Figure 9 - Lux Perpetua on display at Epcot® following Sunrayce 99
In the past ten years, Team Lux has built four solar cars;[40] while they all follow the general overview described above, each has been shaped by different design choices into a unique car. The incomplete Handsome Dan and then Lux Aeterna provided the team with learning experiences where basic solar car techniques were developed and learned. During this time, much effort was spent creating the basic infrastructure to support a solar car team. From this starting point, and with numerous former members available to provide advice, the Lux Perpetua project was able to have more ambitious technical goals. This building upon prior cars would continue - sometimes literally, as with Lux Aeterna's steering wheel being used in both Lux Perpetua, and Lux Millennia - leading the team in new directions. The specifics of the design and construction process for each car warrant closer examination.
| Race | Car | Notes |
| Sunrayce 95 | Handsome Dan | Not completed |
| Sunrayce 97 | Lux Aeterna | Placed 9th out of 36 entries |
| Sunrayce 99 | Lux Perpetua | Placed 15th out of 29 entries |
| ASC 2001 | Lux Millennia | Not completed in time for ASC 2001 |
| ASC 2003 | The John Lee | Did not qualify for ASC 2003 |
| Phaethon 2004 | The John Lee 1.5 | Placed 7th out of 14 entries |
| NASC 2005 | The John Lee 2 | Did not qualify for NASC 2005 |
Before work started on Lux Aeterna, the Team Lux built a prototype car to develop its skills. This prototype was outfitted with a brushed DC motor that had been purchased for Handsome Dan. With the assistance of Professor Peter Kindlmann, two team members constructed a simple motor controller. This was useful for testing, but when it burnt out, the team was ready to purchase a new motor and controller to be used in the actual car. For Sunrayce 97, NGM introduced the NGM SC-M100 motor along with the matching EV-C150 controller. This motor was the logical choice for Lux Aeterna due to its high efficiency and the fact that NGM would have staff at the race to provide support. This support ended up being critical to the team's success. While on the qualifying track in Indianapolis immediately before the race, rainy conditions caused the motor controller to short, disabling the car. Water from the road's surface was being thrown into the car by the motor and landing on the motor controller. NGM's engineers assisted the team with repairs and Lux Aeterna was able to compete.
When evaluating motor choices for Lux Perpetua, NGM was still the clear leader in solar car motors, and although Team Lux considered some other options, they purchased NGM's second-generation motor/controller combination, the SC-M150 and the EV-C200. The new motor controller was longer and thinner than the previous model, and the increased length matched well with a cross-flow blower donated by supporter Jim Moore. This allowed the controller to be cooled more efficiently than with the pancake fans supplied by NGM. The new motor increased peak efficiency to 95% and added an external gear to adjust the rotor-stator air gap (NGM 2002). By changing the air gap, the motor's torque constant can be changed dynamically. This is similar to changing gears on an automobile transmission and allows shifting the motor's peak efficiency range from a high torque/low RPM mode, suitable for starting and hill climbing, to a low torque/high RPM mode for highway travel. Matching the air gap to driving conditions allows for considerable energy savings and Team Lux was very interested in using this new feature. Unfortunately, time constraints and higher priority projects interfered, so a system of allowing the driver to adjust the air gap was never implemented. Even without this addition, the new motor provided considerably improved performance over the prior one, which was brought to Sunrayce 99 as a backup.[41]
With the investment already made in NGM products, Team Lux decided to continue using their existing motors. NGM did not release a new motor for ASC 2001, although the 1999 products received incremental upgrades. A year before ASC 2001, NGM announced a program to service existing motors and controllers. The products would be inspected, any faults repaired, and would be brought up to the 2001 specifications, which included better waterproofing and increased power capability. Team Lux took advantage of this upgrade offer, however, it was several months before NGM returned the upgraded and repaired motor to the team.
Although Team Lux had used NGM motors for all of its previous cars, there was considerable interest in using a motor from CSIRO for the John Lee. CSIRO, the Australian Commonwealth Scientific and Research Organization, had designed in-hub motors for several generations of the Aurora solar car team's entries in the WSC. This motor was nearly 5% more efficient and almost half the weight of the NGM motor; additionally it could be outfitted with a mechanical brake, unlike the NGM motor, and its smaller size would allow a smaller wheel fairing and thus less aerodynamic drag. While these gains were very appealing, CSIRO is a research laboratory, not a commercial venture, so complete motors were not readily available. CSIRO would sell a complete set of windings and stator, but Team Lux would need to design and build a housing, drive shaft, and mountings. As this integration would be difficult and time consuming as well as expensive, the project was put on hold, pending future fund-raising efforts and the possibility of some members working on the motor as an independent study project for academic credit. Neither of these came to pass, so the Team Lux would again use a NGM motor for ASC 2003, though for the first time it would be mounted in the front of the car, rather than the rear.
Upgrading to NGM's latest controller, the EV-C402 was also considered. This new model had even higher efficiencies, additional data acquisition features, and sine-wave controlled switching that eliminated the characteristic 'quacking' noise of NGM motors. The new controller used a different serial interface protocol than Team Lux's existing motor controller, so considerable reprogramming would be needed to substitute the new controller. This reprogramming, along with financial reasons, led to Team Lux staying with their existing motor controller. During scrutineering for ASC03, the controller started behaving erratically and considerable time was spent trying to diagnose and repair it. It was suspected that the motor controller circuitry had been burnt out by operating it with the power input wiring only hand tightened, causing a high resistance connection, and it was determined that the controller would need to be returned to NGM, who, for the first time, did not have a representative at the race. After the race, when investigating the problem further, the damage was actually traced to a loose wiring harness inside the motor controller, which was easily repaired. The same motor and controller were used for both the John Lee 1.5 and the John Lee 2, though the motor was moved from the front wheel to a more conventional rear wheel location.
For Sunrayce 95, Delphi Electronics, a division of AC Delco, donated packs of their valve regulated lead acid (VRLA) electric vehicle batteries to any entering team that wanted them. These batteries, originally developed for GM's Impact electric vehicle, which later became the EV-1, were arguably the best lead acid battery on the market for solar racing at the time, and numerous teams, including Team Lux took advantage of this offer. Although Team Lux did not complete a car in time for Sunrayce 95, the 17 batteries received from Delphi were the obvious choice to use in Lux Aeterna for Sunrayce 97. Beyond the clear advantage of being free, they also had a high energy density and a particularly good mechanical design.[42]
In order to make the best use of the batteries, they needed to be matched into packs of seven batteries, the maximum number which could be use in the car and remain below Sunrayce 97's 140 kg battery weight limit. While Team Lux already had a power supply capable of charging the batteries at low currents, but to properly characterize and condition the batteries, it was necessary to put them through several charge/discharge cycles. In order to do this more efficiently, a higher current power supply and an electronic load was required. Sponsorships were sought, and a 15 kW (100 A, 150 V) power supply and a 4 kW (400 V, 600 A) electronic load were obtained. All of the rooms on the first floor of Dunham Labs were outfitted with a flexible power distribution scheme, making 240 V 3-phase AC electrical power readily available in the Team Lux office and allowing the team to set up the new power supply there. This was not necessarily ideal from a safety standpoint as it put exposed high voltage electrical wiring in close proximity to the everyday activities in the office and there was no exhaust fan to remove any hydrogen gas that might be released by the batteries during the charging process, but sealed lead acid batteries vent very little hydrogen.
Team Lux investigated several different connector possibilities for the batteries, including slide on terminals that would allow easy battery changing. Although convenient, these were determined to have too high a contact resistance for efficient use. The lowest resistance connection was actually achieved by using a hydraulic press to flatten a thin piece of copper tubing over the stripped end of the cable. This terminal was then soldered in place and a hole drilled in it to fit over the threaded studs on the batteries. When charging the batteries in parallel, braided grounding strap worked well as a flexible bus bar into which each battery could be tied.
After consultation with Delphi's battery engineers, the batteries were put through several charge cycles and a boiling cycle (Ankher 2001). With the batteries connected in series, they were purposefully overcharged. At the higher voltage of the overcharge cycle, hydrolysis of the water in the batteries occurred, 'boiling' the batteries as free hydrogen and oxygen were produced and vented from the battery. While boiling permanently damages the batteries and can greatly reduce their life span, these batteries were intended only for racing use with a limited number of charge cycles required. Removing water from the batteries increased the acid concentration, and effectively raised the energy density of the batteries. As an additional benefit, this process helped match the batteries. The power used in hydrolysis increases exponentially with voltage, so as the batteries were charged in series, power was diverted into hydrolysis in some batteries, while 'weaker' batteries continued to charge.
With the batteries conditioned for racing, the only remaining task was to secure them in the car. Space had been allocated in the frame directly in front of the driver, in between the two front wheels. As the heaviest component in the car, this would help distribute weight to the front wheels. In addition, as the mass of the batteries would generate considerable force during sudden decelerations, the central location in the frame was an easy place to reinforce. For safety, and to comply with race regulations for a non-conducting, acid resistant material, the battery box was constructed out of composite panels bonded together with epoxy and secured to the surrounding frame members; another panel served as the lid and was held down by a pair of nylon straps with ratchet mechanisms. Because there was sufficient thickness in that portion of the body, the batteries were oriented vertically, allowing all seven batteries to fit in a very compact space. A very low-tech solution to securing the batteries within the box was solved by a pair of plywood dogs.[43] Cut out to match the spaces between the batteries, the dogs aligned with grooves in the top and bottom of the battery cases and prevented any movement.
While the small space between Lux Aeterna's front wheels was the logical place to locate the batteries, it made loading the batteries into and out of the battery box quite challenging. The batteries' vertical orientation meant one terminal of each battery would be at the bottom of the box, and given the recessed nature of the terminals and the limited space between each battery, accessing those terminals was impossible when all of the batteries were in the box. Additionally while it was possible to access the top terminals with the batteries in the box, this could be done only with a wrench from above with a very limited arc of movement, making it a slow and frustrating process. Since they could not be connected with the batteries in place, all of the cables that went to the lower terminals and most of those going to the upper terminals had to be connected before the batteries were placed in the box. This resulted in a precarious endeavor where several of the 19 kg batteries had to be perched on the rim of the box and then lowered into place simultaneously. This got easier with practice and, although only a major catastrophe would require the batteries to be replaced during a race where speed would be essential, it remained a time consuming annoyance. For example, when several new team members who had not been involved in Lux Aeterna's construction were loading the battery box for testing the summer after Sunrayce 97, the voltage sensing wires that went to each battery were accidentally omitted and given the effort required to remove the batteries and then install them again, it was decided that it would be sufficient to know only the entire pack voltage and not the voltage of the individual batteries, which is certainly not an ideal situation.
New batteries were required for Lux Perpetua, as the regulations governing batteries for Sunrayce 99 had changed considerably from the previous race. Instead of a weight limit of 140 kg there was an energy limit of 5 kWh. Additionally, multiple battery chemistries were now allowed, with nickel-cadmium and nickel-metal hydride joining the previous lead acid (Sunrayce 1999). Team Lux was very interested in NiMH batteries, as they offered nearly twice the energy density of lead acid. GM Ovonic was producing NiMH batteries specifically designed for use in Sunrayce, but they were extremely expensive.[44]
Because of this high cost, a special fundraising effort was planned. The road surrounding the Yale Bowl was used as a test track and Lux Aeterna, after some repairs to the telemetry, was raced for an afternoon. Current measurements were taken both for Lux Aeterna's normal configuration and with 68 kg (150 lbs) of additional weight added to the car. This data was then used in a fundraising brochure showing the power savings of removing 68 kg of weight from the car, such as by replacing a 150 kg lead acid batteries pack with an 82 kg NiMH one, and the increased performance that could be achieved by this.
Unfortunately, insufficient funds were raised and the team had to proceed with lead acid batteries. The Delco 'white case' batteries were no longer produced, as GM's EV-1 had switched from lead acid to NiMH batteries and Delco had discontinued the model. Initially the most suitable lead acid battery appeared to be the Optima Yellow Top series of batteries, which had a slightly higher energy density than the Delco 'white case' batteries. However, at the pre-Sunrayce planning conference in March 1998, Delco introduced a new solar car battery. This would be a 16 kg, 12 V, 46 Ah SLI[45] lead acid and would be marketed under the Delphi brand as part of their Freedom series of automotive batteries. While the 5 kWh maximum would allow only 7 Optimas for a 4.55 kWh pack, 9 Delphis could be used for a 4.95 kWh pack. Although a SLI battery would not have as long a life span as a deep cycle battery like the Optimas or the old Delcos, at just over $3000 for a pack of 9, a backup set could be acquired. Unfortunately, the new Delphis were designed as standard automotive batteries and did not retain the form factor or recessed terminals of the solar racing specific 'white case' batteries. Connections were 'GM style' side terminals - effectively a 3/6"-16UNC threaded nut - so hex bolts would be used to bolt lugs to the batteries. Using a string of 9 batteries also had the disadvantage that the nominal bus voltage would be 108 V, with peak voltages reaching 135 V. Many of the general wiring components in Lux Aeterna - switches, fuses, and the like - had been rated for only 100 V, so less common 150 V parts would need to be found. Commercial wiring devices rated for 300 V are readily available, but these are generally considerably heavier than what is desired for a solar car.
Once the battery pack was finalized, a new problem was found. Due to coordination and communication issues within the team, the battery compartment integrated into the frame was sized around a pack of 8 Optimas. Although the Delphis were not much larger than the Optimas, their additional width was sufficient to prevent them from being arranged into two rows as was originally intended. To accommodate two rows of Delphis, the battery box compartment of the frame would have to be extended from 14.5" (37 cm) to 15.75" (40 cm) deep. While this 1.25" (3 cm) would have made little difference to the car before the frame was constructed, modifying the completed structure would require removing and remanufacturing nearly 20% of the frame. As this was not considered practical, alternate battery locations were explored. Stacking the batteries vertically was also ruled out by space constraints, and it was determined that only a single row of 6 Delphis would fit in the space currently allocated. It appeared that some of the battery pack would need to be further forward in the frame. From a mechanical point of view, this was desirable as it helped equalize weight between the wheels, but it also added weight from additional battery box materials and the longer cables needed to connect the distant batteries. In addition, with batteries located throughout the car, they would not all be maintained at the same temperature, which could lead to unmatched batteries.
A solution was found with five of the batteries located in an enclosure in the original battery compartment behind the driver and the other four located in a pair of smaller enclosures on each side of the driver, effectively creating a "U" shaped battery box. These enclosures were made out of prefabricated 3/8" (9.5 mm) thick fiberglass and Nomex composite panels that were secured together with epoxy and then strapped to the frame. Fans were added to the box ends inside the driver compartment and flexible ducting connected these boxes to the main battery box in the rear of the frame. A temporary battery box was put together just in time for preliminary qualifiers at the end of April 1999, but when the car was taken to the GM Proving Grounds in Milford, MI, problems with this box were one of the reasons Lux Perpetua was not cleared for dynamic qualifiers. The scrutineers were concerned with the sealing of the box, the strength of the joints, and the limited airflow around the batteries.
The battery boxes were reconstructed more carefully, additional fans were added, and load-rated straps were procured. While the final battery box was workable, it was never ideal and was a compromise that slowly evolved from things not working as planned. The battery box was still following the model of a single row of batteries with the two terminals at each end of the chain, but with the batteries laid out in a "U" around the driver, the ends of the chain were far from the switches to which they connected. After a brief inspection of the car, Bennett Sprague, who had been the frame group leader for Lux Aeterna and was now at the race working for Ovonic, immediately noticed a way that several feet of battery cable could have been eliminated. No one involved in the construction of the battery box had stopped to realize that by running a short piece of cable from one leg of the "U" to the other, a loop could be formed and the positive and negative terminals of the battery pack could be located directly underneath their switches.
The 2001 American Solar Challenge had significant changes to the regulations governing batteries. The race would now feature a stock class, limited to lead acid batteries, and an open class with no limitations on battery technology. Capacity would return to being determined by weight, ranging from 165 kg of lead acid batteries to 30 kg of lithium ion batteries. As a further change, the entire battery box was now required to be removed from the car as a single unit so that the team's observer could impound it overnight (ASC 2000).
After evaluating the budget for the car, Team Lux decided that NiMH or lithium ion batteries were still too expensive, so it was decided to build a stock class car. The Delphi Freedom batteries used in Lux Perpetua were still being sold,[46] and with the new limit of 165 kg, 10 of them could be used, resulting in a 120 V, 5.5 kWh battery pack weighing 164 kg. If the higher capacity Optima Yellow Tops, the batteries which were originally going to be used for Lux Perpetua, were chosen, eight of them would be used, for a 96 V, 5.2 kWh pack, weighing 158 kg. While slightly more total capacity was available from the Delphis, whose energy density was 2% greater than that of the Optimas, there were mechanical and electrical drawbacks to them. Since the Lux Perpetua frame would be re-used for Lux Millennia, the problem of only six of the Delphis fitting in the rear battery box compartment remained. This was a more significant problem than it had been in 1999, since the batteries now had to be removed from the car each night. This would require that the power and telemetry wires, as well as the air ducts, which ran between the three separate battery boxes be readily disconnectable. The simplification resulting from all eight Optimas fitting in the space behind the driver was reason enough to choose them, but their being a pack of 8 batteries with a nominal voltage of 96 V was also advantageous. Team Lux's NGM motors and controllers have an input voltage range of 84 to 108 V (a 48V version is also available), which correlates to between 7 and 9 12 V batteries in series. Using 10 Delphis would require both converting their 120 V back to a lower range that could power the motor as well as stepping up the array voltage so that it could charge the batteries. Doing this would add cost, complexity, and electrical inefficiency, so a pack of 8 Optima Yellow Tops was the clear choice for Lux Millennia.
Other than the model of battery, Lux Millennia's high voltage system and that of Lux Perpetua were very similar, but Team Lux would greatly depart from this for the John Lee. Thanks to dropping prices of batteries and solar cells as well as a very generous cash donation from the Stiefel Corporation, Team Lux now felt it could look beyond lead acid batteries and silicon (Si) solar cells to the far more efficient combination of lithium ion batteries and gallium arsenide (GaAs) solar cells. A number of different products were looked at and it seemed that for batteries, the best combination of energy density, price, and availability were 16650 cells (a size approximately halfway between 'AA' and 'C') manufactured by LG and distributed along with protection circuitry by Worley. In one of the oddities of the modern global economy, an American solar car team was purchasing batteries manufactured by a Korean firm along with German protection circuitry that was integrated at the Singapore division of an Australian company.
The 166 Wh/kg energy density of the LG lithium ion cells was considerably higher than the approximately 33 Wh/kg of the lead acid batteries used by Team Lux in the past, which potentially allowed for a considerable increase in performance, while bringing many new problems for the team to deal with. To start with, the battery pack cost $10,000, instead of around $3000 for lead acid batteries. This made the purchase of a backup battery pack impractical and removed a safeguard that Team Lux had always had in the past. Additionally, instead of eight or nine large 12 V batteries, there were instead 650 small 3.7 V batteries. Where assembling a lead acid battery pack is basically as simple as making a fiberglass or Kevlar box, dropping the batteries in, and bolting on some power cables, assembling a lithium ion pack from individual cells is a considerably more complex procedure. There are also difficulties associated with the lithium ion chemistry itself. Lithium ion batteries are more sensitive to temperature shifts and operate at a smaller range of voltages than lead acid batteries. Where the failure mode of an overcharged lead acid battery is merely to vent hydrogen gas, which, with proper ventilation, is not particularly dangerous, overcharging a lithium ion battery can cause it to explode turning the casing into shrapnel and spraying the toxic contents of the battery. Because of this, lithium ion batteries are required to have protection circuitry to disconnect the battery pack if voltages or temperatures reach dangerous levels.
Assembling a lithium ion pack was entirely new to Team Lux, and early in the design process a prototype battery box was built. Unfortunately, construction of the actual box kept being delayed and was not returned to until a finished battery box was needed to complete the car. The basic parts of the box were cut out of fiberglass panel and the end pieces were drilled with a hole pattern that would allow air to flow down the channels that would be formed when the cylindrical batteries were stacked. These pieces were turned over to the body group, who assembled the battery box using epoxy and reinforced the joints with prepreg Kevlar cloth. In addition to the composite panels forming the structure of the box, three contact plates were needed: two to interface the positive and negative terminals of the batteries with the system wiring and one to connect the two rows of batteries that would be installed in the box.
The batteries were arranged into cubes of 25 batteries connected in parallel - five packs of five pre-connected batteries stacked on top of each other - and then 26 of these cubes were connected end-to-end in series, for a total of 130 five-packs, 650 individual batteries. To avoid having a battery box that was five and a half feet long, the batteries would be arranged in two parallel rows, with one end connected with a flat plate, making an electrical circuit in a U shape, with both the positive and negative terminals of the assembled pack being at the other end. These plates needed to be machined with individual contact points for each of the 25 (or 50) batteries they would touch, as well as holes matching the battery box end plates, to allow air to flow through the stacks of batteries. Additionally, the terminal contact boards each needed a protruding section with a hole drilled in it, so that the main power cables could be bolted to the battery box, and the far end plate needed recesses to hold the springs that would maintain contact pressure on the batteries.
These plates were surprisingly difficult to machine, because of the softness of the very pure 1000 series aluminum. It had been chosen for its high electrical conductivity, 50% higher than the 6061 and 7075 aluminum used elsewhere in the car (NDT 2002), but because of its purity, it was very malleable. This led to the thin contact plates deforming when they were clamped down for machining or bending under the pressure of being drilled through. Once the plates had been machined, they, along with some high voltage distribution blocks and bus bars, were sent out to be electroplated with 0.7 mil (17.8 mm) tin coating, protecting the aluminum from oxidization and further enhancing their conductivity.
While waiting on the return of the plates, attention was paid to the batteries themselves. The batteries had been in storage for much of the winter and spring, and about a quarter of them were found to have self-discharged to a rather low level. Since each of the 650 batteries in the car would need to be closely matched in voltage for maximum efficiency and to prevent thermal runaway, the status of each of the batteries needed to be determined. This required that each pack of 5 batteries, soldered in parallel by the manufacturer, be uniquely labeled, have its resting voltage measured and recorded, and then be individually charged to capacity. After all the battery packs were charged, a tedious process tying up all of the electronics work stations in the Morse Teaching Center for several days, they could again be metered over a several day period to monitor their rate of self discharge and identify any suspect packs. With each pack of batteries now characterized, the next step was to attempt to match them to each other through several cycles of charging and discharging in parallel. A battery-paralleling fixture was built, essentially a wooden box just wide enough for one pack of five batteries to stand upright and long enough to accommodate sixty-six packs. The batteries sat upon a heavy brass plate and had another plate placed on top of them. The remaining sixty-six battery packs were then placed on top of the second plate, with their positive contact down, and a final plate placed on top them before the box was closed.[47] The top and bottom of the box were then clamped together to ensure that good electrical contact was being made between the batteries and the brass plates. By using a thick cable to bond the top and bottom brass plates together, the entire pack of batteries was now in parallel - all of the positive terminals were electrically connected to all of the other positive terminals and all of the negative terminals were electrically connected to all of the other negative terminals. The batteries could then be charged together, a much faster process than handling them individually, and it was ensured that they would all reach the same voltage. After being charged in parallel, the batteries were removed from the paralleling fixture and after some time to rest, were again metered individually to determine if all of the batteries were matched to within 1 mV.
In addition to simply holding the batteries, the battery box integrated additional components to make an entirely self contained battery pack with only three connectors on it: main power output; relay switching control and fan power input; and serial telemetry data on the batteries voltages and temperatures. Lithium ion batteries are required to be supplied with protection circuitry and race rules also required the positive terminal of the battery pack be fused and have a disconnect switch before any other connections to the car's electrical system. All of this was located in a small alcove at one end of the battery box sized just large enough to hold the manufacturer's supplied protection circuitry, the fuse, disconnect relay, and a pair of printed circuit boards to interface everything. The protection circuitry had four modules, each having a nine-pin connector to measure the voltage of seven batteries. The twenty-six cubes of paralleled batteries required twenty-seven connection points, as one wire would serve for both the positive of one battery and the negative of the next battery in series for all but the first and last connections. This led to three nine-wire ribbon cables being an optimum solution and an adapter board was manufactured that fit over the protection circuitry's four nine-pin connectors and routed the signals to three nine-pin latching headers on the opposite side of the board. Additionally, those twenty-seven signals were passed through miniature switches that protruded from the battery box, allowing the protection circuitry to be disconnected from the batteries and preventing any possible discharging of the batteries when they were not in use. Finally, the protection circuitry was modified so that the data output, emergency disconnect switch, and relay control output all used the same nine-pin connector so that a single cable could be used between it and the interface board. This board held the connectors for connecting to the rest of the car and routed those signals to the protection circuitry, the relay, and the battery box fans - which race rules required to be operating whenever the car was switched on. The main power connector was on a short cable that protruded from the box, with the negative lead bolted directly to the battery pack's terminal plate and the positive lead bolted to the relay's output.
All of these supplementary parts were carefully packed into the battery box alcove and the battery box was then assembled. As the batteries were laid into the box, the four temperature probes from the protection circuitry were routed through the spaces left between the round batteries and distributed throughout the pack. In between each cube of twenty-five batteries very thin strips of brass foil were run vertically, perpendicular to the manufacturer welded bussing strips that formed the packs of five batteries, to electrically connected each battery in the cube (Watkinson 2001b). Additionally one of these foil strips in between each cube was left long as a connection point for the voltage sensing lines. A tiny fuse was soldered to the foil strip and one of the wires from the ribbon cables that led back to the protection circuitry was soldered onto the other end of the strip. All of these wires were carefully routed on top of the batteries and the top was placed on the battery box. Foam insulation placed around the top's rim provided a tight seal as nylon straps were ratcheted around the box to hold it shut. The assembled battery box was located along the right side of the frame and carbon fiber blocks were bonded into the body to prevent the box from sliding. Additional ratchet straps secured the box to the adjacent frame member.
A few days before departing for ASC 2003, the protection circuitry failed, rendering the battery pack unusable. The cause of this failure was unknown and although two spare modules had been purchased along with the protection circuitry, they were not useful in attempting to repair it.[48] The manufacturer was contacted, and there was a possibility that a replacement could arrive in time for the race, so it was ordered and would be shipped to the team at scrutineering in Chicago, IL. As the car could not be tested without a battery pack, the electrical group leader shifted his efforts from fine-tuning the telemetry to constructing an alternate protection circuitry.
The primary function of the protection circuitry is to monitor the battery pack and to open the relay isolating the pack if conditions become dangerous: a cell dropping below 2.7 V or rising above 4.2 V or the temperature in the pack rising above 50¼C. Designing a new protection circuitry seemed feasible as this functionality was very similar to the telemetry systems that Team Lux had implemented for previous cars, although now there were 26 3.5 V battery cubes instead of 8 12 V batteries. The most challenging aspect was isolation between the high voltage battery pack and the low voltage telemetry. Each cell is only a few volts, but the offset from one end of the pack to the other is more than 100 V. Team Lux had a number of photovoltaic relays on hand that had previously been considered as part of a telemetry system that were used to prototype a new protection circuitry. These MOSFET based relays were triggered through a photovoltaic switch so the control side and measurement side were completely isolated. Each sense line was connected to the input of a relay, and then starting at one end of the battery pack, the outputs of all of the odd-numbered relays were bussed together and the outputs of all of the even-numbered were bussed together. These two busses then connected to the input of a -5 V to +5V analog to digital converter. Each battery sense line functioned alternately as the positive terminal of one battery and then the negative terminal of the next battery, so wiring was simplified considerably by connecting every other relay together and allowing the measured voltage to alternate polarities. The output of the analog to digital converter was fed to a microprocessor, which also received the signals from the four thermistors salvaged from the old protection circuitry. The microprocessor controlled the switching of the relays so only one pair was open at a time, passed the measured data to the telemetry system via a RS232 connection, and generated the control signals for the main battery disconnect relay (Glickman 2004).
Although the prototype relay-based protection circuitry was working well, the new protection circuitry ordered from Worley arrived just before electrical scrutineering and was installed in the car. Unfortunately, Worley had changed the shape of the plastic enclosure that held their protection system[49] and it would no longer fit inside the battery box. The new enclosure had to be strapped to the outside of the box and four DB9 serial cables were necessary to extend the voltage sense cables from their previous termination inside the battery box to the new protection circuitry, which was a cumbersome solution.
Later versions of the John Lee continued to use the same lithium ion battery pack, although a new box was manufactured to better fit around the John Lee 2's frame. As Team Lux's custom-built protection system measured battery voltage with greater accuracy than the one purchased from Worley and would also be easier to troubleshoot and repair, the team decided to continue development of their own protection system and relegate the Worley system to backup status. The protection circuitry design was further refined and de-multiplexers were added to control relay selection, which would prevent the voltage sense lines from possibly shorting during switching, and then a custom printed circuit board was ordered for the design (Glickman 2004). The most significant change was the addition of four electric hair dryers to the car just before the Phaethon. These hair dryers served to dissipate excess electrical energy by converting it to heat. Ideally, a solar car takes advantage of all of the energy available to it, but with the protection circuitry of a lithium ion battery pack disconnecting it when fully charged, this is not always possible. For a highly efficient car with a powerful GaAs array, the race speed limit becomes a constraint and additional power from the array or regenerative braking that cannot be used for charging an already full battery pack must simply be disposed of as heat.[50]
As Sunrayce 97 regulations limited solar cells to those manufactured in North America that were available to all teams at a price of $10/watt or less, there were only a few possibilities for Lux Aeterna's array. All of the cells that met these requirements were silicon based, primarily monocrystalline cells of approximately 14% efficiency. Cells manufactured by Siemens[51] were ordered and sent outfor cutting. Unfortunately, a greater percentage of cells than expected were broken during cutting. More cells were ordered and the task of chaining the cells began. A wooden jig was constructed that would align the cells into chains and then allow the assembled chain to be lifted from the jig as a single unit. Thin strips of tinned metal, referred to as tabs, were donated by E. Jordan Brooks, Santa Fe Springs, CA. These tabs were cut into pieces twice as long as each solder cell and then soldered to the two contact strips on one side of each cell. The remaining portion of each tab could then be soldered to the opposite side of the next cell, connecting them in series. After some experimentation to determine the minimum spacing between cells, chaining became a routine, but delicate operation. For the cells at each end of a chain, thicker, longer tabs were applied which would be used to interconnect the chains and attach them to the power-point trackers.
Once all of the cells were assembled into chains and carefully cleaned, they could be assembled into the solar array. The substrate for the array was a flat piece of Kevlar composite panel that had small stiffening ribs on the underside and a piece of steel angle iron along the front edge that aligned the array with the car. Holes were drilled in the panel for the tabs at the end of each chain to pass through to the underside of the array. A number of Dow Corning products were used to encapsulate the cells. Sylgard® 184 silicone elastomer was used to attach the cells to the car, with a flat sheet of acrylic used to hold the cells in place while the elastomer cured. The array was then cleaned with OS-20 and a final coating of 1-2477 conformal coating sealed the array. To complete the array, the chains were wired into three groups with a long wire tail to connect the array to the solar car. The cut ends of common orange 3-wire extension cords[52] were spliced into the wires between the array sections and the PPTs to serve as disconnects. This was an expedient solution, but had some significant drawbacks. As a separate, and identical, connector was used for each of the three array sections, it was easy to cross-connect them. Also, as this style of connector has exposed male terminals, there was a potential shock hazard. If a PPT is switched on without solar cells attached to the input, the input terminals are energized to the battery voltage and shorting them will blow a fuse on the PPT. This was dealt with by inserting the male plugs into a small block of foam when they were disconnected from the array, but some shorts still occurred, causing a stock of the replacement fuses to be acquired.
The final component of the Lux Aeterna solar array was the stand used to support the array while it was not on the car and to orient the array perpendicular to the sun. A lightweight metal frame was constructed out of aluminum tubing and the array could be simply attached to it with plastic cable ties around the tubes and through holes in the array's ribs. This frame was supported by a leg attached with a swivel bearing at either end of the frame. Nylon straps tied to the top and bottom edges of the frame could then be used tilt the array to the appropriate angle and then tied off to secure the array.
For Lux Perpetua, with similar race regulations, cell selection was again straightforward and Siemens cells were again used. Team Lux tried to implement the lessons learned about cutting cells for Lux Aeterna, but breakage and delays were again a problem. Once the cells did arrive, tabbing and chaining them took nearly three months. Although this was an extremely time consuming process, it proceeded relatively smoothly. Lux Aeterna's array assembly process had been documented in an instruction guide by team member Lutz Berners (BK '99), so repeating the process for Lux Perpetua was straightforward. By mid-March nearly 100 10-cell chains had been completed and each was then evaluated in the Morse Teaching Center using an array of halogen lamps and an automated LabVIEW test routine with computer controlled power supplies and voltmeters. With this test data, the location of each chain on the car was plotted for maximum efficiency and chains needing repair were identified.
Attaching the chains to the body was a new challenge, unlike Lux Aeterna's array where all of the cells lay flat and parallel with the ground, the chains of cells for Lux Perpetua needed to conform to the curved body and not slide down the side of the car. Although the same technique used on Lux Aeterna was tried initially, a great deal of trial and error was required to develop a process whereby the cells could be securely held in place while the conformal coating set, yet not be damaged by excessive force. By mid-April, one-quarter of the array had been attached to the car, but the body group leader desired a much smoother surface finish to achieve suitable aerodynamic performance. As many of the solar cells had been damaged during testing and assembly,[53] additional cells were ordered and chained and additional array work was put on hold until after the race qualifiers and final exams.
With the beginning of summer, classes no longer distracted from work on the car, but there was a great deal of work remaining. After studying other teams' arrays at qualifiers and further experimenting with attachment techniques, the decision was made to completely redo the array, requiring removing the quarter array that had already been applied, purchasing additional solar cells, and soldering them into chains once they arrived. Plans were made to have the chains of cells that had been soldered commercially encapsulated, but in the end a simple method was discovered where each chain was attached directly to the body using 3M Very High Bond Adhesive, or VHB™, an industrial strength double-sided tape. The tape securely held the cells smoothly against the car surface without any of the difficulties of the liquid adhesives that had early been tried. Once the cells were applied, the entire array was covered with silicone elastomer, creating a single smooth surface. Once this process had been worked out, the entire array was attached to the car in a single day, a much faster process than before with considerably fewer toxic chemicals.
Once all of the cells were applied to the body, the chains were interconnected into three groups. Originally, four areas were going to be used, but once it was discovered that there were only four working PPTs on hand, this was reduced to three areas so that there would still be a spare PPT. As a principal goal of the Lux Perpetua electrical group was to produce a more neatly wired system than that of Lux Aeterna, a metal box was obtained that held four PPTs. The PPTs were mounted with standoffs in the box, along with fuses for the PPT inputs and the necessary connectors. This neatly wired box was an attractive and easy to work with solution, but created unanticipated thermal problems. Although the PPTs were 97% efficient, with a 1200 W array, 36 W was generated as heat in a box with almost no free air space. As the efficiency of the PPTs decreases at high temperature, three fans were mounted to the enclosure.
The connector choice for the Lux Perpetua array differed considerably from Lux Aeterna. Instead of the spliced extension cords for the input and a hard-wired output, connectors designed for speakers in touring sound systems were chosen. The Neutrik Speakon® NL-8 had desirable characteristics, as it was lightweight, durable, latching, and the contacts on both the plugs and receptacles were shielded from touch. As the NL-8 had 8 poles, each rated for 30 A at 250 V AC, the four originally planned array sections could all be handled by a single connector. Output from the PPT box to the rest of the car was done through a similar connect, the three pole PowerCon®. Although these connectors worked well electrically, neither was designed for the use with the types of cable Team Lux was using in the solar car. For the Speakon connector between the array and the PPTs, the use of individual cables without an overall outer jacket merely meant that the connector's cable clamp did not grab effectively, but the cables connecting the PPTs to the batteries and motor did not fit in the shell of the PowerCon at all. Because of a misunderstanding with a sponsor, this cable had a 1000 V rating and thick dual layer insulation. By removing the outer layer of this insulation, it was possible to squeeze two of these cables into the PowerCon connector, but this was an awkward solution.
The array was completed before Team Lux had to leave for the start of the race, but time had not been found to construct an array stand. While Lux Perpetua was undergoing scrutineering at Summit Point Raceway, Summit Point, WV, two team members constructed an improvised array stand out of pine 2x4. Three cradles were built to support the array and these cradles could then be tilted by changing the angle of a pivoting support leg. This array stand had some significant drawbacks - it could not pan to follow the sun, did not raise the array over ground level obstacles, and it was cumbersome to set up and transport - but it was functional and served for the duration of the race.
Once it was decided that Lux Millennia was going to be a stock class car, it was necessary to select from the ASC 2001 approved cell list. This limited choices to cells from Siemens, BP Solar, ASE Americas, AstroPower, and SunPower. As cells were being ordered relatively late, only BP Solar, Frederick, MD, could provide sufficient cells immediately, but given the relative equivalence of all of the stock class cells, this was not particularly problematic. The BP Saturn 125 was a 16.25% efficient monocrystalline silicon cell and was 125 mm square with clipped corners. Cutting the cells down to rectangles was investigated, and a quote for saw cutting the cells was received from SunWize. The Lux Millennia body could fit 550 uncut cells or 850 cut cells. Cutting the cells would increase effective array area by 4.9%, resulting in additional 65 W of power output. It would take at least two weeks for the cells to be cut and the actual cutting, along with additional cells to account for breakage, would cost at least $3000. Given both the tight timeline and budget, the team decided the additional 65 W of array power was not compelling and the decision was made to use uncut solar cells. Once the 900 solar cells[54] arrived in mid-March, the group began tabbing them. This process continued throughout the semester, and layout boards were fabricated by Ed Kinter, the chemistry department carpenter, in preparation for cell chaining. There were some initial difficulties in soldering the cells and the water-soluble flux based solder that had been used for Lux Aeterna and Lux Perpetua was now performing poorly, possibly due to is age or extended time spent un-refrigerated. However, after some experimentation a paste flux purchased locally was found that worked well, although additional cleaning of the cells after soldering was required.
Once Lux Millennia was completed, it underwent track testing, where a serious problem with the array was discovered. There was an unanticipated voltage mismatch between the power-point trackers and the battery pack, preventing the array from recharging the batteries. Upon evaluating the data acquired on the first day of racing, it was discovered that the array was only supplying power when the car was driving at full speed. The problem was that each array section was only producing 100V, less than the normal voltage of the battery pack, so it was only when the motor drew large currents, causing pack voltage to drop below 100V, that the array could supply power. In the short term, this was fixed by combing the two array halves in series and using only a single PPT. This raised the array voltage to around 200V, allowing the array to charge the batteries under normal conditions. When time permitted, the plan was to remove one of the eight batteries, which would reduce the total storage capacity of the car, but also drop the battery pack voltage sufficiently so that the array could charge it.
The major drawback to connecting the entire array to a single PPT was the distribution of sunlight. Given the shape of the car, the front of the car received maximum sunlight with the sun directly overhead, one side with the sun at an angle on that side, and the other side the opposite. Since some portion of the car would always not be in direct sunlight and since each series chain of cells connected to a PPT is limited by the lowest performing cell, having the entire car connected to one PPT created a situation where the entire array would function as if was not receiving direct sunlight. With so little power coming from the array, the team decided not to enter the May 2002 Formula Sun Grand Prix, as it was unlikely the car would have sufficient power to complete the race.
While other solutions to Lux Millennia's array problem were considered, the majority of the team's effort transitioned into design for the John Lee. Decisions were made on solar cell suppliers and purchase orders for the cells as well as their cutting, tabbing, and encapsulation were shepherded through the engineering department's business office. As Team Lux was now planning on open-class car, the first choice was triple-junction GaAs solar cells from Emcore, Somerset, NJ, but when these were unavailable, suitable dual-junction GaAs cells were located from SpectroLab, Sylmar, CA. An initial order of approximately 3000 21% efficient cells was placed, which would result in a 1900W array. The final array would have 3051 cells, covering 7.97 m2 total area. Three sample panels were tested, yielding an average of 23.7% efficiency, which would give the total array a peak theoretical power of 1891 W.
Past experience had shown Team Lux that hundreds of man-hours were needed to solder individual solar cells into chains, and as GaAs cells are considerably smaller than the silicon solar cells previously used by Team Lux, there would be three times as many cells on the car. With the team's relatively low membership at that time, it was decided to outsource the soldering of the cells. Quotes were solicited from several companies to chain the cells and encapsulate them into modules that could then simply be attached to the car. To accurately quote this however, the actual layout of the cells on the car and how they were broken down into modules needed to be known, which had not yet been determined. Nevertheless, using preliminary numbers, SunWize Solar, Kingston, NY, was chosen to saw cut the cells and then tab and encapsulate them.
SunWize expedited the three smallest modules so that they would be on hand for testing. One of these modules was placed on an overhead projector and its power output was measured using a four-wire active load borrowed from an Electrical Engineering lab. With that known baseline, the module was then taken to the top body mold and applied to the region of greatest curvature. The solar cell module was then returned to the test bench and re-measured. There had been some concern that in some areas of the car the curvature would stress the cells, reducing their efficiency, however no such reduction was apparent. The majority of the solar cells, grouped into large modules for the relatively flat portions of the car, arrived at the end of February, at which point the order for the remaining modules needed to be finalized. Unfortunately, complete documentation of the cell layout done in September could not be found and it appeared that certain planned sections would not meet the electrical input requirements of the power-point trackers.
Because of this, the cell layout was redrawn in AutoCAD with a corresponding spreadsheet tracking the number of solar cells in series chains that would be connected to each power-point tracker. The test data from the sample cell modules showing their maximum power-point and voltage drop due to increased temperature was used to determine the minimum and maximum number of cells that could be in each to chain to meet the power-point trackers input voltage requirements even when solar cell output voltage dropped due to high temperatures. With this revised cell layout, new full-scale drawings of the solar cell modules were plotted and the paper cutouts were checked against the physical top of the car. Some minor revisions were made to accommodate the curvature and it was decided to add an additional power-point tracker. Previously the rear left and right halves of the body were each on a single power-point tracker with each accounting for more than a quarter of total array power. However, since during much of the day the central hump for the driver's head would shade a portion of the rear half, it was decided to separate the modules that were most likely to be shaded and place them on their own power-point tracker so they would not draw down the performance of the rest of the cells. With the array layout finalized, the final module order was sent to SunWize and the modules on hand were stored in the office for safekeeping until the completion of the body.
The modules received from SunWize were in general satisfactory, but the cell layout was slightly out of tolerance. The modules had been quoted as having a 0.070" (1.78 mm) horizontal spacing and a 0.035" (0.89 mm) vertical spacing between each cell and an overall border of 0.0625" (1.59 mm). The spacing between individual cells was close to this specification, but the borders around each module were considerably larger than this. There were also two modules that were so far out of alignment that they were unusable. These modules were to go along the front edge of the car and were only 1 cell high by 12 or 13 cells wide. These long lines of cells had curved while being encapsulated, so they would not fit on the car. For the remaining modules to be used, the overhanging encapsulate needed to be removed without damaging the fragile cells. A precision artist's mat cutter was purchased that allowed the modules to be carefully trimmed.
To attach the cells to the body, first the outline of each module was traced directly onto the car using full-scale paper templates. Each module would be attached to the car with 3M VHB. Unlike previous cars, where nearly the entire surface of the cell was taped down, the modules could be attached with tape only along their perimeters and two diagonal strips crossing the center of each module. All of the places where tape would be applied were scuffed with sandpaper and holes were then drilled for each module's contact tabs to pass through the body. The surface of the car was then cleaned with acetone and VHB was applied to the car without removing the paper backing. While the tape adhered easily to the outer edges of the car, which were carbon fiber, and along the central spine, which had been painted, significant pressure was required to achieve a proper bond for the bulk of the body, which was Kevlar. Once the tape was secure, the backing could be removed and a module applied.
Once the modules were attached, they needed to be wired into the appropriate series and parallel groups to connect to the various power point trackers. The bare tabs were passed through the drilled holes in the body and along the edges of the car where the body was made from carbon fiber, the tabs were insulated with electrical tape. The majority of the array sections were simply parallel groupings of like size modules, although along the spine of the car smaller modules were connected in series and the center rear of the car was covered with four parallel connected groups of series connected modules. Because of the high cost, surface mount bypass diodes were not installed inside the modules when they were manufactured, but where entire modules were series connected Team Lux did install bypass diodes in parallel with each of those modules. Additionally blocking diodes were installed in series with each parallel connected module.
After experimenting with several different connection types, common insulated crimp butt splices were found to securely attach to the tabs, the diode leads, and the interconnecting wire while providing a suitably low resistance connection. For the large sections connected in parallel, positive and negative "bus" wires were run the length of these sections and the wires from the individual modules were spliced into the bus using compression type tap connectors. All of these bus wires were brought together at the front right corner of the array where they were securely fastened to the underside of the body and then covered with an expandable mesh sleeving. This sleeved cable ran for several feet, enough distance to remove the array from the bottom of the car but maintain an electrical connection, and was terminated with a MIL-C-5015 series circular connector. A 19-pin connector was chosen so that the nine array sections could remain isolated all the way to the power point trackers in the bottom of the car. This connector was weatherproof and locked securely with a threaded collar, but unfortunately, this collar tended to be over tightened, jamming the connectors together.
The same AERL PPTs that Team Lux had used in all of its previous cars were used again for the John Lee, but since they were optimized for use with silicon solar cells, they needed to be reconfigured for use with GaAs cells. While there were some other PPT possibilities that might be one or two percent more efficient, Team Lux already had a significant investment in AERL PPTs and they had proven reliable in the past, so the seven AERL PPTs Team Lux already had were shipped to Australia for repair and upgrading and an additional five new units were purchased. The array was ultimately divided into nine sections, so nine PPTs were mounted to a Kevlar board with standoffs for a Plexiglas cover that would prevent inadvertent contact with the high voltage wiring. A sheathed wiring harness that mated with the array wiring connector was fabricated for the PPT inputs, while the PPT outputs were bused together at two small aluminum distribution blocks that were also mounted on the Kevlar board. The final component of the PPT board was another wiring harness connecting all of the PPT's on/off control terminals to a DB37 connector that went to the car's switchbox.
During road testing after ASC 2003, the array wiring caught fire, melting the two connectors together. The fire most likely started within the connector, from heat resulting from the high resistance of a poor solder joint, which in turn was caused by the difficulty of soldering the 18 array wires into the small 1" (25.4 mm) diameter circle of the connector.[55] To avoid such a problem in the future, it was decided to move the PPTs out of the bottom of the car and instead mount them on the underside of the array, immediately adjacent to the solar cells they were connected to. This would also save on cable weight, as the outputs of the PPTs could be bussed together into two larger cables, rather than many smaller cables. While distributing the PPTs around the car simplified their power wiring, it made control and telemetry wiring more difficult.[56] This problem was avoided by simply eliminating those wires and instead using a wireless connection. Small electronic modules were designed and fabricated that plugged into each PPT and communicated to a master controller over a 900 MHz radio link.
The low voltage electronics for Lux Aeterna were designed primarily for simplicity. A large military surplus linear potentiometer served as the regen control and a single turn potentiometer was attached to a flat piece of aluminum as the accelerator pedal. These potentiometers were then wired directly to the motor controller. A small panel contained switches to precharge the motor controller, enable the power point trackers, turn-on the accessories, and activate the horn. The only other control was a small toggle switch on the steering wheel for the turn signals. Initially a small timer circuit had been built to flash the turn signals, but once this burnt out, flashing was done manually by the driver.
The telemetry systems were equally simple and consisted of individual modules designed for industrial data acquisition. The modules communicated through a RS485 serial bus, which was converted to RS232 and sent to the chase vehicle by a 900MHz serial modem. The majority of the modules were analog to digital converters, which were connected to a shunt to measure the current out of the battery pack and to each battery to measure their voltage. These modules worked initially, but were designed for stationary applications and were unreliable in a moving vehicle. Vibration would loosen the screws connecting the wires to the modules, so after a short period of driving, portions of the telemetry would stop functioning. With insufficient road testing, this was not discovered until race time. During the race, Team Lux attempted to solve this problem by soldering the wires to the module's screw terminals, but the heat from soldering damaged the modules. As a temporary solution, a Cruising Equipment eMeter (now sold as the Xantrex Link 10) was installed in the car, which measured battery pack current and voltage and calculated an approximate state of charge. Additionally, the fuses in the original battery voltage measurement lines were readily accessible, and individual battery voltage could be measured across those fuses with a small digital volt meter.
A number of electrical engineers were recruited at the beginning of the Lux Perpetua project and there was a great deal of interest in developing advanced telemetry systems for Lux Perpetua. The first effort however, was to use Lux Aeterna as a test platform. While the majority of Lux Aeterna's telemetry no longer functioned, the eMeter was reconnected and linked via RS232 to a laptop to log current consumption. Lux Aeterna was then taken to the Yale Bowl to run test laps and gather data on how weight reduction affected the car.
One of the primary goals in the design of Lux Perpetua's low voltage electronics was that they be better organized than those of Lux Aeterna. After attempting to troubleshoot Lux Aeterna's tangled nest of wires, orderly wiring seemed highly desirable. The electrical components were broken up into modules, each of which could be enclosed in a small box fitted with multi-pin connectors that would link the modules. This had the drawback of adding numerous connections, which meant additional construction time and possible failure points, but allowed sections to be built individually and for the overall system to be more easily comprehended and debugged.
Several commercial data acquisition systems were investigated for use in Lux Perpetua, but all were found to be either too heavy for use in a solar car or too expensive. The team again decided to build its own data acquisition and development centered around the use of Parallax BASIC Stamps® and National Instruments LabVIEW software, which were being used in several Electrical Engineering courses. A sponsorship was obtained from Parallax, which provided several development boards and small LCD displays. These were then used to build a prototype current measuring system using precision shunts.
To monitor the battery pack, the voltage of each battery was measured, as was done for Lux Aeterna, but this only provided a rough measure of the battery's state of charge. To more accurately determine the amount of energy remaining in the battery pack, Ji-Jon Sit (BK '00), a junior electrical engineering major, designed and built as an independent project for course credit a sigma-delta charge quantizer which served as an integrating state-of-charge meter. This analog circuit measured all power going into and out of the battery pack in order to provide accurate knowledge of how much energy remained for the solar car. The individual battery voltages would still be useful information and several methods were considered and attempted to measure them. Measuring a single 12 volt battery is not a particular difficult task, but nine of them in series create a nominal pack voltage of 108 volts and measurements of up to 135V. The majority of integrated circuits will be damaged by voltages this high, so voltage measurement for each battery needed to be optically isolated so that the battery could be treated in the 0-15 volt range and not affected by a high voltage offset from previous batteries in series. The initial plan was for a small module to be mounted on each battery that would measure the voltage, isolate and fuse it, and then convert it to the correct scale for transmission to the central data acquisition unit. Eventually this plan was discarded as needlessly complex and only a fuse was mounted at each battery with all isolation and measurement handled centrally. This voltage measurement, the state of charge meter, and two precision shunts to measure battery and array currents, were connected to a basic stamp which collected data and output it to several small LCD screens for the driver and also to a serial modem for transmission to the chase vehicle.
As had been the case with Lux Aeterna, data acquisition that worked well on the lab bench did not fare so well during the actual race. Between the vibration of driving, moisture from the continual rain during Sunrayce 99, and electromagnetic interference produced by the motor controller, data collection in the car was highly unreliable. Even when it was functioning, there were also problems with the laptop used in the chase vehicle to receive telemetry, causing most of the race to be run without accurate information. During the race, the eMeter from Lux Aeterna was installed to measure the battery pack, providing an additional source of information. The most reliable indicator available in the car was of current consumption. This was measured through a precision resistance shunt, processed through a basic stamp, and then displayed on a small LCD. Unfortunately, there had not been time to calibrate this measurement, so instead of displaying actual Amperes it displayed a scaled number referred to as 'jamps,' short for JJ Amperes, named after the lead telemetry designer, Ji-Jon Sit.
Even though not properly calibrated, this current indicator did provide enough information for the car's driver to optimize driving style, but without accurate information on the battery pack's state-of-charge, it was difficult to formulate overall strategy. On several occasions, it was necessary to stop driving and manually meter the batteries, as malfunctioning telemetry indicated the batteries had been discharged to an unsafe level. Energy management was extremely important in the low-sun conditions of Sunrayce 99, and these problems with telemetry contributed to Team Lux's middle-of-the pack placement.
In the year following Sunrayce 99, there was a great deal of interest in developing improved telemetry. A number of different systems were acquired and tested, including a Hewlett Packard palm-top computer and a single-chip PC running Linux. While the single-chip computer was very promising, it was eventually abandoned due to its high power consumption. A more energy efficient embedded computer was found, this time based on the PC104 architecture, which uses a stack of small modules linked by an ISA bus and is commonly used for data acquisition in extreme environments.
Lux Millennia retained the primary high voltage enclosure from Lux Perpetua, which contained switches, fuses, and the DC-DC converter, while much of the low voltage wiring was replaced. The PC104 microprocessor boards were fitted with amplifying isolators to measure the battery voltage. The digital speed output from the motor controller was also connected to the microprocessor. Current measurement relied on the same pair of shunts used in Lux Perpetua, connected in the battery and array negative lines.
When Lux Millennia was taken to West Virginia for track testing at Summit Point Raceway, Team Lux had, for the first time, reliable telemetry under race conditions. Useful data was obtained and array current measurements allowed the team to diagnose a design flaw in the solar array. The one fault that did occur in the telemetry system can most likely be attributed to mechanical error. After a period of heavy racing, the plastic box containing the two shunts started to melt. For this much heat to be generated, the very low resistance of the shunt must have been greatly increased, likely through a loose bolt allowing one of the ring terminals to only make an intermittent connection with the shunt as the car vibrated.
Using Lux Millennia's telemetry as a starting point, a next generation system was planned for the John Lee. The design of the data acquisition system was performed as an Electrical Engineering department independent research project by the electrical group leader, Michael Glickman (JE '04). Control and telemetry were designed together as a comprehensive integrated system and for the first time was located on a custom designed printed circuit board. This greatly simplified the wiring of electronic components and improved both mechanical and electrical reliability.
The low voltage electronics in Team Lux's previous cars were primary analog, with digital systems being limited to the conversion of analog measurements for display on driver readouts and transmission to the chase vehicle. While the John Lee maintained analog motor control, the heart of the system was three digital microcontrollers. These three Microchip PIC® microcontrollers were linked through the serial peripheral interface (SPI™) and monitored all of the car's systems. The primary microprocessor received data from the other two PIC chips and had RS232 serial communication links with the battery protection circuitry, the driver's display, and the 900 MHz radio link to the chase vehicle. One PIC chip was devoted to communication with the motor controller[57] and the remaining chip communicated with the car's sensors: current, acceleration, and tilt.
Current measurement was also done differently than in any of Team Lux's previous cars, with non-contact measurement being implemented for the first time. Past attempts by Team Lux to implement such a system had produced wildly inaccurate results. The most common sensor style takes advantage of the Hall effect, however even when setups using Hall-effect sensors worked on the lab bench, they failed to function reliably when actually in the solar car. The likely causes for this were high temperatures in the car, which skewed the calibration of the sensors, and large quantities of electromagnetic noise produced by the switching action of the motor controller.
To avoid these problems in the John Lee, the Hall effect sensors were isolated on their own circuit board. The use of a printed circuit board allowed very short lead lengths between the sensors and their accompanying analog/digital converters, which helped minimize the possibility of interference. When originally tested, the sensors consumed a large amount of power and took a long time to stabilize. The power consumed was being used to slowly heat the sensing element to a stable temperature. Instead of compensating for temperature by heating the sensor, a thermistor was placed inside each sensor to measure the ambient temperature so the compensation could be performed in software. Power to the sensor could then be switched off except when actually taking a measurement (Glickman 2003).
The increased complexity of the John Lee's electronics required careful documentation and organization. Cadence Design System's OrCAD® software was used to draw all of the circuits and then perform the physical layout of the circuit boards before they were sent out to be manufactured. Beyond the circuit boards themselves, there were hundreds of smaller components - wires, connectors, switches, fuses, LEDs, resistors, ICs, etc. - that needed to be specified and acquired. Some of these were available in small quantities from the Morse Teaching Center stock or had already been purchased during the development of the low voltage system, but many had not yet been identified beyond the point of "we are going to need some wire." Working from the OrCAD layouts of the printed circuit boards, their component requirements were tabulated and an inventory was taken of stock on hand. While the basic wiring of the car had long since been decided upon, it was finalized in a CAD drawing so that all of the component requirements could be verified. A master list of electrical components was generated and considerable time was spent online confirming specifications and finding suppliers and prices. With the majority of the components specified, the layout of the two main circuit boards could be finalized and sent out for manufacture. Parts to populate one set of those boards were ordered as well as samples of the parts that would be needed for the high voltage wiring.
With regards to the actual wire, the wire sizing spreadsheet used on Lux Perpetua to balance the lower resistive loss of large wire with its increased weight was updated to reflect the wire types to be used in the John Lee and expanded to average power loss for several different configurations. The spreadsheet had previously looked only at one current and one value for aerodynamic drag, but currents would be quite different when driving in traffic, climbing hills, or using regenerative braking going down a steep hill. In addition, drag is speed dependent and with the more advanced modeling used on the John Lee's body, more accurate figures were available for power loss at different speeds. To take advantage of this, an estimate was made for what percentage of the time would be spent stopped (charging), at high speed, at low speed, and using regenerative brakes and the spreadsheet weighted the different currents and aerodynamic losses for those periods accordingly. While it would be possible to take a model of the race route and calculate power consumption and drag along the entire route and apply an estimate of the solar power available to come up with a perfect wire size, it was felt that the four period approximation was sufficient and effort spent on optimization would be more fruitful if applied to other parts of the car.
The design of the John Lee's low voltage electrical system was fundamentally sound and was retained in future versions of the car. The primary circuit board underwent another design revision, further simplifying its design. The separate board for the Hall effect sensors was eliminated,[58] moving all of the telemetry to a single board, and the three separate PIC chips were replaced by one higher speed processor. Communication between components was improved by replacing the RS232 serial links with a CAN (controller area network) bus which was highly noise resistant and easily expandable.
The telemetry also now took advantage of the data acquisition capabilities built into the AERL power point trackers that the team used. The PPTs had provisions for the connection of a small LCD to display input and output current and voltage. Team Lux designed an interface module which would connect to a PPT in place of this LCD and transmit the data to the CAN bus for use by the rest of the telemetry. As the PPTs were now distributed around the underside of the solar array, it was decided to equip these interface modules with 900 MHz wireless communication to a central receiver connected to the CAN bus (Glickman 2004).
For Lux Aeterna the primary goal for frame construction was a simple to build rugged structure. To make the frame as strong as possible, yet still lightweight, Team Lux found a sponsor willing to donate W7A ceramic matrix reinforced aluminum. This donation was of raw aluminum that would need to be extruded into usable shapes for the car. The frame design originally contained several sizes of tubing, but the sponsor only wanted to setup a single extrusion, so the design was quickly reworked to use only a single size of tubing.[59] While the ceramic matrix aluminum had a very good strength to weight ratio, it presented continued difficulties. The material was extremely tough and mill bits wore out rapidly. Machining concave ends for the tubes so they would align properly with the adjacent tubes was time consuming and error prone. Once all of the frame members were properly cut, they were welded into a simple rectangular truss in Yale's Gibbs machine shop. Lux Aeterna was not an elegant car, but this frame served reliably.
For Lux Perpetua, with its more sculpted body and more ambitious design goals, a considerably more complex frame was used. After evaluating several materials, the team decided to use the ceramic matrix aluminum tubing remaining from Lux Aeterna's construction. Working from the outside in, wheel locations, driver position, and battery box placement were determined and the structure was designed to adequately support them. A representative person was measured, to determine the biometric constraints to the frame's design. Unlike Lux Aeterna's planar design, very few of the surfaces of Lux Perpetua's frame were parallel to any others. This less traditional shape, with many compound angles and multi-input joints, was chosen to accommodate the more aerodynamic body shape. It was more difficult to analyze and construct than Lux Aeterna's basic box, but the resulting frame worked well.
At the beginning of the Lux Millennia design process a thorough design review of all components of the car was conducted. The frame group constructed one-eighth scale models of possible frames from brass rod and conducted training sessions on the Finite Element Analysis (FEA) software that was available in the engineering computer lab. Initially another metal tube frame was planned, but with the assistance of Richard Askew a composite monocoque design was considered. Cardboard mock-ups of monocoque cars were created and there was extensive discussion over whether the car should be built out of prefabricated flat panels or laid-up in a mold. A frame constructed from flat composite panels would be a new challenge for Team Lux, but after the departure of some key personnel from Team Lux, this project was dropped and Lux Perpetua's frame was reused.
Team Lux had ambitious goals for their next car, the John Lee, so a great deal of planning and analysis was necessary. To achieve the team's weight goals, every component in the car needed to be optimized and pared down to its minimum form while still maintaining a reasonable margin of safety. To evaluate designs and verify dimensional constraints, a detailed three-dimensional computer model was created in SolidWorks®. This model contained parametric information on the materials such as weight, elasticity, and tensile strength that could then be exported to ANSYS DesignSpace® for FEA. In addition, by adding placeholders for the driver and various electrical components, the car's center of mass and wheel loading ratio could be determined with great accuracy and the location of the electrical components could be optimized for maximum balance.
Composite construction was again considered for the John Lee, this time in the form of carbon fiber tubes. Eventually titanium tubes were chosen over carbon fiber, as using carbon fiber would have required constructing time-consuming molds and it would be difficult to verify that the tubes would be structurally sound. A semi-monocoque design was retained, with the bottom panel of the car bonded to the frame as a structural member, reinforcing the sparse titanium tubes. Whereas Lux Perpetua was welded out of 75 individual sections of tube, the John Lee had only 14. With the body providing additional stiffness, computer analysis showed that only a single ring of large diameter tube was required to provide sufficient strength.
As titanium needs to be welded in a very controlled environment to maintain its strength, the frame was welded by certified welders at Sikorsky Helicopter's Stratford, CT facility. While the welding itself was not particularly time consuming, the set up was. Cutting the angles on the large diameter frame tubes so that there would be a tight fit around the entire circumference of the joint required careful machining, and proper alignment of the suspension mounting brackets would be critical. This accuracy was achieved through the use of Sikorsky's CNC milling machines, so that the finished parts matched the CAD files and would properly align. In order to hold all of the components in the correct locations for welding, several jigs had to be built. The primary frame tubes were supported on v-blocks that were bolted into measured locations on a base plate. After an initial attempt to manufacture v-blocks failed, commercial blocks were purchased. However, once they arrived, it was found that there bases were insufficiently level to maintain the desired tolerances in the frame joints, so a great deal of additional time was spent machining the hardened steel v-blocks so that they were sufficiently accurate. In addition to the frame tubes, 14 flat brackets were machined from titanium plate to match the curvature of the tubes and provide mounting holes for the suspension components. For efficient suspension geometry to be maintained, the brackets had to be matched carefully to each other and properly oriented on the frame tubes. Aluminum blocks with holes matching those on the suspension components were laid out in SolidWorks, machined, and then mounted to the jig base plates, so that the mounting brackets could be bolted to them, ensuring that the mounting holes would remain in the proper locations as the brackets were welded to the tubes. The process of building the jigs and fine-tuning their alignment took several weeks, but proceeded steadily.
Once the titanium tube frame was completed, there was some concern about how thin the wall thickness was on the main titanium tube members. Thin-wall, large diameter tube had been chosen to optimize the materials strength to weight ratio, but even minor scratching could significantly weaken the tube's strength. It was eventually decided to wrap the frame's joints with carbon fiber tape. This would, with little added weight, protect the titanium tube and provide increased strength. The tape was applied and then the entire frame was vacuum bagged and then put in the autoclave to allow the prepreg carbon fiber to cure. Once the frame was complete, it was leveled in the body on foam blocks and secured with low-temperature cure epoxy and carbon fiber tape. When the John Lee was modified to become the John Lee 1.5, the frame remained unchanged.
Although the titanium frame performed as designed, the tricycle wheel layout was insufficiently stable at high speeds, as discussed in the next section. It was therefore decided to reconstruct the John Lee as a four-wheeled car for the American Solar Challenge 2005. To accommodate this redesign a new frame was required, and Team Lux took advantage of the experience in composites that had been developed during the original John Lee construction to build a new frame out of carbon fiber tubes. Flat prefabricated panels were cut into strips to form the outside edges of rectangular cross-sectioned tubes. Reinforcing bulkheads were also made out of flat panels and installed perpendicular to the length of the tube. The panels were joined with epoxy and then reinforced by wrapping the entire assembly in prepreg carbon fiber cloth. The frame was laid-up directly into the body, integrating the two as a structural member. Hard points for the attachment of suspension components, the seat belt, and the battery box were created by installing aluminum plates while the frame was being laid-up. The roll cage was also constructed of carbon fiber, utilizing additional layers for strength and two ribs made from Rohacell®, a very high density foam. The resulting composite frame was both lighter and more rigid than the previous titanium frame.
For their first solar car, Team Lux followed the model of previously successful cars and built Lux Aeterna with an inverted tricycle design: two front wheels for steering and braking and a single rear drive wheel. The front suspension used steel a-arms that were manufactured at Alinabal, Milford, CT, and designed with their assistance. A trailing arm manufactured from a steel box beam was used for the rear motor mount.
Lux Aeterna's inverted tricycle configuration had worked well, but at the beginning of the Lux Perpetua design process, all aspects of the car were reevaluated and numerous other wheel possibilities were considered. Lux Aeterna was reconfigured to drive in reverse so a single front wheel design could be tested and a temporary mounting bracket was created to allow testing of an asymmetric design with a single rear wheel in-line with one of the front wheels. This testing got new members involved and generated discussion into the design of the next car, but ultimately the Sunrayce 99 race regulations were published, limiting options. A new rule permitted only four wheeled cars; this was a response to the possibility of dangerous spinouts occurring after a blow-out on a single rear wheel car. Once race officials made this decision, it was clear that Lux Perpetua would have four wheels, distributed in the traditional locations for an automobile. Alinabal again assisted with suspension design and construction, as well as supplying the necessary rod ends and spherical bearings. Lux Perpetua used steel a-arms similar to those of Lux Aeterna, but the upright was constructed out of aluminum plate. The plate was laser cut at Alinabal and slotted together into a curved box shape, which was then welded. The trailing arms were also constructed out of welded aluminum and all four wheels used oil-damped coil over shocks. As the manufacturer supplied springs were not stiff enough, custom springs were wound to support the car's weight.
After it was decided to reuse the Lux Perpetua mechanical systems, they needed to be restored to race condition. The Lux Perpetua frame was removed from the body and completely disassembled. All components were cleaned and lubricated as necessary, new bolts and nuts were ordered, and the brakes were upgraded. ASC '01 regulations required mechanical breaking on all four wheels, so new brake mounts were created to mount a single caliper to each wheel, rather than Lux Perpetua's dual front calipers.
A four-wheeled chassis served quite reliably for Lux Perpetua and again for Lux Millennia, but a four-wheeled car is heavier and has greater rolling resistance than a three-wheeled car. As the four-wheel requirement was dropped after Sunrayce 99 and considering the John Lee's ambitious design goals going back to a three wheeled design was the logical choice. The John Lee was originally intended to have a single rear wheel, like Lux Aeterna before it, but while experimenting with the car's 3D-computer model, it was discovered that inverting the wheel configuration showed a significant reduction in drag. The outer shape of the body was able to accommodate either frame configuration and single front wheel steering would also eliminate the heavy linkages and alignment issues of dual front steering. This weight reduction fit in well with the John Lee's overall design philosophy and it was felt that the final car would be sufficiently light that braking could be safely handled by the rear wheels. With aerodynamic styling already modeled after Aurora's WSC winning car, adopting their tricycle design proceeded naturally.
As suspension and frame design took shape, sample components were ordered from various bicycle manufacturers with the intent of performing verification of the computer models with real world mockups. Suspension members were constructed out of wood reinforced with fiberglass and the composite panel frame[60] was modeled with sheets of plywood. A local bike shop provided mechanical components for the model, it was outfitted with suspension, tires and braking. Eventually it was determined that the wood mockup would not sufficiently mimic true composite construction for strength testing, but the full scale model still allowed verification of dimensions and component spacing and the completed mockup was quite promising. With the suspension geometry verified by the mockup, CAD drawings were finalized and component sponsorships were pursued. Quotes for machining the suspension components were obtained, but ultimately it was decided that the team would machine the suspension components itself by having several members enroll in a fall semester machine shop class taught by David Johnson[61] in the department of Chemistry Department instrumentation shop. This would save the team several thousand dollars and team members would gain valuable machining experience that would be highly useful in optimizing and repairing the suspension components.
In the John Lee's original design conception, suspension, braking, wheels, and rims were all to be high-performance bicycle components. Air-damped bicycle shocks were retained for both front and rear wheels, as were mechanically operated bicycle disk brakes, but to achieve minimum rolling resistance, Team Lux returned to the standard solar racing tires that had been used on all previous cars. Team Lux already had numerous NGM aluminum rims as well as a stock of unused Bridgestone Ecopia tires, so these were used for the rear wheels. The NGM hub motor would be mounted on the front wheel and the NGM rim/Ecopia tire combination was the typical wheel choice for use with the motor. However, this configuration left little room for adding a mechanical brake to the front wheel and, early in the design process, substituting a CSIRO motor for the NGM was still being considered. Ideally a Michelin solar radial, with its larger diameter and lower rolling resistance that the Ecopia, would be used, but these tires were difficult to obtain and the only suitably lightweight rims for the Michelin were extremely expensive.
Two team members undertook a semester long independent design project with the Mechanical Engineering department to create a custom carbon fiber rim for the Michelin tire. This rim was fully dished, allowing the motor to mount flush with one side of the rim and hiding the main body almost completely inside the rim. This was an improvement over the NGM rim, where the motor mounts past the centerline and the motor protrudes 2" (5 cm) from the rim, creating an assembly nearly 4" (10 cm) wide. Minimizing the front wheel width allowed reducing the front fairing width from 13" (33 cm) to 9" (23 cm), which would be a significant reduction in aerodynamic drag. Production and use of the rim was made possible through a partnership with Stanford University's solar car team who would provide Team Lux with carbon fiber prepreg raw material and Michelin tires in exchange for a number of completed rims for their own use. Machining the complex tire profile into a mold for the rim and allowing for mold disassembly to release the rim was an engineering challenge, however with the assistance of Sikorsky's computer controlled machining equipment, the team were able to create a functional mold. A test rim was laid-up and destructively tested which showed ample strength against static loads, but the testing did not take into account the long-term effects of road vibration and repeated loading. Manufacturing of the rims was also very labor-intensive, with each rim requiring hundreds of different pieces cut from carbon fiber cloth and then worked smoothly into the mold, which in some places required up to sixteen layers of material. While the rims were usable, the project was marred by lack of long term testing data as well as confusion over which rim was which as the manufacturing process was refined. This led to ASC 2003 dynamic qualifiers being run on a rim that had missed an autoclave bake cycle and the performance of that rim was of concern to the ASC scrutineers.
By mid-June all of the remaining steering and suspension components were complete and the first carbon fiber rim had been successfully laid-up, had mounting and valve stem holes machined in it, and a tire mounted on it. In preparation for assembling the frame, all of the necessary nuts and bolts had been ordered and bushings had been machined for the suspension mounting bolts. With all of this complete, it was time to put everything together. During one long evening at Sikorsky, the frame had the suspension mounted on it for the first time and was rolled around. For the most part this process went smoothly, with the largest problem actually being presented by the nylon bushings needed to reduce the half-inch holes in the frame's mounting flanges to accommodate the 10-mm bolts needed for the shocks. These had been machined by Team Lux on a lathe, but once they were installed, it was found that they were too small for the 10-mm bolts to fit through them. Initially it was suspected that the drill bit that had been used to bore out the bushings had been miss-filed and was not actually 10mm. One team member returned to New Haven to make new bushings while the rest of the assembly continued. Unfortunately, this led to a fair bit of downtime waiting for the bushings to arrive and once they finally did arrive, only a couple of them were useful. The final diagnosis was that while being drilled the nylon heated and expanded, causing the hole to be undersized once the nylon cooled. Eventually aluminum bushings were made that would not be susceptible to this issue.
Once the suspension was assembled, a new problem was discovered: wobble. The front of the car was disconcertingly unstable. Initially the rod ends for the front a-arms were replaced with stiffer units, but there was still significant play in the assembled suspension. A great deal of time was spent simply bouncing the front of the car and observing the wobble. Several stabilization techniques were tried still including small auxiliary shocks, but ultimately the wobble was not a major factor once the car started driving.
A more significant problem was braking. Initially the car was outfitted with three Avid mechanical bicycle disk brakes. Braking controls were bicycle handgrips mounted on the end of the push-pull steering rods. One control was connected to a linear potentiometer for regen and the other connected to a splitter. Three brake cables went from that splitter to each of the mechanical brakes. As mechanical brakes, braking force was highly dependent upon the strength of the driver's grip. Braking was a significant problem in qualifying for ASC03. During the brake test, the brake mount for the front brake bent.
After failing to qualify for ASC 2003, the mechanical brakes were replaced with Hope C2 hydraulic bicycle brakes. This significantly improved braking performance, but further mechanical problems appeared during road testing. The titanium suspension had considerable undesirable flex and there were concerns about the long-term durability of the carbon fiber rims. Ultimately nearly all of the mechanical components, apart from the frame itself, were redesigned and replaced.
The rear trailing arms were replaced with more conventional aluminum ones and the front suspension was completely reconfigured. An aircraft style telescoping strut was used instead of a-arms and torque control arms were added. Wheel configuration was also changed, with all wheels now using the same NGM/Bridgestone combination and the motor moved to the left rear wheel. Moving the motor considerably simplified the front suspension and allowed for better mounting of the front brake disk. For maximum braking performance, Team Lux designed its own hydraulic braking system with large, but lightweight, dual-piston calibers. Further modifications, such as three-wheel steering and regenerative shock absorbers were also considered, but were determined to be impractical.
While driving during Phaethon, drivers found they had a hard time controlling the John Lee at highway speeds. With one year until the 2005 North American Solar Challenge, the team decided to give up on the aerodynamic gains of only having a single front wheel and to instead rebuild the car as a more traditional four-wheeled vehicle. In conjunction with a new frame, an entirely new suspension would be required. Each of the four wheels would carry 25% of the car's weight and a single suspension design, double wishbone a-arms, would be used for all of the wheels. All of the suspension components, upper and lower a-arms and the upright, were machined by Team Lux from solid billets of 6061T6 aluminum. The only exceptions to this were the mounting tabs on the lower a-arms, which were TIG welded on, as this allowed a considerably smaller aluminum billet to be used. The suspension was completed with 4130 steel rod ends from Alinabal and Rock Shox pro series coil-over shocks. To achieve greater suspension stiffness, the standard springs on the shocks were replaced with custom wound springs from Risse Racing.
The hub, wheel, and brake configuration for the John Lee 2 was very similar to that of the John Lee 1.5. The same hubs and axles were used, but the John Lee 1.5's aluminum brake disks were upgraded to steel. This allowed the use of higher friction brake pads and significantly improved braking performance. In order to comply with NASC 2005 regulations on redundant braking, each front wheel had two of Team Lux's custom built calipers connected to a commercially manufactured dual-bore master cylinder.
Implementation of the new suspension design was slower than planned. This was largely due to difficulties with fundraising, which prevented the purchasing of needed raw materials. Because of this delay, the suspension was not complete until shortly before NASC 2005. There was little time to test the new suspension and the welded aluminum components were not heat treated. It was believed that the suspension was still sufficiently strong, but while driving the car during scrutineering, one of the front a-arms broke. Without a spare and with no way to replace the a-arm, this failure prevented Team Lux from racing.
Lux Aeterna's shape has affectionately been described as a 'cheese wedge.' With a design emphasizing simplicity, most of the body was constructed from flat prefabricated panels. These panels were manufactured by Hexcel and had either a ¼" (6.35 mm) or a ½" (12.7 mm) Nomex honeycomb core between skins of carbon fiber. After being cut to size, the panels themselves were bonded together with prepreg carbon tape and a layer of fiberglass. Longitudinal and transverse ribs made from ½" panels were also attached in this manner. The curved nose was a custom wet lay-up part, made from fiberglass over a foam plug. With the exception of the nose, the entire top of the car was a removable Kevlar-Nomex panel. This flat panel contained the array and had an opening in the middle to allow for the driver and roll cage. The bubble canopy to enclose the driver sat on top of the array and was heat formed out of Celcast acrylic and coated with a UV blocking agent by ThermoTec, Grass Valley, CA. While the body made few accommodations to aerodynamics, it was easy to construct and the open design greatly facilitated work on the car.
For Lux Perpetua, Team Lux wanted to pursue a more aerodynamically optimized body design. Freshman Jonathan Burt took on the leadership of the body group and did extensive research into aerodynamics and the bodies of previous solar cars. To evaluate construction techniques and a potential body design, a prototype body was started at the beginning of the spring 1998 semester.
Donations of foam to construct a plug were sought, and 3" thick sheets of 4' by 8' (1.2 m by 2.4 m by 76 mm) pink foam were obtained from the Connecticut Insulation Distribution Corporation, Oxford, CT. Using Tony Massini's machine shop in Sterling Chemistry Laboratory, these sheets of foam were then cut with a band saw into crosswise sections of the car. Critical points on the outline of each cross-section were determined mathematically and then laid out by hand. With each cross-section cut to approximate size, sections were then joined with Great Stuff™ foam sealant and sanded smooth. While this was a relatively easy method of constructing a plug, the hand done measurements were not particularly accurate and converting between the decimal inches used in the formulas that determined the car's shape and the fractional inches used by the tape measures and rulers used to draw the outlines was a continual annoyance. There was also little way to ensure that the sanding was done evenly or that the final plug complied with the intended design.
Nevertheless, work proceeded on this plug, and contact was made with H&L Plastics, North Haven, CT, which would be an important sponsor for many years to come. The plug was taken to H&L to be sprayed with fiberglass, but it was determined that the unprotected foam surface would be damaged in the spraying process. The plug was therefore returned to Yale where it was covered in epoxy to protect it, before taking it back to H&L and finally being covered in fiberglass. This back-and-forth process took considerable longer than anticipated, which meant that no complete body was ready for the planned unveiling at the end of the school year. To represent the progress that had been made, the fiberglass-coated plug was spray-painted blue with a silver 'windshield.'
It is unclear whether the original intent was to use the fiberglass covered foam plug to create a mold, or whether it was to cut the plug open and remove the foam, transforming it into a mold. The former process would have increased the final car size by the thickness of the fiberglass coating the plug and the unknown, and possibly variable thickness, of that fiberglass would further contribute to inaccuracies in the car's shape. The latter technique would have been quite difficult, both to evenly cut the plug open and to achieve a smooth finish on the inner surface of the fiberglass once the foam was removed. In either case, nothing further was done with this plug and it sat next to Team Lux's fleet of vehicles on top of the Pierson-Sage parking garage for the next several years.
Despite the final product being unusable, the effort spent on this prototype body was not wasted. The contacts made at H&L would be used for the construction of all of Team Lux's cars, the need to pursue a different plug construction technique was discovered in plenty of time to apply it to the construction of the final car, and the actual process of constructing the plug interested and involved members during what would have otherwise been a semester spent solely on design. Further refinement of the body shape continued over the summer, starting with a 3-D model being drawn in AutoCAD and then used for extensive analysis. This was done through EDS, who, as they had for previous Sunrayces, donated computational fluid dynamics (CFD) analysis to all Sunrayce participants. 3-D CAD files of body shapes were transmitted electronically to EDS, where they were processed with Analytical Methods, Inc. VSAERO™ nonlinear aerodynamic analysis software, and the resulting images were then available to the team on an EDS web page. Through several iterations of this analysis, different shapes as well as different wind speeds and body configurations were evaluated and a final body design was chosen.
As it was clear that the mold construction process needed revision, Team Lux set out to construct a 1:3 scale body as a new prototype. A plug was constructed out of foam sections, much as before, but this time the plug was only of the top half of the car and was secured to a flat base. The surface of the plug was sanded smooth and patched with drywall mud. Once this process was complete, the plug was then waxed and transported in the team's trailer to H&L, where it was sprayed with fiberglass; this fiberglass, once released from the plug became the mold - an inverse image of the body shape.
While the design was being finalized, the team was also looking into companies that had CNC milling machines of sufficient size to create a plug of the entire car. There were only a limited number of these machines in the country and several companies were approached about becoming a sponsor. William Kreysler & Associates in American Canyon, CA, was willing to donate machine time and also arranged for matching donation of sufficient foam for the plugs from one of their suppliers. With this arranged, a great deal of time was spent learning to use the CNC Software MasterCAM® software that controlled their CNC machine, and then created a tool path file to guide the CNC machine in cutting the correct body shape. Once this file was sent to Kreysler, the body group simply needed to wait for a dimensionally perfect plug to be delivered. It was the delivery itself however, that turned out to be the problem. As the crates for the top and bottom plugs each measured approximately 6 by 2.5 by 1 m and weighed more than 1000 kg, shipping from California to Connecticut was quoted at several thousand dollars. Given this high cost, Team Lux initially tried to have a member tow them cross-country on a flatbed trailer behind his jeep. Once the crates were loaded on the trailer however, it was determined that this would not be a safe option. A shipping company was quickly located, and during the second week of January, the plugs arrived by truck freight.
The plugs themselves were exactly as specified; unfortunately, the tool path file sent to Kreysler included only the exact shape of the car, so there was no data on how the foam was to be cut surrounding the car. Therefore, foam was removed only where it interfered with the cutting process and the smooth shape of the car was surrounded by an irregular area of foam. In order to create a usable mold however, the plugs needed to be surrounded by a smooth lip several inches wide. As this was the first time Team Lux had used a CNC cut plug, the oversight was understandable, but resulted in a considerable delay in body construction. The foam surrounding the plug needed to be removed down to at least the level of the edge of the plug, which was accomplished through a labor-intensive and time-consuming combination of cutting, prying, grinding, and sanding. To then create a smooth lip, a 6" (15 cm) wide band of 1/4" (6.35 mm) thick hardboard was secured around the plug and then surfaced with Lexan® (polycarbonate) to ensure mold release. After the plug's surface was patched and sanded, it was transported to H&L where, as had been done for 1:3 scale plug, the top mold was made from it. While the top plug was at H&L, the lip creation was repeated for the bottom plug. Once the bottom plug was ready, it was also brought to H&L and the top mold was returned to Mason Garage, ready for body construction.
Simultaneously with the ongoing task of completing the molds, the body group was developing their composite lay-up skills. Richard Askew, who in addition to being a skilled welder was also familiar with composites from his job in ship construction, gave a tutorial in wet lay-up and then, using the 1:3 scale mold that had been built the previous summer, supervised the team in the lay-up of a 1:3 scale top body. Additionally, sample flat panels were made in order to evaluate the optimal number of layers of Kevlar and fiberglass for each side of the Nomex core. Richard's considerable practical experience saved Team Lux a great deal of time and effort. For example, Team Lux was planning to support the molds by constructing plywood frames matching their curvature until he provided large garbage bags filled with shredded paper. These conformed to the mold and eliminated the need for any more complicated structure.
Once the top mold was created and returned to Mason Garage, it received several coats of wax in preparation for lay-up. All of the materials - Kevlar and fiberglass cloth, Nomex core, breather cloth, and vacuum bagging - were cut to the appropriate shapes and sizes and the top of the body was laid-up in a single session lasting several hours. After a day under vacuum, the body was released, but it was unusable; the surface was quite rough, most likely from insufficient waxing of the mold. New materials were prepared and the mold received additional waxing before lay-up was tried again. This time lay-up was a partial success. The body did release well, but the inner layers of Kevlar and fiberglass did not securely laminate to the core. Generally, the outer layer of the body (the side in contact with the mold) and the core are laid-up and left under vacuum to cure and then the inner layers are laid-up in a separate session. In an effort to recover some of the time lost on the first lay-up, both layers were applied at one time, but by the time the inner layers were applied, the resin had already started to set and a proper bond was not formed. The poorly laminated inner layers were removed and new ones applied during a third lay-up session. At the end of spring break, this final lay-up was removed from the mold and the top of Lux Perpetua's outer shell was complete.
After the body was released from the mold, ribs were needed to stiffen it. Full-scale cardboard templates were cut and fit into place and then used to cut the ribs from prefabricated sheets of composite panel. These ribs were then attached to the inside of the body using epoxy with added filler. The top of the half body was then moved to the Morse Teaching Center for the electrical group to apply solar cells to it and the bottom mold was retrieved from H&L and moved into Mason Garage in place of the top mold.
Although lay-up of the bottom half of the body would go far more smoothly than that of the top, there were initial problems with the bottom mold. The mold would not release, as the sprayed fiberglass had stuck to the plug, apparently due to insufficient waxing of the plug. To salvage the mold, the plug had to be carefully cut out from the mold's interior. After the majority of the foam plug was removed, the remainder was scraped and sanded off. The surface of the mold was then patched with filler where necessary, wet sanded smooth, and covered with ten coats of wax in preparation for lay-up. Lay-up for the bottom was then completed without incident. Ribs were also cut and fit for the bottom body and the basic structure of the body was then complete and ready to be integrated with the frame.
After carefully measurement, openings for the wheels were cut into the bottom of the body and holes were drilled to bolt the frame to the ribs. After cutting an opening for the windshield, which was a piece of Lexan secured by Velcro, remaining body work concentrated on finish work, particularly smoothing the nose. When the attachment of the two body halves was deemed inadequate by the scrutineers at the May qualifiers, an improved system was installed, using side release buckles inside the front of the car and small external bolts at the tail. The seam running the length of the car was sealed with 'removable duct tape,' with a new tape being applied whenever the two halves were separated. As the race approach, final details were completed, including tinting the windshield and treating it with Rain-X®, using plastic sheeting on the interior of the car to seal the wheel openings, attaching small half fairings around the front wheels, and then painting the car white and applying team and sponsor decals.
During the race, the only problems that developed with the body were water related. Like most solar cars, Lux Perpetua was not watertight, and there was heavy rain for the first day of Sunrayce 99. Water collected in the bottom of the car, and while waiting in the starting line, drain holes were drilled in the car's underside. Rain-X works best at high speeds, as wind clears the rain from the windshield. With the steady rain during the race, there was little sunlight, and speeds were low. To supplement the Rain-X, Team Lux's mechanical director, Edward West (ES '01), improvised a windshield wiper out of the elastic waistband from a pair of underwear.
Overall, the body process had gone quite well for Lux Perpetua, but for Lux Millennia, the team wanted to avoid another lengthy search for a CNC sponsor as well as the machining and shipping costs associated with CNC cut plugs. Because of this, the body group, still under the leadership of Jon Burt, decided to produce the plugs in house. A scheme was devised where critical sections of the car were cut out of plywood and attached to a base platform. These sections were then covered with thin sheets of tempered hardboard to create the surface of the plug. In theory, this should have been quick and efficient but ended up taking more than six months and involved countless repairs, revisions and corrections. A fundamental problem was that rigid hardboard sheets were unable to conform to the desired body shape the way composites, which have some stretch before they are cured, could. Considerable time and effort was spent on the plugs before this truth was finally accepted and the body redesigned into planes that could be built with rigid sheets. The plywood and hardboard were also subject to swelling due to shifts in humidity and were difficult to securely fasten to each other given the small contact area. Additionally, the hardboard was insufficiently smooth, so the plugs were also covered with Lexan, which created additional difficulty in attachment and smooth seams.
Lay-up of the Lux Millennia body was done in the new WNSL workspace.[62] Although the lay-up process itself went fine, no test lay-up had been performed with the specific combination of resin and the polyurethane coating used on the surface of the mold. The body did not release well from the mold and thus the body was unusable and the mold suffered significant damage.
After classes were over for the year, the condition of the top mold was reevaluated. While initially it was planned to return to the plug and create a new mold, it was decided the mold was repairable. A section of the mold was heavily sanded to remove the residue from the previous lay-up, filler was applied as necessary, the surface was again sanded smooth, and then sealed with several light coats of resin. After five coats of wax, a test panel released well from that section of the mold and the team went ahead with preparing the remainder of the mold for a second lay-up. While the time consuming sanding process went on, a more refined mold preparation process was developed involving a spray-on gel coat to smooth the mold and a final coat of PolyVinyl Alcohol (PVA) as a release agent.
A further impediment to body construction was discovered during the spraying process: the ventilation exhaust from Team Lux's work space had been located too close to the fresh air intake for the WNSL control room, causing fumes from Team Lux activity to contaminate the control room. While these ducts would eventually be rerouted, this presented a short-term logistical challenge. Spraying late at night, when WNSL occupation would be minimal was attempted, but it was further discovered that that the ventilation fan that had been installed was suitable for evacuating only dust contaminants, and not flammable gasses. This ruled out nearly all coatings and paints, making the ventilation system of little use to Team Lux. As a result of these discoveries, all spraying was moved outdoors, which added further delays in having to move the mold each work session and adding daylight and weather as constraints on working. Additionally, greater care was required to prevent surface contamination outside. After all of these issues had been worked through, lay-up was able to proceed and a Kevlar top of the car was created without incident.
The body group was successful in properly preparing the bottom mold to avoid the trouble presented by the top, laying up the bottom of the car, attaching the body to the car, installing supporting ribs and a window in the top of the car, and sanding, priming, and painting the entire car in time for its unveiling. However, significant body touch up work was still required to repair some problems and improve aerodynamics. A miscalculation resulted in the body not fitting around the frame as planned. The top point of the frame prevented the top body half from fully lowering onto the bottom, so a foam filler for the seam - the entire perimeter of the car - was fabricated and installed. Lux Millennia also received body attachment upgrades to attach the body more stably to the frame and to secure the top of the bod