Thread 'YARROW'

Message boards : Puzzles and science : YARROW
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 · Next

AuthorMessage
Alperen Yavuz
Project developer
25 years old

Send message
Joined: 27 Sep 26
Posts: 349
Credit: 5,210,013
RAC: 137,017
Message 861 - Posted: 19 Oct 2026, 18:07:17 UTC
Last modified: 1 Nov 2026, 14:59:52 UTC

☢️ Yarrow: a live map of natural background radiation, built from volunteers' Geiger counters ☢️

Yarrow is the newest project on bitboinc, and it is a little different from the others: it does not crunch numbers for you, it measures something. You plug a Geiger counter into your computer, BOINC reads it, and every ten minutes a count of real radiation events goes to the server. Together the counts become a live, open map of the natural background dose rate, city by city, around the world. Here is what it does, how it keeps the data honest, and how to join.

🌍 What is background radiation, and why map it?

Everywhere on Earth, all the time, a little ionizing radiation passes through you: gamma rays from uranium, thorium and potassium in rocks and soil, from the building you sit in, radon in the air, and cosmic rays from space. It is harmless at these levels, and it is different from place to place:

  • The ground. Granite and some volcanic rocks contain more uranium and thorium than limestone or sand, so a granite town can read twice as high as a town on sediment.
  • The building. Stone and concrete add a little; wood and a wide-open field add less.
  • The altitude. Cosmic rays get stronger the higher you go, which is why a mountain village reads higher than the coast.
  • The weather. Rain washes radon decay products out of the air, which can lift readings for an hour or two.


Typical values are about 50 to 200 nanosieverts per hour (nSv/h). A dense network of cheap detectors, each watching its own corner, shows the normal level of every place, and, because it never sleeps, makes any real change stand out. That is the whole idea: a citizen network that knows what "normal" looks like, everywhere, so that unusual can be seen.

This idea is not new. Radioactive@Home was a BOINC project of the 2010s with its own USB counter. Its website is gone now, but the idea deserves to live on, so Yarrow picks it up with modern, cheap hardware.

🔬 How a Geiger counter measures

A Geiger-Müller tube is a small gas-filled glass tube with a high voltage (about 400 V) across it. When a gamma ray or a fast electron passes through, the gas conducts for a moment and the tube gives one electrical pulse. The counter just counts the pulses. At background level that is roughly one pulse every two seconds, which is why a single reading means little and many readings mean a lot.

Radioactive decay is random, so the count in any period varies by about the square root of the count (Poisson statistics). A ten minute bin at 0.2 µSv/h holds about 350 pulses, which is a 5% uncertainty at best. Averaging bins over an hour, a day and many detectors brings it down. Yarrow's server uses exactly this fact, and it is the reason the project only accepts raw pulse counts: real counts must be a little random.

🧰 The hardware: the FNIRSI GC-03 and the Yarrow firmware

The first supported detector is the FNIRSI GC-03, a pocket Geiger counter with a J321 tube, a colour display, a rotary knob, a battery and a USB port. It is cheap and easy to get. Yarrow comes with its own open firmware for it (GPL-3.0, written from scratch), because the original firmware gives nothing a computer could use. With Yarrow on the detector you get:


  • a dose rate with an adaptive averaging window: quiet at background level, quick to react when the level rises;
  • counts per minute, a cumulative dose and a daily dose budget, alarms with levels and hysteresis, and a log that survives a power cut;
  • dead-time correction, so a strong source is not read too low;
  • a calibration wizard: put it next to any source whose dose rate you know, and it finds the right sensitivity for your tube (the default is 180 counts per minute for 1 µSv/h, the value of the original firmware);
  • electric and magnetic field readings, a clock, a battery gauge and 16 languages (Türkçe, English, Deutsch, Français, Español, Italiano, Português, Nederlands, Polski, Русский, Українська, Bahasa Indonesia, Tiếng Việt, 日本語, 한국어, 中文);
  • a simple USB serial protocol (raw pulse counter, identity, calibration, clock), which is what the BOINC monitor talks to.


Installing is the same as updating the original: hold OK, press power, plug in the USB cable and a drive appears; copy the .bin of your language onto it and the device restarts. The original bootloader and the factory tube calibration are never touched, so the original firmware can be copied back the same way. Files and step by step instructions: github.com/alplix/yarrow.

🖥️ The monitor: what BOINC runs

The BOINC part of Yarrow is a small program, the Yarrow monitor. It uses almost no CPU and one task per computer stays running as long as BOINC does.


  • It looks for a detector every 15 seconds, reads it while it is plugged in, and cuts the pulse counter into exact bins of 10, 30 or 60 minutes (you choose in your preferences). Bins always start on the clock: 12:00, 12:10, 12:20 and so on.
  • A bin is only kept when it is exact: both edges were read in the right second, the tube voltage was on the whole time, the counter did not restart and the clock did not jump. If anything is off, that bin is dropped. Nothing is ever estimated or padded.
  • About every ten minutes the finished bins are sent to the server as a BOINC trickle, together with the detector's identity, calibration and the place you chose. Bins that could not be sent yet are saved and sent later, even after BOINC restarts.
  • If the detector's clock is wrong, the monitor sets it to your computer's time (UTC) before counting.
  • No detector for five minutes? The task puts itself aside for six hours so it never wastes your CPU. Missing settings? It waits an hour and asks you to fill them in.
  • Builds for Windows (x64, x86, ARM64) and Linux (x86-64, ARM64, ARMv7).


Support for other USB Geiger counters is being added to the same monitor, starting with the GQ GMC family and the USB counter of the old Radioactive@Home project. The rule stays the same: only raw pulse counts from the tube are accepted, never a rate the counter has smoothed itself.

🛡️ Keeping the data honest

Any network of home instruments needs to trust the numbers, and Yarrow cannot lock a volunteer's computer. So it makes the checks that fake or broken data cannot pass:


  1. Physics at the door. A bin that is impossible is refused at once: a dose rate over 1,000 µSv/h, a bin from the future or from the distant past (a computer clock that runs a few minutes ahead is fine), an impossible dead time or sensitivity.
  2. The statistics test. A day later, the whole stream of a detector is tested. Real counts are random: their spread must match Poisson statistics. A stream that is too smooth (a made-up constant), or wildly too spread, does not pass. This needs enough data to mean something: 72 bins, or 24 bins with at least 2,000 pulses.
  3. Identity. Every detector has its own identity, the hash of its chip number. A volunteer can have up to 5 detectors, and a detector that is switched off by the project earns nothing.
  4. The crowd. The map shows the median of a city, so one wrong detector cannot move it.


Credit comes only after a bin has passed the checks.

🏅 Credit

Every valid bin is worth 10 credits, up to 150 bins (a bit over a day of 10-minute bins) per detector per day. Credit is granted about a day after the data arrived, when the bins have been checked. A detector running day and night earns about 1,440 credits a day. The pending bins are visible on the map at once; they are only credited after the check.

🗺️ The radiation map

Open the Yarrow radiation map. It is a pixel-art world with country borders, and each point is a city with its median dose rate:


  • Choose the window: the last 1 hour, 3 hours, 24 hours or 7 days.
  • The colour tells the level (in nSv/h): blue under 80, green 80 to 150, violet 150 to 300, pink 300 to 600, red 600 to 1,500, white above. Red or white on a quiet day deserves a second look, most often a detector that needs calibration or sits next to something radioactive, like old luminous dials or a granite counter top.
  • Click a point for the city's history graph over the last 24 hours. Search a city, zoom, and hover for the country.
  • A "higher than usual" notice appears when the last 3 hours of one detector are clearly above the 24 hours before (at least 1.5 times and more than 5 counting-noise sigmas). It is information from hobby instruments, not an official warning system.
  • A table shows who sends how much, with BOINC user names only.


The page refreshes itself every 30 seconds.

🔒 Privacy

A point on the map is a city, the name you type in your preferences. Nothing finer than a city is ever sent or shown: no address, no GPS, no street. The volunteers table shows BOINC user names and amounts, never places. City positions come from GeoNames (CC BY 4.0), the world map from Natural Earth (public domain).

🚀 How to join


  1. Get a FNIRSI GC-03 and put the Yarrow firmware on it (see above).
  2. Join bitboinc and attach your computer in the BOINC Manager.
  3. Open your project preferences: tick Run test applications (Yarrow is still in test), select the Yarrow monitor, and enter your country (two letters, e.g. TR), your city and the bin length. One setting for all your computers.
  4. Plug the detector into the computer, switch it on and leave it on the main screen. The monitor finds it by itself; the first bin arrives at the next 10-minute mark.
  5. Leave it running. The longer, the better: the checks like 72 bins (12 hours) or more, and the map likes long histories.



❓ Questions you may have


  • Does it use my CPU? Practically not. The monitor only reads a counter once a second and sends a small message every ten minutes.
  • Where should I put the detector? Anywhere that stays the same: a shelf or a desk is fine. Move it rarely, because the building and the surroundings are part of the reading.
  • My value is different from my neighbour's. That is normal. Ground, building and tube differ, and a single detector has a counting noise of several percent. Compare the 24-hour or 7-day value, and read the pattern, not the exact number.
  • The map shows my detector as red. Check the calibration with the wizard (and look for luminous clocks or minerals next to it).
  • I do not have a GC-03. Other counters are on the way (see above). Tell us which one you own.



📊 Progress and links


Questions, ideas and your first measurements are welcome below. ☢️


Note for Linux USERS :

Linux and Raspberry Pi
On Linux the BOINC client runs as its own user and may not open USB devices, so the task says No detector found although a test program finds the counter. One command lets BOINC open the counters Yarrow reads (it only touches those):

curl -fsSL https://raw.githubusercontent.com/alplix/yarrow/main/linux/yarrow-linux-setup.sh | sudo sh
sudo systemctl restart boinc-client

Then unplug the counter and plug it in again. Works the same on Raspberry Pi OS.
ID: 861 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilerilian
Volunteer moderator
Code reviewer
Bug hunter
Early supporter
Avatar

Send message
Joined: 28 Sep 26
Posts: 261
Credit: 43,482,494
RAC: 474,864
Message 865 - Posted: 19 Oct 2026, 18:34:01 UTC
Last modified: 19 Oct 2026, 18:34:14 UTC

> Tell us which one you own.

i dont own any but would like some recommendations on the cheapest one on Amazon that will work ?
I crunch for Ukraine

ID: 865 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Alperen Yavuz
Project developer
25 years old

Send message
Joined: 27 Sep 26
Posts: 349
Credit: 5,210,013
RAC: 137,017
Message 868 - Posted: 19 Oct 2026, 19:16:11 UTC - in response to Message 865.  

In reply to rilian's message of 19 Oct 2026:
> Tell us which one you own.

i dont own any but would like some recommendations on the cheapest one on Amazon that will work ?


Yes, I will need to research all devices of this type and compile a list. Essentially, we need devices that provide raw data via a USB serial connection.
ID: 868 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mmonnin

Send message
Joined: 28 Sep 26
Posts: 14
Credit: 20,207,102
RAC: 1,209,823
Message 871 - Posted: 19 Oct 2026, 21:00:52 UTC
Last modified: 19 Oct 2026, 21:04:25 UTC

Multiple users have a GQ GMC-300S (or other variants) from the time when Radioactive@Home was active.

The custom app_info made to work that Geiger counter manufacture needed to have a port ID (which USB port) pass along to it once plugged in. Will that be needed with Yarrow?

A FNIRSI GC-01 is a cheaper version. Would that work?
ID: 871 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Alperen Yavuz
Project developer
25 years old

Send message
Joined: 27 Sep 26
Posts: 349
Credit: 5,210,013
RAC: 137,017
Message 872 - Posted: 19 Oct 2026, 21:12:55 UTC - in response to Message 871.  
Last modified: 19 Oct 2026, 21:24:48 UTC

All models supported by Rad Pro should work in theory; they are exactly the same.
I am currently integrating the components used for radioactive@home into the code. I am trying to implement a fully plug-and-play, auto-detection method.

I could also create a custom operating system for the FNIRSI GC-01; however, since there are two different models, I would need to develop two separate versions depending on which one comes in the box. The same model uses two different chips.

GC-01, GC-02, GC-03 using same tube. GC-02 is good model but not supporting usb serial and firmware update etc.
ID: 872 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mmonnin

Send message
Joined: 28 Sep 26
Posts: 14
Credit: 20,207,102
RAC: 1,209,823
Message 873 - Posted: 19 Oct 2026, 22:24:37 UTC
Last modified: 19 Oct 2026, 22:29:24 UTC

Some tasks downloaded.
Is there a benefit to running more than 1 task at once?
It's set to have BOINC allocate 1 whole thread to it. Will this be updated to an NCI app so it can run alongside a full allotment of CPU threads?
Example: a CPU has 8 threads so it could run 8 CPU tasks plus an NCI Yarrow task. Like Radioactive/WUProp and the couple of other monitoring/non computation projects in the past.

Edit: Well the progress bar is only progressing on one task and the 1st one is very slowly progressing past 10 min run time.
ID: 873 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilerilian
Volunteer moderator
Code reviewer
Bug hunter
Early supporter
Avatar

Send message
Joined: 28 Sep 26
Posts: 261
Credit: 43,482,494
RAC: 474,864
Message 874 - Posted: 19 Oct 2026, 22:35:25 UTC - in response to Message 873.  

@mmonin welcome back to forums :)

i also have a question and i hope tasks only will give credits if they returned actual device readings, empty tasks should not be credited, to avoid cheating (someone can spawn many small virtual machines with this app)
I crunch for Ukraine

ID: 874 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Alperen Yavuz
Project developer
25 years old

Send message
Joined: 27 Sep 26
Posts: 349
Credit: 5,210,013
RAC: 137,017
Message 875 - Posted: 19 Oct 2026, 22:38:44 UTC - in response to Message 874.  

This trick will not work; the application reads raw data directly from the device—such as tube voltage, battery information, battery voltage, etc.—and all tasks are initially marked as invalid. Once a day has passed, the algorithm compares the data across two days; only then are they validated and points awarded.
ID: 875 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mmonnin

Send message
Joined: 28 Sep 26
Posts: 14
Credit: 20,207,102
RAC: 1,209,823
Message 876 - Posted: 19 Oct 2026, 22:41:07 UTC
Last modified: 19 Oct 2026, 22:50:17 UTC

I noticed a bit later that the two running tasks are v 0.05 and 0.06. Maybe thats why they are running concurrently.

Sadly, that was kind of allowed at radioactive and people ran dozens of clients w/o even having a device to gain WUProp hours and having zero credit at radioactive.

Status 99.96 % after 23min

Edit: It reached 100% around 30 min and is still running at 33 min.
ID: 876 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileNichan
Avatar

Send message
Joined: 6 Oct 26
Posts: 16
Credit: 2,177,741
RAC: 56,607
Message 894 - Posted: 20 Oct 2026, 14:32:31 UTC

Just wanted to let you know that a task was triggered for me, even though I don't have a detector.
ID: 894 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileOffDutyTaoist
46 years old

Send message
Joined: 19 Oct 26
Posts: 2
Credit: 94,725
RAC: 1,462
Message 897 - Posted: 20 Oct 2026, 17:27:16 UTC - in response to Message 894.  

I keep getting assigned tasks for this as well. I have it disabled through the settings, but the scheduler keeps grabbing tasks for it.
ID: 897 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
signore2

Send message
Joined: 7 Oct 26
Posts: 11
Credit: 6,000
RAC: 17
Message 904 - Posted: 20 Oct 2026, 20:19:13 UTC

Do you have this option as yes?
If no work for selected applications is available, accept work from other applications?
ID: 904 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileOffDutyTaoist
46 years old

Send message
Joined: 19 Oct 26
Posts: 2
Credit: 94,725
RAC: 1,462
Message 915 - Posted: 21 Oct 2026, 1:48:46 UTC - in response to Message 904.  

For some reason my account_bitboinc.xml was not updating with the check boxes in the project settings. I made sure YARROW was disabled project side in settings, went into the account_bitboinc.xml and deleted <app_id>16</app_id> from the apps_selected section of the project_specific section. After that, I haven't seen another YARROW task. It might be voodoo, it might Maybelline, but it seems to be working as of now.
ID: 915 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[DPC] hansR
Early supporter

Send message
Joined: 29 Sep 26
Posts: 4
Credit: 32,109,742
RAC: 166,702
Message 919 - Posted: 21 Oct 2026, 12:26:05 UTC

My GMC-300E+V4Re is not detected (yet?)
ID: 919 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilerilian
Volunteer moderator
Code reviewer
Bug hunter
Early supporter
Avatar

Send message
Joined: 28 Sep 26
Posts: 261
Credit: 43,482,494
RAC: 474,864
Message 925 - Posted: 22 Oct 2026, 2:46:25 UTC
Last modified: 22 Oct 2026, 2:47:21 UTC

i think there is an issue with yarrow tasks

i've got few on different computers yesterday even on those that has no radioactivity monitoring devices

name: yarrow_1792502405_0_0
WU name: yarrow_1792502405_0
project URL: https://bitboinc.athena.org.tr/
received: Tue Oct 20 13:23:02 2026
report deadline: Wed Oct 20 13:23:01 2027
ready to report: no
state: downloaded
scheduler state: preempted
active_task_state: UNINITIALIZED
app version num: 7
resources: 1 CPU
estimated CPU time remaining: 0.000000
elapsed task time: 0.000000
slot: 32
PID: 5348
CPU time at last checkpoint: 0.000000
current CPU time: 0.000000
fraction done: 0.000000
swap size: 0 MB
working set size: 0 MB


My computer has 32 cores, and preferences had max tasks = 32 (by the way this is the max value, maybe it will be helpful to allow more) because i did not want to have any buffer for testing purposes

because yarrow took 1 "max cpus" slot, my computer had 1 core idle

this is an interesting situation that usually never happen as there are no other projects with cpu and non-cpu apps at one time

i think this project's tasks must be clearly marked non-CPU and should not count towards the max cpus limit setting.

Also deadline of 1 year seems too big for me
I crunch for Ukraine

ID: 925 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profiledskagcommunity
Bug hunter
Early supporter
46 years old
Avatar

Send message
Joined: 29 Sep 26
Posts: 64
Credit: 21,977,371
RAC: 136,334
Message 926 - Posted: 22 Oct 2026, 7:25:18 UTC
Last modified: 22 Oct 2026, 7:34:29 UTC

Before i test it with my multiple radioactive@home sensors, its absolutly nessecary they dont peep. Does anybody using them right now and can say something about that?

Plus..the project needs multiple sensor per User support in the future, i have two of them, on different locations ^^ (one close (500-1000m?) to a 250kW research Reactor) and the second 10km away :3 So i can measure the inner and the outer exposionring of that thing (in case it would be faulty we never hope it happends) ^^
---------- 24/7 Crunching since 2011 -----------
DSKAG Austria: http://www.dskag.at

ID: 926 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
signore2

Send message
Joined: 7 Oct 26
Posts: 11
Credit: 6,000
RAC: 17
Message 929 - Posted: 22 Oct 2026, 9:28:07 UTC

Yarrow is not setup as an NCI app. It could be basically forced with an app_config and ncpus.
ID: 929 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Aurimas

Send message
Joined: 1 Oct 26
Posts: 8
Credit: 203,623
RAC: 786
Message 930 - Posted: 22 Oct 2026, 9:48:30 UTC - in response to Message 926.  

In reply to dskagcommunity's message of 22 Oct 2026:
Before i test it with my multiple radioactive@home sensors, its absolutly nessecary they dont peep. Does anybody using them right now and can say something about that?


Yeah, I tried to connect my old Radioactive@Home sensor for several hours, but turned it off.

First, because of constant beeping and second, because it not appears on map and not reporting hours, so looks like it not working with Yarrow correctly.

One more suggestion from me - tasks need to be NCI, now I have 1 core idle, when Yarrow is working.
ID: 930 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Alperen Yavuz
Project developer
25 years old

Send message
Joined: 27 Sep 26
Posts: 349
Credit: 5,210,013
RAC: 137,017
Message 932 - Posted: 22 Oct 2026, 11:02:04 UTC

I do not have the source code for these devices. The source code is required to resolve this beeping issue and others.
ID: 932 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JagDoc
Source finder
Platform tester
Bug hunter
Early supporter
72 years old

Send message
Joined: 30 Sep 26
Posts: 73
Credit: 214,207,614
RAC: 5,206,790
Message 938 - Posted: 22 Oct 2026, 12:07:38 UTC - in response to Message 932.  

In reply to Alperen Yavuz's message of 22 Oct 2026:
I do not have the source code for these devices. The source code is required to resolve this beeping issue and others.

Try this:
https://web.archive.org/web/20161030162704/http://radioactiveathome.org/files/application-sources.zip
ID: 938 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · 3 · 4 · Next

Message boards : Puzzles and science : YARROW