Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Date

Participants

Goals

Review the following boards/topics and have a discussion about design

  • DCDC

  • Power Select

  • MCI

  • Telemetry

  • Carrier

  • Steering (did not get to)

Discussion topics

Item

Presenter

Notes

MCI

James

  • change ultrafits to a different HV power connector

  • talk to Josh Rong about connector choosing

  • added another load swithc to turn off isolated DCDC

  • another thermistor comparator circtuit

  • precharge mosfet, discharge mosfet, discharge and charge reistors have temp control with GPIOs sent to controller boards

  • precharge is much quicker with decreased res values

    • might not need heat sinking with quick precharge and discharge

    • will not buy a new heat sink

  • add more mounting holes (other than just the four corners)

  • standoffs on controller board holes

  • temp sense connectors - combine into one 8 pin connector

DCDC

Aashmika

  • sim surfing tool on murata for ceramic cap capacitance curves at different voltages at different pacakges

    • look at for RLC filters

  • look at switching frequencies for motors/motor controllers, mppts, dcdc, etc. to find out the frequency you want as baseline for cutoff

  • will ICLs get hot enough now that they are in series

    • yes but double check and make note in schematic

    • how hot will ICLs be?

  • Make fault LED yellow or red and add power led as green

  • change power terminal to power dist. to whats agreed on with josh

  • Maybe dont have pullup resistor on optoisolator

    • should be fine though because not actual NPN

Power Select

Josh

  • Having trouble finding the actual load capacitor value based on equations

    • vout droop is set to 0.2V - should be able to handle droop of 10V

    • was unclear about Vdroop, but now is fine

  • old PD current sense resistor was rated for 7W,

  • place passives before opamp close and after close to op amp (dont have to worry then about the difference of impedances across nets)

  • maybe use op amp array

  • fuse was getting really hot because of the possible resistance of the fuse holder

    • maybe go with minifuse

    • current fuse is tin-nickle which may be the issue?

Other notes

Taiping

  • Make sure to follow battery regs and labeling and isolation and UV protection for additional battery packs for fans and horns

Telemetry

Pei

  • CAN-wifi device

  • edited confluence page

    • best option is ethernet to wifi router

    • other option is using LTE (we will have coverage for FSGP, and most of the route on ASC, so theory we can use the LTE as backup during ASC portion, and can use for FSGP)

    • LoRa has performance issues

  • Same solution as a combo of ASC 2018 and FSGP 2019

  • Verizon has good coverage over the rural site

  • ___ (something other teams use apparently) is lower power than LTE

  • can get a rf/wifi amplifier (found some on ali baba)

  • want a CAN-wifi solution, b

  • Need to keep in mind that we want to use minimum power for the telemetry solution

  • maybe on term get access to esp32? and test out

  • UART-LTE module (maybe easier for programming wise, but limited with speed (text message once in a while))

  • Apparently theres open source for using router to phone connection with data through ethernet to phone - teams use this solution

  • wanted to use FiPy but it might use a lot of power (espunning on python

    • nominal is 200mA, btu more with LTE running at 3.3V (similar to esp32)

    • about a half a watt → maybe 5 watt/hours a day

    • Are we willing to spend this on the car? car capacity is 11KWhrs

      • apparently an extra 20W on low voltage will drop 200KM in the distance the car can run (graph made by clarke)

    • easy to use ESP32 and micropython setup and working on it

      • has can compatibility all would need is SD card and CAN transceiver

      • may not get lower power consumption if using LTE or wifi

  • ubiquity consumes about .1A for 12V (M1 bullet)

    • can put bullet in the chase car, but will need a directional antenna with sufficient gain in the car

    • chase car windshield should be fine for transmission if there is no metallic tinting on the shield, otherwise

    • might need telemetry stuff in Lead car

    • bullet M2 has been tested in e5 parking lot and worked fine (antenna was on the roof in the solar car, and chase car at few car lengths was able to receive signal)

  • Wants to test with GPy and antenna and follow with this otherwise backup is using the bullet (mechanical may not want something to stick out)

  • to test needs usb-can dongle and transceiver on breadboard with GPy and antenna

    • realistically dont need CAN, maybe hookup to laptop for UART with USB32 and sense with another laptop with wifi

BMS Carrier

Liam

  • currently has pi connector (maybe not use)

  • all the signals have added input output expanders, so takes up less space

  • keyed connectors

  • USE the same connector for all 12V power connectors so that testing is much easier

  • put kill switch on a 3pin and have a ground in case you want to run a shielded wire

  • relay has economizer that advertises coil suppression (may not have it, double check)

    • adding the TVS help opens the contactor faster, because it decreases the voltage required to open, but if you don’t need this advantage, no other use for the TVS diodes

  • why aren’t using the interrupts for the MCP238 → should have it if using the surrounding inputs

  • if want to know how fast kill switch was hit on the car, then want to connect the kill switch directly , otheriwse if nothing depends on the time at which the killswitch was hit, then you can have the delay in time through the expander (not time critical)

  • usually the port expander is not good for transients (like duty cycles or PWM) - depends on what you are trying to measure

    • so the fan sense may not be optimized through the expander

    • might just see changing values in registers, and not see in time whats going on

    • have to see how quick things being sent

    • the point was to check if it’s running vs. stalled

      • not critical to know speed, but is helpful for airflow modelling and testing at different speeds (not critical when the car is assembled but is useful for modeling)

      • fans are being PWMd

      • 4wire fan 3V PWM

      • ADT7470 ← fans or similar that monitor speed

  • vias on pads required the via to be drilled then plated then copper will flow

    • sometimes fab does not support vias in pads

    • might be better to not do that

    • usually fill the via then cap the bottom and top then reflow

  • if want low esr caps, can get hybrid polymer caps to replace with aluminum caps ← would get higher inrush, but have a precharge anyways so

  • move the green mechanical layer bodies so they are seperate for each component

  • Pi is only connected to UART and power on this board

    • is that enough bandwidth to transmit what you need to (Arshan said it was fine)

    • CAN gets more overhead but UART logging quicker may be faster than using CAN then transmitting to center console

    • issues with flooding CAN bus

    • 1second delay for isoSPI overhead should still be fine

    • isospi takes about 1second per isospi IC

      • if the time is going to be around the same with using the extra pi and not

Connectors

Josh

  • current microfit has no mechanical relief

  • ended up deciding on gold plating with kinked pins and board lock footpring on microfits

    • the board lock has a different tab footprint tho then the current microfit

  • right now use non board lock

    • board lock has the metal pegs on board mount, and can they be soldered?

      • probably not what designed for (press fit retention kit)

      • but looks like the same material as pins so maybe can solder this

    • josh will make footprints for all new connectors then we will send out board that has all of them on it so we can test before implementing

      • get different sizes and pin numbers

Micah’s notes (we never decided who was going to take notes. Much of this is duplicate of above, but there may be some small details that either of us missed)

DC-DC
Murata Simsurfer - DC Bias of capacitors (150V = 0.4uF)
Larger package, typically smaller
Noise attenuation - switching frequencies from DC-DC, MPPT, motor controllers
To do a first order estimation for attenuation
Parallel NTC ICLs - how hot for normal operation
2 in parallel to reduce the power

Output connectors Automotive qualification recommended Fault LED color change Power Indicator LED Adjust the pullup/pulldown for the optoisolator

Power Selection
LTC4417 Calcs
Vdroop clarification
CL(esr) - is definitely given in the datasheet (0.36Ohm)
Current Sense Resistor
Typically rated for 2x power (in this case is 1.6W ish)
Chose a higher power value originally because we didn't want it to heat up and we had the board space to do so
Schematic is 'all good' according to Josh
Temp sense opamp
Voltage divider closer to the source or the Opamp?
Could put closer to the controller board
Impedance looking back towards the power supply will be larger from the opamp
Keep the signal as low impedance for as long as possible. This way there's less chance of noise coupling in to the system?
R37 etc, close to the opamp, R38 etc close to controller
Connector for the DC-DC to include fault + temp (4 pin required)
ATO vs ATR fuses
Contact resistance + solder joints + fuse cold resistance seems like a reasonable amount of resistance given the voltage drop that we saw
Make sure that the fuses are standard and easily available

Telemetry
It looks like the WiFi will be the best option for speed
Verizon claims pretty good coverage
ESP32 and external antenna
WiFi power booster
We don't want to add extra power draw on our system (Pis, routers, etc)
We want enough RF power
UART LTE Modules - speed might be a bit low
Telemetry to laptop inside car, then broadcast from there
Hook up your phone to the CAN bus
Tritium CAN-Ethernet to router to phone (see the data on the phone) - this was done previously
pycom.io - fipy - ESP332 running python with Lora and LTE
ALso have boards with just WiFi and LTE (GPy)
Ubiquiti Bullet M2 uses about 100-200mA max (as Taiping remembered) - max 7W.
Telemetry stuff should be in lead car potentially (since the back of the car is carbon fiber) and the antenna can be mounted to go through the windshield of MSXIV
Mesh network with chase and lead car?

BMS Carrier
SoC modelling on the Pi header currently - work on getting this onto micros
Sparkfun logging board
Standardize for a single 12V connector on most boards
Consider moving the kill switch to a 3 pin, add GND for shielded wire (if we want)
Current will not be an issue
Added TVS to cut the current off quicker, but may not be too useful since the economizer is already built in (and coil suppression)
Interrupt pin on MCP23008 - best used for inputs, make sure it is connected - INT will change state
Consider directly connecting critical signals to the controller board - how fast do we want to react to the killswitch?
What else is time-critical (opening solar and power relays)
Typically GPIO expanders are not good for transients
We would want it connected directly to the micro to use the input capture registers
What is the timescale that we want to measure?
Controllers for fans that monitor the speed - ADT7470 temperature sensor and fan controller
Vias in pads - professional places will fill the via and plate it
we don't want to have the solder flow through the via and not connect to the pins.
Also some vias were between pads
Low ESR capacitors - hybrid polymer capacitors - might be a good idea for transient loads
Some parts are too close together - keep the green boxes (courtyard layer) not overlapping.
Pi is connected to power and UART -
UART is typically 115200 (115.2kbps) - wouldn't CAN be faster (at 500kbps)
We are limited by the ISOSPI likely - 10s of milliseconds per cell measurement
Arshan had done some original calculations, but maybe we should re-evaluate.

Connector Standardization
Molex microfit, kinked pin, gold plated, board lock.
The board lock is supposed to be a press fit. If we make this a plated through hole then we will be able to solder it.
Josh will create a test board as he is creating the footprints for the microfit
We will send this out quickly to test if the board lock is actually useful - 2 pin, and many pin connectors, single and double row.