[Paper Review] Development Of Raspberry Pi-based Processing Unit for UV Photon-Counting Detectors

Bharat Chandra P, G. Binukumar+7 more|arXiv (Cornell University)|Jan 2, 2024

CCD and CMOS Imaging Sensors11 references4 citations

TL;DR

This paper presents a low-cost, Raspberry Pi-based processing unit for UV photon-counting detectors using a Photek MCP image intensifier and RPi V2 camera, leveraging the Pi's GPU for inbuilt readout and CPU for centroiding. The system achieves 30 fps in both 3×3 and 5×5 window modes, with double-event detection in 5×5 mode, enabling affordable, customizable UV astronomy instrumentation for CubeSats and small satellite missions.

ABSTRACT

In ultraviolet (UV) astronomical observations, photons from the sources are very few compared to the visible or infrared (IR) wavelength ranges. Detectors operating in the UV usually employ a photon-counting mode of operation. These detectors usually have an image intensifier sensitive to UV photons and a readout mechanism that employs photon counting. The development of readouts for these detectors is resource-intensive and expensive. In this paper, we describe the development of a low-cost UV photon-counting detector processing unit that employs a Raspberry Pi with its in built readout to perform the photon-counting operation. Our system can operate in both 3x3 and 5x5 window modes at 30 frames per sec (fps), where 5x5 window mode also enables the provision of detection of double events. The system can be built quickly from readily available custom-off-the-shelf (COTS) components and is thus used in inexpensive CubeSats or small satellite missions. This low-cost solution promises to broaden access to UV observations, advancing research possibilities in space-based astronomy.

Motivation & Objective

• To develop a low-cost, customizable alternative to expensive FPGA- or ASIC-based readout systems for UV photon-counting detectors.
• To reduce development time, complexity, and cost in space instrumentation by replacing complex logic with an embedded Linux platform.
• To enable affordable access to UV astronomical observations for small satellite and CubeSat missions.
• To demonstrate the feasibility of using Raspberry Pi's integrated GPU and CPU for real-time photon counting and centroid calculation.
• To implement system resilience through single-event recovery and automatic reset mechanisms for space mission reliability.

Proposed method

• Utilized a Raspberry Pi Zero 2W as the embedded microprocessor platform with an integrated GPU for camera readout.
• Integrated a Photek NUV image intensifier with a UV-sensitive photocathode and phosphor screen to convert electron events into visible photons.
• Employed a Raspberry Pi V2 camera module to capture the visible light from the phosphor screen for event detection.
• Implemented photon counting and centroid calculation using custom Python code running on the RPi CPU, processing data from the GPU-accelerated camera readout.
• Configured the system to operate in 3×3 and 5×5 pixel window modes, with the latter enabling double-event detection via threshold-based clustering.
• Enabled real-time data transmission of centroid coordinates via UART at 115200 baud to a PC for image reconstruction and analysis.

Experimental results

Research questions

• RQ1 Can a Raspberry Pi-based system with integrated GPU and CPU replace traditional FPGA-based readout electronics for UV photon-counting detectors?
• RQ2 What is the achievable frame rate and spatial resolution of a low-cost photon-counting system using off-the-shelf components?
• RQ3 Can double events be reliably detected and resolved in a 5×5 window mode using a low-cost embedded platform?
• RQ4 How does the performance of the Raspberry Pi-based system compare to conventional high-cost detector electronics in terms of development time, cost, and flexibility?
• RQ5 Can system-level fault tolerance, such as single-event recovery and automatic reset, be effectively implemented on a low-cost embedded platform for space applications?

Key findings

• The system successfully achieved real-time photon counting at 30 frames per second in both 3×3 and 5×5 window modes.
• The 5×5 mode enabled detection of double events, with centroiding accuracy sufficient to reconstruct the Air Force test target image.
• Reconstructed images from centroid data in both 3×3 and 5×5 modes clearly resolved the target features, demonstrating effective spatial resolution.
• A direct long-exposure image of the phosphor screen confirmed UV illumination from a 275-nm UVC LED, validating the system's sensitivity.
• The system demonstrated robustness through implemented single-event recovery and automatic reset functionality in case of malfunction.
• The entire detector system was successfully assembled and tested in a cleanroom environment for the SING UV spectrograph mission, confirming hardware feasibility.