Blackout Mitigation - Senior Design Project Lab | ECE 445, Study Guides, Projects, Research of Electrical and Electronics Engineering

Download Blackout Mitigation - Senior Design Project Lab | ECE 445 and more Study Guides, Projects, Research Electrical and Electronics Engineering in PDF only on Docsity!

Blackout Mitigation By Michelle Graden Edgar Roman ECE 345, SENIOR DESIGN PROJECT Spring 2004 TA: Joseph Mossoba 04 May 2004 Project No. 15

ABSTRACT

A line fault and a load fault are the two major types of blackouts that can occur in any power grid. The circuit in this experiment demonstrates a load fault. A common load fault occurs on hot summer days, usually in the evening, when people come home from work and turn on their air conditioners. These air conditioners increase the amount of current that is drawn out onto the lines resulting in a voltage dip. The current must increase in order to supply the voltage of the load, which leads to another voltage dip. This cascading effect could potentially cause a voltage collapse. The circuit in this experiment prevents such a voltage collapse from occurring by using a local controller. To simulate a power grid, four power supplies were assembled with various output voltages. The power supplies are regulated by the controller outputs, which are determined by five different inputs. Four of the inputs are user-defined priority settings and the fifth input is the input voltage to the power supplies. Finally, to simulate a disturbance in the system, a disturbance controller activates individual transient loads, which are then connected in parallel with each power supply load. ii

1. INTRODUCTION

The circuit proposed in this report is for blackout mitigation. There are three parts to this circuit: a controller, four power supplies with loads, and a disturbance controller. The controller is a microcontroller PIC that monitors the system by receiving five different inputs and using them to regulate which power supplies remain on or off. The four power supplies in this circuit are buck converters. Two of the converters step down a 42V source to a 5V output. The other two converters step down the same 42V source to a 12V output. All of the converters have feedback so that its respective output voltage is maintained regardless of the input voltage. The disturbance controller is also a microcontroller PIC and is used to cause stress on the system. The disturbance controller’s inputs are connected to a 5V logic source through a push-button. This controller’s outputs are attached to four different gate drivers, which are connected to transient loads. 1.1 Purpose The purpose of this project is to design a circuit that would exhibit blackout mitigation properties. The project involves building a local controller that is capable of avoiding a blackout situation when severe conditions are imposed on the simulated power grid. The project should also include commercial dc power supplies connected to various loads at the output. This project should include hardware that is capable of creating a stress on the system. This should cause power supplies to drop out and reduce the load on the system, allowing it to recover. 1.2 Specifications The circuit of this project needs to simulate a power grid capable of handling stresses imposed on it by external hardware. There is a local controller connected to each of the power supplies that regulates which supplies remain on or off. The determination of a power supply’s state is based on two factors: the voltage at the input of the power supplies and the user-defined priority settings. These two factors are further explained in section 1.3 of the report. 1.3 Subprojects The design of the project was made up of three parts with the addition of a transient load. The transient load was connected in parallel with the output load and controlled by a MOSFET switch. 1.3.1 Controller Module The controller module has control over which power supplies are taken out of the system and added back into the system based on user-defined priority settings. A high priority setting implies that particular load is of great importance and should be left on for as long as possible. A low priority setting, however, implies the load is of little significance and can be taken out if the soft bus voltage is low enough. The priority settings, one for each load, make up four of the five PIC inputs. The fifth input to the PIC is a stepped down version of the soft bus voltage. The step down in voltage is accomplished by using an op-amp circuit (for details of the op-amp circuit, see section 4). 1.3.2 Power Supply Modules Four dc-dc buck converters were needed to show a multi-load system of power supplies. The four converters received their input voltage from a soft bus, which was built with a 0.5  resistor in series with the 42V supply source. Two of the converters took the soft bus voltage input and delivered a 5V output. The other two converters took the same bus voltage input and delivered a 12 V output. Each

converter needed feedback to regulate the output voltage so that they simulated real power supplies. A transient load was added as a special modification made to each converter for testing purposes.

2.3 Disturbance Controller Design...........................................................................................

The disturbance controller has four outputs, one for each MOSFET of the transient load configuration. There are three different waveforms for the outputs, which will be deferred to section 2 of the report. The disturbance controller tests the functionality of the controller by causing a stress on the system.

2.4 Transient Load Design........................................................................................................

The transient load is made up of a resistor in series with a MOSFET switch. The four transient loads are connected in parallel with each of the converter output loads, but are not actually “in parallel” until the MOSFET is turned on, which is indirectly accomplished with the disturbance controller. This load will be used in the testing process to test controller functionality.

the MOSFET. The gate-driver used for the MOSFET switch was the MIC4423 instead of the IR because the MIC4423 was cheaper. Also, the complexity that came with using the IR2127 was not needed to turn on the MOSFET in the transient load.

2.5 Priority Settings and Push Button Design..........................................................................

The design of the priority settings was to use variable resistors. The variable resistors were connected to a 5V supply. By turning a knob on the resistor, a different voltage value was obtained. This voltage was then sent to its respective input on the controller module. There were four priority settings, one for each load. The push buttons were selected so that they made a connection between their two terminals only when the button was pressed. The terminals were hooked to a 5V supply and to the inputs of the disturbance controller. There were three push buttons, one for each disturbance waveform.

3. DESIGN DETAILS

3.1 Controller Module..............................................................................................................

The first step in designing the controller module was to pick out the appropriate microcontroller PIC to use. Based on availability and performance capabilities, the PIC16F872 was chosen to perform the tasks required of the controller. The key characteristic of the chosen PIC was the number of I/O pins available to the programmer. The controller required nine I/O pins to perform its required functions. The PIC has 6 A/D (analog-to-digital) pins and 8 I/O pins that could have been used. The A/D pins would be used for the five inputs to the PIC and the remaining I/O pins would be used as outputs to the inputs of the power supplies. The next task was to program the PIC. Before programming the PIC, it was necessary to create a flowchart that would bring structure to the code. The flowchart used for the controller program can be found in the Appendix under “Controller Flowchart”. Figure 3.1.1 shows the principle followed by the controller. If the soft bus input voltage went below the priority setting for one of the power supplies, then that power supply would be turned off. The soft bus voltage must be raised some delta Vi above the turn off value before the power supply would be turned back on so that the system would not re-trip. Figure 3.1.1: Flow Diagram of the Controller Turn On/Off Principle Originally, it was thought that the PIC would be programmed in assembly code. However, a software interface was used that had the capability to compile a C program into assembly code. The complete program used in the controller PIC can be found in the appendix under “Controller Program” (NOTE: the numbers on the left hand side of the program were not in the original code, but are there now for reference). The five expected inputs to the PIC are listed in line 10 of the code. The first variable represents the input voltage from the soft bus. The remaining four variables in line 10 represent the four user-defined priority settings. Lines 11-15 are used in the program to define which supplies are on. A “1” indicates that the power supply is on and a “0” indicates that it is off. Looking at Figure 3.1.1, a horizontal line test will show two possible values of the soft bus voltage. If the left-hand-side path is being followed, this particular power supply should remain on. Conversely, if the right-hand-side path is being followed, this particular power supply should be off. Hence, it is necessary to include these indicators in the program for proper controller functionality. From line 30 through line 50, every other line contains the same code: delay_us(10). This segment of code is interpreted in the program as a 10-microsecond delay. This line of code allows the A/D converter to read in the input and convert it to a digital value that the ALU (Arithmetic and Logic Unit) can understand. Leaving out the time delay during the initial coding of the PIC resulted in inaccurate outputs. V_bus V_off V V_on

R MOSFET Diode C1^ Vout L VIN From gate driver Figure 3.2.1: Basic Schematic of a Buck Converter D = Vout/ Vin (3.2.1) Iout = Pout/Vout (3.2.2) Iin = DIout (3.2.3) Rout = Vout/Iout (3.2.4) Vin (V) Vout (V) D Iin (A) Iout (A) Rout (Ω) 42 5 0.119 0.476 4.000 1. 42 12 0.286 1.190 4.167 2. Table 3.2.1: Calculated D, Iin, Iout, Rout values for 42-to-5 and 42-to-12 V Converters Equation 3.2.5, relating Vout and D, and equation 3.2.6, relating efficiency and D, were derived to see the general behavior of the converters given a certain duty ratio. (For the complete derivations of and η, please see Derivation 1 and Derivation 2, respectively, in the Appendix.) The graphs are shown below in Figure 3.2.2 and Figure 3.2.3 for the 42-to-5 volt and 42-to-12 volt converters, respectively. = (DVin + (D-1)Vd)R/(D(Rds+RL)+R+(1-D)RL) (3.2.5) η = /(42*D) (3.2.6) Figure 3.2.2: 42-to-5 Volt Converter Figure 3.2.3: 42-to-12 Volt Converter

After the basic calculations above were made, the switching frequency of both converters was chosen to be 35 kHz (why?). Calculations were then done to pick out specific components for each converter, making sure that each component had the necessary ratings to operate in the circuit. The ripple current, Δii, was specified as 1 A to calculate the inductance, L, of each converter. Load regulation was limited to ± 2 % and line regulation was limited to ≤ 2 % output ripple voltage to help eliminate some of the circuit noise. Equation 3.2.9 was found directly by the line regulation specification. (For derivation of equations 3.2.7, 3.2.8, and 3.2.10, look in the Appendix under Derivation 3, Derivation 4, and Derivation 5 respectively.) Using equations 3.2.7-3.2.10, the Lcrit, L, Δiv, and C values were found and are summarized in Table 3.2.2. The value of Lcrit was calculated to compare to the calculated inductance of each of the two converter types. The chosen inductance must be considerably larger than Lcrit to ensure continuous mode operation of the converters and positive inductor current at all times. Notice that the C value actually chosen, Cchosen, is much larger than the value calculated. This was done to limit any possible ESR jumps that could come into the converter modules, which can be a serious problem if not limited properly. Lcrit = Rout(1-D)/(2fswitch) (3.2.7) L = (1-D)Vout/(Δiifswitch) (3.2.8) Δiv = .02Vout (3.2.9) C > (1/(fswitch)^2)D(1-D)Vin/(8ΔivL) (3.2.10) Vout (V) Lcrit (μH) Δi (A)i (A) L (μH) Δi (A)v (V) C (μF) Cchosen (μF) 5 15.731 1 125.850 .10 35.714 100 12 29.388 1 244.898 .24 14.881 47 Table 3.2.2: Calculated Lcrit, L, Δi (A)v, and C values for 42-to-5 and 42-to-12 Volt Converters Now that all of the preliminary calculations of the converters were complete, properly rated components had to be picked out. A program from the Micrometals™ website was used to size the inductor and obtain the proper gauge wire. The program gives multiple choices of inductances, but leaves the final decision up to the user. It was previously decided that the inductors would be toroidal-shaped powdered- iron cores of type material 26. This material was chosen because it is the cheapest of all the given materials. Comparing the choices between the 42-to-5 volt and 42-to-12 volt converters, a T200- toroid was picked for both converters since it was available in the parts store. This choice made it necessary to purchase 12 AWG and 14 AWG wire for the 42-to-5 volt and 42-to-12 volt converters respectively. The program calculated the need for 37 turns for the 42-to-5 volt converter and 52 turns for the 42-to-12 volt converter. Obtaining the mean length per turn (MLT) of the material, it was found that approximately 8 feet of wire would be needed to wrap an inductor of the 42-to-5 volt converter and 12 ft. of wire would be needed to wrap an inductor of the 42-to-12 volt converter. See equation 3.2. below for the wire length of a given inductor. Wire length = (MLT) * (# turns) (3.2.11) The MTP50NO6V MOSFET was chosen based on its voltage and current capabilities as well as its low resistance while in use. This power MOSFET has ratings up to 42 A, 60 V, and Rds(on) of 28 mΩ, which is the drain-to-source resistance. A MBR2545CTP shottky diode was chosen because of its voltage and current ratings, which were better than a typical diode. This diode can handle 30 A, 45 V and was readily available in the parts shop. The power resistor across the output had to be rated for the proper wattage at the output. Since the exact values in Table 3.2.1 could not be found for Rout, a 1.5 Ω resistor

Originally, the resistor in the transient load was approximately .2. However, calculations showed that such a resistance would require an input current of about 3.37A for each of the 5-volt converters and about 18.3A for each of the 12-volt converters. The current ratings of the input supply and the power boards would be exceeded with these calculated currents. The large currents would have also exceeded the current capacity of both gauge wires used to form the inductor. Therefore, the transient load resistance was increased to draw safer levels of current from the power supply.

4. DESIGN VERIFICATION

4.1 Module Testing.................................................................................................................

Modulized testing was performed to help debug the overall circuit more quickly. The three modules that were tested separately were system control, power supplies, and disturbance control. 4.1.1 System Control The controller PIC controls the state of the system by regulating which power supplies are on or off. When a controller output is high, that particular power supply is on. Conversely, when a controller output is low, that particular power supply is off. The determination of how long a power supply is on depends on its given priority setting. To test the controller, four priority settings were set at the priority inputs of the PIC. Table 4.1. tabulates the voltage values at each input. Input (Pin #) Voltage 1 (3) 2. 2 (4) 2. 3 (5) 3. 4 (7) 1. Table 4.1.1: Voltage Value at Priority Setting Pin of Controller A variable voltage source was at pin 2 of the controller. The voltage source at pin 2 was varied between 5 and 0 volts. Looking at Table 4.1.1, the voltage value at the pin indicates the priority level of the power supply. Pin 3 represents a power supply with the lowest priority setting of the four power supplies. This is known because its voltage value is the highest, meaning that it will be turned off first. Pin 4 represents a power supply with the highest priority of the four power supplies because its voltage value is the lowest, which implies that it will be turned off last. The voltage at pin 2 was varied slowly during the actual testing of the controller. Figure 4.1.1 shows the outputs of the controller in a small time frame (NOTE: Channel 1 represents the output with respect to input 1. The other channels were connected in the same fashion). In Figure 4.1.1, the outputs of the controller drop from logic level high to logic level low in the order of highest input voltage to lowest input voltage. Figure 4.1.1 shows that the controller works properly by shutting off power supplies in the correct order. Figure 4.1.1: Controller Outputs as Pin 2 Voltage Is Varied From 5V to 0V

1 2 3 4 5 6 7 8 9 18 17 16 15 14 13 12 11 10 5 kOhms 47 Ohms (^) 1 nF 12 V 1 2 3 4 8 7 6 5 2.2 uF 2.2 uF 1N 12 V from dual supply PWM Gate Drive To MOSFET gate To MOSFET source Figure 4.1.3: Open-loop of PWM and Gate Drive Circuits of the Converters Figure 4.1.4 shows open-loop operation of a working 42-to-12 V converter. The light blue line is the current going through the inductor of the converter, which shows the switching of the MOSFET. The dark blue line is the output of the SG3526. The light green line is the clock of the SG3526. The dark blue and light green lines show that the PWM chip is working. The pink line shows that the IR2127 is working and following the PWM chip waveform correctly. Figure 4.1.4:Open-Loop Working Buck Converter Once each individual power supply was working, feedback circuitry was added. Figure 4.1.5 shows the complete schematic for a working buck converter with feedback. Table 4.1.3 shows the values of the variables in Figure 4.1.5 for the 42-to-5 V and 42-to-12 V converters respectively. The output of the converter is put through a voltage divider connected to an op-amp, which is internal to the SG3526. This feedback circuitry replaces the 5 kΩ variable resistor. The only difference between the 42-to-5 V and 42-to-12 V converters, other than the values of the components in Table 4.1.3, is an extra voltage divider at pin 1 of the SG3526. This voltage divider is shown as using the same R3 and R4 values in Figure 4.1.6 as it does in Figure 4.1.5.

Vout (V) R 1 (kΩ) R 2 (kΩ) R 3 (kΩ) R 4 (kΩ) R (Ω) C 1 (μF) L (μH) 5 100 1 10 10 1.5 100 130 12 100 1 6.8 4.7 3 47 300 Table 4.1.3:Chart of Values for Converters Represented in Figure 4.1. 1 2 3 4 5 6 7 8 9 18 17 16 15 14 13 12 11 10 R 47 Ohms (^) 1 nF 12 V 1 2 3 4 8 7 6 5 2.2 uF 2.2 uF 1N 12 V from dual supply PWM Gate Drive 2.2 mF R R R MTP50N R R MTP50N MBR2545CTP C1 Vout L 42 V MIC4423CN Transient Load External source From PIC Buck Converter Figure 4.1.5: Schematic of Buck Converter with Feedback R R 5 V Reference from pin 18 To pin 1 of SG Figure 4.1.6:Voltage Divider of 42-to-5 V Converter at Pin 1 Feedback adjusts the duty ratio autonomously so that the output remains at a constant voltage. To test feedback capabilities of each converter, several data points were taken while varying the input voltage. Table 4.1.4 shows the tests done for each converter with input voltage ranging from 25 V to 42 V. Notice that the two 42-to-5 V converters have feedback capabilities with input voltage as low as 25V, but the 42-to-12 V converters don’t have feedback capabilities until around 27.5 V. Notice also that converter 1 of the two 42-to-12 V converters has lower values compared to converter 2. This could be due to testing this particular converter at a different lab bench on a different day. Not all power sources in the lab will output an identical current or voltage because everything has a certain margin of error when it is made.

42-to-5 V 1

42-to-5 V 2

42-to-12 V 1

42-to-12 V 2

Table 4.1.4: Feedback Test of Single Converters 4.1.3 Creating a Disturbance The job of the disturbance controller is to turn the transient loads on and off in three different patterns. All four outputs of the disturbance controller were attached to an oscilloscope to test the program capabilities. A 5-volt signal was applied individually to each of the three push-button inputs. A description of the three output waveforms can be found in the “Disturbance Flowchart” located in the Appendix. Figures 4.1.7 through 4.1.9 show the output waveforms of the push buttons 1, 2, and 3 respectively. The actual time that each transient load is on in the figures, where an output is logic level low, is 50 ms for the first two figures and 25 ms for the last figure. Later on, the time that a transient load was on for changed to 2 s and 1 s, respectively, so that the effects could be more noticeable. Waveform plots of the change were not taken, but the disturbance controller was retested to verify the waveform outputs.

Figure 4.1.7: PB1 Waveform Figure 4.1.8: PB2 Waveform Figure 4.1.9: PB3 Waveform

4.2 Integrated Testing.............................................................................................................

Since the modules worked properly, the next step was integrated testing. One 42-to-5 volt converter was connected between the controller and the disturbance controller. One output from the controller was connected to the PWM chip of the converter. One output of the disturbance controller was connected to the gate-drive of the transient load MOSFET switch. The outputs of both controllers were logic level high so that the buck converter was on and the transient load was off. In order to verify that the controller and converter worked together properly, the 42V input supply was manually lowered. Instead of testing all three units at once, only two of the units were tested together to limit damaging the whole circuit if a serious problem arose. This would also save time debugging the circuit. The priority setting for this converter was set at 4.2V using a voltage divider. The power supply turned off when the input supply was manually decreased to approximately 34V. The voltage at the soft bus input pin also read 4.2V. To conclude the test, the input supply was increased back to 42V. The oscilloscope showed that the converter was trying to turn back on when the input supply was around 39V. Unfortunately, the converter could not stabilize itself and never fully turned back on. Even when the input voltage was increased a bit further but to no avail. Figure 4.2.1 shows a waveform plot of the converter trying to turn on. Figure 4.2.1: Integration Flaw Between the Controller and the Converter The channel 1 waveform is the output of the controller, channel 2 is the soft bus input to the controller, and channel 3 is the current coming out of the input supply. Figure 4.2.1 shows the current from the input supply increasing when the converter is commanded to turn on. The jagged plot is due to the MOSFET switching action of the buck converter (see Figure 4.1.2). As the current out of the input supply increases, the voltage at the soft bus begins to drop. The largest soft bus voltage drop is 0.96V, which is large enough to turn off the converter. The difference between the turn off voltage and the turn on voltage for this particular priority setting is only 0.42V (NOTE: Line 62 in the “Controller Program” or any similar line has still not been added yet). The current decreases to 0A and the soft bus voltage returns to its nominal value when the converter turns off. The controller reads the new input and turns the converter back on only to experience the same problem. Two key improvements to the project were made after this flaw was encountered. The first correction removed the voltage divider as a way of stepping down the soft bus voltage to the controller. At 42V, the input to the controller was approximately 4.94V using the voltage divider. When the input voltage was decreased to around 34V, the controller input voltage was 4.2V. After an 8V drop at the soft bus, only 0.7V had dropped at the input of the controller. Therefore, the priority setting voltages would have to be set within this small voltage range. To correct the problem, an op-amp circuit was used instead.