main
Download ZIP Propose a change
Name Last commit message Last updated
aiflow Public-facing docs: final-board images, Fields numbers from the final solve, clips merged into one page · John Lauer
design README: probing simulation video, 2D gallery, hi-res 3D render, parts with thumbnails, verification log; sync the redes… · John Lauer
docs README: Adom Fields tour video · John Lauer
ecas-capacitors ECAS study: Adom Fields' own hot-plug check agrees · John Lauer
fusion-360 Fusion files: rebuild the library and board with 3D in your own Fusion (bundle, rebind script, tested recipe) · John Lauer
kicad README: probing simulation video, 2D gallery, hi-res 3D render, parts with thumbnails, verification log; sync the redes… · John Lauer
molecule README: probing simulation video, 2D gallery, hi-res 3D render, parts with thumbnails, verification log; sync the redes… · John Lauer
screenshots hero: framing with margin · John Lauer
sim Public-facing docs: final-board images, Fields numbers from the final solve, clips merged into one page · John Lauer
page.json Edited via adom-wiki: title, brief · John Lauer
README.md ECAS vs MLCC: a study of moving the buck to polymer capacitors, with its section card · John Lauer
verification.json verification: signed off by John · John Lauer

12 V to 5 V / 1 A Buck Molecule (TPS54202)

The buck molecule in 3D, rendered by KiCad 10: the TPS54202, the 15 uH inductor and the 47 uF input can on a 28 x 20 mm board, with VIN, GND and EN contacts on the left, VOUT, GND and VMON on the right, and a medium machine pin at each corner

A 12 V to 5 V, 1 A synchronous buck on the TI TPS54202, designed as a 28 x 20 mm Adom molecule.

It takes the brief from the YouTube video "Hardware Design with GPT-6: Schematic, PSpice & PCB" by Inside Embedded (12 V in, 5 V at 1 A out, terminals, test points) and designs it again from scratch. Every value comes from the datasheet equations. The loop and the switching behaviour were simulated in ngspice, and the result was checked against TI's own PSpice model. The board was placed, routed, poured and analysed in KiCad 10 with Adom AI Flow, and its copper was checked for current density and temperature with Adom Fields.

It was built with a set of Adom tools working together; here is the full list.

The board has not been fabricated yet. Everything below is design and simulation.

The videos

AI PCB Design for Engineers: watch on YouTube Where the Current and Heat Go: watch on YouTube
AI PCB Design for Engineers (about 4 min). The whole build: the prompt in Adom Hydrogen, the circuit, three SPICE engines in agreement, the live layout in KiCad 10, the current and heat analysis and the rework it caused, and the part models. How it was made. Where the Current and Heat Go (8 min 44 s). This board solved in Adom Fields, in eleven narrated views: current density, heat flow, temperature, the current and voltage flowing, energy, how fast a switch-on travels, the energy cycle, charge, the magnetic field and EMC. The numbers.

Fusion 360: the same buck, built by AI in Autodesk Fusion

The buck in Fusion 360
    Fusion 360    
The same brief, run a second time in Autodesk Fusion 360 instead of KiCad, by Claude on Adom AI Flow with the Fusion Bridge driving Fusion. The Fusion 360 build walks through all 21 steps with screenshots, from the one-line prompt through the library, schematic, SPICE, placement loop-backs, routing, current and heat, and silkscreen, to the exported fab files. It also covers what the run cost on each Claude plan and how the video was made.

The Fusion 360 build, step by step

ECAS vs MLCC: should this buck move to polymer capacitors?

ECAS vs MLCC
    ECAS vs MLCC    
A study of swapping this board's ceramic capacitors for Murata ECAS polymer capacitors, using measured part data, ten capacitor sets through the board's own SPICE models, and Adom Fields on the real copper. The answer is not a straight swap: an ECAS with a 0.22 Ω damping resistor earns the input can's place, cutting the tallest capacitor from 5.8 to 2.0 mm, while ceramics stay on the output. The study also found that the as-built feed-forward capacitor should be 47 pF, not 68 pF, once the ceramics' real DC-bias loss is counted.

The study, with every chart and number
top bottom (seen from below)
Top side: copper, pads and silkscreen of the buck molecule Bottom side, mirrored as seen from below: the GND pour, the contact labels and the board name

Copper, exposed pads and silkscreen from the KiCad 10 board, drawn to scale. The bottom is mirrored, so it reads as you would see it with the board flipped over.

Key specs

input 12 V nominal, designed for 9 to 16 V; EN undervoltage lockout starts at 8.58 V and stops at 6.87 V
output 5 V (4.98 V nominal from stocked divider values), 1 A continuous
controller TI TPS54202DDCR, 500 kHz synchronous buck, internal compensation, SOT-23-6
inductor 15 uH Abracon AMPLH5030S-150MT (Isat 3.7 A)
output capacitance 3 x 22 uF X5R, 36.3 uF effective at 5 V
loop (ngspice, 12 V, 1 A) 28.6 kHz crossover, 110 deg phase margin with 68 pF Cff
output ripple (ngspice, 12 V, 1 A) 3.2 mV peak to peak
load step (TI PSpice model) about 185 mV (3.7 %) for a 1 A step
efficiency (estimate) 93 % at 12 V, 1 A
current density (Adom Fields) 50 A/mm2 peak, at the SW node beside U1
thermal (Adom Fields, still air) U1 +45 C at its GND pin, L1 +41 C; 55 C junction-based budget
board 28 x 20 mm, 2 layers, 0.5 oz copper, 4 medium machine pins on a 24 x 16 mm grid, 7 contacts, 7 probe pads
DRC (KiCad 10.0.5) 0 errors, 0 unconnected; the board matches the schematic
sourcing Mouser plus Adom stocked basic parts, for a 3rd party fab

Molecule interface

28 x 20 mm, 2 layers. Positions are in mm from MP1. The molecule's GLB, STEP, footprint and symbol JSON are in molecule/; the KiCad 10 project is in kicad/.

contact position net notes
J1 (0, 12) VIN 9 to 16 V, 12 V nominal; UVLO starts at 8.58 V, stops at 6.87 V
J2 (0, 8) GND input return
J3 (0, 4) EN open drain low turns it off; released, the UVLO divider decides
J4 (24, 12) VOUT 4.98 V, 1 A
J5 (24, 8) GND output and signal ground
J6 (24, 4) VMON VOUT / 2 (2.49 V) for a 3.3 V ADC, 5 k source
J7 (5.25, 2.25) GND spare ground contact for test equipment
MP1 to MP4 (0, 0), (24, 0), (0, 16), (24, 16) GND medium machine pins

Probe pads: TP1 SW (15.6, 4.9), TP2 FB (9.4, 5.6), TP3 VOUT (21.6, 4.9), TP4 GND (15.6, 2.8), TP6 VIN (2.05, 1.95), TP7 VIN at C2 (7.0, 16.75), TP8 VOUT at L1's output (22.5, 14.0).

Parts

Every part on this board has its own page on the Adom wiki, with its symbol, footprint and 3D model, ready to reuse in your own designs. Click a part to open its page. The BOM with every distributor number is design/bom.csv.

part qty refs value what it does source
TPS54202DDCR
TPS54202DDCR
Texas Instruments
1 U1 TPS54202DDCR sync buck 4.5-28 V 2 A 500 kHz Mouser 595-TPS54202DDCR
AMPLH5030S-150MT
AMPLH5030S-150MT
Abracon
1 L1 15uH output inductor Mouser 815-AMPLH5030S-150T
EEE-FK1V470P
EEE-FK1V470P
Panasonic
1 C1 47uF 35V input bulk (cable damping) Mouser 667-EEE-FK1V470P
CC1206KKX7R8BB106
CC1206KKX7R8BB106
Yageo
2 C2, C3 10uF 25V X7R input ceramic Mouser 603-CC126KKX7R8BB106
0603B104J500CT
0603B104J500CT
Walsin
2 C4, C5 100nF 50V X7R VIN pin HF bypass Adom stock
CL21A226MPQNNNE
CL21A226MPQNNNE
Samsung
3 C6, C7, C8 22uF 10V X5R output ceramic Adom stock
0402N680J500CT
0402N680J500CT
Walsin
1 C9 68pF C0G feed-forward across Rtop Adom stock
CR0402-FX-6802GLF
CR0402-FX-6802GLF
Bourns
1 R1 68k FB top (series part 1) Adom stock
CR0402-FX-5601GLF
CR0402-FX-5601GLF
Bourns
1 R2 5.6k FB top (series part 2) Adom stock
CR0402-FX-1002GLF
CR0402-FX-1002GLF
Bourns
3 R3, R7, R8 10k FB bottom Adom stock
CR0402-FX-1004GLF
CR0402-FX-1004GLF
Bourns
1 R4 1M EN UVLO top Adom stock
CR0402-FX-1503GLF
CR0402-FX-1503GLF
Bourns
1 R5 150k EN UVLO bottom Adom stock
CR0402-FX-2201GLF
CR0402-FX-2201GLF
Bourns
1 R6 2.2k LED resistor Adom stock
IN-S63BTG
IN-S63BTG
Inolux
1 D1 green LED 5 V power-good indicator Adom stock
MachineContactMedium
MachineContactMedium
Adom
7 J1 to J7 MachineContactMedium molecule contact: VIN, GND, EN, VOUT, VMON, and a spare GND Adom stock
MachinePinMediumShort
MachinePinMediumShort
Adom
4 MP1 to MP4 MachinePinMediumShort corner machine pin (GND) Adom stock

Feature guides

guide what it covers
The Adom tools Every Adom tool used on this board, what each one did, and a link to its page.
Circuit design The brief, why the TPS54202, and every calculation: inductor (15 uH vs Lmin), COUT, the 68 pF Cff, the stocked-value divider search, EN UVLO, bootstrap, losses.
Simulation ngspice loop gain across Cff, the switching transient at 9, 12 and 16 V, and the check against TI's PSpice model.
Sourcing The BOM with Mouser part numbers, the 3rd party fab sourcing policy, and where every 3D model came from.
Board Molecule format, design rules, placement, routing, pours with the SW patch, DRC, copper and silkscreen.
Thermal and current Adom Fields current density (SW peak 50 A/mm2), where the heat goes, and the still-air budget.
How the video was made Capturing the footage while the board is built, the shots and scenes, and putting the video together.
Probing before ordering The probing workcell's test plan, each test a copy of a SPICE run, the simulated scope captures, and the loops that sent the board back to layout for J7, TP6, TP7 and TP8.
Build clips Watch the board being built in KiCad: placement, routing, pours, the analyses and the silkscreen.
Fusion 360 build The same buck built by AI in Autodesk Fusion 360: every step with screenshots, the cost per Claude plan, and how the video was made.
ECAS vs MLCC Swapping the ceramics for polymer capacitors: measured DC-bias data, ten capacitor sets in SPICE, hot plug, rail impedance in Adom Fields, board size, cost, and the 47 pF Cff finding.

What is in this repo

folder contents
design/ calcs.py (every datasheet equation, run it to regenerate calcs.json) and bom.csv
sim/ the ngspice loop and switching models with their results and plots, and the TI PSpice comparison
kicad/ KiCad 10 project: schematic, board, design rules, the netlist the board was generated from, the generator scripts, project libraries and every 3D model
aiflow/spec.json the layout constraints the board was built to: wide nets, Kelvin taps, loads, hot parts, pours and the thermal budget
molecule/ the molecule STEP (KiCad 10 export), the anchored GLB and the footprint and symbol JSON
verification.json the verification log behind the badge at the top: the checks, and the SHA-256 of every file they cover