VOLANTI: AN ESP32 BOX THAT HEARS DRONES BY THEIR PROPELLERS, SO FIBRE-OPTIC FPV CANNOT HIDE
A box that listens for propeller harmonics instead of radio — the gap fibre-optic FPV leaves wide open.
by Agam Rossen
Open-hardwareAutomation
Built withESP323D printing
- difficulty
- ●●●○○
- time
- a weekend
- license
- Apache-2.0 (software), CERN-OHL-W-2.0 (hardware), CC-BY-SA-4.0 (docs)
- repo
- repo ACTIVE403 stars
●●●○○ · a weekend · Apache-2.0 (software), CERN-OHL-W-2.0 (hardware), CC-BY-SA-4.0 (docs) · 403 stars · repo ACTIVE
WHAT YOU’LL NEED
Jump to section
COMPAREE VERDICT
VolAnti fills a real gap: most drone detection listens for the radio link between aircraft and pilot, but fibre-optic FPV drones trail a spool of optical fibre and transmit nothing at all, so radio detection finds nothing. Propellers leave a harmonic comb — blade-pass rate and every multiple — and this box listens for that instead. The hardware is an ESP32-S3 with four ICS-43434 MEMS microphones summed on one I2S clock. The 79 mm array buys about 6 dB of signal-to-noise for free but gives no directionality; the author tried beamforming and reports it is useless at this size. The firmware runs a 2048-point FFT every 32 milliseconds with four detectors per frame, pinned by golden test vectors so the same WAV always scores the same. In a field test it detected a hovering quad at 104.2 metres on a busy brick street, zero false alarms outdoors so far. An assembled unit is 50 to 80 pounds in parts with JLCPCB doing most of the work; a breadboard version is 35 to 45 pounds and an evening. The one thing most likely to go wrong is expectation: this detects, it does not track direction or jam, and 104 metres is one test against one known airframe, not a promise. If you want to know when a drone is near and you understand that acoustic detection has limits, this is a weekend well spent.
IN THE REPO
GOOD TO KNOW
- —Gerbers, BOM with LCSC part numbers, STLs, firmware, and golden test WAV files are all present.
- —Apache-2.0 for software, CERN-OHL-W-2.0 for hardware, CC-BY-SA-4.0 for docs — all permit commercial use.
- —JLCPCB assembly quote covers all but four through-hole components; breadboard route is fully documented.
- —Detection only — no jamming, no countermeasures, no directional tracking at this array size.
- —104.2 m detection is one field test against a known quad, not a certified range, and zero false alarms is the author's outdoor count so far.
- —The browser simulator at volantitech.com runs the same detector logic live; you can test with your own audio before building hardware.
Parts to buy
4 itemsFrom our check of the build. Exact quantities and part numbers are in the creator’s BOM.
Can I build this?
Build at your own risk. Projects involve tools, electronics and sometimes mains voltage — follow the creator’s safety notes.
More builds like this
All projectsGallery
Start here
Navigation into the creator’s own docs — we don’t rewrite the guide, we route you to the source.
- 1.Read the README and run the browser simulator at volantitech.com with your own audio to see detection logic live before buying parts. (The simulator runs the same FFT and detectors; if you cannot hear the harmonics, the hardware will not either.)
- 2.Pick breadboard or assembled PCB route — breadboard is faster and cheaper for proof-of-concept, assembled is cleaner for field deployment. (BOM has LCSC part numbers and JLCPCB assembly quotes for the PCB route; breadboard parts list is in the same section.)
- 3.Flash the firmware — ESP-IDF v6.0 with idf.py, instructions are in docs/flashing.md. (Golden test WAV files are in the repo; flash, play a test file, and verify the score matches the expected output before field testing.)
- 4.Field test with a known quad — the 104.2 m result was four 2807 motors on 7-inch tri-blades, hovering, in a noisy urban environment. (Start with line-of-sight and known propeller specs; acoustic detection range collapses with wind, foliage, or building reflections.)
Resources
Documentation, files and community threads for this build — we link straight to the original sources and never rehost the creator’s files.
KNOWN ISSUES
- The 79 mm array gives no directional information — it detects presence, not bearing, and beamforming is explicitly stated as useless at this size.
- 104.2 metres is one field test against a hovering quad with known motor and prop specs on a brick street; detection range will be shorter in wind, with foliage, or against smaller or quieter props.
- Zero false alarms is the author's count from field sessions so far, not an independent evaluation. Keep the unit a few metres clear of air conditioning, generators and transformers: the author says they will not trigger it, but they raise the noise floor and cut the detection range.
- Detection only — this is not a jammer, not a tracker and not a countermeasure; the author is explicit that it will never be one. It tells you a drone is near, not where it is or how to stop it.
- The ICS-43434 microphones are I2S and require correct clock and data wiring; a miswired mic will pass firmware checks but produce garbage FFT output.
- Golden test vectors pin detection to known WAV files; if you modify the FFT size, sample rate, or detector thresholds, you must re-validate with field recordings of known aircraft.
Can it tell me which direction the drone is coming from?
No — the 79 mm array is too small for beamforming, and the author tried it and reports it is useless at this size. It detects presence, not bearing.
Will it jam or stop the drone?
No — detection and alert only. The author is explicit that there is no jamming, no interception and no countermeasure of any kind; the only radio is a LoRa link that passes alerts between units.
What counts as a fibre-optic FPV drone?
An aircraft that trails a spool of optical fibre for video downlink and transmits no radio signal at all, so RF-based drone detection sees nothing. The propeller harmonics remain.
Can I use different microphones?
The firmware expects I2S MEMS mics with the ICS-43434 pinout and sensitivity; swapping to analog or PDM mics requires firmware changes and re-validation with golden test vectors.
How do I know it is working before I take it outside?
Flash the firmware, play one of the golden test WAV files from the repo, and verify the detection score matches the expected output. If the score is wrong, check I2S wiring first.
What happens in wind or rain?
Wind is the big one: the author reports range moving by an order of magnitude with wind, and above 8 m/s detection should be assumed degraded. The full unit is a sealed box with grille cloth and silica sachets, so check the sachets at service time — a saturated one means the seal has failed.
Community builds
No community builds yet — be the first, we feature the best ones.
Discussion1
FROM THE COMPAREE TEAM
104.2 metres on a brick street with cars passing — and the author says beamforming is useless at 79 mm, so it detects but does not track. Would you deploy this as-is, or would you want a second unit for triangulation first?
Agam Rossen
Agam Rossen designed and tested VolAnti at the University of York to address the gap fibre-optic FPV drones leave in radio-based detection. The repository went public in September 2026 with full hardware, firmware and test data.
Keep building — related projects
Same category or platform, picked by how close they are to this build.
FULL CADA do-it-yourself fingerprint unlock for a Mac, built to avoid Apple's 149 dollar Touch ID keyboard
Open-hardwareFor intermediate builders
FULL CADYou can build a new brain for an e-bike its maker abandoned
AutomationFor intermediate builders
FULL CADA braille display that costs a fraction of the commercial ones because it uses a single moving cell
Open-hardwareFor advanced builders
FULL CADA card in a spare PCIe slot presses your PC power button for you, from anywhere
AutomationFor advanced builders
FULL CADA 350 dollars machine prints books blind kids can read — the shop version costs 2,500 dollars
Open-hardwareFor advanced builders
FULL CADA module that reads your electric car directly, so the tracking and preheating do not go through the manufacturer
AutomationFor advanced builders
DISCLAIMER
- Comparee is not the author of the projects featured here. All rights to each project belong to its creator — every page links to the original source, and we never host creators’ files.
- Information is provided without warranty and may become outdated as projects evolve. Prices are indicative bands only — always check the creator’s parts list for current costs.
- Building and operating any project is at your own responsibility. Protective equipment, safe workshop practice and compliance with local regulations are the builder’s responsibility.
CompareeTEAM28d agoedited
Practical notes from our verification: the browser simulator at volantitech.com runs the live detector logic and is the fastest way to test before buying parts. The 104.2 m field test was against a hovering rig with four 2807-class motors on 7-inch three-blade propellers, the same airframe class as fibre-optic FPV drones, but that is one test in one environment; the author says range moves by an order of magnitude with wind, from 100 to 200 m in still air down to much less on a breezy or noisy site. The author is unusually explicit about what does not work — beamforming with the 79 mm microphone array buys nothing, so the unit detects but does not find direction — and that honesty makes the project more trustworthy, not less. Golden test vectors pin the behaviour of the first detection tier, so if you modify it, re-validate against them and against field recordings. Correction (4 October 2026): we re-checked this page line by line against the project's own repository, documentation and videos, and fixed errors in earlier versions.