Design Project - Purdue University College of Engineering



Homework 3: Design Constraint Analysis and Component Selection Rationale

Due: Friday, February 10, at NOON

Team Code Name: H.E.A.D. Gear Group No. 12

Team Member Completing This Homework: Eric Aasen

Evaluation:

|Component/Criterion |Score |Multiplier |Points |

|Introduction |0 1 2 3 4 5 6 7 8 9 10 |X 1 | |

|Analysis of Design Constraints |0 1 2 3 4 5 6 7 8 9 10 |X 3 | |

|Rationale for Component Selection |0 1 2 3 4 5 6 7 8 9 10 |X 3 | |

|List of Major Components |0 1 2 3 4 5 6 7 8 9 10 |X 1 | |

|List of References |0 1 2 3 4 5 6 7 8 9 10 |X 1 | |

|Technical Writing Style |0 1 2 3 4 5 6 7 8 9 10 |X 1 | |

| |TOTAL | |

Comments:

______________________________________________________________________________

______________________________________________________________________________

______________________________________________________________________________

______________________________________________________________________________

______________________________________________________________________________

______________________________________________________________________________

1. Introduction

This design is implementing a digital control system with a CAN interface for an existing DC-DC power converter board that is on the Purdue Solar Racing solar car. The design will replace an existing board and as such, will be tightly constrained both in form and function. The design must have a board area of no larger than 1.8” by 3”. The design must retain all functionality of the existing board while including additional interfacing, lower power consumption, and more software functionality all while maintaining a reasonable cost.

2. Design Constraint Analysis

The major design constraints for this project will be board size, DC accuracy, efficiency, low cost, and high reliability. Since this design will have to fit within an already established small form-factor and connector positioning, PCB design for high board utilization is key. There will be up to 8 of these boards on the solar car to convert the solar array energy to the battery voltage, designing for the lowest power consumption is critical, since every mW this design uses is multiplied by eight on the actual car. High DC accuracy (2% over the full input range on all signals) is critical, not only to finding the maximum power point of a solar array, but also for any data that will be gathered by the vehicle’s telemetry system as to array power output and converter efficiency. In addition, cost must always be considered in the design and part selection since the Purdue Solar Racing Team is a completely sponsor-funded organization that must save money wherever possible. Lastly, this design will be subjected to a very harsh operating environment where temperatures are regularly above 120F and high-vibration and high electrical noise is common, so reliability and component ratings are of critical importance.

1. Computation Requirements

The initial tracking algorithm this design will use will operate at approximately a 20Hz refresh rate. This means that the algorithm will read all input and output voltages and currents 20 times a second and compute the input and output power at the same rate. To actually calculate the power, the measurements of voltage and current are multiplied by scaling factors to yield their true value. They will then be multiplied together to get a power measurement. These calculations will involve 3 floating point multiplies for the input power computation, 3 multiplies for the output power computation, and one floating point multiplication to scale the temperature measurement of the main DC-DC switch. This yields a total of 7 floating-point multiplies at a rate of 20Hz. When the algorithm is fully implemented on the microprocessor, an analysis will be performed to determine if the refresh rate of the algorithm can be increased to further enhance the power-point tracking capabilities of the design.

2. Interface Requirements

This design will not only provide interfacing to and control of a DC-DC power converter board, but will also provide visual indication of proper operation and any error conditions that may exist and provide for MPPT cooling via a connector for an external fan. To interface with the DC-DC power board, only one GPIO pin is required. This pin will provide a control signal to a BJT on the power board that controls the input activation relay. This BJT/relay combination can be driven by a port pin with an output capability of 5mA. To provide visual indication of the status of the converter, a minimum of 3 LEDs will be used; one for a software heartbeat, one as an error indicator, and one to indicate the converter is operational, but not outputting any power. To power the fan, a standard 3-pin PC fan header will be utilized, with the 12V to the fan being controlled by a port pin via a low-side power MOSFET. The tachometer output of the fan will also be input into an input capture pin on the microcontroller for potential future fan feedback control. Since the I/O requirements of this design are well defined and will not require any additional interfacing to external devices such as LCDs, keypads, or any other external I/O devices and given the space constraints, it was determined that a PLD would not be optimal to include in the design.

3. On-Chip Peripheral Requirements

This design will require a multitude of on-chip peripherals to interface to analog signals, a DC-DC converter MOSFET, a CAN network, and an RS-232 network. To measure the input and output voltage and current, in addition to at least one temperature sensor, a minimum of 5 A/D channels are required. To control the DC-DC converter board’s conversion ratio, a single PWM channel is needed, running at as high a resolution as is possible, given a 20kHz switching frequency. The new communications network being implemented in the solar car is CANOpen. The easiest way to interface to this network is with an on-board CAN controller with one channel for bi-directional communication. For debug purposes and to reduce the necessity of having a device on a laptop which can interface to CAN, a single RS-232 port will be necessary for debug and to put the converter into special conversion modes.

4. Off-Chip Peripheral Requirements

The TX and RX outputs of a CAN controller are generally not in the proper CAN differential voltage ranges. As such, a CAN transceiver is necessary to translate the half-duplex signals from the CAN controller to the full-duplex, differential signaling required by the CAN physical layer interface.

5. Power Constraints

The solar array on the solar car puts out a maximum 1200W of power. It is absolutely critical for as much of that power to be transformed into the mechanical motion of the car as is possible. Given that, low power consumption is crucial. Also, since there are 8 converters necessary for the solar array, for every 1mW consumed by this design, 8mW will be consumed on the solar car.

6. Packaging Constraints

As stated previously, this design must interface with an existing PCB and fit within the form factor of 1.8” by 3” that is able to fit on the power board. This is additionally important as ventilation enclosures for groups of power trackers are being designed by Purdue Solar Racing’s Aeronautical team to help cool them and overall MPPT dimensions must not change from the current values to prevent enclosure interference.

7. Cost Constraints

Commercial devices with a similar controller design, but a much more complicated power converter design retail for $780 with a CAN interface [1]. With Purdue Solar Racing being a very cost-sensitive group, this design cannot cost nearly this much. With 8 modules on the car, if the MPPT cost $780, the total cost would be $6240, which is approaching the total cost of the solar cells that are used on the car. A more reasonable cost for this design would be $40, which, when combined with the power board cost, brings the total unit cost to $80. This brings the total of 8 units to $640, or less than the cost of one commercial converter. With that estimate in mind, the goal will be to have all 8 converters cost less than one commercial MPPT with CAN, i.e. this design cannot cost more than $57 in raw parts (not including PCB, since PSR generally gets those for free).

3. Component Selection Rationale

1. Microcontroller

The particular manufacturer and series of microcontroller were design constraints, as Microchip PIC18F series microcontrollers are being used elsewhere on the car. Within this series of microcontrollers and given the I/O and on-chip peripheral constraints listed previously, there were four choices that fit the minimum requirements: the PIC18F2480, PIC18F2580, PIC18F2585, and the PIC18F2680. Since these parts were all pin-compatible with each other and their only major difference was the amount of on-board memory, the part with the largest memory, the PIC18F2680, was chosen. This microcontroller has a program memory size of 64kB, a RAM size of 3328B, and a data EEPROM of 1kB. It also contains an on-board CAN controller, 1-Ch PWM output, 8-Ch of 10-Bit A/D, and a USART [2]. The price of this microcontroller is quite reasonable, with a budgetary price of $5.54. There are two package options for the microcontroller, a 28-pin 300 mil PDIP and a 28-pin 300 mil SOIC. To keep the size as small as possible, the SOIC package option was chosen.

If complete freedom were given to choose a microcontroller manufacturer and device, there are many more options. One such series is the AT90CANXXX series from Atmel. These devices also include an onboard CAN controller, PWM output, Dual USARTs, and 8-CH of 10-Bit A/D [3]. Unfortunately, these devices only come in an LQFP 64 or an MLF 64 package option, of which the LQFP would have had to have been used given its relative ease of layout and assembly when compared to the MLF style.

Another option would be a microcontroller from Freescale. Using the Freescale Microcontroller selector guide[4], it was found that multiple options exist for microcontrollers with CAN from their 8-Bit HC08 series up to their 32-Bit PowerPC MPC processors with three CAN channels and 1Mbit of on-board Flash. Looking further into the 8-Bit HC08 controllers, two controllers stand out, the XC68HC08AZ32 and the MC68HC08AZ60, which both meet the minimum on-board peripheral requirements, but both of which have only an 8-Bit A/D converter, a maximum frequency of 8.1MHz, and come in large, 64-pin QFP packages like the Atmel.

Between the three sets of microcontrollers listed above, the PIC would still have been chosen for many options:

1. Smaller footprint and less pins than either the Atmel or Freescale microcontrollers.

2. 10-Bit A/D converter vs. the 8-Bit converter of the Freescale.

3. Higher maximum clock frequency.

4. Availability of CAN-Specific development board (Both Atmel and Microchip have a development board, but Freescale does not)

5. Low price (~$16 for Atmel, ~$16 for Freescale, ~$12 for PIC from Digikey)

2. Power Supply

The power supply is of critical importance to this design to keep power consumption to an absolute minimum. Two major types of power supplies were considered, Linear Regulator based and switch-mode DC-DC. Linear regulators offer design ease, low part counts (generally just the input capacitor, the regulator, and the output capacitor), very high noise rejection and very low output ripple. Unfortunately, given an input voltage of 12V, an output of 5V and a load of 100mA, these power supplies are only ~42% efficient and have to dissipate 0.7W of power. At lighter loads of say 10mA, the power dissipation is lower, but the efficiency is still low. Given these efficiencies, it was decided that a switch-mode DC-DC converter would provide a better solution and would achieve low output ripple if care is taken. One way of providing this would be to have the DC-DC converter provide a slightly high voltage, say 5.5V and then use an LDO linear regulator for final power supply regulation, but this option seemed to be overly complex given the requirements. The other way of ensuring the lowest possible output voltage ripple is to utilize very low ESR capacitors on the input and output. For this power supply, solid Tantalum capacitors were chosen to provide very low ESR (approximately 40mΩ for a 10V, 33uF capacitor [5]) The most efficient and lowest part-count DC-DC controller chip was found to be the MAX1684 by MAXIM. This converter operates at 300kHz with an efficiency of up to 96% at a full-load output of 1A. Oftentimes this is misleading as DC-DC power supplies are notorious for having poor low-load efficiency. However, this converter boasts an efficiency of ~86% at an output current of only 10mA and with an input voltage of 12V, will maintain an efficiency of 90% for an output current range of 30mA to 400mA [6], which is where the expected current consumption of this design is estimated to lie. This converter is available in a 16-pin QSOP package. The inductor for this converter must be 47uH for continuous conduction at a load current of 1A. Multiple surface-mount inductors of various current capabilities and packages sizes were sampled from Coilcraft using their selection guide [7]. The most appropriate option for the inductor will be chosen via experimental comparison between the different inductor once a more accurate power budget for this design is determined.

3. Op-Amp/CAN Transceiver

Given the accuracy, power consumption, and cost restrictions in this design, the MCP6004 from Microchip was selected that had a quiescent supply current of 100uA/amplifier, a typical input bias current of 1pA, typical input offset voltage of 4.5mV [8] and a budgetary cost of $0.39, it was a good fit. This package houses 4 amplifiers and comes in SOIC or TSSOP surface-mount variations.

To further ensure accurate data measurement and conversion, a 4.096V voltage reference will be used to supply both the op-amps and the A/D converter voltage reference input. One option for this purpose is the MAX6241 that has up to a 10mA output with an output voltage drop of less than 50uV [10], though final component selection for this part has not been determined as of yet.

The CAN Transceiver is the device that translates the TX and RX signals from and to the microcontroller, respectively, into differential CANH and CANL signals. The transceiver selected is the MCP2551 from Microchip. It has a multitude of features such as 250V transient capabilities on CANH and CANL, externally controllable slew rate, ground fault detection, and full 1Mb/s capability [9]. Since there are no specific requirements as to the features of this device, the Microchip part was selected to essentially guarantee proper operation with the PIC microcontroller when the device is used as directed. This device comes in a SOIC surface mount package.

4. Summary

The interfacing, peripheral and I/O requirements of this design are very well defined because of the necessity to replace an existing design without modification to the power board. The constraints of this design, size, accuracy, and cost are well defined and as such, component selection was quickly narrowed-down to very few possible candidates. Then, efficiency, cost, and other requirements of a more nebulous definition were used to further narrow down the selection list. Of these candidates, the components that fit the requirements and constraints the best were chosen, with major component selection being discussed in detail.

List of References

1] Biel School of Engineering and Architecture, “MPPT New Generation Order Form,” 2001, .

2] “PIC18FXX8 Datasheet,” Microchip Technology Inc., 2004, .

3] “AT90CAN32/64 Summary,” Atmel Corporation, 2005, .

4] “MCU Selector Guide,” Freescale Semiconductor, Inc., 2006,

5] “Kemet Surface Mount Capacitors,” Kemet Electronics Corporation, 2005, $file/F3102.pdf

6] “MAXIM Low-Noise, 14V Input, 1A, PWM Step-Down Converters,” Maxim Integrated Products, 2001,

7] “SMT Power Inductors,” Coilcraft, 2005,

8] “MCP6001/2/4 Datasheet,” Microchip Technology Inc., 2005,

9] “MCP2551 Datasheet,” Microchip Technology Inc., 2003,

10] “MAX6225-MAX6250 Datasheet,” Maxim Integrated Products, 2001,

Appendix A: Parts List Spreadsheet

|Vendor |Manufacturer |

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download