FSGP 2019 Retrospectives

Owned by Pei Liang Guo
Last updated: Jul 14, 2019

Broken Controller Boards and Programmers

  • Could not find a working controller board and a working programmer
  • Bring up backup CBs and flash programmers before the race
  • Identify the causes of broken CBs and programmers

BPS

  • The carrier and the AFEs should be fully tested before the race
  • We should have at least one backup carrier and one or two backup AFEs for quick replacement
  • Backup ICs are necessary

Flashing the Controller Boards

  • Experienced difficulty in flashing controller boards in hard-to-reach places (e.g. pedal board)
  • Investigate firmware update OTA via CAN

Motor Controller Interface

  • Unable to read motor temperature and other data from the motor CAN bus on system CAN
  • Investigate MCU with two CAN controllers
  • Redo MCP2515 driver and motor interface code 

Waterproofing

  • The thunderstorm was a concern in the beginning of the race
    • Enclosures should be waterproofed 
    • Boards can be conformal coated
      • Makes rework much more difficult 
  • The motor controller interface was damaged due to water

Telemetry

  • The LTE dongle provided a very reliable internet connection for telemetry purpose
  • Considered using long-range Wi-Fi for telemetry but the equipments are very expensive

Solar

  • Solar sense was not functional during the race due to firmware issues
  • Investigate behaviour of the MPPTs (i.e. pack voltage vs. charging current)
  • Did not have any documentation regarding the solar enclosure

Pedal Calibration

  • Needed to calibrate the brake pedal via a programmer multiple times
  • Investigate calibrating the brake via CAN

Components & Assembly Quality

  • We should try and design out as many non-leaded (QFN, DFN, etc) components as possible for ease of inspection and rework. Significant time was spent before and during FSGP to rework ICs such as the LTC6820 or the load switches on Power Distribution/Lights. 
  • Poor assembly (reflow/soldering) quality have plagued us since before ASC 2018. We need to either fix how we're doing things (play with stencil openings for example), train the team better to bring up higher quality boards, and functionally test every aspect of every board before putting it into the car. Most boards in the car are either reading inputs (driver controls, etc), or outputting power (lights, power distribution, etc), so perhaps we can set something up to test things? 

Organization

  • It was hard to keep track of where everything was
  • It took 20-30 mins to pack and unpack everything when we had to move to the trailer park. How can we do better at this? 

Raspberry Pi Soft Shutdown

  • The Raspberry Pi is connected to the main power supply at the moment, which can be shutdown at any moment without notice
  • Powering off the Pi while it is recording data to the SD card might corrupt the filesystem and the data 
  • Consider using a super capacitor as a uninterrupted power supply for a brief moment before properly shutting down the Pi