New Images for all recent Cosmic Pi units + Data so far.

Following some fine tuning on the latest V1.8 design, we have now produced updated images for all of our detector versions from V1.5 up. Here are the links to the image files, guidance on which units to install them on and an explainer on how to upgrade.

What’s new?

If you have a V1.8 unit already, then the only significant change is that the MQTT connection now persists. When we shipped the V1.8 the python script was prone to disconnection from our server - meaning that your point on the map at https://data.cosmicpi.org would stop flashing even as you detected new events locally. So far we have collected over 22 million muons, but with persistent connections we can collect a lot more in less time!

For all units, we’ve modified the boot configuration for the Raspberry Pi Zero to turn of the power/activity LED on the PCB. This means one less light pollution source inside the enclosure. It will help to ensure that the signals we detect are muons and not spurious escaped photons.

If you have a version 1.5, 1.6 or 1.7 unit, then you will get our latest user interface which is based on Grafana. It provides a responsive and customisable view into the data you collect and stores it persistently in an InfluxDB instance on the device. You can create new visualisations of your data via the web interface, as well as saving and sharing them. It’s also very easy to extract the raw data from behind the visualisations by simply clicking “export”.

Note: For version 1.6 and 1.5 there are some hardware differences, so the binary firmware images for the Cosmic Pi main board cannot be uploaded. Due to some changes in the Arduino Core, used to compile firmware for 1.5 and 1.6 hardware, we currently don’t have a working binary that compiles from source for either of these designs. If you try to run the flash routines on a 1.5 or 1.6 unit, they will fail - this is normal, and your board firmware will not be changed.

Images!

Here are the latest images:

For Hardware Version 1.5

For Hardware Version 1.6

For Hardware Version 1.7 & 1.8

When you have downloaded the images you will need to unzip them and flash them to the SD card in your unit. You will need to open the unit to remove the SD card. You can find flashing instructions here, remember to use the expanded image file as the source.

Cosmic Ray Data - Download the story so far

We’ve now collected over 22 million cosmic rays with our growing network. If you want to explore the dataset you can download a portable version of the master InfluxDB files here. If you want to explore the dataset directly, the easiest way is via a local InfluxDB installation. You can find instructions for how to restore the dataset here.

An update from the world of Cosmic Pi!

What have we been up to for the last little while? Apart from avoiding Covid-19 as best as we can, it’s been time to finally get our database and dashboard online!

The new dashboard, shipping on all 2021 units. If you already have a cosmic pi it can be upgraded!

We’ve revised the software that runs on the Cosmic Pi unit’s Raspberry Pi Zero. Using the latest version of Raspberry Pi OS (formerly Raspbian), we benefit from a number of security and performance updates due to improvements in the Linux kernel. From a user perspective, this translates to having less to worry about and getting more performance from the same hardware. The biggest change is that there is now a compelling reason to leave your Cosmic Pi detector turned on and plugged in!

Big improvements:

• A map showing all online cosmic pi detectors
• Global count of all muon detections
• Easy to export and analyse data from your unit and the whole network

We will soon be organising a webinar explaining how the dashboard is put together and how you can build your own visualisations. You can follow us on Facebook or subscribe to our YouTube Channel to stay informed. The best part is that it is now really easy to share your cosmic rays with the world and to analyse your data via the web! Check out our new data portal on http://cosmicpidata.mooo.com where, at the time of writing, we’ve logged more than 3 million muons already. The architecture used on each cosmic pi unit (Python scripting + InfluxDB + Grafana) is the same as that used on the server, with an MQTT protocol link. This means that you can easily develop and share dashboards (visual data analytics, based on InfluxDB queries that come out as numbers of plots), between Cosmic Pi units, and have them shared on our data portal. If you can’t wait to look under the hood, check out our code repositories on GitHub now.

The data portal. Connect your unit to the internet to see it appear on our map!

It also makes all the data accessible to everyone via the internet. From our data portal, you can click on “Inspect -> Data” for any of the panels to export the raw data underneath into a CSV or Excel file for local analysis. There are some limits on the query size, so you can’t all download all 3 million cosmic rays at once (as this would stall the server!), but this shouldn’t be an issue for most analysis.

The main innovations for the onboard software are:

• Simple wi-fi hotspot or network configuration (no more dual hotspots, it was too unstable)
• A new on-device database, using InfluxDB
• A new on-device dashboard, using Grafana
• Modified MQTT parameters, with password protection for server access

What can I do with this system?

We have massively simplified the architecture that makes a Cosmic Pi run. We now use a single python script to read data in from the detector circuit board, store it in the local database and share it online via the MQTT protocol. This means increased reliability for the single core CPU in the Raspberry Pi Zero, and the system overall.

InfluxDB is a dedicated time series database, designed to be efficient in storing large quantities of data. It runs well on the Raspberry Pi Zero and integrates with Grafana, our chosen visualisation platform. With the admin credentials for Grafana, you can build your own visualisations, to look at the relationships between different parameters collected by your detector. Here are some examples from the global data portal.

Altitude vs Frequency for a static detector - as you would expect there is no relationship!

Cosmic ray frequency vs Air pressure over a 7 day period.

What about confidential data and my identity?

Only the unique identifier of your unit, GPS location and data you collect is shared via our platform, there is no personal identifying information stored or transmitted. You can even observe the flow of data directly on your smartphone with an MQTT client.

What does this mean for you if you have already have a Cosmic Pi?

If you turn it on occasionally, then not much will change. You can update your SD card following the instructions at the bottom of this article. This update is backwards compatible with all our old units from V1.5.

You can find instructions on how to re-image your SD card here.

If you have just received a new Cosmic Pi unit here are some instructions to get you up and running. If you prefer to watch rather than read, you can check out our video here:

Video!

Step 1:                            

Remove from packaging

Step 2:

Place on a flat, non-metallic, surface

Step 3:

Connect the GPS antenna – screw the connector in carefully. The antenna should be able to see as much sky as possible, so consider putting it out of a window, on a windowsill, or securing it to a window. The antenna module is magnetic, so can stick itself to a metal surface.

Step 4:

Connect the power. Plug in the power adapter to the wall and put the jack carefully into the front of the comic pi unit.

Step 5:

The power light should come on (sorry, it’s rather hard to see!) and after a few minutes the event light will start to flash too. The power light will start to blink once a GPS fix has been acquired, depending upon the weather and how well your antenna can see the sky this can take a while – typically it’s 5 minutes, but in poor conditions it can take much longer.

Step 6:

Connect to the Cosmic Pi wifi network, there is no password. You may need to persuade your device to remain connected even when there is no internet (since our cosmic ray detector doesn’t provide its own internet connection) – typically there is a box to tick saying ‘remain connected even if this connection does not provide access to the internet’.

Step 7:

Open a web-browser and navigate to http://192.168.0.1 and it should present you with your cosmic ray data! Sometimes the graph isn’t up to date as the GPS lock is required to put the correct timestamp data into the database – so make sure the GPS antenna can see the sky and try again in 5 minutes. If your computer supports the bonjour protocol, you can navigate to http://cosmicpi.local as an easier to remember alternative.

Please note that the times on the graph are UTC, therefore may not be synchronous with the time zone in your location.

Step 8:

You can read all of the sensor data off the web-page and download the cosmic ray event graphs directly. A full extraction of the database (i.e. all cosmic ray events from the device) can be extracted using the advanced, password protected, feature set accessible by the menu on the left hand side of the web page. If you want to access this please get in touch with us.

Addendum:

We noticed late in production that there is a small software bug which prevents some units from showing the altitude (pressure readings are unaffected). If you see that the altitude is showing 0, this is likely due to the bug. You can always manually calculate your altitude either via the GPS position or the pressure reading (https://keisan.casio.com/exec/system/1224585971)

Two summer students joined us over the last two months. Darko from Serbia and Tong from China. Both were part of the Geneva Tsinghua Initiative Sustainable Development Goals Summer School. And since August another student, Hendrik, has joined us and will be with us until mid of October.

Summer student group with version 2 mockup at Ideasquare. From left to right: Shrey, Hendrik, Darko, Faouzi, Oday (missing: Tong).

So what have we worked on over the last few months?

First of all, we moved our website from an old WordPress instance to GitHub. Next a lot of time has been spent on designing the next version of CosmicPi, as well as an intermediate step to allow the electronics and performance to be tested. For the near future (i.e. the end of this year) an intermediate is planned, which we're calling V1.5. This version will improve on the existing design using the Arduino DUE. Mainly by replacing the analogue amplifier board and improving the power supply for the silicon photo-multipliers (SiPM's). In order to improve the power supply we decided to use two low noise boost converters, providing an independent supply for each sensor on our version 1.5 release. Our new team member Louis has designed a completely new power supply circuit featuring an increased voltage range and higher resolution, which we'll be sharing the schematic once we've integrated it into a prototype PCB. During the same timeframe, another new team member Nikolai has re-designed the analogue amplifier board from the V1 release to employ transimpedance amplification principles. The new design can be found here.

A first batch of the new analogue board has already been assembled and is currently being tested by the team. So far the results are promising.

And you can read more about it in our last post!

Recorded spike of a triggered SiPM with the new analogue board. The spike is nicely amplified into an area of a few 100mV and minimal additional noise is introduced. Note that the timebase is wrong, it should be 200ns.

At the same time we are working on a Version 2 of CosmicPi. The Version 2 will build on many of the design elements of the Version 1.5, such as the improved analogue board whilst introducing our new power supply design. The Version 2 will move away from the Arduino DUE and the current digital hardware model for data acquisition. Instead a FPGA will be used to read out two high speed analog to digital converters (ADC), which run at approximately 100 MHz. With these ADCs the amplified signal from the SiPM's can be read directly, without the need for a pulse shaper as in our Version 1 design. This will improve the precision of the received data as well as providing more options to analyze the data we get from the SiPM's. The FPGA is then directly connected to a single board computer (SBC) such as the Raspberry Pi (or the lower cost Orange Pi) via GPIO-Pins. All other sensors, such as GPS, accelerometer and the temperature sensor will be read out by the single board computer, since this is not time critical like the SiPM readout.

Darko has been working a lot on the software for the FPGA. He has taken the first steps in validating a design for the software and implementing this into VHDL. The high level design can be seen below and the corresponding code can be found here.

It is fair to say that this version will still take some time until it is up and running. At the moment we are working from two Lattice FPGA iCE40HX8K development boards. However over the last few weeks we already designed a first version of the hardware main board for the Version 2 CosmicPi. Six prototypes of this board are now being produced by Seeed Studio in Shenzhen and will be arriving any day at our lab. Having a Raspberry Pi (or equivalent SBC hardware) as an integral part of the detector gives us more options in developing the user experience and interaction possibilities. Hendrik has started designing the software that is going to run on the Raspberry Pi.

In order to test and develop this software without having the FPGA ready yet we decided to build a mockup of the Version 2 CosmicPi. The mockup uses a Raspberry Pi 3 to run the software and has a GPS, temperature sensor, pressure sensor and a 9 DOF integrated measurement unit connected to it - the only thing missing is the custom cosmic ray detection hardware. In this way, most of the software can already be developed for the Version 2 design, with only the connection to the FPGA to add when the hardware is available.

Version 2 mockup with Surface showing the user interface

You can take a look at the software here.

A part of the GTI SDG-Summer school was as a two weeks trip to China for Darko and Tong, with Hendrik staying for a total of four weeks. During this time a lot of engagement work with the Chinese general public was done. This is why the Version 2 mockup is completely bilingual, supporting both English and Chinese! In Beijing and Shenzhen we gave multiple presentations on CosmicPi and in Shenzhen had a poster session with students and teachers from the area.

Poster session in Shenzhen

Shenzhen is a great city for hardware development and we wanted to see if an extended stay there could be useful for the project. In general hardware and hardware production means are readily available there. While James, Darko and Tong stayed in Shenzhen for one week, Hendrik was able to stay two weeks longer.

Rather disappointingly "special" parts, such as scintillators and SiPM's (which are only really used for High Energy Physics applications) are very hard to find there. So we decided to explore different ways of detecting cosmic rays without scintillators, using silicon - of which there is an abundance in Shenzhen. One approach we evaluated was to use CMOS or CCD sensors for directly detecting ionizing particles. These sensors can be extremely cheap, with CMOS image sensors as used in smartphones available from as little as 50 cents each. However the cheaper sensors require advanced manufacturing methods for the PCB (4 layers with blind vias), making small volume prototyping for this approach very expensive.

Another detection technology that has been popular among enthusiasts for some time now is the use of pin-diodes. As an example, the BPW34-diode have been used for particle detection in many projects and the needed parts are readily available in Shenzhen and through the international electronics supply chain. Over the last few weeks Hendrik and James have thus been discussing and designing such a detector. The resulting board design can be seen in the picture below.

Back of the pin-diode detector

The detector has 16 diodes on the top of the PCB and 16 on the back to allow coincidence triggering. After the signal of the diodes is amplified the readout is done by an Atmel 328 microprocessor, exactly the same chip that is the core of every Arduino UNO. The final data acquisition and user interface is then again provided via a single board computer. The circuit design can be found here.

Some prototypes of this detector are also currently in production at Seeed Studio. We hope that we can start testing it soon, and of course we will share the results with you as soon as we can!

Three of our prototype new analogue amplifier boards

One SiPM pulse measured on the oscilloscope. Note that the timebase is wrong, it should be 200ns

Two peaks from two SiPMs inserted into a single scintillator slab. It might even be a muon!

Stacking the slabs vertically to produce simultaneous coincidences in both sets of slabs.

Two peaks from two SiPMs connected to the Cosmic Pi Analogue amplifier board, occuring at the same time as a coincidence on the reference scintillators. This is very likely to be a muon!

A simple counter collects events on the reference detector, the Cosmic Pi analogue circuit with a discriminator and coincidence unit, and finally the coincidence of the Cosmic Pi with the reference detector.

Here we plot the timing distribution of signals captured by the Cosmic Pi and the reference detector. Both are the type of distribution we would expect, however we still have more work to do.

Here we plot the timing distribution of signals captured by the Cosmic Pi and the reference detector. Both are the type of distribution we would expect, however we still have more work to do.