2024-05-07 EE Architecture
- Daniel Puratich
ToC
During Meeting Notes
Discuss features we want for the next rev of hardware.
was intended to be zp4 focussed, got out of hand quickly
Autonomy, EFS, & EE Leads present
chose two board
daniel gave the same pitch he’s been giving before, i’ll write this later
integrate compute and flight controller (rpi and pixhawk)
having like elrs transciever as a castellated board
phone connection and usb to it
Why RPi 5
130 cad
rpi vs jetson
scrws & standoffs in the stack
phase 1
using supported proccessor, are they available commercially
rpi5 hat pixhawk
includes all pixhawk functionality
should be careful on imu and barometer mount
everything is in the little cube
stack would be
rpi5
lte hat
zp
phase 2
~50W 5V power supply
everything else that’s an objective toard final goal
Final Goal
everything on drone comes off except for
ESCs
rc link (ELRS)
telemetry link (right now lte via phone, will be added to the stack)
video system (vtx + camera)
GPS
Main power current sensor (bc it’s a part of the hv system)
Rpi + LTE hat + ZP all connected in one big stack
power input from batt, couple sensors come out
separate LV and HV systems
Super Far Goal
final goal + combined RF link for rc link telem and video
Insane Goals
custom rpi
rf heartbeat
todo
what do we not like
what are we trying to optimize
ok so here we go
no zp4, its not needed
no need for the extra auto stuff
focus on tangible objectives
monster mount CAN device ← Michael Botros
making ESCs support CAN ← Meghan Dang
possible make the GPS (which are uart) a CAN thing so we can space them a ton
Video system unified board reduce harnessing, CAN to pixhawk
combinging power supplies: already in BEC project
servo module CAN support
having the stack will reduce system complexity and assembly complexity
a problem we have is assembling small connectors, but let’s come up with a way to reduce those
zp is a solution kind of, but there are better solutions to worse problems that are simpler
new pixhawk is a fun and interesting project, not very high value, be upfront about it
zp4 solves integrating everything into a single stack with would simplify failure points and making it a single easy to work with assembly.
Going all in on CAN is a worthwhile endeavor because
makes electrical harnessing easier in exchange for firmware complexity
we want EFS team to have comp related projects
it is doable and made up of lots of smaller projects
it is easily reversible if we don’t like it architecturally
Summary
We should only pursue projects that are very tangible and clearly optimize something. The following are in order of priority and should be pursued by EE for these reasons:
todo
design board with sensors and a CAN node
use case
future vehicles
why
CAN node for minimal harnessing connection of a lot of sensors to the FC
we commonly use these sensors
who
michael botros
todo
design
use case
future vehicles
why
CAN node for minimal harnessing connection to the FC
we need to space out ESCs on larger aircrafts
who
Meghan Dang
todo
make the board smaller
use case
for competition
why
improved rf link quality
doesn't exist commercially
todo
assemble and use
use case
for realism points for competition and cool ig posts
pegasus
future vehicles
who
Parker
todo
assemble flight interface board
fix oscillator on primary
use case
for EFS M2 and beyond ZPSW objectives
never be competition viable
who
todo
design, test, etc.
use case
all future vehicles
why
for reducing the number of power supplies we have on the drone right now
simplify all the power harnessing we have
who
Ishman
why
for reducing harness count on the antenna head
todo
combine nathan’s and Michael’s board, best of both worlds
todo
assemble and test
add CAN support:
use case
future plane or vtal
why
for spacing out servos
minimizing wire complexity on upcoming airframes
more stable power to far away servos
todo
design
test
make smaller
use case
all future vehicles
why
improved rf link over gemini
who
Nolan Haines
todo
make GPS’s work with CAN
use case
all future vehicles
why
allows us to space them more without SI concerns
busability
todo
make video system sit on one panel
use case
all future vehicles
why
improves packaging
reduces harnesses
reduces emi problems
use case
for ground validation
unlikely to be competition viable
why
fun
use case
for ground validation
unlikely to be competition viable
why
fun
maybe viable for competition
needs to offer some improvement on COTS to be viable
validate and test our copy of the COTs for now
plan for the future here, we need to think of exactly what functionality we want and what’s achievable.
todo
design
use case
future competition vehicles
why
for some integration improvement in the future
for fun
never be competition viable
for ground validation
for future vehicles i guess but we dont have any vehicles of this size planned
never be competition viable
for fun
never be competition viable
for fun
lots of integration overhead
who: would need to have someone ambitious, shouldnt be pursued for now
Insane ideas that are years out but possible in Daniel’s opinion (and will stay on Daniel’s list of personal projects and if I get anywhere in the next few years I will merge back into WARG projects):
RF Unified Link
HAM band (70cm or 2m)
RF Heartbeat initially, then more comes later
CAN interface to FC
variable airside power, very high power ground side
not started but lots of COTS hardware could be leveraged
More efficient conversion from 48->12
I have started this as a personal project, finishing is a whole nother challenge
fully custom including magnetic