
Telescope and Photodetector Array FieldOfViewsRadobs 20
The following is a discussion on the form of focal plane silicon PIN (positiveintrinsicnegative) photodetector array that will be used in the optical (visible and nearinfrared wavelength) heterodyne SETI receiver. The unobstructed linear fieldofview (FOV) of a typical large telescope is given by: FOV = tan1(1/f#) degrees (1) = 1/f# radians where f# = focal length/diameter (150/10). For the new 10 meter diameter Great Keck telescope on Mauna Kea in Hawaii, with an f/15 (fnumber or focal ratio) secondary mirror effective focal length of 150 meters: FOV = 3.8 X 3.8 degrees * * Cone of light * . * . * * * * . * * . > / * * . / * *  Photodetector > THETAt * THETAa   Array \ * *  N X N Pixels > \ * * . * * . * * . * * * . * . * Focal Plane * THETAa << THETAt Diagram (not to scale) illustrating how photodetector array only occupies a small fraction of the telescope's fieldofview. This is the theoretical FOV if there are no obstructions, limiting apertures in the optical path, or considerations of peripheral field quality. However, because of limits on the Keck's fieldofview, the actual (specified) FOV is 0.33 X 0.33 degrees. The FOV of the array will be very much less, and at any instant of time after being centered on a target star, is likely to contain only a single star that is within 1,000 light years. We would probably want to use the rest of the telescope's fieldofview for other purposes, such as laser guidestar sensors, course spectrographic systems, cameras, and perhaps a twodimensional (incoherent) photoncounting array in addition. The latter could be large 2048 X 2048 pixel devices. This is really no different to what is done in the Hubble Space Telescope (HST), where the fieldofview is divided up into sectors for different instrumentation packages. There are some very good system reasons involving telescope repositioning or slew time overheads, and which will be elaborated upon later, for having a much larger telescope FOV that can be scanned with a mirror across the array. We now arrive at the basic design for the array through the following analysis. It is assume than an adaptive or spacebased telescope will be employed so that diffraction limited performance is achieved. In this situation, the photodetector element size, or pixel size, corresponds to the minimum spot size, and may be set to onehalf the diameter of the Airy disk. This is a factor of 1.2 times the Full Width Half Maximum (FWHM) spot size, and is half the size of the Airy disk which contains 84% of the energy (see RADOBS.14). If we set the pixel size to correspond to the diameter of the Airy disk to include more signal energy, we would likely degrade the sensitivity by having some of the localoscillator power fall outside the strong part of the spot, thus producing extra shotnoise with no corresponding contribution to heterodyne signal gain. This pixel size is given by the Rayleigh criterion: pix = (1.22).Wl.(f#) microns (2) where Wl = wavelength (656 nm). pix = 12 microns Now we need to determine the number of linear photodetector elements that we need in order to efficiently observe the sky. We note that the Rayleigh criterion produces an angular resolution given by: (1.22).Wl THETA =  radians (3) d where Wl = wavelength (656 nm), d = diameter (10 m). THETA = 0.0165 arcseconds Thus, the FOV of each pixel, 12 microns in diameter, is 0.0165", and the corresponding Image Scale is 1.38" per mm. As previously indicated, for the moment we shall assume that for the highest received signaltonoise ratio (SNR), the pixel size should be matched to the Rayleigh criterion size; this being slightly larger than the FWHM (Full Width Half Maximum) diffraction limited beamwidth (0.0138"). Again, see RADOBS.14 for a more detailed discussion on beamwidths. A rigorous analysis may show that for heterodyne detection, a slightly different pixel size will maximize the SNR. The optimum pixel size will also depend on whether we illuminate the entire array with a collimated localoscillator (L.O.) laser, or scan a focused (Gaussian beam waist) L.O., row by row or pixel by pixel. For the nearest stars, say at range R = 10 light years, we would want the entire stellar biosphere to fall within this fieldofview. The beamwidth at that distance based on the Rayleigh criterion, is given by: (1.22).Wl D = R.tan[] (4) d where the angle is in radians. D = 0.0506 A.U. Assuming that the maximum outer diameter of a stellar biosphere is 5 A.U., this means that the linear dimensions of the beam must illuminate at least 5 A.U. The ratio of 5 A.U. to 0.0506 A.U. is 99. Thus, a 128 X 128 pixel array would allow us to simultaneously view the entire stellar biosphere when the image of the star was centered on the array. The angular FOV of this array would be given by 128 X 0.0165", i.e., 2.1" X 2.1", which is about 3.2 X 10^6 of the area corresponding to the telescope's 0.33 X 0.33 degrees actual fieldofview. The array FOV at 10 light years range would have dimensions 6.5 X 6.5 A.U. This gives us some margin for the relatively few stars at ranges 5 to 10 light years. We might possibly want to consider a 256 X 256 pixel array with a 4.2 X 4.2 arcseconds fieldofview, but this would probably be the largest we would want to use. See RADOBS.22 for concerns about array optical and electric power dissipation. Note that to improve the efficiency of a targeted optical search, the computer controlling the receiving system will be programmed to avoid sampling pixels which are very unlikely to contain an ETI signal. When we target the more distant stars, the stars and their biospheres will be contained within the array's central pixels. Hence, there is no point in spectrum analyzing the outputs of the peripheral pixels. At a distance of several hundred light years, Planckian starlight will arrive at the same pixel as the signal from the ETI transmitter, and they will no longer be separately resolved. In this case, if the adaptive optics works very well, only one pixel need be sampled. See RADOBS.21 for a discussion about using the array for spacediversity reception. Let us assume that an imaging lens is used that magnifies (linear dimensions) the focal plane image by a factor of 16.7. If the linear dimensions of the array are increased in the same ratio, the angular FOV remains the same, the corresponding Image Scale drops to 0.083" per mm, and the new spot size is given by: pix = 0.2 mm This pixel diameter ought to provide enough active area to avoid photo detector current saturation effects and help dissipate heat, yet at the same time present a very small capacitive load to minimize the RC timeconstants. Clearly, we can only optimize the physical pixel size for one wavelength. However, we could change the imaging lens and hence the magnification (and array FOV) for different parts of the optical spectrum. As far as presentday commercial photodetectors are concerned, Ortel Corporation manufactures a stateoftheart single photodetector device which has an active region 25 microns in diameter, and a 3 dB bandwidth of 12 GHz with a 50ohm load. Optoelectronics Inc. also manufactures single element 50ohm Si photodetectors with active areas equivalent to about 100 microns in diameter, and bandwidths of the order of 10 GHz. Antel Optoelectronics Inc. produces devices with bandwidths up to 20 GHz. There are no 10 GHz photodetector arrays commercially available today, though we may expect that they will be available in a few years for fiberoptic communication applications, and for internal highspeed communications within large computers. Since we propose to use a 128 X 128 element array, with a total of 16,384 pixels, the minimum size of the array is: 25.6 mm X 25.6 mm 1"  o o o o  o o 0.2 mm dia. pixels  o  o       Alien Transmitter (100 L.Y.)   .  1"  *   Star           Alien Transmitter (10 L.Y.)   .     128 X 128 Array FieldOfView = 2 X 2 arcseconds < 6.5 A.U. at 10 Light Years > 65 A.U. at 100 Light Years Diagram showing the image of a nearby star system superimposed on the outline of the heterodyning photodetector array. This 1" square array is quite a reasonable size, though it has linear dimensions an order of magnitude greater than typical CCDs with the same number of pixels. In this custom array, the pixels could be square or round, with a small inactive barrier (mask) between adjacent pixels. The array size could be increased further if it is desired to increase the masked area, provide more room for more complex interface circuitry, or just to improve heat dissipation. January 24, 1991 RADOBS.20 BBOARD No. 333 * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Dr. Stuart A. Kingsley Copyright (c), 1991 * * AMIEE, SMIEEE * * Consultant "Where No Photon Has Gone Before" * * __________ * * FIBERDYNE OPTOELECTRONICS / \ * * 545 Northview Drive  hf >> kT  * * Columbus, Ohio 43209 \__________/ * * United States .. .. .. .. .. * * Tel. (614) 2587402 . . . . . . . . . . . * * skingsle@magnus.ircc.ohiostate.edu .. .. .. .. .. * * CompuServe: 72376,3545 * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
