Team 2 Final Report



Solar Powered, Multi-seated, Internetted Computer System

[pic]

Final Report

December 3rd, 2008

Sponsored By:

[pic]

In Cooperation With:

[pic] [pic]

Michigan State University University of Dar es Salaam

ECE 480 - Design Team #2 - Fall 2008

Management Jakub Mazur

Web Josh Wong

Document Ben Kershner

Presentation/Lab Eric Tarkleson

Executive Summary

With the increasing proliferation of affordable, reliable personal computers into the marketplace, there is a great demand to develop affordable personal computers for remote and underdeveloped areas. Before deploying a computer system into such harsh conditions several obstacles must be overcome, including source of electricity, telecommunications, and the savannah climate. The Lenovo Corporation has tasked this team to develop a solar-powered computer workstation that can accommodate up to eight users. The solution must be robust enough to withstand the harsh environment with as little technical maintenance as possible, yet still be affordable for rural schools.

A bank of solar panels charges a small battery bank using a commercially available charge controller. A custom engineered battery management system monitors voltages and currents from the solar panels and the battery to calculate system data. The system also monitors case temperatures to insure everything is operating within a safe range. Should the battery become discharged or temperatures rise above the safe limits the system will automatically shut itself down. All power data is logged to the computer and stored for system optimization. A small LCD panel displays pertinent system information

A single computer can connect up to 8 monitors, keyboards and mice. Each terminal allows independent simultaneous logons. The operating system is Linux based and uses only free and open source software (hereinafter referred to as FOSS).

Since this project has three vastly different, but equally important main tasks that needed to be researched, tested, and built this report is split into three main sections. First is solar power system design, second is power management, and third is Multi seat hardware and software. Each section will address the research, design issues, testing, and conclusions that were found. The last section will deal with system integration and project findings and conclusions. Since we did not deal with the telecommunications issues these will not be addressed in this report.

Acknowledgement

[pic]

We would like to thank our sponsor, Lenovo, for their help in making this project a success. We also would like to thank our team members in the Telecommunications department as well as the faculty and students working with us at the University of Dar Es Salaam, Tanzania.

Also we would like to thank LEM and Man vs. Machine for their donations and samples. Their engineers and salespeople were very helpful in guiding us in the right direction.

This project would not have been possible without the help of our facilitator Dr. Jian Ren. Also we would like to thank Dr. Goodman for helping keep us on track and helping us plan our deployment of this project in Tanzania.

[pic]

Figure 1 - Power system block diagram.

Table of Contents

Solar Powered, Multi-seated, Internetted Computer System 1

Executive Summary 2

Acknowledgement 3

Section 1 - Photovoltaic System Design 7

Determining Size of Photovoltaic Panel Array: 7

Determining Load Power Consumption: 7

Determining Solar Insolation Levels: 8

Sizing Battery Array 11

Wire Sizing and Connections: 12

Maximum Solar Power Output: 13

Inverter to Battery Wiring 14

DC-AC Inverter 15

Charge Controller 16

Conclusion 17

Section 2 - Power Management 18

Research 18

Voltage Measurements 19

Control Board 21

Power Management Board 21

Conclusion 22

Section 3 - Multi Seat System Hardware and Software 23

Thin Clients 27

Multiple-Motherboards 28

Multi-User 29

Testing/conclusion 30

Section 4 - System Integration and Conclusion 31

Integration 31

Safety 31

Future Improvements: 33

Intelligent Monitoring 34

Conclusion: 36

Appendix 1 – Technical Roles and Responsibilities 38

Eric Tarkleson 38

Joshua Wong 39

Jakub Mazur 40

Ben Kershner 41

Appendix II - Schematics 42

Appendix III - Nomenclature 44

Appendix IV - References and Recommended Reading 45

Section 1 - Photovoltaic System Design

[pic]

There are many types of solar systems but most can be categorized into a variation of the following: A “grid-tie” system where there are no batteries and the power grid provides back-up power. A hybrid “grid-tie” system where the power grid provides back-up for the solar panels and batteries act as a backup for the grid. In cases where there is no access to grid power an “off-grid” system is used, in which the battery bank stores and provides all the energy for the system without a backup. Since this is generally the case in under-developed areas this will be the system discussed here. There are also systems with generators as backups, they are comparable to “grid-tie” systems and will also be omitted from discussion here.

Determining Size of Photovoltaic Panel Array:

There are several steps involved in sizing the PV array, determining load power consumption, accounting for losses and dividing by solar insolation levels for deployment region.

Determining Load Power Consumption:

The first step is to determine how much power the total system load will draw. Power is measured in Watts:

P = V ∗ I (Joule’s Law)

However, the power rating is more useful when looked at in terms of time, this is indeed how electric companies charge consumers. For example a 200Watt light bulb running for 24 hours uses 4.8 KWh.

200Watts ∗ 24hrs = 4800 Watt-Hours or 4.8 KWh

A list of all devices connected to the system should be made with their appropriate power draw available from specifications sheets or better yet, actual measurements.

|Component |Power (Watts) |

|Lenovo S10 (Idle) |91 |

|Lenovo S10 (Full Processor + Hard Drive) |116.76 |

|Lenovo S10 (30% Duty Cycle) |98.73 |

|Satellite Router (Idle) |53.8 |

|Satellite Router (Busy) |72.5 |

|Est. Typ. Satellite (30% Duty Cycle) |59.41 |

|17” LCD Screen x 4 |(20*4) = 80 |

|Total |238.14 Watts |

Figure 2 - Power measurements on 11/11/2008. Note: Headsets, Keyboards & Mice are currently not included in calculations because the team is not in possession of them and their power consumption should be minimal.

Since these devices are designed to plug into AC power, a DC-AC power inverter is needed. The power inverter ideally operates at 90% efficiency. Therefore the maximum inverter draw from batteries is:

238 Watts / 0.90 = 264.60 Watts

This system power draw is then multiplied by the amount of hours per day that it will operate.

264.60 Watts ∗ 8hrs/day = 2116.80 Watt Hrs/day

To compensate for system losses during battery charge/discharge cycles the total system power consumption is multiplied by a 20% compensation factor (Sunwize).

2115.52 Watt Hrs/day ∗ 1.2 = 2540.16 Watt Hrs/day

Determining Solar Insolation Levels:

In order to determine a good approximation of how much power the PV panels will output, solar insolation levels need to be considered. Solar insolation is the amount of incoming solar radiation incident on a surface, for PV applications the surface of interest is the earth’s surface. The values of solar insolation are commonly expressed in kWh/m2/day, which is the amount of solar energy that strikes a square meter of the earth's surface in a single day. This is commonly referred to as a “Sun-Hour-Day”. The amount of insolation received at the surface of the Earth is controlled by the angle of the sun, the state of the atmosphere, altitude, and geographic location.

[pic]

Figure 3 - World insulation levels.

This map divides the world into six solar performance regions based on winter peak sun hours.

It is important to keep this in mind when designing the system because as seen below in Figure 4, during the winter you have a much smaller ‘Solar Window’. Worst case scenarios should be calculated as it is better to have extra energy in the summer than not enough in the winter. Therefore the “Sun-Hour-Day” values for December (since December days are shortest) are generally used.

[pic]

Figure 4 - Sun path chart.

Solar Insolation Levels for Arusha, the prototype deployment area, are seen below in Figure 5.

The compensated total power consumption per day value calculated above is then divided by the solar insolation values for given deployment region to determine minimum PV panel array power output requirement:

2540.16 Watt Hrs/day / 5.5 = 461.84 Watts

Sizing Battery Array

Nearly all large rechargeable batteries in common use are Lead-Acid type, although there are three variations, flooded, gelled electrolyte (“Gell Cells”) and absorbed glass matt (“AGM”). Flooded is the oldest and cheapest technology used but can be dangerous, in case of a malfunction acid can spill. Gell Cells contain acid that has been "gelled" by the addition of Silica Gel, turning the acid into a solid mass, therefore even if the battery where cracked open, no acid would spill. Gell batteries need to be charged at a slower rate (capacity / 20) but this is not a concern in the PV setup as the panels will not be outputting nearly this much current. AGM batteries are the newest technology and have all the advantages of Gell Cells without the charging limitations.

All deep cycle batteries are rated in amp-hours. An amp-hour (Amps x Hours) is one amp for one hour, or 10 amps for 1/10 of an hour and so forth. The accepted AH rating time period for batteries used in solar electric and backup power systems is the "20 hour rate". This means that it is discharged down to 10.5 volts over a 20 hour period while the total amp-hours it supplies is measured (Windsun).

The compensated total power consumption per day value is used again to calculate minimum battery array size.

2540.16 Watt Hrs/12 Volts = 211.68 AmpHrs/day

Number of days of autonomy to support: 1 (8hrs)

211.68 ∗ 1 = 211.68 AmpHrs

“Battery life [deep cycle] is directly related to how deep the battery is cycled each time. If a battery is discharged to 50% every day, it will last about twice as long as if it is cycled to 80% DOD [depth of discharge]. If cycled only 10% DOD, it will last about 5 times as long as one cycled to 50%. Obviously, there are some practical limitations on this - you don't usually want to have a 5 ton pile of batteries sitting there just to reduce the DOD. The most practical number to use is 50% DOD on a regular basis (Windsun).”

Depth of discharge for battery: 0.5

211.68 / 0.5 = 423.6 AmpHrs

This means that after 8 hrs of use without sun the battery will be discharged to 50%

8 Hrs of autonomy and battery depth of discharge at 0.80 (Half the life-span of 0.50):

264.60 Amp Hrs

6 Hrs of autonomy and battery depth of discharge at 0.50:

264.6 Watts ∗ 6 Hrs = 1587.6 Watt Hrs / day ∗ 1.2 = 1905.12 Watt Hrs / day

1905.12 Watt Hrs/12 Volts = 158.76 AmpHrs

142.8 AmpHrs / 0.5 = 317.52 Amp Hrs

4 Hrs of autonomy and depth of discharge at 0.50:

238 Watts ∗ 4 Hrs = 1058.4 Watt Hrs / day ∗ 1.2 = 1270.08 Watt Hrs / day

1270.08 Watt Hrs / 12 Volts = 105.84 Amp Hrs

105.84 Amp Hrs / 0.5 = 211.68 Amp Hrs

Wire Sizing and Connections:

Another important consideration for the system is the electrical wiring. All wiring needs to safely accommodate the amount of current draw of the system with an acceptable amount of losses. In a DC system losses quickly become an issue. This is especially a concern PV systems as they can only handle a small voltage drop as there must be enough potential to charge the battery array, and of course it is good practice to keep energy loss sourced from the sun to a minimum. Generally a 3% drop between PV array and batteries is acceptable. Also, “any type of connection bigger than AWG 10 should have a proper compression connector, with appropriate joint compound and preparation. This does require special tools and dies. Otherwise you are running the risk of burning up your connections if you get any kind of heavy current flowing. (SolarForum)”

Losses associated with transmission of DC power:

CM = (22.2 ∗ A ∗ D)/VD

CM = Circular Mills In Copper

A = current in amps

D = one-way cable distance in feet

VD = Voltage Drop

22.2 = Constant for Copper

For wiring from the PV panels to charge controller the maximum PV short circuit current specification (from PV data sheet) is used.

Maximum Solar Power Output:

24 Volt Systems:

|Configuration |Max Current Out (Amps) |

|6 x PW080 |3 x (5.14A-ISC) = 15.42 |

|3 x ST-165 |20.63 |

|4 x KY125 |20.83 |

Figure 6

12 Volt Systems:

|Configuration |Max Current Out (Amps) |

|6 x 80 Watt |6*(5.14A-ISC) = 30.84 |

|3 x PW165 |41.25 |

Figure 7

Using the loss equation above the following result was obtained for the selected system:

Distance: 50ft

Voltage Drop: 0.72

Current: 15.42 Amps

Circular Mills: 23772.5

AWG: 6

Inverter to Battery Wiring

For current level estimates from the battery to inverter maximum power draw levels are used although this distance is generally short and maximum available wire gauge is recommended. This is also due to the fact that the system will encounter surge currents as various components are ‘turned on’. Since the system used as an example here is not continuously running and is to be turned off every night and back on in the morning this was a serious issue that needed to be tested. (Refer to Figure 9).

Maximum Power Draw:

|Component |Power (Watts) |

|Lenovo S10 (Full Processor + Hard Drive) |116.76 |

|Satellite Router (Busty) |72.5 |

|17” LCD Screen x 4 |(20 ∗ 4) = 80 |

|Total |270 Watts |

Figure 8

Assuming the inverter that will be sourced in deployment area is operating at 90% efficiency:

270 Watts = 300 Watts x 90%

Maximum current draw in 12 Volt system = 300 Watts / 12 Volts = 25 Amps

Maximum current draw in 24 Volt system = 300 Watts / 24 Volts = 12.5 Amps

[pic]

Figure 9

Figure 9 shows DC current draw as measured during power-up of Lenovo S10 Workstation (custom configuration) and L193p Monitor. Although the system is only drawing 5 amps while running the surge current spikes are clearly visible. This is indeed one of the reasons why proper electrical connections are crucial.

DC-AC Inverter

Since the computer and monitors are designed to plug into AC power and accessory plugs for phone charging are a project specification an inverter is necessary. There are two types of inverters, pure sine wave and modified sine. Most devices will work from modified sine, this is what common uninterrupted power supplies provide and what was selected for this system. It is important to make sure that the inverter is rated to provide enough power for everything running off of it.

Charge Controller

The charge controller chosen for this system is the Outback Power FlexMax 60. This decision was based on versatility, efficiency, robustness, and availability in deployment area. The Outback can accept a wide range of voltage inputs as well as various battery arrays, this was important for this specific system as ultimately whatever solar panels are in stock at the time of deployment in the region will be used.

Note, the efficiency curves (Figure 10 and Figure 11) are for the Flexmax80, they are identical to the Flexmax60 other than the fact that the FX60 does not accept 85 and 100V.

The highlighted area on the graph represents the highest efficiency while charging a 12V battery array. The charge controller is operating at about 95.5% efficiency with an input Voltage between 17-34V. Typically a 12V PV panel's Voltage at Peak Power is around 17 Volts.

The highlighted area in this graph represents the optimum efficiency if the system where charging a 24V battery array. The charge controller is operating at about 98% efficiency with an input Voltage around 34V. Two 12V panels in series will typically have 34 Volt equivalent Voltage at peak power.

In an ideal setup the FlexMax 60 would operate at 98.1% efficiency with an input of 68V while charging a battery array at 48V. This would be the case with the optimum PV panel chosen in section 1, the Kaneka G-EA060 as the VPM is 67Volts.

Conclusion

Designing an off grid photovoltaic system involves many steps and although the math is simple all calculations should be double checked. If the calculations for one component are off chances are the whole system will not work, every stage relies on the previous one. Designing the system for worst case scenarios is good practice, it is better to have extra energy than not enough. All safety precautions should be followed especially on electrical connections that have a possibility of carrying a lot of current. Breaker boxes before and after battery connections for easy power disconnect should be implemented. These breakers should be rated for DC voltages.

Section 2 - Power Management

[pic]

The core of the power management system is the data-gathering module. The module uses a PIC microcontroller to measure voltages, currents and temperatures and send the results to the computer via USB.

The end result of this system is similar to the power management utility in your laptop. Just as with a laptop this is a self contained system that will not have access to external power some of the time (when there is no sunlight). We feel that because of the complexity of the multi user computer as well as the photovoltaic system we needed to design a system that not only watches battery voltage but also stores and interprets data for intelligent power decisions.

Research

The current sensors used are LEM FHS series sensors. They are Hall effect sensors which operate by measuring the electric field produced by the current carrying wire. The field is proportional to the current. These current sensors are isolated from the actual power path and can be turned off to conserve power. Different models of the sensors are capable of measuring anything from milliamps to over 50 amps. The biggest advantage of these sensors is scalability. A single chip can handle any range of current by clever PCB board design. We acquired several sample units from LEM and put them into small project boxes that can be used anywhere. They attach to the power management system via d-sub cables.

Temperature measurements use a small probe, which outputs a voltage proportional to the ambient temperature. We chose the national instruments LM335z. These are Zener diodes with a linear internal voltage drop proportional to the temperature of the device. Our design allows for up to 4 sensors.

Because the PIC uses a very small amount of power, it is continuously operating to ensure system safety. The PIC is connected to a small LCD screen which displays pertinent information about the condition of the battery and solar panels.

Two buttons mounted beside the screen allow a user to scroll through the information showing them battery voltage, current, and amp hours remaining. Also there will be displays for voltage and current from the solar panel. This information will be used to compute time and percent remaining before system power down. There are also three status LED's that inform the user of system power, battery good, or battery low.

The PIC communicates to the computer via USB. This is done using the FTDI UM232R. This device automatically converts the USB signal to a standard serial signal. It also registers with the computer as being a serial port. With the connection, the computer can query the PIC about the status of the battery and can shut the system down safely when the battery is low. The computer can also query the PIC to get real time data on the voltages, currents and temperatures. The information is stored in an XML file and can be easily reprocessed into any file format for analysis of the data.

Voltage Measurements

Voltage measurements require the addition of a voltage divider into the circuit because the PIC has a maximum dynamic range of 5 volts. An advantage of using the PIC to measure voltage is that the value of the resistors in the voltage divider can be accurately measured and the program can use that information to improve the accuracy of the voltage measurement.

The built-in 10-bit ADC of the PIC 18F4520 microcontroller is a cheap way to measure analog voltages. A drawback is that it is limited to measuring only 0-5VDC. By adding a voltage divider, customizing the software slightly, and using a digital IO pin as a current sink, we can efficiently and accurately measure much larger voltages.

[pic]

Figure 12 - Voltage divider with selectable current sink.

The 10-bit ADC on the PIC18F4520 is implemented as an RC circuit inside of the PIC. Because of this, there are certain parameters to be aware of, all of which have been taken from the PIC18F4520 Datasheet.

By adding a voltage divider to the input pin of the PIC, we are now able to scale the input voltage. All that is left is the selection of resistor values. Starting with known values, we can select vout to be 5V, as that is vref+ on the PIC. Also, recalling from the first section, the nominal input impedance to the PIC is 2.5kΩ, and as such will be selected for R2. The node labeled vin is the high voltage input to the voltage divider, and should be selected as the highest voltage desired to be measured. For this example we will say 50V.

As can easily found by inspection, the formula for a voltage divider is:

vout = vin ∗ (R2 / (R1 + R2))

This gives us a formula, three knowns, and one unknown. Solving for R1:

R1 + R2 = vin ∗ R2 / vout

R1 = R2 ∗ ((vin / vout) – 1)

R1 = 2.5kΩ ∗ ((50V / 5V) – 1)

∴ R1 = 22.5kΩ

We can now build our new voltage divider and accurately measure voltages up to 50V! To convert the ADC result to actual voltage, the following equation is now used:

van0 = ((VDD ∗ ADC_output) / 1023) ∗ (R2 / (R1 + R2))

Control Board

This is the schematic and for the control board. It is mounted behind the control panel and connects directly to the LCD header. There are headers for connection to the status LEDs as well as the control buttons. A ribbon cable connects this board to the rest of the circuit

[pic]

Figure 13 – LCD control board schematic.

Power Management Board

This circuit allows connection of up to 4 current sensors, 4 temperature probes, and 4 independent voltages. The voltage inputs have a voltage divider circuit that can be configured to accept voltages for systems running at 12, 24, or 48 volts. A USB connection allows data sharing with the computer.

[pic]

Figure 14 - Power system monitor schematic.

Conclusion

Our power management system performs as it was designed to. Each channel accurately reads a voltage from the sensor and converts the voltage to a value. The PIC then uses this information and displays this information clearly on the LCD panel. Both boards are mounted in an enclosure with connectors for each sensor. The board will be mounted in the main enclosure and will allow a user to turn of the entire system with the touch of a button. The sturdy PCB design and construction will stand the test of time. The PCB board also allows for on site reprogramming if it is deemed necessary.

[pic]

Figure 15 - Power system monitor and faceplate.

Section 3 - Multi Seat System Hardware and Software

[pic]

The primary goal of this project is to help promote education in developing countries by providing grade schools with electronic resources. There are a variety of other groups that have already initiated solutions to this problem. The most prominent group is the One Laptop Per Child Association (hereinafter referred to as OLPC), which has created a cheap, durable laptop known as the XO-1. Other groups such as the Center for Scientific Computing and Free Software (hereinafter referred to as C3SL) have made significant strides in reusing older computers for schools; however, both of those programs have some significant drawbacks.

The primary competitor identified is the OLPC. The OLPC Association is dedicated to producing low cost laptops and distributing them to low-income areas. There exist several problems with the program, including the per-deployment cost and deployment into rural areas. The original intent was to deliver a laptop to every child for a cost of $100 per device. The program, however, is unable to deliver the laptop at the $100 target; in fact, the cost to donate a system is almost $200. Deployments also require a minimum commitment of 100 laptops. This represents a very significant financial burden, though once deployed, it is difficult to integrate multiple PCs into a cohesive learning environment, and this takes away from educating the students. However, the XO-1’s are extremely rugged PCs and do not depend on any external power sources except to charge the batteries.

C3SL’s solution integrates into school systems better, and was widely deployed in the Paraná Digital project. This project involved having multiple terminals running off of a single computer in multiple schools. This program has been very successful and shows great promise, but there is a critical flaw. The program is entirely software, and this software was intended to run in a classroom equipped with basic utilities, such as power and internet connectivity.

Our solution was to integrate the OLPC's ruggedness and the C3SL's novel software solution into one robust package. The design team preceding ours built a solar powered computer system that can be deployed in a relatively durable building. They assembled a solar panel, battery, and a charge controller into a self contained solution, such that deployment in a wide variety of climates and locales is possible, but they were unable to decide on the computer system. Our primary goal for this project is integrating the solar charging and battery system with a computer system that is suitable for educating youth, regardless of regional or socio-economic boundaries.

The requirements for the computer system were fairly simple; four seats, low power. This could be established very easily with four laptop computers, but there were design issues that had to be considered. First, the system will be deployed in a relatively extreme environment. It also will not be running off of mains power, and will therefore have to be low power. It should also be as cost efficient as possible, i.e. lowest cost per seat. Four architectures were discussed for the computer system: laptops, thin clients, multiple-motherboard, and multi-seat.

| |Laptops |Thin Client |Multi-User |Blade Client |

|Cost Outline: |Baseline: |Baseline: |Baseline: |Baseline: |

| |- Router with advanced |- Server ($500) |- Powerful Server |- Server ($500) |

| |features ($200) | |($800) | |

| |Per Seat: |Per Seat: |Per Seat: |Per Seat: |

| |- Lenovo IdeaPad S10 Latop |- Diskless Workstation |- Lenovo L197 Monitor |- Small Motherboard |

| |($439) |LTSP 1220PXE Thin Client |($239) |with RAM & CPU ($100) |

| |- Mount ($50) |($285) |- Keyboard/ Mouse ($30)|- DC-DC Power Supply |

| | |- Lenovo L197 Monitor |- Video card ($30) |($50) |

| | |($239) |- Optional Software |- Keyboard/ Mouse ($30)|

| | |- Keyboard/ Mouse ($30) |($100) |- Lenovo L197 Monitor |

| | | | |($239) |

|Total Cost: |Base: $200 |Base: $500 |Base: $800 |Base: $500 |

| |Per Seat: $489 |Per Seat: $554 |Per Seat: $399 |Per Seat: $419 |

|Pros: |Easy, Reliable, Server- |Easy, Reliable, Stable, |Cheap, Lowest Power |Possibly Cheaper than |

| |less, Redundant, Low |Low |Consumption, Single |Thin |

| |Power Consumption |Power, COTS |Point of |Client, 100% Lenovo |

| | | |Maintenance, 100% |Hardware |

| | | |Lenovo | |

| | | |Hardware | |

|Cons: |Small Screens, Defeats |Relatively Expensive, |COTS Software is |Lots of Enclosure Work,|

| |Purpose of Designing a |Lenovo |Expensive and | |

| |New system as |Does not Make a Thin |Open-Source is |Reliability |

| |Opposed to Donating |Client |Immature, | |

| |Laptops, Security | |Reliability is Main | |

| |Concerns | |Concern | |

Figure 16 - System architecture prototypes.

Laptops[pic]

Figure 17 - Laptop-type system architecture mockup.

The first prototype was a simple laptop-server setup. Each workstation would consist of small laptop (a 10” form factor, such as the Lenovo S10). The laptops would be connected to the Internet either by an Ethernet cable, or even Wi-Fi. Laptops would be run without being directly connected to AC power; a charging station would be setup next to the server.

This style of architecture would be very simple to configure. The server and the laptops would all be off-the-shelf Lenovo products. The workstations would have low power consumption, given the fact that the monitor, computer, keyboard, and mouse are all combined into one device. Should a laptop be damaged, it would also be very easy to replace, requiring little re-configuration, and no custom engineering. The laptops, however, would be most susceptible to accidental damage given that they would be out in the environment, rather than safe inside the case.

Thin Clients

[pic]

Figure 18 - Thin client-type system architecture mockup.

The thin client architecture is much closer to a traditional desktop user experience. A thin client is little more than a small form factor motherboard, processor, and RAM (no HDD necessary) loaded with a special BIOS. The server does most of the processing and the think client simply helps to shuttle data back and forth. With this setup, four thin clients are each connected to their own monitor, keyboard, and mouse. All of the applications and processing tasks are handled on a central server.

This is another easy solution that uses all commercial off-the-shelf (COTS) hardware and software. It also has much lower power consumption when compared to installing four individual desktops, but the power consumption is not as low as other methods. The key drawback is price. Since Lenovo does not produce a thin client, they would have to be purchased at full market price. This brings the total per seat cost up to around $700.

Multiple-Motherboards

[pic]

Figure 19 - Blade client-type system architecture mockup.

One method of avoiding this high per seat cost would be to build custom thin clients, very similar to the blades of a blade server. The team estimates that even paying market prices, a hand-built thin client could be built for around $100, compared to the nearly $300 for third-party thin client.

The pros are the same as that of the thin client, with the added benefit of being cheaper and being made totally of Lenovo hardware. The key drawback though is the packaging. Since these “thin clients” are just a motherboard, they would need to be put into a proper case, first. This would add to the cost and could cause lifecycle issues (a poor case would lead to high failure rates of the motherboard).

Multi-User

[pic]

Figure 20 - Multi-user desktop-type system architecture mockup.

The most attractive architecture is the Linux based multi-seat system. It is implemented by building a central PC with multiple video cards (2-4), multiple keyboards, and multiple mice. Each workstation consists of a monitor, keyboard, and mouse plugged directly into the PC (acting as an X window server).

The cost for such a setup is low. Since no thin clients are required, the per-seat cost consists of only a monitor, keyboard, and mouse. The power requirements are also lower, given that the CPU and all of its resources are be shared by all of the users. The system would also respond much quicker than a thin client, without the LAN bandwidth and latency issues. With significant customization of the operating system (UNIX based), this architecture was completed using only free, open-source software.

This is the architecture that has been chosen by the design team. The low cost and low power requirements make it an ideal solution. Also the server can be safely mounted inside a case and only the monitors, keyboards, and mice need to be outside of the case.

Testing/Conclusion

To test the computer terminals, four accounts were created and each account attempted to do the following tasks:

• Write a document using a word processor

• Watch an offline video

• Play an educational game

• Browse a simple webpage

• Browse a rich webpage

• Watch an online video.

These tasks represent what a typical student might be doing on a school computer. Support for advanced programs, such as MatLab, were not tested. Each station was able to perform the tasks listed, and therefore the computer terminals are ready to be deployed.

| |CPU usage |Video |Audio |Comment |

|Open Office |low |good |n/a |Basic document preperation. Simulates multiple students writing a paper |

|Firefox basic |low |good |n/a |Basic web surfing. Simulates multiple students reading static webpages, ie |

| | | | |wikipedia |

|Firefox rich |high |good |good |Advance web surfing. Simulates multiple students browsing flash enabled |

| | | | |websites, ie Youtube |

|Totem |mid |good |good |Video/audio player. Simulates multiple students watching an offline video |

|Gcompris |low |good |good |Education games. Simulates multiple students playing non accelerated games |

Figure 21 - Programs used to test multi-seat system.

Section 4 - System Integration and Conclusion

[pic]

Integration

Our final project consists of a new more robust charge controller and a custom engineered battery management system. New solar panels have been chosen for our deployment area but due to their similarity to our current components they were not ordered here. We decided to change our solar panel array and battery bank from a 12V to a 24V system. This lowers our current throughout the system which leads to lower losses and smaller wire gauges. We purchased a 24V inverter to replace our 12V inverter that was left to us from the previous team.

The final architecture that we chose was the single workstation multi-seat system. This allows for relatively low power consumption and the lowest cost per seat of the other architectures we researched. We were successful in having up to 4 independent simultaneous logons from a single desktop computer with multiple graphics cards. With our current hardware we can easily support up to 8 seats with only the addition of more graphics cards and accessories.

We were successful in combining our objectives into a complete and functioning prototype that can and will be deployed in a rural village in Tanzania for testing. Our finished prototype has 4 user terminals connected to a large case that will hold the communications equipment, workstation, batteries, charge controller, and battery management system. This system will be hooked up to a solar panel array that should provide enough power for the system to function throughout a normal school day.

Safety

Since we are designing a system that will be deployed in a school with no preexisting power, and around people without much experience with electrical equipment we need to take some extra precautions to make sure that the entire system is as safe as possible. There will be solar panels that will either be mounted on poles outside of the building or on the roof so we will have to make sure to protect these cables because they will carry high currents. The battery bank also needs extra safety measures because it will have the potential to unleash huge amounts of current should a short, or other low resistance connection (ie. body) cross the terminals.

The solar panels will be mounted outside. If we mount them on the roof then we need to make sure that the cable runs go into the structure through a watertight seal. Also there will need to be an earth ground installed in case lightning hits the building. The building itself must be structurally sound enough not only for the solar panels but also for several people so that installation and maintenance is safe.

If the building is orientated in such a way that the panels can not face the sun while they are on the roof then they will have to be mounted on a pole outside of the school building. In this case extra precaution must be used to insure that a person is unable to access any wire leads because the current can easily kill a person. Also the cabling will have to be run underground so the wires will have to be enclosed in the proper type of duct to that they cannot be cut by someone digging in the area. They also must be completely weatherproof to avoid the wire insulation breaking down from the elements. One advantage to this setup is that the structure that the panels hang on can be used as an earth ground for the entire system.

Once inside the building all wiring outside of the case should be in ductwork. This prevents a person from easily cutting the wires and also prevents animals from chewing through the insulation. The system case must have a hole that is not sharp so that the wires cannot fray over time. Also cabling should be secured to the case so that it cannot easily be pulled out which could cause short circuits or broken hardware. Any short unintentional short circuits have a large probability of destroying the expensive equipment nearby or causing a fire.

The battery bank must be carefully installed so as to minimize any possibility of short circuits. Each battery terminal should have a cover installed so that it is difficult to touch the electrodes accidently. All batteries should also be secured to the casing so that they cannot move if the whole case is tipped or moved.

There must be circuit breakers at several key points in the system to insure that if there is a problem the system will cut off power to itself and hopefully save the equipment from being ruined. There will be a circuit breaker between the solar panel array and the Charge Controller and also between the Battery Bank and the Inverter. These will trip if current exceeds expected ratings and will also allow the system to be manually shut down if the need arises.

Following these simple precautions should allow the system to be safe from prying fingers, weather, animals and equipment failure. Also in the case of a failure troubleshooting should be easy because if our battery management system is running it will provide feedback about which components are not working. If a breaker is reset then it will also provide clues as to which part of the system failed.

Future Improvements:

A major improvement to the efficiency of the system could be accomplished by using a computer system that is optimized for minimal power consumption yet within requirements for the ‘Multi-Seat’ setup. The computer used in the prototype is a Lenovo S10 workstation. This computer is designed to be a state-of-the-art processing power house, efficiency was near the bottom of requirements during its design. However, it is the only hardware that Lenovo currently offers that would support the ‘Multi-Seat’ architecture; therefore it was used in this proof of concept prototype. After returning from Tanzania the team is traveling to North Carolina to give a presentation to Lenovo Corporation about requirements and suggestions for developing a more efficient computer system for use in PV systems. Amongst specific suggestions of low-power components include using the Intel Atom processor and lower power hard drives, this will also include a design modification to their power supply. The power supply is designed to transform and rectify AC power to DC which is used by the computer. Since power sourced from PV systems is already DC unnecessary losses can be avoided by using a DC to DC power supply. Most modern power inverters optimally operate at 90% efficiency; hence eliminating these conversions from the system cuts losses significantly.

Although the team used the most efficient PV panels on the market at the time of deploying the prototype, a close eye should be kept on advancements in PV technology. Current panels are still very inefficient and expensive. This is an area that is heavily researched and as advancements develop in this field they will greatly benefit the system.

Intelligent Monitoring

One of the goals of developing the power monitoring unit is to collect usage data. This includes computer and telecommunication equipment duty cycles as well as specific component power consumptions and PV array and battery efficiencies at measured operating temperatures. Before deployment many factors are estimated, once experimental data is available the system can be designed within tighter tolerances. As part of the product life cycle management minor changes can be made at identified weak links in order to maximize system efficiency.

[pic]

|Item |Cost |Quantity |Total Cost |

|Miniature Pushbutton |3.40 |2 |6.80 |

|Metal Binding Posts |5.47 |2 |10.94 |

|Liquid Tin |31.94 |1 |31.94 |

|Project Box |6.99 |1 |6.99 |

|Outback Flexmax60 |545.22 |1 |545.22 |

|PSE-24125A Inverter |350.00 |1 |350.00 |

|Winford BCF9 Adapter |6.50 |2 |13.00 |

|LEM |37.20 |1 |37.20 |

|Calling Card |10.00 |1 |10.00 |

|Mobile Computer Cabinet |1871.00 |1 |1871.00 |

|VGA & USB extensions |54.35 |1 |54.35 |

|i-rocks USB Hub |11.99 |4 |47.96 |

|SYBA USB Audio Adapter |7.99 |4 |31.96 |

|Lenovo S10 Workstation |1190.60 |1 |1190.60 |

| | |Total: |4207.96 |

Figure 22 - Team budget.

Conclusion:

This project is special because the driving force behind it is not artificial. The team members do not care about their grade. They do not care about the outcome of the design day competition. This team is driven by passion, and the sole thing this team cares about is the successful deployment of a system that will provide hundreds of kids living in a world of great hardship with a priceless tool, a tool that all of us grew up taking for granted. A tool that has the ability of transforming people, a tool that transforms kids into students, a tool that transforms thoughts into dreams… this tool is access to information.

We live in the information age. We live in a society where the term “Just Google it” is used on a daily basis. Where we are two clicks away from the answer to any question we have. So, what is it that separates us? What makes us different? How do we have access to this tool when there are people living in this world that are deprived of something so fundamental to us? We are no different. We are just lucky to be born into this society.

This team is fortunate enough to have the ability to give our most prized tool to hundreds of under-privileged children. Children that go to bed at night and dream, and we want to let them dream big and turn those dreams into realities.

The team members participated in every aspect of engineering a product and service from start to finish. The team not only accomplished everything outlined in their initial project description but everything that was thrown at them along the way, and what is yet to come.

This December, Michigan State University’s Design Team 2 will deliver the most priceless present anyone can ever give, this is all that we care about and there is nothing that can stand in our way.

Appendix 1 – Technical Roles and Responsibilities

[pic]

Eric Tarkleson

Eric Tarkleson was this semester’s lab facilitator and document preparer. He worked with Ben on the current and voltage monitoring system. Specifically he researched and acquired and built the current sensing equipment used in the final prototype.

Eric also worked with the team to decide on casing options. He chose the final case design that will be used on design day. Eric worked with Jakub on designing and manufacturing the faceplate and box that will house the battery management system. The end result is a sleek easy to use interface to the system that was custom designed for monitoring a solar powered computer system.

Almost all of the PCB design was completed by Eric. The final project includes two small PCB boards. The main board includes a microprocessor controlled data acquisition system that monitors the various components power usage. It then sends this data to the computer via USB for datalogging. The second PCB mounts to the LCD and routes the appropriate signals to the status LEDs as well as the power, reset and menu buttons.

Testing was performed by the whole team and Eric was no exception. Much testing needed to be performed to insure that the correct solar panels, battery banks, and charge controllers were purchased and ready in Africa. Much power testing was done on both the Lenovo S10 PC and the monitors. This data will be useful in determining how to streamline the system in the future.

Joshua Wong

Josh’s primary duties for the semester were to find and implement a multi-terminal solution. The first attempt to implement a multi-terminal solution was installing and testing an LTSP server. LTSP allows Linux based system to run thin clients which support PXE booting and uses the x86 instruction set. This was rejected due to the use of non-lenovo hardware. The second attempt was to implement a multiseat system by modifying the login manager and the X windowing system. During the course of searching for up to date documentation on Xephyr and gdm a new program called MDM was discovered. MDM provided scripts to do most of the work referred to by the older documentation. However, the software was immature and several patches were needed to ensure it functions correctly. Specifically, there were several bugs which the developers never tested for, including a bug where the program terminates prematurely.

Josh was able to analyze the codes and fix most of the bugs. The software required major retooling to be compatible to with the Nvidia drivers. Fortunately one of the developers uses an Nvidia system and was able to provide valuable assistance in adapting the existing code to use the Nvidia drivers. Another major development was discovering how to enable audio devices for the system. Enabling audio devices system wide required solving a problem where certain programs would work properly while others failed. This was eventually traced to an issue where a file was not getting processed correctly. Other tasks included locating a USB serial interface and helping my teammates develop software for the USB interface. Overall, most of my tasks involved dealing with software, or interfacing hardware with software.

Jakub Mazur

Jakub Mazur was the Project Manager of the team. He was heavily involved in photovoltaic component selection including: Determining the size of the PV array, sizing the battery array, wiring gauge selection, and the DC-AC inverter selection.

Jakub acted as supply chain manager since there needed to be a serious amount of logistics to be taken care of to source parts in Africa and communicate and assign tasks to the Tanzanian students. This included plenty of research, keeping up with emails on a daily basis and making phone calls at 2am to motivate third parties to do their job. This task turned out to be extremely time consuming and quite frustrating. The time-difference also played a role in this as often only one communication per day is possible.

Jakub was also heavily involved in testing stages, he performed load current measurements using data acquisition hardware and programming in LabView. This provided critical power consumption data as well as capturing surge current data.

His technical roles also included helping all the other team members with any issues they where having. He worked with Ben and Eric on the microprocessor power monitoring circuitry. He also worked with Eric to design and physically build the faceplate of the power monitoring equipment.

Ben Kershner

Ben Kershner, the document prep, handled the collaboration, formatting, and finalization of the team documentation. He also spent several days researching the types and prices of solar panels and batteries, and building a spreadsheet that helped to organize and analyze this data, so that the team could choose the most cost efficient parts of the power system.

The majority of Ben’s time was spent designing and programming the circuitry for the power system monitor. He built the prototype on a breadboard and wrote a series of proof-of-concept coding examples to test the various digital and analog I/O features of the PIC, including a custom LCD driver library to interface the PIC to a Motorola 44780-type chipset.

Ben’s application note in measuring high DC voltages with the built-in ADC of the PIC in a power efficient manner was implemented in the final design of power system monitor.

Ben also wrote the server side code to accompany the power system monitor. Written in C++, and taking advantage of UNIX system calls, it enables the server to pull metric data off of the PIC. It then writes this to an XML file, which is uploaded to an offsite server on a daily basis.

Appendix II - Schematics

[pic]

Figure 23 - LCD control board schematic.

[pic]

Figure 24 - LCD control board layout.

[pic]

Figure 25 - Power system monitor schematic.

[pic]

Figure 26 - Power system monitor layout.

Appendix III – Gantt Chart

[pic]

Appendix IV - Nomenclature

[pic]

• C3SL – Center for Scientific Computing and Free Software.

• CPU – Central Processing Unit, refers to the main processor chip on a computer motherboard, not the computer as a whole.

• COTS – Commercial Off The Shelf, describes hardware or software that may be purchased rather than designed and built.

• FOSS - Free and Open Source Software.

• MPPT – Maximum Power Point Tracker, a style of solar charge controller.

• multi-seat – A type of system architecture in which many workstations are built onto a single machine.

• PIC – The company that produces the microcontroller used, may also refer to the microcontroller itself.

• PV – Photo-Voltaic, i.e. solar panel.

• PXE – Pre-boot eXecution Environment, a manner of booting computers over the network without a locally installed operating system.

• OLPC – One Laptop Per Child.

• OSS – Open Source Software.

• system architecture – The term used within this document to describe how the style in which the workstations are deployed.

• thin client – A client computer that relies on a central server for a majority of its processing tasks.

Appendix V - References and Recommended Reading

[pic]

• Hankins, Mark, and Francis Njeru. Solar Electric Systems for Africa : A Guide for Planning and Installing Solar Electric Systems in Rural Africa. Ed. Timothy Simalenga. Beverly: Commonwealth Secretariat, 1995. 7-8.

•

• World Insolation Map:

• Sun path chart:

• Outback Power Flexmax User Manual:

• Wiring Safety Concerns:

• Also,

• ** Source: Sunwize PV installation guide as well as other installer personal experiences

•

•

•

•

•

•

•

•

•

-----------------------

Figure 5 - Hankins.

Figure 10 - Outback charge controller.

Figure 11 - Outback charge controller.

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

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

Google Online Preview   Download