables outdoors testing

Easy board mounting

Solar panel adjustment

Received funds from UROP and EEF

Total funds: $3200

Obtained some parts for free, some on loan
Category
New Expected
Spent
Solar
60
0
Load Monitoring
300
10
Controller
370
0
Rectifier
210
0
Converter
310
20
Inverter/Converter for Loads
120
0
Energy Storage
60
60
User Interface
40
0
Web Interface
20
0
Loads
260
0
Total
1750
90
Task
Primary
Secondary
Network Interface
John
Kit
Load Monitoring
Kit
None
Controller H/W
Kristine
John
Solar Converter
Josh
Kristine
Grid Rectifier
Tim
Neil
Power Point Tracking
Tim
Josh
Controller S/W
Neil
Kit
Battery Management
Neil
None

Dragan Maksimovic

Robert Erickson

Trojan Battery

Advanced Circuits