6S Servo Module Rev 1
- 1 Introduction
- 2 Engineering
- 2.1 System Block Diagram
- 2.2 Power Block Diagram
- 2.3 Needs Assessment
- 2.4 Requirements
- 2.5 Components
- 2.5.1 Buck IC
- 2.5.2 Current Sense
- 2.5.3 Current Limiting IC
- 2.5.4 CAN Circuit
- 2.5.5 Thermistor
- 2.5.6 Diodes
- 2.5.7 Connectors
- 2.5.8 Voltage Sense dividers
- 2.5.9 LDO
- 2.5.10 LED
- 2.6 Layout
- 2.6.1 Placement
- 2.7 Simulations
- 2.7.1 Input filtering sim
Introduction
Background
Target use for Fixed Wing 2025
Buck converter can be designed smaller than 12S Servo Module because of lower input voltage.
What
Same current limiting circuits as 12S Servo Module
4 Channel
Leverage the validation we’re already doing on this PCBA as much as possible
Common CAN circuit from ESC CAN Adapter
Buck Converter for 6S to 5V
Could be copied from Tracking Antenna Nucleo Shield Arch using AOZ2261NQI-12 though maybe higher switching frequency. Test this chip on the previous board if possible, cc @Jerry Tian.
Keep 30.5x30.5mm mounting pattern from Mounting Hole & Pattern Specifications
using 61x30.5mm mounting pattern for ease of routing and to keep all components on one side
Current Sense
just on the input current would be nice, get some metrics as to how much current the servos require during maneuvers and we already have a CAN line so why not!
Target all components on a single side of the PCBA for easier assembly & mounting
Why
12S Servo Module has some room for optimization for this airframe
no CAN support
buck converter could be made smaller and lighter since lower max input voltage
Who
@Meghan Dang to take this up July 2024 as lead engineer
@Daniel Puratich guidance and review for architecture & design
Link
Engineering
@Meghan Dang to fill in details
System Block Diagram
Power Block Diagram
Needs Assessment
Steps down 6S battery voltage to 5-6V to supply servos
Cut off power to servo if it fails
Convert CAN from pixhawk to PWM for servos
Requirements
Supports 18V~28V for 6S battery (rough calcs, 3V-4.6V max per cell)
3x4 header inputs & outputs
Input:
CAN, GND from PIxhawk (creating own 5V PWR)
Ouput:
PWM, PWR, GND
30.5mmx30.5mm board to comply with Mounting Hole & Pattern Specifications
Components
notes:
input cmc ← perhaps not needed, check 12s doc to read why they implemented
current sense
use INA180 unless determined otherwise
shunt resistor: calculate for it? to be looked up
buck converter
check if existing solves, check for board space
if not, make buck with lower input voltage & smaller inductor to maximize board space
LDO
does current one from can circuit solve? if not look for other one
purpose is to reduce noise from buck.. consult esc can doc for details
diode
what kind of diode… to be looked up
current limiting eFuse
current ones probably solve, however finding a singular IC with 4 channels instead could be better → maximizes board space
try and do some component selection here; will probably ask more questions
determine whether 4channel efuse vs 4 individuals will be better
things to consider:
cost
board space
power dissipation probs not super critical
if 4 channels exist for this purpose
thermistor
not sure how to select one, will see later, not super high priority
Buck IC
AOZ2261NQI-12
does this solve??
yep
inductor selection
switching freq
600-800kHz
inductance
anywhere from 4.6-5.5 ish
lower value, higher fsw based on EQ below
range also changes with input voltage, using typ of 22.2V
Current Sense
INA180A3
100V/V gain
shunt resistor
220mOhm
should do additional calcs?
ask abt this
RC filter (output) .;
1kHz cutoff
use formula from datasheet
RC filter (input)
f = 1kHz
Cf set to 1uF
size cap to be
Current Limiting IC
| ||
---|---|---|
Choice Summary | 4 channel, only need 1, fewer passives, 2.5A current limit | single channel, will need 4 individual ICs |
Operating Input Voltage Range | 8-60V | 4.4-18V |
went with NIV3071 for like a week bc no reading comprehension (buck delivers 6V, servos need ~6V, NIV3071 doesn’t operate with <8V ; chose diff IC
TPS4H160AQPWPRQ1 | DigiKey Electronics choice (for now)
Source
6V from buck
4-channels
Passives
15k pullup on fault pin
Cin
TBD
Cout
TBD
50k to gnd on RLIM
^for old IC
Overcurrent limit for eFuse
if possible, implementing adjustable current limit using microcontroller could be the play
this would require using a combination of EN and ILIM pins (perhaps also FAULT pin)
put notes from datasheet here:
Passives
R(CL)
= 0.8 x 2500 / 1A
V(CL) & K(CL) can be found in data sheet
R(cl) = 2k
CAN Circuit
lowkey look into how microcontrollers work
how do they get info from voltage input?
using ADC pins
connect voltage divider VOUT to ADC pins then efs interprets that data omg so awesome
i removed OSC_EN and
Thermistor
NTC Thermistor
bc they’re pretty standard and faster than alternatives
10k
because it’s pretty standard?
makes readings more accurate probably
used voltage divider to get 3.6 or 5 for mcu pin
0603
operating temp
-40 ~ 125 C
based on this graph, selected 10k and another 10k normal resistor for voltage divider, @ 25C outputs 3V which the MCU pin can read, at ~40C voltage divider outputs 4V, max that MCU can read
Diodes
Purpose
OR-ing bc two sources, don’t want to do current sharing
max reverse voltage VR
around 10V
selected diode array to save space
selected 20V and 2A max bc they didnt have anything ~1A or higher at 10V, 2A good safety margin
Connectors
XT30
for VBATT input voltage bc will probably use a harness anyways & takes up less board space
TSW-104-08-F-T-RA
for servo connectors bc 4 channels 1 component, makes life easy
taken from 12S Servo Module
SM04B-GHS-TB(LF)(SN)
for can line
Voltage Sense dividers
Voltage Divider Practicals | Wilderness Labs Developer Portal ← good resource, explained a lot of assumptions that were made throughout design process
general notes:
generally higher resistance resistors make for better power losses and minimal impact on voltage due to load resistance
kinda made assumptions before reading this and now everything makes sense
pins should be chosen based on what voltage divider ranges will be
because thermistor and vbatt voltage dividers have varying VIN/R1 values, they should be on FT pins (5V tolerant) because it gives more room
ranges from 3V-7.3V
added caps to all to make them low passes with ~1khz cutoff
LDO
putting two LDOs in series to drop 6V-5V, and 5V-3V3
for 6V-5V
for 5V-3V
reuse LDO from ESC CAN Adapter
LED
not entirely sure what goes into selecting addressable LEDs, are there any firmware impacts i should be considering?
https://www.digikey.ca/en/products/detail/harvatek-corporation/T3A33BRG-H9C0002X1U1930/15930508
seems fine?
VIN is 3.3-5.5 but abs max is 6.5V so can be run on 3V3 rail
it’s kinda chonky
going with the one from SSD for now, using logic modifier also from SSD
wanna know more abt how the logic thing works
Layout
Placement
following layout guidelines for the most part
choosing not to use separate PGND and AGND pours on buck
hopefully won’t be too noisy
will probably be fine
Simulations
Input filtering sim
resonance peaks at ~10dB which is chill, much better compared to https://uwarg-docs.atlassian.net/wiki/spaces/EL/pages/2539980265/RCA+Fried+Boards?search_id=2fb11de2-5048-4262-8172-71d5792f14ca which was ~60dB
won’t explode