| Table of Contents | ||
|---|---|---|
|
Introduction
WARG-Altium365 Link: https://warg.365.altium.com/designs/74D65A2E-05AC-4B8C-AE55-D8D307F760C7#design
Who
Reserved for Kevin Li co-op student
Robert Tang doing the engineering
high level architecture defined by Daniel Puratich
What
Buck Converter
6S LiHv input
27V absolute max on LiHv
5V output voltage
rpp
Kind of nice because autonomy software people like to use this hardware
5 A output current
30.5x30.5 mount pattern
usb-c (for rpi) and jst (for pixhawk) output conns
USB-C connects to the raspberry pi easily, can copy circuit from 12V->5V @ 5A Buck Converter Board
Pixhawk standard connector for POWERx style port so this can be used as a redundant supply
To be clear, both of these output connectors should be present. Only one will be used in any given system though (so both should be sized on the PCB to handle the full 5 A of current).
xt30 input connector or just solderpads for input
both in library already (2.54mm pitch solderpads)
if solderpads are used, give space for epoxy so we can strain relieve
clear polarity label would be nice if solderpads are used
Why
Powering the Raspberry Pi 5 for https://uwarg-docs.atlassian.net/wiki/spaces/ARCHS22/pages/2556133415/Fixed+Wing+2025?search_id=772ded46-862b-41b6-8115-093586821d0d if we decide we want to mount the RPi into this system and dont want to use https://uwarg-docs.atlassian.net/wiki/spaces/EL/pages/2701197313/RPi+Interface+Rev+C?search_id=b1053a4c-4ec7-47de-9489-d50daf49510a&additional_analytics=queryHash---7d2c48e78a27d21f150b620681696259928cd14daa6e640c84eaf507c582ec26 . Currently there is no plan to do this but it is nice. Ground testing power supply for the raspberry pi is also a use case, but 12V->5V @ 5A Buck Converter Board kind of accomplishes this already.
Redundant Pixhawk power supply for https://uwarg-docs.atlassian.net/wiki/spaces/ARCHS22/pages/2556133415/Fixed+Wing+2025?search_id=772ded46-862b-41b6-8115-093586821d0d . See https://docs.px4.io/main/en/flight_controller/pixhawk6x.html for details.
...
As for how to design a buck converter, I referred to Kevin’s documentation on bucker converter IC Buck Converters
Synchronous Buck Converter
For maximum efficiency the buck converter we selected will be synchronous, with a little bit of research, I discovered that the main difference is that the diode is replaced/connected in parallel with another FET. The turning on and turning off of the switch is controlled by the on/off of the two FETs.
The benefit for that is the increased efficiency and reduced diode loss (R_dson*I_d<Vf*I)
Power losses
Inductors: has DC Resistance (DCR) loss, P_L=I_0^2*R_ESR
Diode: forward voltage drop (note that it is only conducting for (1-D)*T seconds) P_D=I_0*V_f*(1-D)
MOSFET: 1. Conduction loss, P_Cond=I_0^2*R_ds*D
2. Switching loss, P_sw=[V_ds*I_0*(t_on+t_off)]/2T_s
Component selection
Buck Converter IC
This is the most important part, once we have selected the IC we can select other components (inductors, capacitors, resistors) based off the specs of the IC.
Since we are converting from 24V to 5V, we need to consider a buck converter IC with: maximum input voltage larger than 24V; output voltage adjustable/fixed at 5V; rated output voltage should be larger than 5A; if possible the switching frequency should be a wide range; step-down, buck converter only (no buck-boost).
After applying all of the above constraints to DigiKey search query, my options came down to around 30 ICs, at first I was only considering TI’s IC, and the main IC are the LMR and TPS series. After I enlarge my options to other manufacturers, I found other good chips. Applying more close constraints and looking at the Datasheets more carefully, it comes down to 2 options:
...
With this chip, the footprint configuration looks really complicated, other than that there is no critical weakness of this IC
...
The only problem with this chip, is that the switching frequency is fixed, that may cause some problems when selecting Inductors, luckily, after calculation I can source pretty good inductors based on the fixed frequency, so in the end I chose the MP2491CGQB-Z chip
EN Current Calculation:
R_Pullup +3kOhm (internal pullup) = (24V-7V)/I_EN (7Volt is the internal Zener Diode)
...
In this case, R_Pullup=220kR, so I_EN=62.7uA, smaller than the 100uA, which is good.
Inductor
From the data sheet the design requirements of the inductors is that:
DC current rating 25% larger than output current (5A)
DC Resistance should be smaller thatn 15mOhms
I_sat (Saturation current) should be larger or equal to high side switch current
The current ripple can be calculated in the following equation
...
And the value of inductor can be calculated using the following equation, substitute in all the values.
...
I chose standardize inductor values, and calculate the K backwards (4.7uH, 5.1uH, 5.6uH, 6.2uH)
Inductance | 4.7uH | 5.1uH | 5.6uH | 6.2uH |
|---|---|---|---|---|
K value | 0.343 | 0.316 | 0.288 | 0.26 |
Based on that, I started looking on Digikey for the different inductors. The constraints are that inductors should be shielded, and with current rating and saturation current around 7-10A.
5.6uH:
https://www.digikey.ca/en/products/detail/sumida-america-inc/177CDMCCDS-470MC/9490436 *if set K=0.35, but still bigggg
https://www.digikey.ca/en/products/detail/vishay-dale/IHLP6767GZER560M11/2139396
https://www.digikey.ca/en/products/detail/pulse-electronics/PA4344-563NLT/5641842
https://www.digikey.ca/en/products/detail/codaca/VSAB1770-560M/16731527
https://www.digikey.ca/en/products/detail/coilcraft/XGL1010-563MED/21381839 *Seems to be the best one but not available on DigiKey
https://www.digikey.ca/en/products/detail/shenzhen-sunlord-electronics-co-ltd/MWSA1206S-470MT/14120328 *set at 0.35, relatively smaller
https://www.digikey.ca/en/products/detail/bourns-inc/SRP1265A-470M/4876628 *0.35, smaller (the above one is the smallest available)
4.7uH:
https://www.digikey.ca/en/products/detail/bourns-inc/SRP6060FA-4R7M/9351055
*The rated current and isat are a little bit high, small size (6mmx6mm) (1.86)
https://www.digikey.ca/en/products/detail/abracon-llc/ASPIAIG-F6060-4R7M-T/20096290
*Rated:10, isat:9A, small size (1.57)
https://www.digikey.ca/en/products/detail/w%C3%BCrth-elektronik/744314490/1638566
*4.9uH actually, small size, 6.5A/6.5A, higher price (3.8)
https://www.digikey.ca/en/products/detail/w%C3%BCrth-elektronik/7443340470/3476751
*8mm*7mm, 7.5A/8A, 2.14
The final selected inductor is https://www.digikey.ca/en/products/detail/w%C3%BCrth-elektronik/7443340470/3476751
Capacitor
Several capacitors are needed for this application, and most of the capacitors can be found in the WARG library.
*Several notes includes: use standardize value (e.g. 22uF=10+10+1+1); ceramic preferred, use X5R or better.
Other capacitors includes those such as feedforward capacitors, I just refer the value of them via the datasheet, the details are not expanded here.
Input Capacitors:
For simplification I just used a 20uF capacitor (connected 2 10uF in parallel),
Output Capacitors:
The effective capacitance can be calculated as follows,
...
To compensate the DC-bias and looking at the capacitors the WARG library already have, I selected 2 10uF capacitors with equivalent capacitance around 4.24uF at 5V DC and connect them in parallel.
Resistors
For the ILIM pin, a resistor is needed to set the current limit, according to the graph in the data sheet, to get a current limit of 6A we need 60.4kOhm resistor
For the FB pin, we need to build a voltage divider network to set the output voltage from the feedback voltage.
I have already set V_FB=0.5V by connecting VSEL_1 and VSEL_2 to VCC
...
Updated (10/3): to compensate for the DCR losses, the voltage aimed to be regulated is changed to 5.1V
In this case, R1/R2=9.2, R2=10kOhms, R1=92kOhms, the library values I found are 10k and 91k.
System Block Diagram
...
Reverse Polarity Protection
One feature of the buck converter that is nice to have is RPP. It can be implemented using the following methods:
Using a diode in series, this is the most simple method, however for higher current (>3A) application there will be great power losses.
MOSFETS
P-MOSFET on the high side
More expensive, due to the P-MOSFET
N-MOSFET on the low side
Less expensive, but requiring the power ground and load ground to be different
IC
Small voltage drop across MOSFET
Small current
Overall, choosing a P-MOSFET and an IC are equally favorable, for the sake of learning I have chosen to use IC.
RPP IC selection:
...
RPP diagram:
...
PCB Layout
Some guidelines in the datasheet of the Buck IC:
...
Placement
Initial placement:
...
Kevin’s feedback are as follows:
...
Some notes:
1. Top priority, input caps and output caps, Buck IC, and RPP circuit, then everything else;
2. the distance between the inductor and the switch node pin should be as close as possible;
3. the path for high current traces should be as large as possible (use polygon rather than traces for high current lines)
4. can move the RPP circuit down to leave more room for the BUCK IC
Updated Placements (From 9-30 to 10-2)
...
Routing
...
PDN Analysis
Goal is to calculate loop impedance for the following:
VBUS: Input connector -> NMOS, NMOS -> Buck IC
SW: Buck IC -> Inductor
5V: Inductor -> Pixhawk Conn, Inductor -> USB-C Conn, also do Inductor to farthest decoupling cap down (basically before the via punch down to L4).
Used the Power Analyzer by Keysight tool, not able to calculate the loop impedance directly, use the voltage drop and current simulated to calculated loop impedance.
...
Pixhawk Wiring
The Pixhawk connector that I find is the 5024430670 Molex Clickmate connector, linked below:
https://www.digikey.ca/en/products/detail/molex/5024430670/2380429
The link below is the pixhawk wiring standard
...
The wiring harness that WARG use is 1:1 harness, as pictured, the red cable indicating pin 1
...
Comparing the cable with the pin placements on my PCB shows that it is correct
...
...