6S Power Module
Introduction
Who
@Andrew Chai Engineer
Background
What
30.5x30.5mm mounting pattern
5V @ 4 A buck buck capability
high current pass through with 2x XT60s
100 A pulsed current capability
60 A max continuous current capability
pre accurate current sensing
Use the custom CAN circuit designed for ESC CAN Adapter .
EFS would need to dev for it but nws.
Single JST GH four pin to CAN x port on pixhawk
Balance Lead Connection
Measuring each cell voltage would be nice as well to improve SoC calculation.
would need 2x headers if we run two 6S batts in parallel which seems to be POR as of June 2024 for Fixed Wing 2025
Stretch requirement: Active balancing (Can be done, some extra circuitry required and must be taken into consideration in regards to component sourcing and which IC to use)
could be cool if there’s board space for it, hmm extra weight and risk tho, maybe not, EE lead thoughts
more work for firmware to develop cell balancing algorithm
save for next rev probably
Input Voltage Range: 26V to 12V
Target Use
Would be for Fixed Wing 2025 .
Why
few COTS options with CAN
few COTS options with 30.5x30.5mm mounting pattern
Fun for EFS dev, opens doors for more complex SoC calculation
Electrical Architecture
Features:
Buck + LDO (Power Module)
Current Sense
Active Cell Balancing
Cell monitoring and measurement (SoC calculations)
MCU for custom cell balancing algorithm + cell input calculations + interfacing with IC
CAN circuitry
Elec System Design
Component Sourcing
Buck:
Filter for voltage regulators rated between 30V-40V max input voltage (larger than expected 26V), 4A output current, and 5V output voltage
3V to 36V input voltage range
up to 95% efficiency
4A continuous output current
adjustable switching frequency
adjustable output voltage (0.8V to 30V)
AOZ1284PI Buck Design:
Exposed Pad (Vin Exposed pad - pin number on symbol??? is this ok in general for a vin pin to be an exposed pad?)
What should I use for switching frequency? From what I’ve seen online, higher switching frequency means I can use smaller inductors and capacitors but increases losses.
Going to use 1M switching frequency for reduced board space. Seems to be a typo/mistake in the datasheet as the 5000 should actually be 55000 or 52500, so in theory RF should be around 47.5 to 50 kOhms. I will be using a 47k resistor because Yageo do not offer a 46.6k
Vout = 0.8 * (1 + R1/R2)
5V output requires standard resistor values of: R1 = 52.3k, R2 = 10k (recommended from datasheet) this yields around 4.984V
alternatively, R1 = 53.6k yields around 5.088V (Would it be better to design for slightly higher than 5V for some error compensation?)
For some error tolerance and to compensate for losses, 5.2Vout will be used so R1 = 54.9k and R2 = 10k to produce around 5.19Vout
standard resistor values source: https://www.rfcafe.com/references/electrical/resistor-values.htm
Component Calculations
Buck Converter Calculations ← Calculations can be found here
Input Capacitor
Ideally <1% of Vin for ripple voltage so 0.01* (20V or 26V) ripple range: 0.2V to 0.26V
Cin should be between 3.75uF and 2.39uF, 3.3uF should be fine
Will use 4.7uF to reduce input ripple voltage, this gives around 0.52%-0.82% input ripple voltage
Output Capacitor
From online, it seems around 1% of Vout ripple voltage is desirable (lower is better obv)
Since we’re using X5R and X7R capacitors I’m assuming ESR can be ignored???
Vout is 5V so 1% is 0.05V would be preferred
For ripple current 0.2,0.3,0.4 of Iout = 4A (0.8A, 1.2A, 1.6A → 2uF, 3uF, 4uF)
higher capacitance = less ripple voltage = better
https://rfcafe.com/references/electrical/capacitor-values.htm
Standard cap values within range: 2.2uF, 3.3uF
I will be choosing 3.3uF (at 0.3Iout ripple current, this is around 0.9% of output voltage (45mV)
4.7uF also works and this reduce the ripple voltage even more, this gives around 0.61% output ripple
Should be rated for >4A and >5V
Inductor Selection (Section 3 of the TI document)
Vin = 20 - 26V
Vout = 5V
Fs ~ 1MHz
Ripple Current estimation: IL = (0.2 to 0.4) * 4A = 0.8 to 1.6
Inductor should be rated to minimum 5A to handle the peak current in the above eqn
L should be between 2.34uH and 5uH
After EE meeting, estimation of 0.3Iout for ripple current, inductor should be around 3.125uH to around 3.3uH (same calculations as above)
standard inductor values around this range: 2.4, 2.7, 3.0, 3.3, 3.6, 3.9, 4.3, 4.7, 5.1
https://www.rfcafe.com/references/electrical/inductor-values.htm
Will be choosing 3.3uH inductor to satisfy the requirements
https://www.digikey.ca/en/products/detail/sumida-america-inc/0630CDMCCDS-4R7MC/5812458
https://www.digikey.ca/en/products/detail/sumida-america-inc/0630CDMCCDS-3R3MC/5812457
Loop Compensation Resistor and Capacitor
With 4.7uF cap output, and around 10kHz bandwidth (this is 1/100 of the switching frequency but the stack exchange Low cost buck SMPS with IC AOZ1284 - 5V x 4A - values for COMP pin answer verified with sims that around 10kHz is a more realistic achievement (Need to get this verified)
Rc is about 2050 ohms so 2k resistor should work
RL is the load resistor value, load resistance = Vout/Iout = 5/4 = 1.25Ohms
CC is around 2.9375nF so a 3nF cap (3.3nF will be used)
Others
Enable Pin is high between 1.2V and 5V so 3.3V from the AFE IC will be supplied to this pin (filter cap needed??)
Soft start Pin: “A soft start process begins when the input voltage rises to 3V and voltage on EN pin is HIGH. In soft start process, a 2.5µA internal current source charges the external capacitor at SS. As the SS capacitor is charged, the voltage at SS rises. The SS voltage clamps the reference voltage of the error amplifier, therefore output voltage rising time follows the SS pin voltage. With the slow ramping up output voltage, the inrush current can be prevented. Minimum external soft-start capacitor 850pF is required, and the corresponding soft-start time is about 200µs.”
will be using 10nF cap, was available on the WARG component library (relatively similar value)
Schottky Diode:
To reduce the losses due to the forward voltage drop and recovery of diode, Schottky diode is recommended to use. The maximum reverse voltage rating of the chosen Schottky diode should be greater than the maximum input voltage, and the current rating should be greater than the maximum load current.
Using the SK54A-LTP Diode which is already in the warg component library
rated for 40V @ 5A which should be enough for 26Vin and 4A current out
Cell Monitor IC
https://www.digikey.ca/en/products/filter/power-management-pmic/battery-management/713
https://www.digikey.ca/en/products/detail/analog-devices-inc/LTC6810IG-2-3ZZPBF/9658988
https://www.digikey.ca/en/products/detail/rochester-electronics-llc/AD7280WBSTZ-RL/12102176
3-6s cell monitoring
no ADC (connect to MCU and use that one)
Current Sense
https://www.ti.com/lit/an/slua707/slua707.pdf?ts=1721224314354
AFE Design:
Does D23 need to be implemented if we are using an XT90 Connector?
STM32
Using the STM32L431KCU6 (the same one from Meghan’s CAN adapter circuit)
Will be using the CAN adapter circuit @Meghan Dang, with STM32, 8Mhz oscillator, and connectors
On Pin definitions of the STM datasheet (starting at Page 59), the UFQFPN32 model is the one used for this design
ADC pins used (Voltage, temp readings from AFE)
Pin 8 [PA2] → ADC_REF2 (reference voltage for AFE2)
Pin 14 [PB0]→ ADC_REF1 (reference voltage for AFE1)
Pin 11 [PA5]→ VCOUT_1 (voltage data 1)
Pin 10 [PA4]→ VIOUT (current data)
Pin 15 [PB1]→ VCOUT_2 (voltage data 2)
Instead of hooking the SPI lines from both AFEs up together, 2 separate SPI ports are used on the STM32 (I2C1and I2C3)
Pin 19 [PA9] → SCL1 [I2C1_SCL]
Pin 20 [PA10] → SDA1 [I2C1_SDA]
Pin 13 [PA7]→ SCL2 [I2C3_SCL]
Pin 27 [PB4]→ SDA2 [I2C3_SDA]
Overcurrent GPIO pin connected to pin 9 [PA3] (regular I/O)
Everything else is the same as usual for the CAN adapter circuit except OSC_EN (connected to Pin 7, should be fine)