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

@@ -1,78 +0,0 @@-# Scaffold: the molecule wired to the control panel for live probing--The buck molecule sits on a LrgMed 64 x 64 scaffold on the base scaffold, next to the Control Panel scaffold (V2.1.0). Five 24 AWG wires connect it to the panel: the input from the programmable supply PS1, its return, the enable pin to a GPIO, and the VMON output to an ADC input with its ground. VOUT is left free for an electronic load or a resistor, so the workcell can power the buck, switch it, read its output and step its load without anyone touching it.--![The buck molecule on a LrgMed 64 x 64 scaffold, with five wires to the Control Panel scaffold](docs/img/scaffold-iso.png)--The layout file is `scaffold/buck-probe-layout.json` (Hydrogen 3D Editor v2 format). The renders on this page come from `scaffold/viewer.html`, a three.js page that loads the same layout file with the three published models, so the picture and the file cannot drift apart.--## Mechanical stack--| layer | part | position (layout mm) | why |-|---|---|---|---|-| base | base scaffold, 576 x 576 mm, large contacts at 16 + 32 n | origin | the workcell floor |-| scaffold | [Scaffold LrgMed 64x64](https://wiki.adom.inc/adom/scaffold-lrgmed-64x64): large pins at its corners (64 mm apart), medium contacts on a 4 mm grid | FL large pin at (240, 464) | the buck has medium machine pins, which need a medium grid; this is the smallest square LrgMed |-| molecule | this buck, 28 x 20 mm, medium pins on a 24 x 16 mm span | FL pin (MP1) at (260, 488), rotation 0 | centred on the scaffold; 24 and 16 are multiples of 4, so all four pins land on the scaffold's grid |-| panel | [Control Panel: Scaffold](https://wiki.adom.inc/adom/control-panel-scaffold) V2.1.0, 160 x 45 mm | behind the base's back edge | power, GPIO and ADC for the workcell |--The scaffold has room left over for a load molecule beside the buck.--## Wiring--| molecule contact | net | panel contact | panel net | wire | why |-|---|---|---|---|---|---|-| J1 (0, 12) | VIN | MC2 | +PS1 | red | PS1 set to 12 V (9 to 16 V range), current limit 0.8 A; worst-case input is 0.6 A at 9 V |-| J2 (0, 8) | GND | MC20 | -PS1 | black | the input return, next to the input |-| J3 (0, 4) | EN | MC33 | USER_IO25 | yellow | open drain: drive low to switch the buck off, release (input) and the UVLO divider decides |-| J6 (24, 4) | VMON | MC29 | USER_IO29 (RP2350 ADC3) | blue | VMON = VOUT / 2 = 2.49 V, inside the 3.3 V ADC range; 5 k source impedance |-| J5 (24, 8) | GND | MC15 | GND | black | ADC and GPIO reference, taken at the output side |-| J4 (24, 12) | VOUT | none | | | left for the load |--Contact positions are in the molecule frame (mm from the centre of MP1), from the `step2glb` footprint JSON. Every panel contact was checked against the net on the V2.1.0 `.kicad_pcb`, not against a label. The panel's KiCad file names MC1 to MC4 `+PS1_SENS`; the page's contact map confirms that is the PS1 rail itself. J2 and J5 are the same net on the molecule, so the wiring also joins -PS1 and panel GND. On V2.1.0, jumper JP1 already bonds them once (bridged by default), so nothing changes.--![The wire ends on the control panel's contact row](docs/img/scaffold-panel.png)--![Close view of the molecule on the scaffold: VIN, GND and EN wires on the left edge, GND and VMON on the right](docs/img/scaffold-close.png)--## Probe plan--`scaffold/probe-plan.json` has every test point in both frames and every test with its setup and expected result. The test points in layout coordinates are the molecule's plus its FL pin position (260, 488).--| test point | net | molecule (mm) | layout (mm) | expected |-|---|---|---|---|---|-| TP1 | SW | (15.6, 4.9) | (275.6, 492.9) | 500 kHz, 0 to VIN |-| TP2 | FB | (9.4, 5.6) | (269.4, 493.6) | 0.596 V (use a 10x probe, high impedance) |-| TP3 | VOUT | (21.6, 4.9) | (281.6, 492.9) | 4.98 V; about 3 to 5 mV pp ripple at 12 V, 1 A |-| TP4 | GND | (15.6, 2.8) | (275.6, 490.8) | probe ground next to SW and VOUT |--| test | setup | pass | source |-|---|---|---|---|-| power up | PS1 12.0 V, 0.8 A limit, EN released, no load | VMON reads 2.49 V; TP3 4.98 V; D1 lit | [design](design.md) |-| soft start | pull EN low for 100 ms, release; scope TP3 | 95 % in about 4.8 ms, overshoot under 5 mV | [simulation](simulation.md) |-| ripple at 1 A | 5 Ohm or 1.0 A electronic load on J4 / J5; TP3 AC-coupled, ground spring to TP4, 20 MHz limit | about 3 to 5 mV pp | ngspice 3.19 mV, TI model 5.0 mV |-| load step | 0.5 to 1 A and 0 to 1 A at 1 A/us; scope TP3 | 80 to 95 mV per 0.5 A step, about 185 mV (3.7 %) for 1 A, settling about 0.2 ms | TI PSpice model |-| UVLO | no load; sweep PS1 up from 5 V, then down | starts at 8.58 V, stops at 6.87 V | EN divider 1 M / 150 k |-| input range | 1 A load; PS1 at 9, 12, 16 V | VOUT 4.98 V at each; PS1 current about 0.6 A at 9 V, 0.45 A at 12 V | 93 % efficiency estimate |-| thermal | 1 A, 12 V, still air, 10 min; thermal camera | U1 and L1 about 40 to 45 C above ambient | [thermal and current](thermal-current.md) |--The load step is the test that matters most. The fitted ngspice model and TI's model disagree by about 15 % there, and the design's conclusion rests on TI's figure.--Probe the ripple with the shortest ground you can. A long ground lead picks up the SW edges and measures itself.--## What is not confirmed yet--- **Both molecules are in the Hydrogen library.** The buck is `john/buck-12v5v-1a-molecule` v1 in the Hydrogen library, imported from its wiki page (the page's anchored GLB, footprint and sources). No LrgMed scaffold existed as a molecule, so `john/Scaffold LrgMed 64x64` v1 was made from the wiki's scaffold model, re-anchored at its front-left large pin. Open `scaffold/buck-probe-layout.json` in the Hydrogen 3D Editor to let it snap the stacking height (the file uses 8.4 mm per level, taken from the reference drone layout) and route the wires. (From an Adom container, point `adom-molecule` at `ADOM_HYDROGEN_URL=https://hydrogen.adom.inc`; the container's local proxy does not serve the molecule routes.)-- **The control panel's position in the preview is approximate.** The editor places the panel itself; the wires name panel contacts, not coordinates.-- **Nothing has been built.** The probe plan is what the first powered board will be checked against.--## Files--| file | what |-|---|---|-| `scaffold/buck-probe-layout.json` | Hydrogen 3D Editor v2 layout: the scaffold, the molecule and five wires |-| `scaffold/probe-plan.json` | test points in both frames, tests, expected values and their sources |-| `scaffold/viewer.html` | three.js preview of the layout (serve the folder, open `viewer.html`; `?view=iso`, `close`, `top` or `cp`) |-| `scaffold/wire-purposes.json` | why each wire goes where it goes |-| `scaffold/cp-contacts.json` | control-panel contact positions from its V2.1.0 `.kicad_pcb` |-| `molecule/buck-12v5v-molecule.glb` | the molecule model from `step2glb convert --pin medium` (anchored at MP1, contacts at z 2.8 mm) |-| `molecule/buck-12v5v-molecule_footprint.json` | pins, contacts and test points in the molecule frame |