After we competed in ASC 2016, the electrical team had a quick retrospective. We discussed things that had gone well and things that had gone wrong, as well as possible ideas for the future.

Reflection

Battery

• Battery pack was a nightmare - extremely imbalanced, hard to maintain, and too dense. Use mechanical connectors
• Voltage sense was very sketchy and more of an afterthought. Switch to PCB-based lanes?
• Need to calibrate the thermistors - curve was very off
• Test setup was good - should integrate into actual design
• Current sensor was off by percentage - could be to do with crystal accuracy

Power

• Killswitch didn’t actually kill the car
• The white connectors used for Chaos were terrible. They kept falling off and they’re really hard to crimp. (apparently being sent a list of recommended connectors by adem? - TODO: figure out who has this)
  
• Horn worked okay, but it probably shouldn’t be on the power distribution board.
• The aux battery should not be powering every single LV system in the car - need more granularity over what gets powered when the DC-DC isn’t working
  
• HV Transducers on Chaos were super overkill - ended doing a time-based precharge cause I don’t know how RC circuits work ):
• Powerpath on chaos is cool but pretty black magic
• Integrate flyback diodes into chaos?
• Integrate MPPT enable into a board
  

General

• Boards had a tendency to lock up - possibly because there weren’t any nodes on the network?
• CAN messages weren’t processed fast enough - caused some to be dropped
• Inconsistency in parts and board layout caused a lot of issues for software - spent many hours debugging the same blocks (esp CAN)
• M12 connectors are nice, but they’re hard to find and very expensive. Considering using separate connectors for internal boards (i.e. in a box) and panel-mounts for external systems
• Need to standardize our connectors and parts
• Connectors were hard to glue to teklam
• Move the electronics so it’s easier to work on/remove from the car
• Fewer external connectors - annoying to unplug and plug back in
  
• Need people dedicated to each board, more people who understand how the entire system fits together
• Silkscreens on boards need to be more useful - Most are pretty useless, some are completely wrong
• Getting the programmer to work was pretty inconsistent - didn’t help that the silkscreen was wrong on some boards
• Telemetry should be a focus next time, not an afterthought. Need much more debug info, especially for BMS
  

Driver Controls

• Probably want a better display that doesn’t rotate when you turn the wheel (maybe separate display from steering wheel)
• Wheel mounted throttle and brakes is pretty terrible - makes turning weird (mac uses pedals from a camry - pretty interesting idea)
• CAN cable got caught on the rollcage and was also hard to get into the steering column
• Can we not use teklam and aluminium brackets for everything - super overkill

Ideas

• Controller board design
  • Source connectors and test them
  • Schematic capture and layout of STM32 MCUs
  • Discuss peripherals
• Plutus/Battery Pack redesign
  • Remove/internalize cell balancing
  • Source PMICs/AFEs
  • Put Plutus in battery pack
  • Add test points for strings
  • Add native support for voltage injection, etc.
• Themis redesign
  • Component/wiring reduction
  • NO MORE CRIMPING
• Chaos redesign
  • Make it smaller (look into replacing transducers)
  • Add better control of 12V rail
• CAN network redesign (tbd)