Space Surveillance Sensors: Some Additional Background on GEODSS’ CCD Detector (August 30, 2012)

Modern systems for visible detection and tracking of space objects, such as the Ground-based Electro-Optical Deep Space Surveillance (GEODSS) system are typically built around a telescope equipped with a charge-coupled device (CCD) detector.  This post follows up on the August 20 post on GEODSS by providing some additional information of how the GEODSS CCD operates. In the next GEODSS post, I will make a rough estimate of GEODSS detection capability (in terms of the faintest object it can detect) based on the available information about the telescope and its CCD detector, and compare this to published values.

 

Figure 1. A GEODSS camera.[1] 

 

A CCD is a semiconductor-based device which is structured into a large number individual pixels arranged in a rectangular array.  A photon striking the CCD’s imaging surface CCD can produce, via the photo-electric effect, an electron which is then trapped and stored in the corresponding pixel.  By manipulating the voltages applied to each pixel, the electrons can either be accumulated and stored in a pixel or shifted to an adjacent pixel.  After an exposure, the CCD can be read out by sequentially shifting each row into an output register and (before reading in the next row) sequentially shifting out each element of the row into the detecting electronics.  It can thus produce a two dimensional image, in which the light intensity in each image pixel is proportional to the charge in the corresponding CCD pixel. 

Typical integration times for deep space visible detection and tracking systems can be as short as a few tenths of a second.  Thus, in order to operate efficiently, it is essential that the CCD be read out quickly.  The current GEODSS system uses a 1960 x 2560 pixel CCD, and thus has about 5 million pixels.  At GEODSS’ readout clock rate of 2.0 MHz, about 2.5 seconds would be required to read out the CCD (higher readout CCD rates are possible, but lead to additional heating and noise).  With such a readout rate, for exposure times less than a second, the system would be actually only be able to collect data for a small fraction of the time.  GEODDS solves this problem in two ways.  First, it uses eight different readout ports, each of which reads out a different portion of the CCD.  This allows the readout time to be reduced to about 2.5/8 = 0.31 seconds.  Second, the CCD uses a split-frame architecture.  In such a design, at the end of an exposure, each column in the imaging CCD array is rapidly (in a few milliseconds) read into the corresponding column of a second, light-shielded storage array.  The imaging array can then immediately begin collecting another frame of data, while the storage array is being read out.  The combination of these two techniques allows the CCD to collect data essentially continuously, and allows GEODSS to achieve a frame rate of about 2.7 frames/second, corresponding to a frame time of 0.37 seconds.[2]

In general, the visible brightness of a space object will fluctuate in time, often in a manner that provides a characteristic “signature” for a given object.  Data collected for such identification purposes is known as Space Object Identification (SOI) data, and may require frame rates of up to a kHz.  Such high frame rates can achieved in several ways.  In GEODSS, a separate 32×32 pixel array with a 1.25 MHz readout rate is used both for SOI and for photometric measurements of a target’s brightness.  Since the 1024 pixels can be read out in less than a millisecond, this array can support frame rates as high as 1,000 Hz. 

A key parameter for a CCD is its quantum efficiency, which is the fraction of photons striking the surface of the CCD that produce an electron that is collected.  The quantum efficiency will vary with the photon wavelength, and typically an integrated value across the CCD’s spectral range is given.  For the current GEODDS, the solar-weighted average quantum efficiency of the CCD array is 0.65, a very large improvement from the 0.08 quantum efficiency of its pre-upgrade vidicon detector.[3]

 

Figure 2.  The measured quantum efficiency of a GEODSS CCD.[4] 

There are several important sources of noise in a CCD.[5]  The dark current consists of thermally-generated electrons in the semiconductor material and is generally specified as electron/pixel/second.  It can be reduced to low levels by cooling the detector.  For GEODDS, the dark current noise is about 6 electrons/pixel/second.  The readout noise is the (average) noise created by the output electronics each time the CCD is read out, for GEODSS it is about 12 electrons/pixel.   A third internal noise source results from the quantization performed by the analog-to-digital converter, and increases with the gain of the detector (the gain is the number of electrons required to produce a single count in the A-D converter). Finally, external background noise, such as a diffuse sky background, must be taken into account.

The signal-to-noise ratio for a CCD detector can then be written as:[6]

 

where:

τ = integration time in seconds

NT = number of electrons/second due to target

NB = number of electrons/second due to sky background

ND = number of electrons/second due to dark current

NR = number of readout noise electrons

G = detector gain

σ2 = factor that depends on details of A/D converter, typically about 0.29

 

Typically, a S/N ≈ 4 – 6 is sufficient for automatic detection and tracking. 

 The CCD is placed at or near the focal plane of an astronomical telescope.  Thus the aperture of the telescope combined with the losses within the telescope determine the number of photons reaching the detector from a given target.  The design of the telescope determines the field of view of each pixel and of the array as a whole.

With non-imaging sensors such as those considered here, it is often not desirable to focus the beam to the limit allowed by diffraction or atmospheric turbulence.   Instead, the focus is adjusted so that while most (perhaps 50%) of the target’s photons go into a single pixel, enough go into neighboring pixels to be measured.  This allows the location of the target to be interpolated within the main pixel, typically to about one-third of a pixel.

If a target image is centered in a pixel, then fraction of the photon energy falling into that pixel is known as the ensquared energy.  However, in general the target image will not be centered in a pixel.  Thus the average fraction of the energy in the central pixel must be determined by averaging the fraction of the beam energy in a pixel for all possible beam center positions in the pixel. The result is known as the straddle factor and, as will be discussed in the next GEODSS post, is about 0.4 for GEODSS. 

There are two main operational modes available for detecting and tracking space objects.  In the sidereal mode, the telescope tracks the stars, which thus appear as points.  If the exposure is long enough, a space object will appear as a streak, allowing for its automated detection.  Historically, this was the standard mode of operation for GEODSS, with the target’s position determined at the beginning and end of each streak based on pointing data from the telescope mount.  However, this approach limited the minimum detectable object brightness, since an object would not remain in a single pixel during the integration time.  Thus for dimmer objects, GEODSS could make position measurements by tracking the target itself (rate-tracking).  The upgraded GEODSS apparently operate in a similar manner, primarily tracking in a sidereal mode, but using a target-tracking mode for fainter objects.


[1] Photograph from:  John R. Tower, Pradyumna K. Swain, Fu-Lung Hsueh, Robin M. Dawson, Peter A. Levine, Grayzna M. Meray, James T. Andrews, Verne L. Frantz, Mark S. Grygon, Michael A. Reale, and Thomas M. Sudol, “Large Format Backside Illuminated CCD Imager for Space Surveillance,,” IEEE Transactions on Electron Devices, Vol. 50., No. 1 (January 2003), pp. 218-224

[2] Tower, et. al., “Large Format Backside Illuminated CCD,” p. 220.

[3] Grant H. Stokes, Frank Shelly, Herbert E.M. Viggh, Matthew S. Blythe and Joseph S. Stuart, “The Lincoln Near-Earth Asteroid Research (LINEAR) Program,” Lincoln Laboratory Journal, Vol. 11, No. 1 (1998), pp. 27-40. Available at: www.ll.mit.edu/publications/journal/pdf/vol11_no1/11_1linear.pdf ; Walter J. Faccenda, David Ferris, C. Max Williams, and Dave Brisnehan, “Deep Stare Technical Advancements and Status,” MITRE Technical Paper, October 2003.  Available at: http://www.mitre.org/work/tech_papers/tech_papers_03/faccenda_deepstare/index.html.

[4] Photograph from Tower, et. al., “Large Format Backside Illuminated CCD,” p. 222.

[5] The noise values in this paragraph are Tower, et. al., “Large Format Backside Illuminated CCD Imager.”

[6] Adapted from Steve B. Howell, Handbook of CCD Astronomy (Cambridge, UK: Cambridge University Press, 2000), p. 54.   Additional noise terms can be introduced when making photometric measurements, as these require estimating and then subtracting out the background signal.

About these ads
Leave a comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 55 other followers

%d bloggers like this: