The CUBIC Data Problem

This photograph shows a portion of a raw image from an EEV CCD similar to those used on CUBIC. This image was collected in May 1995 on NASA sounding rocket flight 36.092UH and is indicative of the type of images we can expect to obtain during flight with CUBIC. In this image, the size of the signal in each CCD pixel is color-coded, with purple corresponding to the smallest signals and white corresponding to the largest. The small, insignificant, single pixel dots in this image are the valid X-ray events from the sky, which are obtained on a background of amplifier noise (blue speckles) and "particle background". Particle background events are the prominent blotchy and linear events that dominate the image, and which are due to cosmic ray and gamma ray interactions in the CCD. For example, the tiny white speck in the bottom right corner of the image is a valid X-ray event, while the white blotch halfway between the bottom right corner and the vertical streak (a "hot", or bright, column due to a small defect in the CCD) is a cosmic ray event. The large blotchy event to the right of the picture's center (with the green streak extending to its left) is a large cosmic ray hit surrounded by a shower of secondary particles produced as the primary cosmic ray passed through the payload.

CUBIC collects an image containing 393,216 pixels digitized with 12 bits every 30 seconds, for a raw data rate of over 150 Kbits per second. Since this vastly exceeds the SAC-B telemetry capabilities, we cannot telemeter every image to the ground and must distinguish between valid and invalid events using an on-board event recognition algorithm. CUBIC uses a very simple, but effective, event recognition algorithm that allows it to process the image data in real time using a simple NEC V30 microprocessor running at 10 MHz. In simplest terms, this algorithm selects X-ray events which occupy either a single isolated pixel or two adjacent pixels and stores those data in compressed format for subsequent transmission to the ground. Several candidate event recognition algorithms, including the one we selected for CUBIC, are discussed in a paper by Cawley et al. (1995).

Testing the algorithm

We tested our event recognition algorithm in the laboratory to ensure that it is effective in identifying X-ray events and eliminating cosmic ray events, and that it is fast enough to keep up with the CUBIC data rate. The image above is a portion of a raw frame of our laboratory simulation of flight data, made by illuminating the flight camera with a Bi-207 gamma ray source to simulate the cosmic ray background. This image is color-coded using a color scheme that runs from black (no signal) through red, yellow, and white (brightest pixels). The "stringy" and blotchy events (typically yellow or white) are produced by the gamma rays, and are similar to the cosmic ray events seen during the rocket flight in the top image. The small reddish dots are X-ray events from an on-board Fe-55 X-ray source that produces X-ray lines at 5.9 and 6.4 keV.

The output of our event recognition algorithm is shown in the image below. This image was reconstructed from the normal instrument telemetry data, which contains only the valid (single pixel and two pixel) X-ray events. The X-ray events have been properly selected with excellent rejection of cosmic ray events. We expect this processing to be more than 99% effective for in-flight background rejection. The event recognition algorithm (coded by Doug Kelly) successfully processed these data frames, which contain an order of magnitude more events than anticipated in a normal flight image, in real time during the CCD readout.

CUBIC Home Page Ask Us