Date
Oct 17, 2018 7:30 pm - 9:30 pm
Attendees
@Karl Ding
@Taiping Li
@Robin Pearce
Stanford Solar Car Team
Two of their alumni also attended—Logan Herrera (Arctan BMS HW, 2015), Max Praglin (Xenith driver display, tire pressure monitoring, and telemetry HW in 2011, Luminos BMS HW + Strategy + Systems Integration in 2013).
Goals
Obtain insight into how their car architecture works (considering the previous times we went their Electrical team was not present)
See if they have any innovative ideas for their electrical system
Gain tire data
Understand how WSC scrutineering process works
Network with the Stanford Solar Car team
Understand how they run their team
Learn more about aero
Discussion items
| | | |
|---|
15 min | System Architecture | Sarah Woodard (2019 Electrical Lead) | |
15 min | Lights | Ricardo Iglesias | Currently they use the Cree XLamp ML-E LED They wanted to look into using the Cree XLamp XM-L I assume they have a certifying engineer that validates their documentation demonstrating compliance with the photometric requirements of the UNECE or SAE/DOT regulations, since it doesn’t look like those LEDs are compliant… 2017 lights: resin that they sanded down until clear
Were considering doing PWM in order to brighten/dim the lights Wanted to have a PWM enable instead of running a line all the way (presumably what she meant was like a 555 timer) They scrapped this idea due to complexity, after creating an Action Item to test their lights to make sure it meets regulations
Lights scrutineering process at WSC seems to vary depending on the year For 2017, the scrutineers just eyeballed it and said it looked good One year they took the cars into a dark room, and used a light meter
|
15 min | Driver Controls | Sarah Woodard | Due to packaging constraints, they might need to put throttle on the steering wheel Hall Effect or Rotary Potentiometer Apparently there’s a TI part that is pretty sick that does contactless sense that someone called Harry really liked? They had concerns about reliability and packaging, and didn’t really swing either way since it was dependent on whether or not this needed to be done
They had concerns about whether this would be ergonomic for the driver (especially when turning) WSC is pretty straight tbh They seemed to very much be against “being the first” to do something, presumably after their experience with Xenith. After Logan assured them that this wasn’t by any means innovative, they seemed more open to pursuing it further.
Thoughts into how to make repairs and swapping wheels quick and seamless
|
10 min | WaveSculptor 22 Electrical Efficiency Modifications | Sarah Woodard
| Sasha Zbrozek (the guy that worked on Stanford’s MPPTs back when Sam was designing his, and now works at Waymo as well) had suggested some interesting modifications to the WaveSculptor 22s for better efficiency Get rid of CAN isolator Get rid of CAN linear regulator Short across the isolated supply and run the logic side directly from vehicle power. You'll save slightly more than half of the Tritium's total LV power Remove 2/3 of the parallels FETs in the Tritium to reduce switching losses. Your motors aren't big enough to need so much silicon.
Consensus at the meeting was that these would most likely reduce the reliability of the WS22 without too much gain in power. They did mention the WS22’s cover was a pain to work with since you had to open it to get access to the lugs. Consider adding connectors for that or remake a housing that allows easier access incase the motor controllers have to be swapped during the race. |
10 min | WaveSculptor 22 Information | Logan Herrera | Apparently the WaveSculptor (at least the 22) current sense is awful. He talked to the Tritium guys at WSC when he went in 2015, and they confirmed that yes, it was terrible, and no, they did not plan on fixing it The WaveSculptor 22 casing is not waterproof or dustproof, and could be a relatively easy project to redesign
|
10 min | Battery Box | Temidayo Dairo | They had an interesting proposal for a battery box that got shot down pretty quickly Due to their packaging constraints, they were considering what appeared to be two boxes Constrained by Aero team, and Aero design is not finalized until Winter quarter They designate individuals who are responsible for servicing the pack, and only those people touch it
|
10 min | Pack Voltage | Temidayo Dairo | Overall, this guy seemed very confused about things. In the interests of optimizing for wire weight and efficiency, they were considering whether to run their pack at double the voltage (260 V) This was shot down pretty quickly, as for one, the WaveSculptor 22 only goes up to 165 V They would also need to redesign their BMS, Motors, etc. To be honest, I felt that this was a micro-optimization that wasn’t really necessary, considering the gains you would get from Mechanical reductions are much more significant than anything they can do with their current Electrical system
He was also wondering why one of the teams at ASC ran significantly lower bus voltage They have a spreadsheet somewhere that gives the optimal wire gauge vs losses vs weight They estimated 3 kg of weight = 1 minute of race time, and 1 W of quiescent power = 1 minute at WSC for their car
|
15 min | BMS | Sarah Woodard | LTC6804-1 MCP3903 for current sense + shunt Sits on top of their battery Sounds a lot like they’re considering building a penthouse (like in a Tesla) since they had concerns about servicing the pack safely Not sure if that’s legal due to safe state requirements, if they take out the BMS…
Apparently had concerns about noise that they were seeing on the current sense? Not clear how they determined that this was an issue, since they were comparing against the WS 22 reported values, and never measured the actual current being drawn Designing to mitigate noise is a defense in layers. Obviously the higher up you “fix” the noise issue in the chain, the better. Identify the source of the noise Can you get rid of that noise? Can you reduce the effects of that noise on the rest of the system? Can you make other hardware more noise tolerant? Can you make your software work despite noise?
Alumni expressed concerns about whether their current team understands what happens when they turn the car on Currently their car turns on via a supercapacitor + always on 12V and 3.3 V buck converter. Due to WSC Challenger regulations, an “Aux battery” like we had for MS XII would be considered part of their Energy Storage System, and thus they would lose capacity from their 420 Li-Ion cells. We currently do not have this problem, thanks to the regulations. WSC regulations have suggestions about how to do this without a battery
WSC scrutineers will ask you “How does the car turn on?” in order to verify that your system meets their isolation and safe-state requirements. Be prepared to answer this!
|
15 min | Pack Welds | Temidayo Dairo | 420 cells (as per regulations) spot welded together Sundae’s (2017) pack had issues where they were seeing some modules with significantly lower voltage They didn’t want to “stress” the pack too much, so they limited the amount of testing that they were performing I feel like they could’ve built 2 packs and just used the other pack for the race if they had concerns about SOH.. It’s not like their pack is that big or they’re lacking money…
tbh they could’ve easily detected this when testing with their telemetry and remedied the issue by swapping a module (or fixing the weld) When they probed the welds and measured the resistance, the weld resistance on their bad modules were much higher than the other modules They were considering improving their cell balancing circuit tbh this is probably unnecessary, due to the clear manufacturing error that was present Logan + Max reiterated that since they’re just attending a single race (and some testing), a well manufactured pack should never need to be balanced
Wanted to plot a heatmap in order to analyze their pack thermals IMO this is probably the most useful suggestion they had (besides things that should be common sense), to validate thermal behaviour of your pack, as hotter cells will experience reduced SOC + SOH However, with that being said, this is usually as a result of a pretty significant degree rise in thermals, which seems unlikely to me, given how flat WSC is, as well as how much power their car should be consuming.
Balancing is rarely used during the race, but can be used before the race to make sure your pack is as well matched as possible before the race The alumnis strongly suggested testing modules after the pack is welded to identify weak modules. These can then be replaced prior to the race.
|
1 min | Shipping Logistics | Sarah Spaugh | |
1 min | Tire Data | Sarah Spaugh | |
??? | Aero | Jason Trinidad | While this Design Review was going on, Robin spent the entire time we were there talking to Jason about aero (and drinking Soylent) |
Overall, it was interesting seeing how their team quantified things in terms of how much efficiency they can squeeze out of their car. For Arctan, it was something like a reduction of 3 kg resulted in reducing their WSC race time a minute faster, and a reduction in power consumption of 1 W meant they could get to their destination a minute faster (not sure what the distance metric is). Irrespective of what the actual numbers were, I think having some sort of numbers like this really makes sense, as you can quantify whether or not it’s worth optimizing a particular system, vs focusing on making those optimizations elsewhere that gives you better bang/effort.
Their design review was very much an open discussion, where the people hosting the review would present their work, and then defer any questions they had (about things they hadn't solved yet or weren't sure about) to others (alumni, other leads, etc.). They had a deck of slides that contained any relevant schematics, data, etc to have an overview of what they want to talk about. This made the review process a very systems oriented approach, where everyone who would be responsible for the system would have their input in order to determine what could feasibly be accomplished. Their reviewers were quick to point out potential flaws in their designs, and any other things that perhaps might not have considered. These reviewers were also used to scope out the amount of work, and to determine whether they had concerns about reliability or other potential problems they might run into when selecting a particular approach. People were quick to admit when they did not know something, but also made sure to take note of it such that they could learn it.
Their team also seems to understand what they are strong at, and uses that to reasonably scope out what work they can accomplish in a cycle. Their current members don’t seem to have a very detailed grasp of what’s going on with their electrical system, but their alumnis attend these sessions to help them with the car. On the other hand, they also seem to be very risk-adverse and against introducing potential sources of additional complexity (which perhaps holds them down in Challenger class), and focus a lot on reliability. I think this is something that Rachel Abril’s TEDxStanford talk touches on, where she advises to “Go fast, but not recklessly fast” and warns the team to be wary of overzealous innovation (in reference to Xenith’s ambitious technical design). They’re also willing to not commit to features that might not necessarily be completed in time, and play it safe in favour of having more time to test.
Also, according to Jason, when they needed new motors, they told Sam to interview at Google (back when Waymo was still Chauffeur), and the conversation went something like, “Hey, you know how to build motors. We need new motors. The tech industry here is booming. Would you like to build our motors?” Sam ended up getting the job, and would come in after work and on the weekends, and that’s how they got their motors.
We also got invited to their beach team social at Half Moon Bay on the following Saturday.
Action items
Decisions