Open experimental platform for detecting and studying cosmic-ray particles
First experiment: ground-based cosmic muon detector Status: PLANNING / PROCUREMENT Target hardware readiness: June 2027
CANFLY Cosmic Node is an open experimental project for building a small, reproducible scientific instrument capable of detecting particles originating from cosmic-ray interactions.
The first experiment is intentionally simple:
Detect cosmic-ray muons at ground level, record individual events, characterize detector noise, and build a scientifically usable dataset.
The project does not claim to detect dark matter, dark energy, or particles originating directly from a particular distant astronomical object.
The first objective is much more fundamental:
Learn to build, calibrate, operate and analyze a real particle detector.
Once this capability is established, the same measurement architecture can be expanded into a directional muon telescope, a multi-detector cosmic-ray network, and eventually a spaceborne instrument.
Primary cosmic rays continuously interact with nuclei in Earth's atmosphere.
These interactions produce secondary particle cascades containing pions and kaons, whose decays produce muons.
Many of these muons have sufficient energy to reach the Earth's surface.
Muon detection therefore provides a direct and practical way to build an experimental connection between a laboratory on Earth and high-energy processes occurring in the atmosphere and beyond.
A typical ground-level cosmic muon has an energy of several GeV. CAEN's educational documentation describes an average energy of approximately 4 GeV at sea level and places typical muon production high in the atmosphere, around 15 km.
Important distinction:
The detector does not tell us that a particular muon came from a particular star or galaxy.
The first experiment measures the local flux of secondary cosmic-ray muons.
Sources:
- CAEN Educational — Muons Detection
- CAEN Educational — Muons Vertical Flux
- Particle Data Group references cited by CAEN
The initial scientific question is deliberately modest:
Can a compact, low-cost detector based on plastic scintillators and SiPMs continuously identify cosmic-ray muon candidates while suppressing detector noise and random coincidences?
After this has been demonstrated, additional questions become possible:
- What is the measured muon count rate?
- How stable is the rate over hours, days and months?
- How does the rate depend on detector geometry?
- How does it depend on atmospheric pressure?
- How does it depend on altitude?
- How does it depend on detector orientation?
- What is the detector's efficiency?
- What is the accidental coincidence rate?
- Can particle trajectories be reconstructed?
- Can multiple geographically separated detectors identify correlated events?
- Can the system eventually be adapted for spaceflight?
This project is not initially:
- a dark-matter detector;
- a dark-energy detector;
- a gamma-ray observatory;
- a neutrino detector;
- an astronomical telescope;
- a source-identification system for individual cosmic rays;
- a replacement for professional cosmic-ray observatories.
The project should not make claims stronger than the measurements support.
The first successful result is simply:
We detected cosmic-ray particle events with a reproducible instrument and quantified the detector response.
That is already a real experimental result.
The first detector consists of two independent scintillation detector modules.
Each module contains:
Cosmic particle
↓
Plastic scintillator
↓
Scintillation photons
↓
SiPM
↓
Preamplifier / readout
↓
Discriminator
↓
Digital pulse
The two modules are positioned one above the other.
A particle passing through both detectors produces approximately:
TOP ─────────●
│
│ particle
│
BOTTOM ─────────●
The electronics searches for two detector pulses occurring within a predefined coincidence window.
The event is accepted when:
TOP = 1
BOTTOM = 1
Δt < coincidence_window
This is important because individual SiPMs generate spontaneous dark-count pulses and other electronic noise.
A coincidence between two independent detectors strongly suppresses random single-detector events.
CAEN explicitly uses double scintillator coincidence for cosmic-muon detection and describes its role in reducing spurious events and selecting the detector's effective solid angle.
Initial architecture:
COSMIC PARTICLE
↓
┌─────────────────┐
│ SCINTILLATOR #1 │
│ + SiPM │
└────────┬────────┘
│
analog pulse
│
┌─────▼─────┐
│ READOUT #1│
└─────┬─────┘
│
DISCRIMINATOR
│
│
┌────▼────┐
│ │
│ DAQ │
│ │
│ STM32 │
│ │
└────┬────┘
│
USB / UART
│
Raspberry Pi
│
DATA STORAGE
│
┌────────────┴────────────┐
│ │
RAW DATA ANALYSIS
The second detector has the same architecture:
SCINTILLATOR #2
↓
SiPM
↓
READOUT #2
↓
DISCRIMINATOR
↓
STM32
The final trigger is based on coincidence between the two detector channels.
Quantity:
2
Type:
plastic scintillator
Candidate material:
- EJ-200
- EJ-204
- equivalent plastic scintillator
Initial target dimensions:
approximately:
50 × 50 × 10 mm
Exact dimensions may be changed after component selection.
The two scintillators should preferably be identical.
Quantity:
2
Type:
SiPM
Target:
approximately 6 × 6 mm active area.
The initial implementation should preferably use commercially available SiPM modules or readout boards rather than designing the complete SiPM bias/readout circuit from scratch.
The purpose of the first prototype is to establish the physical experiment before optimizing the electronics.
Quantity:
2
Required functions:
- SiPM bias
- signal extraction
- amplification
- suitable output for oscilloscope/discriminator
The exact board will be selected according to the chosen SiPM.
Compatibility between:
SiPM
+
bias voltage
+
preamplifier
+
connector
+
oscilloscope/discriminator
must be verified before purchase.
Required:
2-channel discriminator
Function:
Convert analog detector pulses into digital events.
The threshold must be adjustable.
The threshold is an experimental parameter and will be calibrated rather than arbitrarily selected.
The detector must characterize:
threshold
↓
single-channel count rate
↓
coincidence count rate
↓
random coincidence rate
CAEN specifically demonstrates measuring dark-count rate as a function of discriminator threshold and adjusting the threshold to suppress random coincidences.
Initial choice:
STM32
The STM32 is responsible for:
- detecting digital pulses;
- measuring event timing;
- counting single-channel events;
- identifying coincidences;
- generating event records;
- communicating with the host computer.
A 2026 CERN conference contribution describes a compact SiPM cosmic-ray system using an STM32 microcontroller and hardware timers for real-time event counting and coincidence processing, making STM32 a reasonable architecture for this class of instrument.
Quantity:
1
Purpose:
The Raspberry Pi is the scientific data logger rather than the fast detector.
Responsibilities:
- receive events from STM32;
- maintain long-term storage;
- maintain local database;
- collect environmental measurements;
- synchronize files;
- generate monitoring plots;
- perform preliminary analysis;
- create backups.
Quantity:
1
Initial purpose:
- UTC time reference;
- geographic position;
- optional 1-PPS timing signal.
A normal GNSS receiver is sufficient for the first experiment.
A precision timing architecture can be introduced later.
Initial sensor:
BME280 or equivalent
Measurements:
- temperature;
- atmospheric pressure;
- humidity.
Atmospheric pressure is particularly important because the amount of atmosphere above the detector affects the secondary cosmic-ray population reaching the ground.
Environmental parameters therefore belong in the scientific dataset rather than being treated as irrelevant housekeeping.
Quantity:
1
Possible devices:
- BMM150
- LIS3MDL
- equivalent 3-axis magnetometer
Purpose:
Record the local magnetic environment.
It is not initially used as a dark-matter detector.
It is an auxiliary variable for later correlation analysis and environmental characterization.
Required:
adjustable laboratory DC power supply with current limiting.
If an appropriate laboratory power supply is already available, a new unit is unnecessary.
Stable detector bias is important for reproducible SiPM operation.
Recommended minimum:
100 MHz bandwidth
Preferably:
≥1 GS/s
Purpose:
The oscilloscope is primarily a development and calibration instrument.
It allows direct observation of:
- SiPM pulses;
- amplifier output;
- baseline noise;
- pulse amplitude;
- pulse width;
- discriminator threshold;
- electrical interference;
- coincidence timing.
The oscilloscope is not expected to be part of the final autonomous detector.
Required:
optical coupling grease
Purpose:
Improve optical coupling between the scintillator and SiPM.
The optical interface must remain mechanically stable and reproducible.
Required as appropriate to the selected electronics:
- BNC cables;
- SMA cables;
- MCX adapters/cables;
- 50 Ω terminators;
- suitable coaxial connectors.
Exact connector requirements will be finalized after the SiPM/readout and discriminator models are selected.
The detector must be protected from ambient light.
Possible construction:
scintillator
+
SiPM
↓
light-tight wrapping
↓
mechanical enclosure
The enclosure must allow:
- optical isolation;
- mechanical stability;
- access for calibration;
- cable routing;
- reproducible detector geometry.
The first detector should have adjustable spacing.
Example:
┌───────────────┐
│ SCINTILLATOR 1│
└───────────────┘
│
│ adjustable
│ distance
│
┌───────────────┐
│ SCINTILLATOR 2│
└───────────────┘
A 3D-printed frame is acceptable.
The geometry must be documented precisely.
Initial procurement list:
| Component | Quantity | Required |
|---|---|---|
| Plastic scintillator | 2 | YES |
| SiPM | 2 | YES |
| SiPM readout/preamplifier | 2 | YES |
| 2-channel discriminator | 1 | YES |
| STM32 development board | 1 | YES |
| Raspberry Pi | 1 | YES |
| GNSS receiver | 1 | RECOMMENDED |
| BME280/environment sensor | 1 | RECOMMENDED |
| 3-axis magnetometer | 1 | RECOMMENDED |
| Laboratory power supply | 1 | YES* |
| Oscilloscope | 1 | YES* |
| Optical coupling grease | 1 | YES |
| Coaxial cables/adapters | set | YES |
| 50 Ω terminators | 2–3 | RECOMMENDED |
| Light-tight material | set | YES |
| 3D-printed mechanical frame | 1 | YES |
* Only if suitable equipment is not already available.
The first successful measurement should produce something like:
TIME TOP BOTTOM COINCIDENCE
2027-06-xx 12:00:01 124 119 8
2027-06-xx 12:01:01 130 125 9
2027-06-xx 12:02:01 127 121 7
...
The exact rates are not assumed in advance.
The detector must measure them.
A scientifically useful result is not simply:
"The detector counted particles."
It is:
"The detector measured a stable coincidence population that is distinguishable from independently measured single-channel noise and accidental coincidence background."
Before hardware assembly:
- freeze the initial detector design;
- record every component;
- record manufacturer and part number;
- record SiPM characteristics;
- record scintillator dimensions;
- record wiring;
- record firmware version;
- create the data schema;
- create calibration procedures.
No scientific measurement should exist without a record of the instrument configuration that produced it.
Before attaching the scintillator:
SiPM
↓
readout
↓
oscilloscope
Measure:
- baseline;
- dark-count rate;
- pulse amplitude distribution;
- response to threshold changes;
- temperature dependence.
This establishes the detector's electronic baseline.
Build one complete detector.
Measure:
single-channel count rate
vs.
discriminator threshold
The goal is to understand the detector before attempting coincidence measurements.
Build the second detector.
Measure independently:
R1 = detector 1 rate
R2 = detector 2 rate
Then activate coincidence:
R12 = coincidence rate
This is mandatory.
Two detectors can generate accidental coincidences even when no particle traverses both.
The experiment must therefore estimate:
R_random
and compare it with:
R_cosmic_candidate
CAEN has a dedicated experiment specifically for random coincidence measurements because this effect is fundamental to coincidence detectors.
Place the two detectors vertically aligned.
Record:
single rate #1
single rate #2
coincidence rate
temperature
pressure
humidity
timestamp
detector configuration
Run continuously.
Initial target:
24-hour continuous dataset
Then:
7 days
Then:
30 days
After basic detection works, change the geometry systematically.
┌───────────────┐
│ #1 │
└───────────────┘
┌───────────────┐
│ #2 │
└───────────────┘
Purpose:
maximize acceptance.
┌───────────────┐
│ #1 │
└───────────────┘
↑
distance
↓
┌───────────────┐
│ #2 │
└───────────────┘
Purpose:
reduce the accepted solid angle and make the telescope more directional.
CAEN explicitly describes changing detector separation as a way to modify the solid angle and directional selection.
Rotate the entire detector.
Measure:
count rate
vs.
zenith angle
This begins transforming the system from a simple particle counter into a directional cosmic-ray instrument.
Every long measurement should include:
timestamp
temperature
pressure
humidity
magnetic field
detector voltage
threshold
detector geometry
Later analysis can test:
muon rate
vs.
atmospheric pressure
and:
muon rate
vs.
temperature
and:
muon rate
vs.
orientation
The purpose is not to assume a correlation exists.
The purpose is to measure whether it exists.
Every detected event should receive a unique identifier.
Example:
{
"event_id": 184729,
"timestamp_utc": "2027-06-18T12:43:17.382941Z",
"detector_top": true,
"detector_bottom": true,
"delta_t_ns": 37,
"top_amplitude": 0.412,
"bottom_amplitude": 0.387,
"temperature_c": 22.4,
"pressure_hpa": 1008.2,
"humidity_percent": 48.1,
"magnetic_field_ut": 48.7,
"threshold_top": 31,
"threshold_bottom": 31,
"firmware_version": "0.1.0",
"configuration_id": "MUON-V1-001"
}The exact schema will evolve.
Raw data should never be destroyed after calibration.
The repository should distinguish:
raw/
calibrated/
derived/
analysis/
plots/
metadata/
Data exactly as produced by the detector.
Data after applying documented calibration constants.
Calculated quantities such as:
- rates;
- coincidence rates;
- efficiencies;
- pressure-corrected quantities;
- angular distributions.
Reproducible analysis code.
Generated figures.
Instrument configuration and calibration information.
Recommended structure:
canfly-cosmic-node/
│
├── README.md
│
├── docs/
│ ├── physics.md
│ ├── detector.md
│ ├── electronics.md
│ ├── calibration.md
│ ├── experiments.md
│ └── data-format.md
│
├── hardware/
│ ├── bom/
│ ├── schematics/
│ ├── pcb/
│ └── mechanical/
│
├── firmware/
│ └── stm32/
│
├── software/
│ ├── collector/
│ ├── database/
│ └── monitoring/
│
├── analysis/
│ ├── calibration/
│ ├── coincidence/
│ ├── flux/
│ └── environmental/
│
├── data/
│ └── README.md
│
├── notebooks/
│
├── experiments/
│ ├── 001-single-detector/
│ ├── 002-two-detector/
│ ├── 003-coincidence/
│ ├── 004-muon-rate/
│ ├── 005-pressure/
│ └── 006-angular-response/
│
└── LICENSE
Large datasets should not necessarily be stored directly in Git.
Every experiment should have a unique identifier.
Example:
EXP-001
with:
date
location
detector configuration
scintillator dimensions
SiPM model
bias voltage
threshold
coincidence window
detector spacing
orientation
environment
firmware version
software version
operator
duration
raw-data location
This makes the experiment reproducible.
The project follows several rules.
Never call an unexplained event "new physics".
First eliminate:
- electronic noise;
- SiPM dark counts;
- accidental coincidences;
- power-supply noise;
- electromagnetic interference;
- temperature effects;
- timing errors;
- software errors;
- mechanical changes;
- environmental correlations.
Never modify raw data.
Calibration creates a new dataset.
Every change to the detector configuration receives a new configuration ID.
Every firmware change receives a version.
Every analysis result must be reproducible from stored data.
Unexpected results must trigger additional measurements rather than immediate interpretation.
Version 1 is considered successful when the system can:
- continuously operate for at least 24 hours;
- record individual detector events;
- measure independent detector rates;
- measure coincidence events;
- quantify accidental coincidence background;
- demonstrate a statistically significant excess of physical coincidences over the measured accidental background;
- reproduce the result on another measurement run;
- store timestamped environmental data;
- preserve raw data;
- regenerate analysis plots from the stored dataset.
The project does not require discovering anything unknown to be considered successful.
Once v1 is stable, the project can evolve.
CANFLY COSMIC NODE
│
▼
┌─────────────────┐
│ MUON DETECTOR │
│ V1 │
└────────┬────────┘
│
▼
┌─────────────────┐
│ MUON TELESCOPE │
│ V2 │
└────────┬────────┘
│
▼
┌─────────────────┐
│ MULTI-DETECTOR │
│ NETWORK │
└────────┬────────┘
│
▼
┌─────────────────┐
│ SPACE-QUALIFIED│
│ CONCEPT │
└─────────────────┘
Potential future instruments include:
- three-layer muon telescope;
- four-layer particle tracker;
- cosmic shower detector;
- geographically distributed detector network;
- high-altitude detector;
- balloon experiment;
- CubeSat experiment.
Project definition.
Physics study.
Hardware selection.
Procurement.
Electronics architecture.
Software architecture.
Data format.
Simulation and analysis development.
Target:
first complete ground detector
Initial tasks:
power-on
↓
SiPM characterization
↓
single detector
↓
second detector
↓
coincidence
↓
first cosmic events
Stable 24-hour operation.
7-day continuous dataset.
30-day dataset.
Directional measurements.
Environmental correlation.
Detector efficiency.
Long-term stability.
The project deliberately follows established detector principles rather than inventing an untested detection method.
CAEN demonstrates cosmic-muon detection using a plastic scintillating tile directly coupled to a Silicon Photomultiplier.
The documentation also describes SiPM dark-count characterization and discriminator-threshold optimization.
CAEN demonstrates measuring the vertical cosmic-muon flux and comparing measured rates with expected flux.
CAEN provides a dedicated experiment for quantifying accidental coincidences between detector channels.
CAEN demonstrates efficiency measurements using multiple scintillator coincidence configurations.
CAEN demonstrates detection of cosmic-ray showers using multiple scintillator coincidences.
Recent detector-development work demonstrates compact plastic-scintillator/SiPM systems with discriminator electronics and STM32-based data acquisition.
The CosmicWatch project demonstrates a compact educational cosmic-ray detector based on a small plastic scintillator, SiPM and microcontroller.
The project begins with a deliberately small question:
Can we build a reliable instrument that sees particles we cannot see?
Then:
Can we measure them accurately?
Then:
Can we understand the detector well enough to trust the data?
Then:
Can multiple instruments see the same physical phenomenon?
Only after those questions have been answered should the project attempt more ambitious measurements.
The long-term goal is not to build a device that produces mysterious numbers.
The goal is to build a device whose numbers can be trusted, reproduced and independently analyzed.
Recommended repository:
canfly-cosmic-node
Possible alternatives:
canfly-cosmic
canfly-muon-node
canfly-cosmic-detector
canfly-cosmic-lab
cosmic-node
Recommended:
canfly-cosmic-node
because the project is intended to grow beyond the first muon detector.
[██████████] Project concept DONE
[██████████] Physics definition DONE
[██████████] Initial architecture DONE
[██████████] Initial BOM DONE
[░░░░░░░░░░] Component selection NEXT
[░░░░░░░░░░] Procurement PLANNED
[░░░░░░░░░░] Electronics PLANNED
[░░░░░░░░░░] Firmware PLANNED
[░░░░░░░░░░] Detector assembly PLANNED
[░░░░░░░░░░] Calibration PLANNED
[░░░░░░░░░░] First cosmic event PLANNED
[░░░░░░░░░░] Long-term dataset PLANNED
Target first complete ground experiment: June 2027.
The initial project design is based on established cosmic-ray and detector methods described by:
- CAEN Educational — Muons Detection
- CAEN Educational — Muons Vertical Flux on Horizontal Detector
- CAEN Educational — Random Coincidence
- CAEN Educational — Detection Efficiency
- CAEN Educational — Cosmic Shower Detection
- CERN / TIFR detector-development work on compact plastic-scintillator + SiPM systems
- CosmicWatch educational cosmic-ray detector project
All quantitative claims and detector parameters should be checked against the primary documentation before being frozen into the hardware design.