Skip to the content
6 layers158 × 174 mmTritonECU Main, the printed circuit board in 3D

TritonECU Main

alpauna/TritonECU10

ESP32-S3 engine control unit — fuel injection, ignition timing, wideband O2 (CJ125/LSU 4.9), dashboard, and MQTT telemetry

Size
158 × 174 mm
Copper layers
6
Thickness
1.6 mm
Parts
457
Nets
422
Pads
1,585
Tracks, vias
8,335, 1,580
Pours
4
Routed
Yes
KiCad
KiCad 10.0 (file 20260206)
Created
13 Feb 2026
Updated
21 Oct 2026

Overview

TritonECU Main is an open-source printed circuit board by alpauna, from the alpauna/TritonECU repository on GitHub. It is a 6-layer board of 158 × 174 mm with 457 parts on 422 nets, routed with 8,335 tracks and 1,580 vias.

It is built around the 74HCT541D, ISL9V3040S3ST and MAX9926UAEE+T, with the OPA2376AQDRQ1, ADS8588HIPM and DRV8837DSGR-MS among its other parts. Its parts include 225 resistors, 114 capacitors, 41 diodes, 34 transistors and 19 ICs. It belongs with the home automation and IoT boards and the ESP32 boards of this library.

Main components

12

README

README.md

TritonECU

Standalone engine and transmission controller for Ford modular V8s of the EEC-V era — replacing the factory PCM.

Developed against a 1999 F-150 4WD, 5.4L Triton 2V, 4R70W, but the architecture is the modular family rather than one truck: 4.6L and 5.4L 2V share the crank trigger, the sensor set, the SCP bus and the connector, so Mustangs and other EEC-V modular applications of the period are the same problem with a different calibration.

Built on an STM32F767ZI. Rewritten from scratch on the rebuild branch; the previous ESP32-S3 firmware is described in docs/legacy-esp32s3-README.md and still lives on main.

What is vehicle-specific and what is not:

Portable across EEC-V modularsSpecific to this truck
36-1 VR crank decode, cam syncfiring order, displacement
EEC-V 104-pin connector and pinout4R70W shift logic (4R100 differs)
SCP / J1850 PWM bus4x4 transfer case inputs
Ignition and injection output stagesVE and spark tables
Sensor conditioning and protection

Status

Milestone
M0 Board bring-up✅216 MHz, 2 MB flash, verified from the chip's own ID registers
M1 Storage + config✅SD on SPI4, JSON config, persists across resets
M2 Analog front end✅AD7606 reading known signals within tolerance
M3 Crank sync◐decode logic proven — 10 native tests; VR board on order, CKP sensor + pigtail ordered
M4 Cam sync◐logic proven — 6 native tests; CMP sensor + pigtail ordered — same connector as CKP
M5 Spark output◐scheduler proven — 12 native tests
M6+ Injection, closed loop, SCP, transmission○
ECU carrier board○the largest piece left. Fully specified in schematic-ecu-v1.txt — 736 lines, every part chosen — but not yet captured as a schematic or laid out. Envelope and the height constraint: carrier-envelope.md
Power board V1🔒released and frozen — tagged power-v1-released, Schematics/README
Power board V2✅V1 + one part: C1 100 nF 1206 X7R 200 V across B+/PGND at the connector — a 19 mm² loop against D1's 131. Reviewed clean, ready to fab
VR board V1✅4 channels, 2 × MAX9926 Mode A2 — reviewed clean, fabricating
VR test rig◐all dimensions measured; 9 of 10 parts ready to print, STLs + BOM. Parts ordered bar the stepper
VR rig firmware✅RP2040 PIO step generator with a cranking profile — builds clean, firmware/
2N7002 driver board✅3.3 V → DM542 opto inputs, reviewed clean from Gerbers — hardware/2N7002 Driver/
Knock front end◐next build after the VR board — schematic, net list and BOM drawn, knock-front-end-schematic.md
EEC-V connector✅sourcing solved — Ranger donor + the TE controlled drawing
Enclosure◐cabin, behind the glovebox, connector face through the firewall — vent on the cabin side, enclosure.md
Carrier template✅printable measurement jig for the donor case floor — captures the mounting-boss pattern the layout is waiting on, hardware/carrier-template/
Harness label✅printed plate for the always-hot battery lead and the four added pins — STL + README
PSU platform✅Z-bracket that stands the supply module off the surface — 74.5 × 92 × 51, hardware/psu-platform/

28 native unit tests passing. The decode, cam-sync and spark-scheduling modules are <stdint.h>-only and testable without hardware — which is how M3–M5 were proven before the sensors existed.

bash
pio test -e native              # logic, no hardware
pio run -e nucleo -t upload     # flash the board

Why it is being rebuilt

The previous firmware worked, but was written against an ESP32-S3 that was never in a truck, and one of its structural decisions turned out to be wrong: coils and injectors were driven through MCP23S17 SPI expanders.

At 6000 rpm one crank degree is 27.8 µs. An SPI transaction is several microseconds at best, is not deterministic, and cannot be driven from a hardware timer. Spark timing cannot live behind a port expander. That is not a bug to patch — it changes ignition, injection and the pin map together.

Proven pieces get ported forward deliberately. Everything else earns its place again. See docs/roadmap.md.

Hardware

Design is frozen — docs/v1-scope.md is the build list.

Two boards are finished. Schematics, BOMs and Gerbers are all version controlled in docs/Schematics/ — the fab outputs, not just pictures of them.

BoardSizeReview
Power V176.58 × 46.74 mm, 4 layerreview-power-v1-bom-gerbers.md
VR V135.43 × 33.91 mm, 2 layerreview-vr-v1.md

Both reviews were measured out of the Gerbers — copper polygon areas, via positions, pad geometry and the flying-probe netlist — rather than read off a render. That is how the MAX25239 footprint, Q2's thermal path and the output-cap pour were actually checked. The earlier schematic-only pass on the power board is schematic-review-power-v2.md.

BlockPart
MCUSTM32F767ZI (Nucleo-144 now, raw chip later)
Input protectionLTC4364-2 — reverse polarity, load dump, overcurrent, brownout holdup
Pass elementIRF540NS (D2Pak) on HGATE — planar, published SOA, 20 tented vias into 1.14 in² of B+
5 V railMAX25239AFFA buck-boost, 2.1 MHz, spread spectrum, 2.2 µH
3.3 V railTLV62085 buck from 5 V, 2.4 MHz, 500 nH
ADC referenceMAX6070AAUT25, 2.5 V, gated for sleep
Battery senseINA238 on the LTC4364's own 10 mΩ / 1 % / 50 ppm shunt
Analog inADS8588H — 8 ch, 16-bit, 500 kSPS simultaneous, ±10 V, 9 kV input clamp, −40…+125 °C
Analog in (bench)Tokmas AD7606BSTZ — same LQFP-64 footprint, 200 kSPS, already in hand
Crank / cam / OSS2 × MAX9926 Mode A2 — 4 channels, adaptive threshold + zero crossing
KnockTLV9064-Q1 differential charge amp + own 2.5 V ref — schematic
BarometricKP497 (Infineon) on I²C/SPI, AEC-Q100 −40…+105 °C, 3.3 V always-on rail — requires a vented enclosure
Ignition8 × ISL9V3040 ignition IGBT + 74HCT541
Injection8 × ZXMS6005DGQ IntelliFET
SCP / J1850 PWMDRV8837 differential TX + TLV7031 comparator RX — the cluster's only link to the PCM
ConnectorEEC-V 104-pin

Decisions worth knowing

  • No Wi-Fi. An always-on radio on an engine controller is a remote attack surface with a path to code execution via OTA. Ethernet needs physical access. docs/platform-decision.md
  • 2.1 MHz switching to clear the AM broadcast band — a 400 kHz converter puts harmonics at 1.2 and 1.6 MHz, inside it. The MAX25239's A and B suffixes are 2100 kHz and 400 kHz respectively, and the board was nearly built with the wrong one. docs/always-on-domain.md
  • The MCU never fully powers down. It sleeps on an always-on 3.3 V rail at ~3 µA and brings the rest up on demand, which removes the graceful-shutdown problem, replaces KAPWR and makes supply-fault latching implementable in firmware. Parked drain is ~950 µA, about 1 % of the battery per month. docs/always-on-domain.md
  • The ADC lives on a daughterboard. Not for proximity — that is a placement problem — but for analog ground plane control on a board that low-side switches eight ignition coils, and so the front end can be respun without the carrier. docs/adc-daughterboard.md
  • The SEPIC was retired. An integrated buck-boost deletes the external FET, the Schottky, both inductors and the coupling-cap bank, and adds spread spectrum the controller design never had. docs/power-supply.md keeps the input-chain analysis and marks the rest superseded.
  • Clamp both loads, dissipate neither. A coil's flyback is the spark (400 V clamp); an injector's is waste but a plain diode makes it close slowly and over-fuel at idle (60–70 V clamp). docs/output-drivers.md
  • The shunt is sensed twice and sets two things. The LTC4364's current limit and the INA238's battery-current reading both come off one 10 mΩ resistor, so its tolerance and TCR are the accuracy of both — which is why it is a 1 %, ±50 ppm/°C, AEC-Q200 metal-element part in 2512 rather than whatever was cheapest.
  • Leave the factory harness alone. Every "this should be referenced differently" question is answered by PCB routing, not by moving wires — the board defines what each of the 104 cavities connects to. An unmodified harness stays diagnosable by anyone with a Ford wiring diagram, and the PCB is the half that can be revised. docs/1999-Ford-F150-4wd-5.42v/connector-sourcing.md
  • The VR trigger point needs no calibration term. MAX9926 Mode A2 arms on 33 % of the previous peak but triggers on the zero crossing (−6.5 / +10 mV) — which is the tooth centre, because a VR sensor outputs dΦ/dt. Amplitude, rpm and acceleration cannot move it; scaling a waveform does not shift where it crosses zero. Propagation delay is 50 ns, or 0.0018° at 6000 rpm. This repo previously had it wrong in three places, claiming the comparator switched at ⅓ of peak and that the offset would need characterising. docs/1999-Ford-F150-4wd-5.42v/vr-conditioning.md
  • MAX9926 over the LM1815, on arming margin. Both arm adaptively and trigger on zero crossing, so timing is a wash. The LM1815 arms at 80 % of the previous peak against the MAX9926's 33 % — tolerating a 1.25:1 amplitude collapse between teeth where the MAX9926 tolerates 3:1. Cranking, slowing hard against each compression stroke, is exactly where that collapse happens. The differential input the single-ended LM1815 cannot offer settles the rest.
  • Keyway to missing tooth is zero, on the rig and on the truck. The bench wheel's keyway lines up with the gap, and the 5.4L was timed to the same datum, so CKP_GAP_TO_TDC_DEG = 0 for both. It stays a named constant rather than a vanished zero — the rig's absolute angle is then the engine's absolute angle, which makes the rig a timing calibration bench rather than a functional check. hardware/vr-test-rig/README.md
  • Stiffness comes from section, not from material. Two 8 mm rods in ribs under the rig's base take it from EI 20.5 to 826 N·m² — 40× — because the offset contributes A·d². A 5 mm rod in a rib beats a 10 mm rod buried in the plate. The same arithmetic found the real soft spot: the printed bearing bracket was less stiff than the shaft it was holding. hardware/vr-test-rig/BOM.md
  • A printed trigger wheel produces nothing. A VR sensor works on reluctance change; printed plastic is magnetically identical to air. The bench rig is printed, the 36-1 wheel is steel, and the cam target is a printed disc with one steel bolt. hardware/vr-test-rig/
  • The knock sensor is differential, and that decides its front end. A two-wire floating piezo (pins 57, 32) into a single-ended knock IC would ground one leg and inject sensor-ground noise straight into a microvolt signal. A differential charge amp is needed either way, so it is the part to build — and it defers the IC-versus-DSP choice rather than forcing it. One TLV9064-Q1 quad holds the lot: two charge amps, a difference stage and a VMID buffer. docs/knock-front-end.md
  • Charge mode, because the harness moves. Read a piezo as a voltage and the cable capacitance divides the signal down — 100 pF of loom against a 1 nF sensor is a 10 % error that changes if the harness is rerouted. Read as charge, both terminals sit at a virtual ground, the cable sees no swing, and sensitivity stops depending on how the loom was dressed. The feedback caps then are the calibration (Q/Cf), which is why they are C0G and not X7R.
  • Analog blocks get their own references. The knock front end has a 2.5 V reference separate from the ADC's. It does not save an amplifier though: the difference stage's reference leg is one arm of a resistor bridge, and CMRR depends on that bridge staying balanced, so it still wants a buffer between the reference and R8 rather than the reference's own output impedance in series with one arm.
  • The enclosure must be vented, and not only for the barometer. A sealed box tracks its own temperature rather than the weather, so the barometric sensor would be useless in one. But a sealed automotive housing also pumps moisture past its own seals — every heat cycle pushes air out, every cool-down draws it and whatever is near the seal back in. A vent membrane gives that breathing a deliberate path; the correct pressure reading is a side effect of doing the housing properly. The vent goes on the cabin side — the box straddles the firewall, and its connector face lives under-hood in the spray. docs/enclosure.md
  • One ADC footprint, five parts, two vendors. AD7606 / B / C and ADS8588S / H are pin-for-pin in the same LQFP-64 — verified pin by pin, not assumed. So the part is a populate-time choice, not a layout one: ADS8588H in production for its 9 kV input clamp on harness-connected pins, and the Tokmas AD7606BSTZ already on the bench for development. (The ADS9324 is not in this set — 16 channels, VQFN, and a 1.8 V rail this board does not have.) docs/adc-front-end.md
  • A "level shifter" is not a current driver. The DM542's step inputs are optocouplers wanting 14 mA; a BSS138 + 10 kΩ bidirectional module supplies 0.38 mA, and an IRF520 module is not logic-level at all — its gate threshold spans 2–4 V against a 3.3 V drive, so it works warm and fails cold. The answer is a logic-level FET sinking the input, with a gate pulldown so nothing moves while the Pico boots. hardware/2N7002 Driver/
  • Model the part you are designing around, then print it. The rig's two VR sensors are modelled from the same parameters the mount uses, so a printed copy held against the real sensor tests every one of those numbers at once — and a mismatch proves the mount wrong in the same way. Cheaper on a 20 g print than after the base. hardware/vr-test-rig/
  • The MAF is a differential measurement. It has a dedicated signal return separate from the ground its supply current flows in — which is the whole reason for the AD7606 over the MCU's own ADC.

The reference vehicle

Every design decision traces to a factory schematic, recorded in docs/1999-Ford-F150-4wd-5.42v/. Several overturned assumptions taken from parts catalogues and forums:

AssumedActually
Wasted spark, 4 coil driversCoil-on-plug, 8 individual coils
ECT coolant sensorCHT — cylinder head temperature
Oil pressure senderA switch. Binary, not analog
No knock sensor on the 2VPresent — C103
4R70W has a TSSIt does not. 4R100 only
GEM is on the SCP busIt is on ISO 9141. Only the PCM and cluster are on SCP

PATS is a non-issue: the anti-theft logic lives in the instrument cluster and authorises the PCM over SCP. A replacement ECU never asks, so there is nothing to defeat.

External references — to read, not yet read

4R100 rebuild manual for the DIY⚠ 4R100, not 4R70W — this truck has the 4R70W. Worth having anyway: the MegaSquirt pinout this project cross-checks against was developed on a 4R100, which is why it listed a coast clutch solenoid that does not exist here (output-drivers.md). If another MS-sheet entry ever looks wrong, this is where the explanation will be
Dorman 917-060 — the rig's 36-1 wheel (Amazon, with the reviews)Steel, 35 teeth, OE cross XW1Z-12A227-AC, modular V8 (2001-2010 F-150 4.6, 2002-2010 F-150 5.4). ⚠ Bench part only — do not fit it to the truck. A review reports its teeth do not match the OEM wheel, giving timing codes and rough running; we measured its keyway 5.00° — half a tooth pitch — from where the factory datum was believed to be (vr-test-rig/README.md)

Layout

text
firmware/ecu/
  src/core/     decode, cam sync, spark scheduling — portable, unit-tested
  src/stm32/    STM32 platform layer
  src/esp32/    ESP32-P4 platform layer (earlier target, still builds)
  test/         native unit tests
hardware/       pin budget and supply calculators
  vr-test-rig/    parametric OpenSCAD for the 36-1 + cam bench rig
  psu-enclosure/  printed box for the bench supply
  harness-label/  printed label - the always-hot lead and the added pins
  carrier-template/ measurement jig for the donor ECU case
  psu-platform/   Z-bracket platform for the supply module
docs/           design documents and vehicle schematics
  Schematics/   board schematics, BOMs and Gerbers — fab-ready
  Datasheets/   every part datasheet the decisions were made against
  1999-Ford-F150-4wd-5.42v/  Ford's own diagrams and service manual
  DonorECU/     the scrap PCM the connector and case come from
tools/          USB permission setup

What gets committed, and why the repo is big

~300 MB, and that is a deliberate trade. The breakdown:

MB
Vendor datasheets79.8re-downloadable, but manufacturers pull obsolete parts
Ford service manual29.5hard to re-source, and getting harder
Ford scans — diagrams, connectors24.4same
Interactive BOMs22.9regenerable
Board design files11.9
STLs4.4regenerable from the .scad
Source and docs — the actual project1.50.9 %

The thinking is 1.5 MB. Everything else is evidence.

Keep the evidence. Design decisions across these documents are justified by "Ford's own sheet says X", and that phrase only means something if the sheet is here. The archive has been re-read repeatedly to correct conclusions, not just to support them — pin 64's function, the MAF's Kelvin wiring, the EPC resistance that showed a suspect part was healthy. A 1999 service manual gets harder to find every year.

Do not commit regenerable artifacts. Interactive BOMs, build outputs, export by-products. The two BOM HTMLs predate this rule; they stay because removing them now would not reclaim anything.

Why "not reclaim anything": git keeps every blob ever committed. Deleting a file from the working tree shrinks the checkout, not the history. Only git filter-repo and a force-push reclaims space — and that rewrites shared history, invalidates the power-v1-released tag's SHA, and breaks every existing clone. Not worth it at this size.

Revisit at ~700 MB. GitHub soft-warns near 1 GB and hard-limits individual files at 100 MB; the largest here is 30 MB. If it ever does need shrinking, the lever is the vendor datasheets at 45 % — not the Ford material, which is the least replaceable thing in the repository.

Bench setup

docs/nucleo-setup.md is a runbook for bringing a Nucleo up from scratch, including the failures that actually cost time — the ST-Link udev rule, programming the ELF rather than the .bin, and -Wl,-u,_printf_float.

bash
tools/setup-usb-permissions.sh      # debug probe access

Parts

457
QtyValueFootprintReferences
50100nCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC101, C102, C302, C305, C307, C309, C312, C314, C316, C319, C321, C401, C402, C411, C412, C1009 +34
36100kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR004, R421, R422, R423, R424, R425, R426, R427, R428, R431, R432, R433, R434, R435, R521, R522 +20
3510kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR109, R110, R201, R202, R203, R204, R205, R206, R207, R208, R305, R310, R315, R320, R325, R330 +19
245kResistor_SMD:R_0805_2012Metric_Pad1.20x1.40mm_HandSolderR301, R302, R303, R304, R306, R307, R308, R309, R311, R312, R313, R314, R316, R317, R318, R319 +8
172.2kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR411, R412, R413, R414, R415, R416, R417, R418, R511, R512, R513, R514, R515, R516, R517, R705 +1
160Resistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR001, R399, R801, R810, R811, R812, R813, R821, R825, R829, R833, R853, R854, R855, R856, R910
15BAT54JDiode_SMD:D_SOD-323FD703, D704, D805, D807, D810, D811, D1002, D1004, D1006, D1008, D1010, D1012, D1014, D1016, D1017
141uCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC303, C310, C317, C807, C808, C817, C819, C822, C824, C834, C835, C902, C909, C911
14NCV8405ASTT1GSamacsys:SOT230P700X175-4NQ401, Q402, Q403, Q404, Q405, Q406, Q407, Q408, Q501, Q502, Q503, Q505, Q506, Q507
1310nCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC301, C308, C315, C710, C836, C1001, C1002, C1003, C1004, C1005, C1006, C1007, C1008
1015kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR915, R1003, R1006, R1009, R1012, R1015, R1018, R1021, R1024, R1026
910.5kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR441, R442, R443, R444, R445, R541, R562, R564, R852
933kResistor_SMD:R_0805_2012Metric_Pad1.20x1.40mm_HandSolderR1002, R1005, R1008, R1011, R1014, R1017, R1020, R1023, R1025
9470Resistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR101, R102, R103, R104, R105, R106, R107, R108, R845
81kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR820, R824, R828, R832, R836, R837, R839, R850
84.7kResistor_SMD:R_0805_2012Metric_Pad1.20x1.40mm_HandSolderR1001, R1004, R1007, R1010, R1013, R1016, R1019, R1022
8ISL9V3040S3STSamacsys:ISL9V3040S3STQ101, Q102, Q103, Q104, Q105, Q106, Q107, Q108
8SMF24ADiode_SMD:D_SMFD1001, D1003, D1005, D1007, D1009, D1011, D1013, D1015
8ZXMS6005DGQTAPackage_TO_SOT_SMD:SOT-223-3_TabPin2Q201, Q202, Q203, Q204, Q205, Q206, Q207, Q208
7100Resistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR002, R003, R846, R902, R903, R909, R918
71nCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC004, C007, C816, C818, C821, C823, C912
6470pCapacitor_SMD:C_0805_2012Metric_Pad1.18x1.45mm_HandSolderC304, C306, C311, C313, C318, C320
510uCapacitor_SMD:C_0805_2012Metric_Pad1.18x1.45mm_HandSolderC005, C008, C805, C809, C810
547kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR921, R922, R923, R924, R925
5STPS3H100UYDiode_SMD:D_SMBD501, D502, D503, D504, D505
410kResistor_SMD:R_0805_2012Metric_Pad1.20x1.40mm_HandSolderR818, R822, R826, R830
41MResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR111, R838, R840, R917
422MResistor_SMD:R_0805_2012Metric_Pad1.20x1.40mm_HandSolderR819, R823, R827, R831
4BAT54SPackage_TO_SOT_SMD:SOT-23D802, D803, D804, D806
4MF-MSMF030/60-2Fuse:Fuse_1812_4532MetricF601, F602, F603, F604
31uCapacitor_SMD:C_0805_2012Metric_Pad1.18x1.45mm_HandSolderC003, C006, C701
3220Resistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR808, R1027, R1028
34.7uCapacitor_SMD:C_0805_2012Metric_Pad1.18x1.45mm_HandSolderC813, C906, C913
374HCT541Package_SO:SOIC-20W_7.5x12.8mm_P1.27mmU101, U401, U501
3BAV199Package_TO_SOT_SMD:SOT-23D808, D809, D812
3MAX9926UAEE+TPackage_SO:QSOP-16_3.9x4.9mm_P0.635mmU301, U302, U303
3TPS22917DBVRPackage_TO_SOT_SMD:SOT-23-6U001, U002, U905
210uCapacitor_SMD:C_1206_3216Metric_Pad1.33x1.80mm_HandSolderC001, C002
22.49kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR814, R815
2220pCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC827, C828
222nCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC708, C709
24.7nCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC908, C910
25.1kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR913, R914
260.4Resistor_SMD:R_1206_3216Metric_Pad1.30x1.75mm_HandSolderR911, R912
2620pCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC829, C830
2DMN6040SVTQ-7Samacsys:SOT95P280X100-6NQ701, Q702
2LM74700QDBVRQ1Package_TO_SOT_SMD:SOT-23-6U703, U704
2MM3Z3V3BWDiode_SMD:D_SOD-323FD902, D903
2NUP2105LPackage_TO_SOT_SMD:SOT-23D901, D905
2OPA2376AQDRQ1Package_SO:SOIC-8_3.9x4.9mm_P1.27mmU803, U804
2SMAJ24CADiode_SMD:D_SMAD701, D702
1100nCapacitor_SMD:C_0805_2012Metric_Pad1.18x1.45mm_HandSolderC601
1104031-0811Connector_Card:microSD_HC_Molex_104031-0811J902
112.4kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR709
115nCapacitor_SMD:C_0603_1608Metric_Pad1.08x0.95mm_HandSolderC831
1180kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR816
118kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR848
122Resistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR920
122uCapacitor_SMD:C_1206_3216Metric_Pad1.33x1.80mm_HandSolderC703
127kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR916
130kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR817
14.7MResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR908
14.7kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR706
174HC4051Package_SO:TSSOP-16_4.4x5mm_P0.65mmU402
18.06kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR707
19.1kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR835
191kResistor_SMD:R_0603_1608Metric_Pad0.98x0.95mm_HandSolderR834
1ADS8588HIPMSamacsys:QFP50P1200X1200X160-64NU801
1BOOT0 DFUJumper:SolderJumper-2_P1.3mm_Open_RoundedPad1.0x1.5mmJP1001
1BSS138Package_TO_SOT_SMD:SOT-23Q109
1DRV8837DSGRPackage_SON:Texas_DSG0008A_WSON-8-1EP_2x2mm_P0.5mm_EP0.9x1.6mmU901
1EEC-V 104TritonECU-Main:EECV_104J1
1MAX6070AAUT18+TPackage_TO_SOT_SMD:SOT-23-6U806
1MAX6070AAUT25+TPackage_TO_SOT_SMD:SOT-23-6U802
1MCU module F767TritonECU-Main:MCUModule_F767_MateDB3
1NCV8408BDTRKGPackage_TO_SOT_SMD:TO-252-2Q504
1NCV8772CDT504RKGPackage_TO_SOT_SMD:TO-252-5_TabPin3U701
1NetTie_3NetTie:NetTie-3_SMD_Pad0.5mmNT801
1OE2 strapJumper:SolderJumper-2_P1.3mm_Open_RoundedPad1.0x1.5mmJP101
1Power board V3TritonECU-Main:PowerBoard_V3_MateDB1

The 444 most common of 457 parts.

More from alpauna/TritonECU

2
  • 4 layers75.3 × 58.9 mmTritonECU, the printed circuit board in 3D

    TritonECU

    alpauna/TritonECU

    ESP32-S3 engine control unit — fuel injection, ignition timing, wideband O2 (CJ125/LSU 4.9), dashboard, and MQTT telemetry

    83 parts10
  • 6 layers46 × 46 mmTritonECU MCU, the printed circuit board in 3D

    TritonECU MCU

    alpauna/TritonECU

    ESP32-S3 engine control unit — fuel injection, ignition timing, wideband O2 (CJ125/LSU 4.9), dashboard, and MQTT telemetry

    38 parts10

Similar boards

  • 2 layers24.6 × 24.7 mmV3 Conditioner Kicad, the printed circuit board in 3D

    V3 Conditioner Kicad

    EFI-Customs/Motronic-1.3-Speeduino-Compatible-Project

    Collection of resources for Speeduino related items & EFI Customs products

    17 parts5
  • 2 layers164 × 118 mmProteusOBD1honda, the printed circuit board in 3D

    ProteusOBD1honda

    rusefi/proteus-Honda-OBD1-adapter

    🟢 See https://github.com/rusefi/proteus-Honda-OBD2A-adapter for a much happier adapter board 🟢

    15 parts2
  • 4 layers182 × 160 mmProteusM54, the printed circuit board in 3D

    ProteusM54

    rusefi/proteus-M54-adapter

    (E39 m54 should be the ms42/43 like the e46, E39 = E46)

    146 parts
  • 2 layers350 × 111.3 mmProteusM73tu, the printed circuit board in 3D

    ProteusM73tu

    rusefi/proteus-M73tu-adapter

    V12 harness uses TWO 134 pin connectors - one with BLACK plugs another with GREY plugs.

    129 parts
  • 2 layers103 × 95 mmDSO, the printed circuit board in 3D

    DSO

    Aryan-Kodan/Two_Channel_DSO

    A custom Multi-Channel Digital Storage Oscilloscope (DSO) designed using the STM32F103C8T6 microcontroller. This project combines a custom analog front-end, trigger circuitry, anti-alias filtering, multi-channel…

    94 parts
  • 4 layers86 × 29.9 mmGeiger probe, the printed circuit board in 3D

    Geiger probe

    bartkepl/CanGeigerProbe

    Polowa sonda promieniowania jonizującego z miniaturową lampą Geigera-Müllera, komunikująca się cyfrowo przez CAN 2.0B.

    116 partsMIT
  • 4 layers76 × 29.9 mmGnss node, the printed circuit board in 3D

    Gnss node

    bartkepl/CanGnssNode

    Polowy węzeł pozycyjny z modułem GNSS Quectel L76-L (GPS/GLONASS/Galileo/QZSS) i czujnikiem orientacji (akcelerometr + magnetometr), komunikujący się cyfrowo przez CAN 2.0B.

    141 partsMIT

Questions

How big is the TritonECU Main PCB?

It measures 158 × 174 mm and has 6 copper layers, 1.6 mm thick.

What parts does the TritonECU Main use?

457 parts, built around the 74HCT541D, ISL9V3040S3ST, MAX9926UAEE+T and OPA2376AQDRQ1. The full list of its parts is on this page, with links to each part's symbol and footprint where PCBKit's library has them.

Where can I download the TritonECU Main KiCad files?

From the alpauna/TritonECU repository on GitHub, at the commit this page was made from: the board is docs/Schematics/KiCad/TritonECU-Main/TritonECU-Main.kicad_pcb and its schematic docs/Schematics/KiCad/TritonECU-MCU/TritonECU-MCU.kicad_sch. PCBKit opens the board in the browser, with nothing to install.

Can I use the TritonECU Main design in my own project?

Its repository states no licence, so its author keeps all rights: ask alpauna before reusing the design.