Ishan

Why so many can bus
"after some recent talking" - Lakshay Gupta
why not just stop isolating

Tianda

• DC/DC converter has poor efficiency at our load

• Adding more contactors will be expensive & power hungry

• Testing on LV supp rather than on DC/DC converter output will still work, rather than devising a system to power our LV system off the battery without turning on motor

• EFuses: be wary that the regs guys don't really like EFuses

  • You can EFuse for light load cases (1A) and physically fuse for smth like 10A

  • Don't go entirely off regs because you will be shot down by officials at comp

  • Double up: EFuse + Normal Fuse

• Fuck isolation for most things

  • We have pretty bad noise from the isolated DC/DC converters

  • You can benefit from a common GND reference

  • It's really hard to analyze all of the places where isolation can be broken

  • a lot of the data acq boards are digital sensors, so as long as you have proper filtering you'll be fine

  • the isolators don't prevent noise

  • prompted a big discussion about where to put isolation and fusing, as well as the usecase for isolation

• Test all of your CAN chips operating ranges to see where it breaks. Ramp from 0 to 100V and record where the CAN chip breaks, and do the same thing for 0 to -100V

• Use loctite to lock the screws on for the board-to-board connector, otherwise if automotive rated you should be fine

• Look into DBCs

Manthan

• something about rotary switch & removal of LV on position

• Is 12V what you're providing to the CAN chip on PSOM?

  • no, it'll be stepped down to 5V for the CAN chip

  • talked about the 6 pin model

Chase

• long and skinny battery module length?

  • 575mm long

• Why go with PCB instead of wires?

  • reduction of # of harnesses on either side of the battery, strengthening the physical structure of the battery (less holes in middle panel)

• How many can buses (and where do they go)?

  • BPS, DataAcq, Controls, Array, CarCAN

  • DataAcq and Controls routed all around the car

  • We think it's better since we're daisy chaining the PSOMs

• Do bandwidth calculations for CAN for each bus

  • without enough bandwidth you may suffocate some lower priority messages

  • this is something i didn't think about. priority will introduce more weird shit

• Keep in mind, you need your CAN buses operating at the same voltage

  • each psom needs to output CAN H/L at the same logic level

• Reducing the number of unnecessary power connections between boards will help you stay sane, so perhaps using non-isolated CAN chips is in your interest

• Try to figure out your bandwidth goals and organization for CAN

• More numbers on the next design reviews

Sidharth

• Sit down and formalize your CAN spec architecture

• Make sure there's a way to detect faults on a peripheral network

• Make sure that the faults can propagate to the DataAcq system at runtime

• No mention of commercial MPPT backups; are y'all still doing that?

Philip

• Will multiple CAN buses make wire harnessing more difficult?

  • The problem wasn't manufacturing/the amount of the wires, it was we didn't know where wires are going

  • CAN forwarding won't be a problem software-wise either

• Manufacturing heat-sinks for battery will be tough, have we found someone to outsource?

  • Parthiv - we've found a few shops through Ops that are willing to manufacture for us