Skip to the content
4 layers128 × 84 mmDronemagnav2, the printed circuit board in 3D

DroneMagNav 2.0 - survey-fluxgate magnetic-anomaly navigation board for drones (IGRF-14 + EMAG2v3 continued to flight altitude)

Size
128 × 84 mm
Copper layers
4
Thickness
1.6 mm
Parts
222
Nets
212
Pads
909
Tracks, vias
2,178, 418
Pours
8
Routed
Yes
KiCad
KiCad 10.0 (file 20260206)
Created
13 Sep 2026
Updated
13 Sep 2026

Overview

Dronemagnav2 is an open-source printed circuit board by ajisrawi, from the ajisrawi/DroneMagNav2.0 repository on GitHub. It is a 4-layer board of 128 × 84 mm with 222 parts on 212 nets, routed with 2,178 tracks and 418 vias.

It is built around the ESP32-S3-WROOM-1-N16R8, MS5611-01BA03 and TPS65131RGER, with the AP2112K-3.3, LT3045EDD and AMS1117-3.3 among its other parts. Its parts include 101 capacitors, 73 resistors, 23 ICs, 8 diodes and 7 connectors. It belongs with the ESP32 boards of this library.

Main components

12

README

README.md

DroneMagNav 2.0 — survey-fluxgate magnetic-anomaly navigation board

GNSS-denied positioning from the Earth's crustal magnetic field, measured with a survey-grade three-axis fluxgate (no magnetometer on the board), plus the phone-operated great-circle route planner from v1.

text
 survey fluxgate on a mast ─ J2 ─ 3x THS4551 ─ 3x ADS1262 32-bit ─┐        ┌─ IGRF-14 core field (degree 13)
   (SENSYS FGM3D/100 LN, ±10 V, ≤8 pT/√Hz)   (shared clock+START) │        ├─ EMAG2v3 2′ anomaly grid (4 km)
 optional 2nd fluxgate ─ J7 ─ 3x THS4551 ─ AIN2/AIN3 (gradiometer)│        │    or local survey grid
 BMI088 IMU ─────────────────────────────────────── ESP32-S3 ─────┼─ map ──┤
 MS5611 baro ───────────────────────── particle filter + integrity│ engine └─ both continued to the drone's
 u-blox MAX-M10S (init, truth, geoid, date) ──────────────────────┘           altitude above WGS84
                                   │
                                   ├─ MAVLink GPS_INPUT → flight controller
                                   └─ WiFi AP → phone: offline map, route, status, integrity, logging

What changed from v1

v1 (DroneMagNav)2.0
MagnetometerLIS3MDL on the board (~320 nT rms raw)none on board; external survey fluxgate on J2 (≤8 pT/√Hz) digitised by 3× ADS1262
Core fieldWMM2025 (degree 12)IGRF-14 (degree 13, epoch 2025.0, SV to 2030)
Anomaly mapWDMAM 3′ at fixed 5 kmEMAG2v3 2′ (3.7 km cells) at 4 km, continued to the drone's altitude (FFT, gain-limited downward continuation); optional local survey with its own reference height
Altitude referenceBMP280, MSLMS5611 (~0.1 m), referenced to GNSS, converted to ellipsoid height with the GNSS geoid separation
Power5 V from USB/FC2S–6S battery (XT30) → 5 V buck; USB-C fallback; ±13.5 V converter → ultra-low-noise ±12 V sensor and +5 V analog rails
Integritynonechi-square gate with bias-step handling, trajectory-consistency check against dead reckoning, optional gradiometer gate (second fluxgate on J7)
Board64 × 60 mm, 4 layers128 × 84 mm, 4 layers, split analog/digital power plane

Deliverables

ItemWhereStatus
Schematics (9 sheets + root)hardware/dronemagnav2/*.kicad_sch, dronemagnav2-schematic.pdfERC 0 errors, 2 intended warnings (shared SPI MISO)
Netlist verificationtools/check_netlist.py77 critical nets + "no analog pin on a rail" check: PASS
PCB 128 × 84 mm, 4-layerhardware/dronemagnav2/dronemagnav2.kicad_pcbsee PCB status below
BOMdocs/BOM.csv74 line items, 218 parts (38 optional gradiometer parts, column Fit), MPNs
Firmware (ESP32-S3, PlatformIO)firmware/builds, 0 warnings (1.10 MB flash, 129 KB RAM)
Integrity checkstools/integrity_check.pynumpy port vs 4 injected faults, 5 seeds: PASS (figure docs/integrity-sim.png)
Map data pipelinetools/prepare_sd.pyEMAG2v3 GeoTIFF → 116.6 MB MAG2 grid (read-back verified); IGRF-14 → .COF
Core-field validationtools/geomag_check.pyWMM2025 official test values 100/100 (worst 0.0007 nT); IGRF-14 vs independent model 200/200 (worst 0.002 nT)
Continuation validationtools/continuation_check.pyanalytic dipoles + EMAG2v3 round trip: PASS (figure docs/continuation-emag2.png)
Fluxgate calibrationtools/fluxgate_cal.py12-parameter fit, self-test PASS

Selected sensor

SENSYS FGM3D/100, low-noise option: ±100 µT, ≤8 pT rms/√Hz (0.1–10 Hz), ±10 V at full scale (0.1 V/µT), ±12…±15 V at ±26 mA, 112 g, IP65, orthogonality option ±0.1°. It was chosen because its full specification is published, its noise meets the 0.01 nT class, it is used on survey drones (SENSYS MagDrone), and its analog interface is the de-facto standard. Bartington Mag-13MS (<5 pT/√Hz, ±10 V, ±12 V) plugs into the same connector with a different cable; set VOLTS_PER_NT in fluxgate.h for other ranges.

J2 (JST-GH 10): 1 +12 V · 2 −12 V · 3 sensor power GND · 4/5 X out/return · 6/7 Y · 8/9 Z · 10 cable shield. Make the cable to your sensor's connector pinout (check the sensor manual); keep the sensor ≥ 0.5 m from motors.

Optional gradiometer: a second fluxgate of the same type on J7 (same pinout), e.g. 0.5 m below the first on the mast. Its drivers U21–U23 feed the spare AIN2/AIN3 inputs of the same three ADS1262, which then alternate between the sensors in pulse-conversion mode. J7, U21–U23 and their passives are marked optional in the BOM; /cal/grad.txt turns the mode on (see firmware/README.md).

How it navigates

  1. Measure B vector at 400 SPS on three simultaneous 32-bit channels, average to 20 Hz, apply the fluxgate calibration.
  2. Compensate the total field for the drone's own magnetism (18-term Tolles-Lawson, fitted in flight with GNSS).
  3. Model core field = IGRF-14 at (lat, lon, ellipsoid height, date); anomaly = EMAG2v3 or local survey window continued to that height.
  4. Update a 600-particle filter with measured − (core + anomaly + bias); predict with flight-controller velocity.
  5. Check every update: gradiometer gate (B − A against the predicted gradient; nearby sources are steep, uniform fields cancel), chi-square gate on the innovation (a persistent offset becomes a bias step), and a trajectory-consistency check (a position jump that dead reckoning cannot explain is rolled back).
  6. Feed the estimate to the autopilot as GPS_INPUT when GNSS is lost.

Expectations, honestly

  • The sensor chain is ~30 pT rms (estimate) — no longer the limit.
  • EMAG2v3 only: the map limits accuracy. Its 3.7 km cells and 4 km compilation height hold no short-wavelength detail; continuation puts what it has at the right height but cannot add detail. Expect hundreds of metres to ~1 km over strong continental anomalies, worse over flat or ocean areas.
  • ~10 m is the target with a local aeromagnetic survey (e.g. Geoscience Australia 40–80 m grid, NRCan 200 m, GTK Finland) loaded as local.bin, a good compensation fit and a mast-mounted sensor. It has not been demonstrated.
  • Disturbances: in simulation over a 100 m survey layer (27 m fault-free), a steel structure beside the track, a +60 nT field step and a current loop at the sensor each pushed the unchecked filter 0.3–3 km off; with the checks the peak error stayed at 39–84 m. A field that imitates another place on the map is only bounded (~400 m). Details in firmware/README.md.

Build / flash / prepare

See firmware/README.md. Short version: pio run -t upload, then python tools/prepare_sd.py --emag2 <EMAG2v3 tif> --igrf igrf14coeffs.txt --coast <geojson> and copy firmware/sdcard/* to the card.

Hardware regeneration: python tools/design.py → kicad-cli sch erc / export netlist → python tools/check_netlist.py → KiCad python tools/make_pcb.py → tools/route.py export/route/import → tools/widen_power.py → tools/route.py drc → python tools/make_bom.py → python tools/make_fab.py → python tools/make_design_doc.py.

PCB status

  • Placed by tools/make_pcb.py (260 plane pads fanned out before routing), autorouted with Freerouting (tools/route.py), supply tracks widened (tools/widen_power.py: 181 segments), last two connections (BMI088 +3.3VA, gradiometer driver U21 OUT−) hand-routed by tools/manual_routes.py.
  • KiCad DRC: 0 unconnected pads, 0 electrical violations; 74 silkscreen notes (reference text over / overlapping copper, cosmetic — fabs clip silk on pads).
  • Fab package in hardware/dronemagnav2/fab/ (Gerbers, drill, pick-and-place, notes, ZIP); renders board-top.png / board-bottom.png; design document DroneMagNav2-Design-Document.pdf.

Caveats

  • Prototype; not certified for any regulated use.
  • Not yet built or run on hardware.
  • LT3045/LT3094 application values follow the typical-application pin guidance (ILIM to GND, PGFB to IN, PG/VIOC open); the ADI datasheets could not be downloaded in this session (server timeouts) — verify before ordering.
  • Datasheets used for pinouts and values: TI ADS1262, THS4551, TPS65131, TPS62933, TPS2116; TE MS5611 (in data/, not redistributed).

Parts

222
QtyValueFootprintReferences
28100nFC_0603_1608MetricC4, C11, C14, C15, C39, C40, C44, C45, C49, C50, C55, C57, C62, C64, C69, C100 +12
1710KR_0603_1608MetricR4, R5, R37, R38, R39, R40, R41, R42, R43, R47, R48, R49, R50, R51, R52, R53 +1
161uFC_0603_1608MetricC7, C12, C53, C54, C56, C58, C60, C61, C63, C65, C67, C68, C70, C72, C73, C76
13100RR_0603_1608MetricR8, R21, R22, R27, R28, R33, R34, R60, R61, R66, R67, R72, R73
121.00K 0.1%R_0603_1608MetricR19, R20, R25, R26, R31, R32, R58, R59, R64, R65, R70, R71
1210uFC_0805_2012MetricC8, C9, C10, C13, C35, C36, C74, C77, C82, C83, C85, C86
124.99K 0.1%R_0603_1608MetricR17, R18, R23, R24, R29, R30, R56, R57, R62, R63, R68, R69
1247nF C0GC_0603_1608MetricC37, C38, C42, C43, C47, C48, C87, C88, C92, C93, C97, C98
610nF C0GC_0603_1608MetricC41, C46, C51, C91, C96, C101
610uF 25VC_1206_3216MetricC18, C19, C20, C21, C29, C32
6THS4551IRGTRVQFN-16-1EP_3x3mm_P0.5mm_EP1.68x1.68mmU9, U10, U11, U21, U22, U23
54.7uFC_0805_2012MetricC16, C17, C28, C31, C34
41KR_0603_1608MetricR7, R44, R45, R46
44.7nF C0GC_0603_1608MetricC25, C52, C59, C66
4M3MountingHole_3.2mm_M3H1, H2, H3, H4
3100KR_0603_1608MetricR9, R11, R13
34.7uF 25VC_0805_2012MetricC27, C30, C33
3ADS1262IPWRTSSOP-28_4.4x9.7mm_P0.65mmU12, U13, U14
2120KR_0603_1608MetricR14, R15
222uF 25VC_1206_3216MetricC5, C6
233RR_0603_1608MetricR35, R36
24.7uHL_Bourns-SRN4018L2, L3
25.1KR_0603_1608MetricR54, R55
2AP2112K-3.3SOT-23-5U4, U20
2B5819WD_SOD-123D4, D5
2LT3045EDDDFN-10-1EP_3x3mm_P0.5mm_EP1.65x2.38mmU6, U8
11.02MR_0603_1608MetricR10
11.10MR_0603_1608MetricR12
1100nF 50VC_0603_1608MetricC2
110nFC_0603_1608MetricC26
110uF 50VC_1206_3216MetricC1
1220nFC_0603_1608MetricC24
13.3uHL_Coilcraft_XAL4030-XXXL1
133nFC_0603_1608MetricC3
1340KR_0603_1608MetricR1
149.9KR_0603_1608MetricR16
152.3KR_0603_1608MetricR3
16.8pF C0GC_0603_1608MetricC22
17.3728MHzOscillator_SMD_Abracon_ASE-4Pin_3.2x2.5mmX1
17.5pF C0GC_0603_1608MetricC23
175KR_0603_1608MetricR2
1AMS1117-3.3SOT-223-3_TabPin2U3
1BATT_2S-6SAMASS_XT30PW-M_1x02_P2.50mm_HorizontalJ1
1BMI088Bosch_LGA-16_4.5x3mm_P0.5mm_LayoutBorder7x1y_ClockwisePinNumberingU16
1BOOTSW_SPST_PTS645Sx43SMTR92SW2
1ESP32-S3-WROOM-1-N16R8ESP32-S3-WROOM-1U15
1FC_TELEMJST_GH_SM06B-GHS-TB_1x06-1MP_P1.25mm_HorizontalJ6
1FIXLED_0603_1608MetricD7
1FLUXGATEJST_GH_SM10B-GHS-TB_1x10-1MP_P1.25mm_HorizontalJ2
1FLUXGATE_BJST_GH_SM10B-GHS-TB_1x10-1MP_P1.25mm_HorizontalJ7
1GNSS_ANTU.FL_Hirose_U.FL-R-SMT-1_VerticalJ3
1LINKLED_0603_1608MetricD8
1LT3094EDDDFN-12-1EP_3x3mm_P0.45mm_EP1.65x2.38mm_ThermalViasU7
1MAX-M10Sublox_MAXU18
1MS5611-01BA03LGA-8_3x5mm_P1.25mmU17
1NAVLED_0603_1608MetricD6
1PWRLED_0603_1608MetricD3
1RESETSW_SPST_PTS645Sx43SMTR92SW1
1SMAJ26CAD_SMAD2
1SS34D_SMAD1
1TPS2116DRLRSOT-583-8U2
1TPS62933FDRLRSOT-583-8U1
1TPS65131RGERVQFN-24-1EP_4x4mm_P0.5mm_EP2.45x2.45mm_ThermalViasU5
1USB-CUSB_C_Receptacle_HRO_TYPE-C-31-M-12J5
1USBLC6-2SC6SOT-23-6U19
1microSDmicroSD_HC_Hirose_DM3AT-SF-PEJM5J4

Similar boards

  • 2 layers80 × 60 mmMEJ2D1205SC, the printed circuit board in 3D

    MEJ2D1205SC

    Pato01010/ECE455_MAG5_captsone-project

    46 parts1
  • 4 layers50 × 100 mmFlight computer one, the printed circuit board in 3D

    Flight computer one

    liquid-rocketry-illinois/flight-computer-one-hw

    SAME51-based flight computer with active control capabilities

    105 partsCERN-OHL-S-2.0
  • 2 layers21 × 54 mmSensorBoardNoMCU, the printed circuit board in 3D

    SensorBoardNoMCU

    Terrapin-Rocket-Team/SRAD-Avionics

    Code and PCBs for the avionics subteam for the SRAD flight computer and side projects

    29 parts17
  • 4 layers64 × 60 mmDronemagnav, the printed circuit board in 3D

    Dronemagnav

    ajisrawi/dronemagnav

    Magnetic-anomaly navigation (MagNav) board + firmware for drones: ESP32-S3, BMI088, LIS3MDL, BMP280, u-blox GNSS, WDMAM map matching, particle filter, MAVLink, phone route planner. Proprietary - all rights reserved.

    60 parts11
  • 4 layers90 × 90 mmPCM1704, the printed circuit board in 3D

    PCM1704

    smileym4n-collab/Kicadprojects

    143 parts
  • 4 layers80 × 120 mmUAV MainBoard V3, the printed circuit board in 3D

    UAV MainBoard V3

    MennoScholten/UAV_Project

    Design and development of a UAV

    231 parts
  • 2 layers77 × 32 mmBananaSchplit, the printed circuit board in 3D

    BananaSchplit

    pms67/BananaSchplit

    Single rail DC (<12V) to split rail DC supply (+/-12V and +5V).

    53 parts66MIT
  • 4 layers100 × 100 mmPhmod, the printed circuit board in 3D

    Phmod

    gallicchio/AMOD-and-RFMOD

    The tutorial's 2x12 adapter (/mudd/133electronics/icepi-zero-dac-adc/adapterboard/) with a 2 x 6 right-angle Pmod socket on a 15 mm extension of its right-hand end, so an Icepi Zero can drive AMOD-A's and RF-MOD's Pmod…

    234 parts

Questions

How big is the Dronemagnav2 PCB?

It measures 128 × 84 mm and has 4 copper layers, 1.6 mm thick.

What parts does the Dronemagnav2 use?

222 parts, built around the ESP32-S3-WROOM-1-N16R8, MS5611-01BA03, TPS65131RGER and AP2112K-3.3. 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 Dronemagnav2 KiCad files?

From the ajisrawi/DroneMagNav2.0 repository on GitHub, at the commit this page was made from: the board is hardware/dronemagnav2/dronemagnav2.kicad_pcb and its schematic hardware/dronemagnav2/adc.kicad_sch. PCBKit opens the board in the browser, with nothing to install.

Can I use the Dronemagnav2 design in my own project?

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